Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effects of cumulative forest disturbances on hydrology in the interior of British Columbia, Canada Zhang, Mingfang 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_fall_zhang_mingfang.pdf [ 8.98MB ]
Metadata
JSON: 24-1.0071984.json
JSON-LD: 24-1.0071984-ld.json
RDF/XML (Pretty): 24-1.0071984-rdf.xml
RDF/JSON: 24-1.0071984-rdf.json
Turtle: 24-1.0071984-turtle.txt
N-Triples: 24-1.0071984-rdf-ntriples.txt
Original Record: 24-1.0071984-source.json
Full Text
24-1.0071984-fulltext.txt
Citation
24-1.0071984.ris

Full Text

THE EFFECTS OF CUMULATIVE FOREST DISTURBANCES ON HYDROLOGY IN THE INTERIOR OF BRITISH COLUMBIA, CANADA  by  Mingfang Zhang A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy in  THE COLLEGE OF GRADUATE STUDIES (Environmental Sciences)  THE UNIVERSITY OF BRITISH COLUMBIA (Okanagan)  May 2013  © Mingfang Zhang, 2013  Abstract Research on forest disturbances and their impacts on hydrology in large watersheds (>500 km2) is limited. Forest disturbances such as harvesting, wildfire and insect infestation are cumulative over space and time in large watersheds. The major objectives of this study were: 1) to assess the impacts of cumulative forest disturbances on annual and seasonal mean flows; 2) to evaluate the impacts of cumulative forest disturbances on flow regimes (magnitude, timing, duration, frequency, and variability); and 3) to conduct an integrated analysis, and discuss their ecological implications.  The six selected large watersheds experiencing various levels of forest disturbances along environmental gradients in the BC interior include the Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River watersheds. Cumulative equivalent clear-cut area (CECA) was used to indicate cumulative forest disturbances. A novel methodology combining advanced statistical methods (e.g., time series analysis) with graphical methods (modified double mass curve and flow duration curve) was employed to conduct statistical and quantitative assessments.  The analysis demonstrates that forest disturbances produced significant impacts on annual mean flows and some of the seasonal mean flows in the intensively disturbed watersheds including Willow River, Baker River, Moffat Creek, and Tulameen River watersheds. Forest disturbances and climate variability have produced counteracting effects or joint positive effects on streamflow in these watersheds. The comparisons suggest that the wetter years or the wetter watersheds were more sensitive to forest disturbances as compared with the drier years or the drier watersheds. Using the long-term data on the hydrological responses in the Willow and Tulameen River watersheds, a pattern of reduced impact strengths over time probably due to watershed resilience was discovered.  Forest disturbances significantly altered the magnitude, timing, variability, duration, and frequency of high flows in the Baker Creek, Moffat Creek, and Willow River watersheds. Less  ii  pronounced impacts on low flow regimes and inconsistent relationships between forest disturbances and low flow regimes were found.  These findings are of great implications for designing forest and watershed management strategies to protect watershed ecological functions and public safety in the context of future forest and climate changes.  Keywords: Cumulative forest disturbances, Large watersheds, Mean flows, Flow regimes, High flows, Low flows  iii  Preface This dissertation is submitted for the degree of Doctor of Philosophy at the University of British Columbia (Okanagan). The research described herein was conducted under the supervision of Dr. Adam Wei in the Department of Earth & Environmental Sciences and Geography, University of British Columbia (Okanagan), from September 2008 to June 2013.  This work is completely original, except where references are made to previous work. Part of this work has been presented in the following publications: •  Wei, X., Zhang, M., 2010a. Research methods for assessing the impacts of forest disturbance on hydrology at large-scale watersheds. In: Chen, J., Chao, L. and Lafortezza, R.(Eds.) , Landscape Ecology and Forest Management: Challenges and Solutions in a Changing Globe. Higher Educ. Press, Beijing & Springer-Verlag, Berlin Heidelberg, pp119–147. (equal contribution, see Chapter 2)  •  Wei, X., Zhang, M., 2010b. Quantifying streamflow change caused by forest disturbance at a large spatial scale: A single watershed study. Water Resour. Res. 46(12), W12525. (equal contribution, see Chapter 1 and 4)  •  Zhang, M., Wei, X., Sun, P., Liu, S., 2012a. The effect of forest harvesting and climatic variability on runoff in a large watershed: the case study in the Upper Minjiang River of Yangtze River basin. J. Hydrol. 464-465,1-11. (see Chapter 1 and 4)  •  Zhang, M., Wei, X., 2012b. The cumulative effects of forest disturbance on streamflow in a large watershed in the central interior of British Columbia, Canada. Hydrol. Earth Syst. Sci. 16, 2021–2034.(see Chapter 4)  •  Zhang, M., Wei, X., 2013a. Alteration of hydrological regimes caused by large-scale forest disturbances. Ecohydrology, DOI: 10.1002/eco.1374. (see Chapter 2 and 5)  •  Zhang, M., Wei, X., 2013b. Contrasted hydrological responses to forest harvesting in two large neighboring watersheds in snow hydrology dominant environment: implications for forest management and future forest hydrology studies. Hydro. Process. (Submitted) (see Chapter 5)  iv  Table of Contents Abstract.......................................................................................................................................... ii Preface........................................................................................................................................... iv Table of Contents ...........................................................................................................................v List of Tables .............................................................................................................................. viii List of Figures............................................................................................................................... xi Acknowledgements .................................................................................................................... xvi 1  Chapter: Introduction .......................................................................................................1  2  Chapter: Literature review ...............................................................................................7  2.1  Research methods for assessing the impacts of forest disturbances on hydrology in large watersheds ............................................................................................................7 2.1.1  Hydrological modelling ..............................................................................................7  2.1.2  Non-modeling approaches ........................................................................................10  2.2  Quantification of forest disturbance level......................................................................13  2.3  Impacts of forest disturbances on hydrology ................................................................17 2.3.1  Impacts of logging on hydrology ..............................................................................17  2.3.2  Impacts of fire on hydrology.....................................................................................24  2.3.3  Impacts of insect infestation on hydrology ...............................................................26  2.3.4  Forest disturbances and flow regimes .......................................................................29  3  Chapter: Research design and study areas ...................................................................32  3.1  Research design ................................................................................................................32  3.2  Study areas .......................................................................................................................38 3.2.1  Willow River watershed ...........................................................................................38  3.2.2  Cottonwood River watershed ....................................................................................43  3.2.3  Baker Creek watershed .............................................................................................48  3.2.4  Moffat Creek watershed............................................................................................53  3.2.5  Tulameen River watershed .......................................................................................58  3.2.6  Ashnola River watershed ..........................................................................................63  4  Chapter: Impacts of forest disturbances on mean flows ..............................................68  4.1  Background ......................................................................................................................68  4.2  Data ...................................................................................................................................70 4.2.1  Hydrological data ......................................................................................................70 v  4.3  4.4  4.5  4.2.2  Climatic data .............................................................................................................71  4.2.3  Forest disturbance data .............................................................................................73  4.2.4  Geographic data ........................................................................................................73  Methods.............................................................................................................................74 4.3.1  Quantification of forest disturbance levels ...............................................................74  4.3.2  Time series cross-correlation analysis ......................................................................80  4.3.3  Quantification of annual mean flow responses .........................................................81  Results ...............................................................................................................................91 4.4.1  Statistical test of cause-effect relationships between forest disturbances and annual and seasonal mean flows ...............................................................................91  4.4.2  Quantification of the impact of forest disturbances on annual mean flow ...............92  4.4.3  Integrated analysis ..................................................................................................119  Discussion........................................................................................................................122 4.5.1  Determination of forest disturbance thresholds for significant changes in annual mean flows ..................................................................................................122  4.5.2  Forest disturbances impact on annual mean flow ...................................................126  4.5.3  Forest disturbances impact on seasonal mean flows...............................................128  4.5.4  Climatic gradients and mean flow response to forest disturbances ........................129  4.5.5  Watershed resilience and the response of annual mean flows to forest disturbances.............................................................................................................130  4.5.6  Joint effects and relative contributions of forest disturbances and climate variability ................................................................................................................132  4.5.7  Single watershed study V.S. quasi-paired watershed study ....................................136  4.5.8  Implication for watershed management ..................................................................137  4.6  Summary.........................................................................................................................140  5  Chapter: Impacts of forest disturbances on flow regimes .........................................142  5.1  Background ....................................................................................................................142  5.2  Data .................................................................................................................................145  5.3  5.2.1  Hydrological data ....................................................................................................145  5.2.2  Climatic data ...........................................................................................................145  5.2.3  Forest disturbances and geographic data ................................................................145  Methods...........................................................................................................................146 5.3.1  Quantification of forest disturbance level ...............................................................146  5.3.2  Definitions of flow regime components and studied hydrological variables .........146 vi  5.4  5.5  5.3.3  Statistical test of cause-effect relationships between forest disturbances and flow regimes............................................................................................................148  5.3.4  Quantitative assessment on the impact of forest disturbances on high flow and low flow regimes.....................................................................................................148  Results .............................................................................................................................154 5.4.1  Statistical relationships between forest disturbances and high flow regimes .........154  5.4.2  Statistical relationships between forest disturbances and low flow regimes ..........155  5.4.3  Quantitative assessment on the changes in high and low flow regimes caused by forest disturbances .............................................................................................157  Discussion........................................................................................................................174 5.5.1  Forest disturbances and high flow regimes .............................................................174  5.5.2  Forest disturbances and low flow regimes ..............................................................179  5.5.3  Return periods and the responses of high flows to forest disturbances ..................184  5.5.4  Ecological implications and watershed management .............................................186  5.6  Summary.........................................................................................................................191  6  Chapter: Conclusions and future studies ....................................................................192  6.1  Conclusions .....................................................................................................................192  6.2  6.1.1  Cumulative forest disturbances and mean flows ....................................................192  6.1.2  Cumulative forest disturbances and high flow regimes ..........................................192  6.1.3  Cumulative forest disturbances and low flow regimes ...........................................193  6.1.4  The offsetting effects of forest disturbances and climate variability on annual mean flows ..............................................................................................................193  6.1.5  Forest disturbance thresholds for detectable changes in annual mean flows..........193  6.1.6  Climatic gradients and hydrological responses.......................................................194  6.1.7  Hydrological responses and watershed resilience...................................................194  Limitations and future studies ......................................................................................195 6.2.1  Limitations ..............................................................................................................195  6.2.2  Future studies ..........................................................................................................195  References ...................................................................................................................................199 Appendices ..................................................................................................................................215 Appendix A: Modified double mass curves with log-transformed data........................215 Appendix B: An example of ECA calculator ...................................................................218  vii  List of Tables Table 2.1 Major hydrologic models used in large forested watersheds ........................................9 Table 2.2 Methods for quantifying forest disturbance levels .....................................................15 Table 4.1 Equations for evaporation estimation .........................................................................82 Table 4.2 Time series cross-correlation between CECA and mean flows (annual and seasonal scales)...........................................................................................................91 Table 4.3 The ARIMA intervention model of the slope in the MDMC for the Willow River watershed ..........................................................................................................93 Table 4.4 The regression model of the MDMC for the Willow River watershed ......................94 Table 4.5 Forest disturbance effects on annual mean flows in the Willow River watershed in different periods (1968-2008) ................................................................................97 Table 4.6 Climate variability effects on annual mean flows in the Willow River watershed in different phases (1968-2008) ................................................................98 Table 4.7 The relative contributions of forest disturbances and climate variability on annual mean flow variations in the Willow River watershed.....................................98 Table 4.8 The ARIMA Intervention model of slope in the MDMC for the CottonwoodWillow quasi-paired watershed study ......................................................................100 Table 4.9 Regression model with autoregressive errors of MDMC for the CottonwoodWillow quasi-paired watershed study ......................................................................101 Table 4.10 Forest disturbance effects on annual mean flows from 1986 to 1995 estimated by the Cottonwood-Willow quasi-paired watershed study ......................................102 Table 4.11 The relative contributions of forest disturbances and climate variability on annual mean flow variations from 1986 to 1995 estimated by the CottonwoodWillow quasi-paired watershed study ......................................................................102 Table 4.12 The ARIMA Intervention model of the slope in the MDMC for the Baker Creek watershed .......................................................................................................104 Table 4.13 The autoregressive model of the MDMC for the Baker Creek watershed ...............104 Table 4.14 Annual mean flow variation and its components in the Baker Creek watershed (1999-2009) ..............................................................................................................107 Table 4.15 The relative contributions of forest disturbances and climate variability on annual mean flow variation in the Baker Creek watershed (1999-2009) .................107 Table 4.16 The ARIMA Intervention model of slope in the MDMC for the Moffat Creek watershed ..................................................................................................................109 Table 4.17 The autoregressive model of MDMC for the Moffat Creek watershed ....................109 Table 4.18 Annual mean flow variation and its components in the Moffat Creek watershed (1998-2009) ..............................................................................................................112  viii  Table 4.19 The relative contributions of forest disturbances and climate variability on annual mean flow variation in the Moffat Creek watershed (1998-2009) ...............112 Table 4.20 The ARIMA Intervention model of the slope in the MDMC for the AshnolaTulameen quasi-paired watershed study ..................................................................114 Table 4.21 The autoregressive model of MDMC for the Tulameen River watershed ...............114 Table 4.22 Forest disturbance effects on annual mean flows in the Tulameen River watershed in different phases (1984-2009) ..............................................................118 Table 4.23 The relative contributions of forest disturbances and climate variability on annual mean flow variations in the Tulameen River watershed (1984-2009)..........118 Table 4.24 Mann Whitney U test on differences in annual mean flow responses to forest disturbances (∆Qf/CECA) between different watersheds ........................................119 Table 4.25 The estimated forest disturbance thresholds for significant changes in annual mean flows in the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds ......................................................................................................124 Table 4.26 The quantification of forest disturbance effect on annual mean flow ......................127 Table 4.27 The relative contributions of forest disturbances and climate variability to annual mean flow variation ......................................................................................136 Table 4.28 Comparison of the quasi-paired watershed and single watershed studies on forest disturbance effect on annual mean flow (1986-1995) ....................................137 Table 5.1 Definitions for flow regime components ..................................................................147 Table 5.2 Canonical correlation analysis between the set of hydrological variables and the set of climatic variables in the Willow River watershed ..........................................150 Table 5.3 Canonical correlation analysis between the set of hydrological variables and the set of climatic variables in the Baker Creek watershed............................................151 Table 5.4 Selected pairs for the Willow River watershed ........................................................152 Table 5.5 Selected pairs for the Baker Creek watershed ..........................................................153 Table 5.6 Cross-correlations between forest disturbance levels (CECA) and high flow regimes .....................................................................................................................155 Table 5.7 Cross-correlations between forest disturbance levels (CECA) and low flow regimes .....................................................................................................................156 Table 5.8 The Mann-Whitney U test on the differences in flow regimes between the paired reference and disturbed years in the Baker Creek watershed ........................158 Table 5.9 The disturbance effects on the magnitude of high flows (Mh) in the Baker Creek watershed .......................................................................................................160 Table 5.10 The disturbance effects on the timing of high flows (TMh) in the Baker Creek watershed ..................................................................................................................162 Table 5.11 The disturbance effects on the return periods of high flows (RPh) in the Baker Creek watershed .......................................................................................................163 ix  Table 5.12 The disturbance effects on the variability of high flows (CVh) in the Baker Creek watershed .......................................................................................................165 Table 5.13 The disturbance effects on the magnitude and CV of low flows in the Baker Creek watershed .......................................................................................................166 Table 5.14 The Mann-Whitney U test on the differences of flow regimes in the paired reference and disturbed years in the Willow River watershed .................................170 Table 5.15 The disturbance effects on the magnitude of high flows in the Willow River watershed ..................................................................................................................170 Table 5.16 The disturbance effects on the durations of high flows in the Willow River watershed ..................................................................................................................171 Table 5.17 The disturbance effects on the return periods of low flows in the Willow River watershed ..................................................................................................................173 Table 5.18 Cross-correlations between forest disturbances (CECA) and hydrological variables in the six study watersheds .......................................................................175 Table 5.19 The responses of high flows with the different return periods in the Baker Creek and Willow River watersheds ........................................................................186 Table 5.20 Forest disturbance-induced changes in the high flow and low flow regimes and their possible effects on critical fish species ............................................................190  x  List of Figures Figure 2.1 An example of percent hydrological recovery with average forest stand height increases ....................................................................................................................16 Figure 2.2 Interactions between forest disturbances and key watershed ecological processes ...................................................................................................................29 Figure 3.1 The distribution of study watersheds in the BC interior............................................35 Figure 3.2 Research design .........................................................................................................37 Figure 3.3 Location of the Willow River watershed...................................................................39 Figure 3.4 Hypsometric curve in the Willow River watershed ..................................................40 Figure 3.5 Cumulative forest disturbances in the Willow River watershed ...............................42 Figure 3.6 Annual disturbed area by different types of forest disturbances in the Willow River watershed ........................................................................................................43 Figure 3.7 Location of the Cottonwood River watershed ...........................................................44 Figure 3.8 Hypsometric curve in the Cottonwood River watershed...........................................45 Figure 3.9 Cumulative forest disturbances in the Cottonwood River watershed .......................47 Figure 3.10 Annual disturbed area by different types of forest disturbances in the Cottonwood River watershed ....................................................................................48 Figure 3.11 Location of the Baker Creek watershed ....................................................................49 Figure 3.12 Hypsometric curve in the Baker Creek watershed ....................................................50 Figure 3.13 Cumulative forest disturbances in the Baker Creek watershed .................................52 Figure 3.14 Annual disturbed area by different types of forest disturbances in the Baker Creek watershed ........................................................................................................53 Figure 3.15 Location of the Moffat Creek watershed ...................................................................54 Figure 3.16 Hypsometric curve in the Moffat Creek watershed ...................................................55 Figure 3.17 Cumulative forest disturbances in the Moffat Creek watershed ...............................57 Figure 3.18 Annual disturbed area by different types of forest disturbances in the Moffat Creek watershed ........................................................................................................58 Figure 3.19 Location of the Tulameen River watershed...............................................................59 Figure 3.20 Hypsometric Curve in the Tulameen River watershed..............................................60 Figure 3.21 Cumulative forest disturbances in the Tulameen River watershed ...........................62 Figure 3.22 Annual disturbed area by different types of forest disturbances in the Tulameen River watershed .......................................................................................63 Figure 3.23 Location of the Ashnola River watershed .................................................................64 Figure 3.24 Hypsometric curve in the Ashnola River watershed .................................................65 xi  Figure 3.25 Cumulative forest disturbances in the Ashnola River watershed ..............................67 Figure 4.1 ECA coefficients of different forest disturbance types for a) the Willow and Cottonwood River watersheds; b) the Baker and Moffat Creek watersheds; c) the Tulameen and Ashnola River watersheds ...........................................................77 Figure 4.2 Cumulative equivalent clear-cut area (CECA) in a) the Willow River watershed from 1953 to 2008; b) the Cottonwood River watershed from 1953 to 1995; c) the Baker Creek watershed from 1960 to 2009; d) the Moffat Creek watershed from 1960 to 2009; e) the Tulameen River watershed from 1957 to 2009; and f) the Ashnola River watershed from 1960 to 2009 ....................80 Figure 4.3 A hypothesized example of MDMC for single watershed study ..............................84 Figure 4.4 A hypothesized example of MDMC for quasi-paired watershed study ....................89 Figure 4.5 The modified double mass curve for the Willow River watershed (Qa: accumulated annual mean flow; Pae: accumulated annual effective precipitation) .............................................................................................................93 Figure 4.6 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf); and b) the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Willow River watershed estimated by single watershed approach ...........................95 Figure 4.7 a) The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances (∆Qf%) in the Willow River watershed .....................................................................95 Figure 4.8 The annual mean flow variation attributed to climatic variability (∆Qc) in the Willow River watershed ...........................................................................................96 Figure 4.9 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); b) The normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA) in the Willow River watershed ......96 Figure 4.10 The modified double mass curve for the Cottonwood-Willow quasi-paired watershed study .......................................................................................................100 Figure 4.11 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf) and b) the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Willow River watershed estimated by single watershed approach .........................101 Figure 4.12 The modified double mass curve for the Baker Creek watershed ...........................103 Figure 4.13 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf); and b) the accumulated annual mean flow variation attributed to climate variability (∆Qaf) in the Baker Creek watershed ...........................................................................................105 Figure 4.14 a) The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances (∆Qf%) in the Baker Creek watershed .....................................................................106 xii  Figure 4.15 The annual mean flow variation attributed to climate variability (∆Qc) in the Baker Creek watershed ...........................................................................................106 Figure 4.16 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); and b) the normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA)in the Baker Creek watershed .......107 Figure 4.17 The modified double mass curve for the Moffat Creek watershed .........................108 Figure 4.18 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf); and b) the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Moffat Creek watershed..........................................................................................110 Figure 4.19 a) The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances(∆Qf%) in the Moffat Creek watershed................................................111 Figure 4.20 The annual mean flow variation attributed to climate variability (∆Qc) in the Moffat Creek watershed..........................................................................................111 Figure 4.21 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); b) the normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA)in the Moffat Creek watershed......................112 Figure 4.22 The modified double mass curve for the Ashnola-Tulameen quasi-paired watershed study .......................................................................................................113 Figure 4.23 The accumulated annual mean flow variation (∆Qa), accumulated annual mean flow variation attributed to forest disturbances (∆Qaf), and the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Tulameen River watershed .....................................................................................115 Figure 4.24 The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances (∆Qf%) in the Tulameen River watershed ...............................................................116 Figure 4.25 The annual mean flow variation attributed to climate variability (∆Qc) in the Tulameen River watershed .....................................................................................116 Figure 4.26 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); b) The normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA) in the Tulameen River watershed ................................................................................................................117 Figure 4.27 The correlation between annual precipitation (P) and annual mean flow responses to forest disturbances (∆Qf/CECA) ........................................................120 Figure 4.28 The comparison of annual mean flow response to forest disturbances in the watersheds with different long-term average annual precipitations .......................120 Figure 4.29 The correlation between annual mean temperature (T) and annual mean flow responses to forest disturbances (∆Qf/CECA) ........................................................121  xiii  Figure 4.30 The comparison of annual mean flow response to to forest disturbances in the watersheds with different long-term annual mean temperatures ............................121 Figure 4.31 a) The conceptualized graph of watershed resilience to the forest disturbanceinduced change in annual mean flow with a moderate to high level of disturbances; and b) with extreme high level of disturbances ................................132 Figure 4.32 The conceptualization of possible joint effects of forest change and climate variability on annual mean flows ............................................................................134 Figure 5.1 Comparison on the magnitude of high flows in the reference years and the disturbed years ........................................................................................................158 Figure 5.2 Flow duration curves for the selected pairs in the Baker Creek watershed: a) 1989-1999; b) 1968-2007; and c) 1964-2009 .........................................................159 Figure 5.3 Comparison on the timing of high flows in the paired reference and disturbed years in the Baker Creek watershed ........................................................................161 Figure 5.4 Hydrographs for the paired years with similar climates to show advancing of high flows after forest disturbances in the Baker Creek watershed: a) 19891999; and b) 1964-2009 ..........................................................................................161 Figure 5.5 The comparison on the return periods of high flows in the paired reference and disturbed years in the Baker Creek watershed ........................................................163 Figure 5.6 The comparison of the variability (CV) of high flows in the paired reference and disturbed years in the Baker Creek watershed .................................................164 Figure 5.7 High flows for the paired years in the Baker River watershed: a) 1976-2007 and b) 1964-2009 ....................................................................................................165 Figure 5.8 The comparison of the magnitude of low flows in the paired reference and disturbed years in the Baker Creek watershed ........................................................167 Figure 5.9 The comparison of the CV of low flows in the paired reference and disturbed years in the Baker River watershed ........................................................................168 Figure 5.10 The comparison on the magnitude of high flows in the paired reference and disturbed years in the Willow River watershed ......................................................169 Figure 5.11 Flow duration curves for the selected pairs in the Willow River watershed: a) 1957-1974 and b) 1960-2007 ..................................................................................169 Figure 5.12 The comparison on the durations of high flows in the paired reference and disturbed years in the Willow River watershed ......................................................171 Figure 5.13 The comparison on the durations of high flows in the pairs of 1957-1974 (a) and 1960-2007 (b) in the Willow River watershed .................................................172 Figure 5.14 The comparison on the return periods of low flows in the paired reference and disturbed years in the Willow River watershed ......................................................173 Figure A-1 The modified double mass curve for the Willow River watershed study with log-transformed data ...............................................................................................215  xiv  Figure A-2 The modified double mass curve for the Cottonwood-Willow quasi-paired watershed study with log- transformed data ...........................................................215 Figure A-3 The modified double mass curve for the Baker Creek watershed study with log- transformed data ..............................................................................................216 Figure A-4 The modified double mass curve for the Moffat Creek watershed study with log- transformed data ..............................................................................................216 Figure A-5 The modified double mass curve for the Ashnola-Tulameen quasi-paired watershed study with log- transformed data ...........................................................217 Figure B-1 The selection of data files ....................................................................................... 218 Figure B-2 The display of data for calculation ..........................................................................218 Figure B-3 The time series of calculated annual equivalent clear-area .....................................219  xv  Acknowledgements  I would like to take this opportunity to thank my supervisor Dr. Adam Wei for his endless support, knowledge, and friendship. I am grateful to my committee members Dr. Dan Moore from UBC Vancouver and Dr. David Scott from UBC Okanagan for their long-term commitment to guide my research.  My thanks also go to Dr. Tongli Wang from UBC Vancouver for help in climate data and Mr. James Wang from B.C. Ministry of Forest, Lands and Natural Resource Operations for assistance in forest disturbance data. I am grateful to Peicong Zhou from Jianxi Agriculture University for developing a program to calculate cumulative forest disturbance.  Drs. Rita Winkler, John Rex, David Spittlehouse, and Mr. Patrick Peti from the BC Ministry of Forest, Lands and Natural Resource Operations provided valuable guide and support in watershed selections as well as interpretations on various stand-level forest hydrological studies in the BC interior. They deserve my sincere appreciation.  I would like to thank all my friends in Canada and China for their help and encouragement.  Finally, I want to thank my dear family for their endless love to support me to conduct this 5year study in Canada.  xvi  1 Chapter: Introduction Forests play an important role in the water cycle by influencing rainfall interception, evapotranspiration, soil infiltration and storage, and streamflow. Forest disturbances such as logging, wildfire, and insect infestation can affect streamflow by altering its regime (i.e., magnitude, frequency, timing, duration, and variability). Numerous studies on the hydrological impacts of forest disturbances have been conducted in small watersheds (watershed area<100 km2) using the paired-watershed experimental approach. These studies have shown that forest disturbances can significantly increase annual mean flow and small to medium-sized peak flow, and increase or decrease low flow (Stednick, 1996; Neary et al., 2003; Bruijnzeel, 2004; Moore and Wondzell, 2005). However, research on hydrological responses to forest disturbances in larger watersheds (watershed area >500 km2) is limited, and inconsistent results have been reported (Ring and Fisher, 1985; Buttle and Metcalfe, 2000; Costa et al., 2003; Tuteja et al., 2007; Wei and Zhang, 2010a; Wei and Zhang, 2010b; Vose, et al., 2011).  Forest hydrological studies in large watersheds are mainly constrained by several challenges. The first challenge is the lack of an efficient, commonly-accepted methodology. The greatest difficulty in a large watershed study lies in separating the effects of forest disturbances and climate variability on hydrology (Zhang et al., 2008; Wang et al., 2009; Zheng et al., 2009; Wei and Zhang, 2010b). Forest disturbances and climatic variability are generally viewed as two major drivers interactively influencing streamflow in large forested watersheds (Buttle and Metcalfe, 2000; Sharma et al., 2000; Blöschl et al., 2007; Ma et al., 2010; Wei and Zhang, 2010b). It is commonly accepted that the effects of climate variability on hydrology must be excluded in order to quantify the hydrological impacts of forest disturbances in large watersheds. The experimental paired watersheds or physically based hydrological models, commonly used to study the hydrological effects of forest disturbances in small watersheds, however, have limitations when applied to large watersheds (Tuteja et al., 2007; Juckem et al., 2008; Scott and Prinsloo , 2008; Zégre et al., 2010; Zhao et al., 2010). The experimental paired watershed approach is generally infeasible for large watersheds given the great difficulty in locating suitable control watersheds (Fohrer et al., 2005). Similarly, physically based hydrological 1  models, such as the Distributed Hydrology Soils and Vegetation Model (DHSVM), MIKE-SHE (an integrated water simulation model designed by Danish Hydraulic Institute) and the Variable Infiltration Capacity (VIC) model are only applicable for the watersheds that are well monitored with extensive, long-term available data on vegetation, soil, topography, land use, hydrology, and climate (Kirchner, 2009; Stednick, 2008; Wei and Zhang, 2010a, b). Moreover, the empirical relationships between different watershed processes and components used in hydrological models are mainly drawn from small watershed studies, and may be problematic when transfered to large watersheds (Kirchner, 2006). In short, the most commonly used methods have limited utility in forest hydrological studies in large watersheds.  Secondly, the lack of a suitable indicator for representing and integrating various types of forest disturbances is another challenge in large watershed studies (Wei and Zhang, 2010a). In a large watershed, different types of forest disturbances are accumulated over space and time. To quantitatively represent cumulative forest disturbances over time at a watershed scale, an integrated indicator other than a simple indicator such as a total disturbed area or forest cover rate is needed. A suitable forest disturbance indicator for a large watershed should not only represent all types of disturbances and intensity ranges, but also include their cumulative forest disturbance histories and subsequent recovery processes following disturbances over space and time (Wei and Zhang, 2010a). Equivalent clear-cut area (ECA) is defined as the area that has been clear-cut, with a reduction factor to account for hydrological recovery due to forest regeneration after disturbances (BC Ministry of Forest and Rangeland, 1999). The indicator of cumulative equivalent clear-cut area (the sum of annual equivalent clear-cut area, hereinafter referred to as “CECA”) has been successfully used in the Pacific Northwest to test watershedscale forest logging or wildfire and their effects on various watershed processes including aquatic habitat, hydrology, and aquatic biology (Whitaker et al., 2002; Chen and Wei, 2008; Lin and Wei, 2008; Jost et al., 2008). In spite of the growing recognition of CECA, to my knowledge, its utility in representing all types of long-term forest disturbances in a single large watershed for hydrological studies has not been found prior to this study.  Finally, the lack of suitable study watersheds can also constrain forest hydrological studies in large watersheds. In order to detect the effects of cumulative forest disturbances on hydrology, a  2  large watershed must experience significant forest disturbances (e.g., >20-30% CECA) and must also include a sufficiently long period without forest disturbances (or with limited forest disturbances) as a comparable reference or control period. In addition, long-term data on forest disturbance history, climate, and hydrology must be available. Given the fact that the majority of large watersheds are regulated or poorly monitored, it is rather challenging to find suitable study watersheds to assess forest changes and their effects on hydrology.  In view of the shortcomings and difficulties mentioned above, efforts have been made to explore alternative methods to study the effects of land use or forest changes on large watershed hydrology (Zhang et al., 2008a,b; McCormick et al., 2009; Wang et al., 2009; Zheng et al., 2009). Advanced statistical methods (e.g., non-parametric tests, regression analysis, and time series analysis), combined with graphical methods (e.g., double mass curve, single mass curve, and flow duration curve) are promising alternatives in light of their fewer requirements for data, and their ability to generate reliable inferences and quicker assessments (Jones and Grant, 1996; Buttle and Metcalfe, 2000; Wei and Lin, 2008; Wei and Zhang, 2010b). For example, Buttle and Metcalfe (2000), by use of double mass curve and regression technique, have examined the impacts of forest disturbances on annual mean flows and peak flows in six small to large sized watersheds in the Moose River basin of Canada. Similarly, Pizarro et al. (2006) have used nonparametric tests to investigate the relation between river flows and vegetation change in the Purapel River basin of central Chile. Recently, Wei and Zhang (2010b) have developed a methodology combining statistical analysis (e.g. time series analysis) with graphical methods (e.g., modified double mass curve), which has successfully quantified the relative contributions of forest disturbances and climatic variability to annual mean flow variations in the Willow River watershed in the central interior of British Columbia (hereinafter referred to as “BC”), Canada.  The ecological role of flow regimes (referring to the magnitude, frequency, duration, variability, and timing of flow) in aquatic habitat and biodiversity has received growing recogniztion recently (Poff et al., 1997 and 2010a). However, the impact of forest disturbances on flow regimes has been seldom studied. Although several studies have examined limited components of flow regimes such as the magnitude, frequency, and timing of peak flows (Troendle and King, 1985; Cheng, 1989; Alila et al., 2009; Kuraś et al., 2012) and the magnitude and duration of low  3  flows (Calder, 2005), the majority of them have mainly focused on the magnitude of mean, peak, and low flows given their significance in water supply and flood control. To my knowledge, a comprehensive study of the effects of forest disturbances on all components of flow regimes has not been conducted even in small watersheds. However, a comprehensive understanding of forest disturbance-induced changes in flow regimes is critical to protect the structures and functions of aquatic, floodplain, and riparian ecosystems (Poff et al., 1997 and 2010a; Pettit et al., 2001). The alteration of flow regimes can negatively affect the diversity and abundance of aquatic and riparian species, channel morphological attributes, and many other ecosystem functions (Poff and Allan, 1995; Sparks, 1992; Stanford et al., 1996). This highlighs an important research gap in forest disturbances associated impacts on flow regimes.  Despite a limited number of studies to date, the topic of forest disturbances and hydrology in large watersheds has received growing attention mainly because many practices and policies of natural resource management are operated on large landscape, watershed or even regional scales. The scientific information on large watersheds is critically needed to support the design of natural resource management strategies, especially given the fact that climate change (e.g., global warming) and anthropogenic activities (e.g., logging, urbanization, and slash burns) are dramatically and extensively altering watershed processes and ecosystem functions, and are leading to more frequent and catastrophic forest disturbances (e.g., insect infestation and wildfire) (Schindler, 2001). A comprehensive understanding of forest disturbance-induced changes in large watersheds is essential for the sustainability of long-term water supply and the protection of watershed ecosystem functions under a changing environment.  Forests in the BC interior play a significant role in timber supply, biodiversity, wildlife habitat, and other ecological services. The majority of the watersheds in the BC interior have experienced forest disturbances such as logging, fire, and insect infestation in the last 50 years. Some of these watersheds such as the Baker Creek, Willow River, Moffat Creek, and Tulameen River watersheds have been disturbed by intensive logging and severe mountain pine beetle infestation and subsequent salvage-logging. For example, in the Baker Creek watershed, a tributary of the upper Fraser River, up to 2009, 70.2% of the watershed area was attacked by mountain pine beetle (hereinafter referred to as “MPB”), and cumulative logged area accounted  4  for about 41.4% of the total watershed area. The forest disturbance level in terms of CECA was up to 62.2% in 2009. Concerns have already been raised over potential flooding in lower mainland area (e.g., Vancouver) due to the large-scale MPB infestation and subsequent salvage logging in the upper and middle reaches of the Fraser River basin. In the southern interior of BC, forest disturbances have been reported to cause drying-up of some streams, which consequently places tremendous pressure on regional water supply and threatens aquatic habitats (Winkler et al., 2008; Nelitz et al., 2011). Although the impacts of forest disturbances on hydrology have received growing attention, scientific information that supports the development of watershed management strategies for the protection of watershed ecosystem functions in large watersheds is very limited in BC.  Some of large watersheds in the BC interior have experienced significant forest disturbances and have long-term data on climate, hydrology, and forest disturbances. This provides a unique opportunity to investigate the possible effects of forest disturbances on hydrology at a large watershed scale. In this research, six large watersheds along environmental gradients in the BC interior including the Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River watersheds were selected. A methodology that combines advanced statistical methods (e.g., regression analysis and time series analysis) and graphical methods (modified double mass curve and flow duration curve) was developed to assess the hydrological effects of forest disturbances in these large watersheds. The time series cross-correlation analysis was first used to determine the statistical significance of cause-effect relationships between forest disturbances and hydrological variables. In addition to the most frequently studied hydrological variables such as annual and seasonal mean flows, flow regimes were included. The single watershed or quasi-paired watershed approaches were then applied to quantify the effects of forest disturbances on annual mean flows that are of significant relationships with forest disturbances according to the statistical analysis. For the components of flow regimes that were significantly altered by forest disturbances, the paired-year approach was conducted to quantify the forest disturbance-induced changes in these components.  The major objectives of this study include (1) quantifying cumulative forest disturbances over time and space in large watersheds; (2) assessing the effects of cumulative forest disturbances on  5  annual and seasonal mean flows; (3) separating the relative contributions of forest disturbances and climatic variability to annual mean flow variations; (4) evaluating the effects of cumulative forest disturbances on flow regimes; (5) comparing the differences in hydrological responses to forest disturbances among the significantly impacted watersheds; and (6) discussing the watershed management implications of those results in the context of future forest and climate changes.  6  2 Chapter: Literature review 2.1 Research methods for assessing the impacts of forest disturbances on hydrology in large watersheds Current approaches on hydrological changes associated with forest disturbances in large watersheds can be classified into two general categories: hydrological modeling and nonmodeling. The selection of a suitable research approach mainly depends upon the purpose of the research, data availability, and the number of available watersheds. 2.1.1  Hydrological modelling  Hydrological models can be classified as lumped, semi-distributed, and distributed in light of their spatial representations. Lumped models are not spatially explicit, but view the watershed as a whole, using the average values of the watershed characteristics and inputs, which consequently lead to the averaging of hydrological processes. The model calibration and computation processes are simple, mainly suitable for a very large watershed without enough detailed spatial data. In semi-distributed models, a watershed is divided into several sub-basins or landscapes, whose hydrological processes are modeled separately as independent response units. Spatial heterogeneity can be expressed to some extent, but not in great detail. Unlike the lumped or semi-distributed models, distributed models can well represent a watershed by assigning input data and physical characteristics to grids or elements (Putz et al., 2003). Physically based distributed models are able to provide distributed approximations or predictions of hydrological variables across watersheds, and thus have a better representation of reality. However, a fully distributed model requires a large dataset involving topography, vegetation, climate, and hydrology. In a large watershed study, a semi-distributed model is commonly used given a general absence of detailed data in large watersheds.  Table 2.1 presents some examples of models applied to large watershed studies to assess the effects of forest disturbances on hydrology. Clearly, different models serve different purposes (subject to data availability), and each one has its own advantages and limitations. For example, DHSVM, a widely used distributed hydrological model, was developed particularly for assessing the effects of forest management on streamflow (Alila, 2001). Like many other distributed 7  models, its ability to simulate groundwater flows in unsaturated and saturated zones is limited. In contrast, MIKE-SHE is able to simulate the entire land phase of the hydrologic cycle including both surface and subsurface flows as well as their interactions (Danish Hydraulic Institute, 2004), but its representation of evapotranspiration process is not as complete as the DHSVM (Thyer et al., 2004; Sun et al., 2005).  In spite of increased applications, hydrological models are still based on current theories that are deeply rooted in the physics of small-scale processes. This gives rise to difficulties in representing nonlinear hydrological processes and their interactions at all scales across heterogeneous landscapes. In addition, calibrating and testing a model may not always assure its validity due to some inherent drawbacks in the approaches of parameter calibration and validation. Models are often over-parameterized to meet high accuracy levels, ignoring the equifinality problem that different parameter sets for a model might yield the same results during calibration (Beven, 1992), but yield distinctly different predictions when conditions are altered (Kirchner, 2006). Ideally, a hydrological model for a watershed can describe the hydrological system well enough even when conditions are changed. A model like a distributed one is characterized by hundreds of free parameters. The tuning process can be very tedious and timeconsuming, and potentially lead to the equifinality problem because of an excessive number of free parameters. This issue can be a major problem for large watershed modeling because the dynamics of vegetation growth over space and time make hydrologic processes more complicated, and thus result in more difficulties in parameter selection.  Despite some limitations, a modeling approach is a good choice for watersheds that have been well observed and monitored, for smaller ones in particular, while its application is largely constrained for large watersheds mainly due to the lack of detailed data and empirical relationships between various processes for model calibration and validation. Therefore, alternative methods must be explored to quantify the hydrological effects of forest disturbances and climatic variability in large watersheds.  8  Table 2.1 Major hydrologic models used in large forested watersheds Name  Spatial  Hydrological  Representation  Process  Lumped  Surface flow  Key Features  HSPF (Hydrological simulation  Water quantity and quality (Choi and Deal, 2008)  program-Fortran)  Available for a large watershed or continent scale  VIC (Variable infiltration capacity)  Semi-  Surface flow;  distributed  Subsurface flow  Water and energy modeling Simple unsaturated zone flow (Thanapakpawin et al., 2007)  Water quality and quantity SWAT  Semi-  (Soil water  distributed  assessment tool)  Surface flow; Subsurface flow  Better for agricultural watersheds Simple unsaturated zone flow (Arnold et al., 1998; Franczyk, 2008; Ma, 2009)  Hydrologic process and ecological process  LHEM (Library of hydro-  Semi-  ecological  distributed  Surface flow; Subsurface flow  (nutrient cycling, vegetation growth, decomposition) Simple unsaturated zone flow (Voinov et al., 2004)  modules)  100m spatial resolution Water and energy balance  DHSVM (Distributed hydrology, soils  Distributed  Surface flow; Subsurface flow  Forest watershed only Simple unsaturated zone flow (Bowling et al., 2000; La Marche and Lettenmaier, 2001; VanShaar  and vegetation)  et al., 2002; Stonesifer, 2007)  Entire land phase of hydrological cycle  MIKE-SHE  Distributed  Surface flow;  Available for wetland, forest, agriculture  groundwater  Completely modeled unsaturated zone flow (3dimention)(Sun et al., 2005)  Patch level resolution Hydrologic process RHESSys  Distributed  Surface flow; Subsurface flow  Ecological process such as nutrient and carbon cycling Simple unsaturated zone flow (Tague and Band, 2004)  9  2.1.2  Non-modeling approaches  Advanced statistical methods (e.g., nonparametric tests, regression analysis, and time series analysis) and graphical methods (e.g., double mass curve, single mass curve, and flow duration curve) are proven to be promising alternatives for forest hydrology studies in large watersheds (Buttle and Metcalfe, 2000; Pizarro et al., 2006). One of the most notable advantages of these methods is their ability to assess the relationship between forest disturbances and hydrology with less data, as compared with hydrological modeling. Non-modeling approaches can be applied in either multiple watersheds (Buttle et al., 1996) or a single watershed (Lin and Wei, 2008; Rodriguez et al., 2010; Wei and Zhang, 2010b).  The paired-watershed experiment is commonly accepted as a traditional, classic approach to study relationships between forest change and water yield through comparisons between two adjacent watersheds (one as a controlled watershed and the other as a disturbed one). In this way, the influences on water yield from factors other than forest change are removed. But this approach is suitable only for relatively small watersheds (<100 km2). For large watersheds (>500km2), it is impossible to locate a controlled watershed to make a pair.  In the absence of strictly controlled large watersheds, Buttle et al. (1996) proposed a concept named “quasi-paired watersheds” to study the effects of forest harvesting on streamflow using several large watersheds in Ontario, Canada. A watershed with a greater disturbance level is defined as the disturbed watershed while its neighboring, intact, or less disturbed watershed is viewed as the controlled or partially controlled one. Under the concept of quasi-paired watersheds, statistical (e.g., linear regression and ANOVA) and graphical methods (e.g., double mass curve and flow duration curve) can be applied. For example, Buttle and Metcalfe (2000) utilized double mass curves to detect streamflow responses to forest disturbances in the northeastern Ontario, Canada. However, their approach is constrained in application due to great difficulties in locating several adjacent large watersheds that have similar physical conditions and comparable long-term hydrological, meteorological, and disturbance data, but with contrasting disturbance levels. Moreover, given that large watersheds show great variations in precipitation, climate variability cannot be completely removed in the quasi-paired watershed study, leading to less reliable results. 10  The application of statistical and graphical methods in a single watershed can be rather challenging. Climatic variability and forest disturbances (or land use changes) are commonly recognized as the two most important drivers influencing streamflow changes. The greatest difficulty in a single watershed study is how to separate the hydrological effect of forest disturbances from that of climatic variability. Zhang et al. (2001) developed a two-parameter model to estimate mean annual evapotranspiration by precipitation, potential evapotranspiration, and plant-available water capacity. The relative effects of climatic variability and forest disturbances on streamflow were estimated based on their respective contributions to climatic variability and forest disturbances on evapotranspiration. Based on the work by Zhang et al. (2001) and Milly and Dunne (2002), Li et al. (2007b) generated an equation to estimate streamflow change attributed to climate variability so that the respective effects of climate variability and human activities on streamflow were quantified in the Wuding River, China. They estimated 87% of the reduction in streamflow was attributed to human activities and the rest was caused by climate variability. By use of a similar approach, Zhang et al. (2008) analyzed the relative contributions of climate and land use/cover change on streamflow variations in 11 large watersheds in the Loess Plateau, China. The relative contributions of climate variability on streamflow reduction differed among 11 watersheds, ranging from 21% to 57%, while the contributions from land use changes or human activities varied from 43% to 79%.  Instead of quantifying the impacts of both forest disturbances and climate variability in a large watershed, some studies focus on forest disturbance impact only. The study on deforestationinduced changes in annual and seasonal streamflow (50-year data) in the Tocantins River, Southeastern Amazonia is a good example of a combined application of statistical and graphic methods including t-test, ANOVA, and hydrograph (Costa et al., 2003). The two-way ANOVA was used to account for climate variability, while the hydrographs were used to interpret streamflow variations caused by land cover changes. Similarly, Rodriguez et al. (2010) used flow duration curves and non-parametric tests to examine the effect of deforestation on streamflow in the Ji-Paran´a River watershed, Southwest Amazonia. The study that used several non-parametric tests to investigate the relation between river flows and vegetation change in the  11  Purapel River basin of central Chile is another good example (Pizarro et al., 2006).  The advantages of non-modeling approaches include fewer requirements for data, quicker assessments, and more reliable inferences. The main disadvantage of a single watershed study lies in the difficulty of revealing the mechanisms responsible for watershed-specific hydrological responses because large watersheds are simply treated as “black boxes.” Another disadvantage is the requirement for long-term data on hydrology, forest disturbances, and climate. As a matter of fact, most large forested watersheds lack long-term data on hydrology and forest disturbances. Nevertheless, non-modeling approaches offer a useful alternative for detecting the effects of forest change on hydrology in large watersheds, and will likely gain more popularity in the future as more long-term data become available.  12  2.2 Quantification of forest disturbance level Forest disturbances can occur naturally or be caused by human or both, and they are normally characterized by disturbance regimes comprised of frequency, intensity, size, pattern, and agent(s). Natural disturbances (e.g., wildfire, flood, drought, hurricane, and insect), as part of natural processes, can play an important ecological role in forest and watershed ecosystems (Dale et al., 2000). They contribute to landscape diversity, nutrient circulation, species evolution, forest succession, and thus more resilient ecosystems. Because of their ecological significance, natural disturbance regimes are, therefore, often viewed as the best model for forest management guides (Roberts, 2007; Bouchard et al., 2008). Natural disturbances generally have large variations in frequency, intensity, landscape patterns, and their consequent effects. In contrast, anthropogenic disturbances (e.g., logging, road construction, agricultural activities, urbanization, mining, and recreation) generally have low variations with relatively uniform patterns, and can be permanent and catastrophic. Understanding the differences between the natural and anthropogenic disturbances is needed for the quantification of forest disturbance levels.  In a large watershed, different types of forest disturbances are accumulated over space and time. To quantitatively express cumulative forest disturbances over time at a watershed scale, an integrated indicator is needed. The most direct way to quantify forest disturbances in a large watershed is to compute the disturbed area (e.g., cumulative harvested area), mainly because these data are normally available and relatively easy to derive (Buttle and Metcalfe, 2000; Jones, 2000; Edwards and Troendle, 2012). However, the disturbed area merely serves as a basic indicator without differentiating forest species and disturbance types, and fails to express the spatial pattern of disturbances and subsequent forest recovery processes. A suitable forest disturbance indicator for a large watershed should not only express all types of disturbances, and their intensities and severities, but also account for their cumulative forest disturbance history and subsequent recovery processes over space and time.  Equivalent roaded acre or area (ERA) and equivalent clear-cut area (ECA) are recognized as better indicators than the disturbed area or forest cover rate because they account for dynamic vegetation condition or change following disturbances. A brief comparison of pros and cons  13  between these indicators is presented in Table 2.2. ERA was originally developed in the early 1980s by Region 5 of the USDA Forest Service to evaluate channel destabilization (McGurk and Fong, 1995). Although ERA has been broadened to include some other cumulative impact sources, it mainly works for assessing sediment and erosion yield. Since ERA quantifies the total disturbances by use of empirical coefficients and recovery curves for each forest activity (Cobourn, 1989), its accuracy relies greatly on professional judgments for each activity from foresters. To some extent, it just indicates a level of risk but does not reflect the actual effect of forest practices. Moreover, ERA is not spatially explicit and the impacts of an activity cannot be tested against its location in a watershed (Cobourn, 1989; Menning et al., 1996). A similar index developed by the USDA Forest Service is ECA (equivalent clear-cut area), which is often used to assess the cumulative effects of forest harvesting on annual water yield. The ECA concept has also been widely used in Canada, particularly in BC and Alberta. Roads, clear-cuts, burned areas, and partial cuts can all be expressed as “equivalent clear-cut area.” There are various revised versions of ECA calculation procedures, but the core concepts are similar (USDA Forest Service, 1974; King, 1989; BC Ministry of Forests and Rangeland, 1999; Silins, 2003). In addition, for fast ECA computation, Ager and Clifton (2005) developed a software program called ETAC (Equivalent Treatment Area Calculator), which has been successfully applied in some US forest management projects.  In a revised version developed by the BC Ministry of Forests, ECA is defined as the area that has been clear-cut, with a reduction factor to account for the hydrological recovery due to forest regeneration (BC Ministry of Forests and Rangeland, 1999). Although it was originally designed for clear-cut areas, ECA can be applied to wildfire-killed areas, roads, and other open spaces. Research has established the relationships between vegetation growth (ages or tree heights) and hydrological recovery rates following logging so that ECA can be derived spatially and temporally in a watershed (Hudson, 2000; Talbot and Plamondon, 2002; Winkler et al., 2005). A simple, generalized relationship is shown in Figure 2.1(BC Ministry of Forests and Rangeland, 1999). A recent study by Lewis and Huggard (2010) has estimated the hydrological recovery rate of forests infested by MPB according to previous stand-level studies in Canada, which makes it possible to calculate ECA for watersheds suffering from disturbances including the MPB infestation in BC. As a matter of fact, ECA has already been successfully used in BC to test  14  watershed-scale forest disturbances (mainly logging) and its effects on various watershed processes including aquatic habitat, hydrology, and aquatic biology (Whitaker et al., 2002; Chen and Wei, 2008; Lin and Wei, 2008; Jost et al., 2008). In spite of a growing recognition of ECA, to my knowledge, its utility in representing various types of forest disturbances including MPB infestation, logging, and fire in a single large watershed for forest hydrological studies has not been applied.  Table 2.2 Methods for quantifying forest disturbance levels Name  Advantage  Disadvantage Only available for single  Disturbed area or forest cover rate  Simple calculation  disturbance; No consideration of hydrological recovery  ERA (Equivalent roaded area/acre)  ECA (Equivalent clear-cut area)  Accounting for various types of disturbances; Assessing erosion risk and sediment yield  Accounting for various types of disturbances; Considering disturbance severity and hydrological recovery  Complex calculation; No consideration of hydrological recovery; Lacking spatial representation (such as the position of harvest) Complex calculation; Lacking spatial representation (such as the position of harvest)(Pike et al., 2007)  15  % recovery  100 90 80 70 60 50 40 30 20 10 0 3  4.5  6 7.5 Tree height (m)  9  Figure 2.1 An example of percent hydrological recovery with average forest stand height increases  It is important to note that the ECA calculation for a large watershed is time-consuming, and involves data collection and calculation over millions of harvested, burned, and infested blocks. Professional judgments are always needed in determining the hydrological recovery rates of different tree species for each disturbance type in different watersheds. In addition, the ECA concept may not differentiate disturbance severity in great detail, and it may not explicitly consider other land uses (e.g., agricultural land) due to the lack of empirical relationships between these land uses and watershed processes. Despite its weaknesses, ECA is, to my knowledge, the best indicator for assessing forest disturbance effects on hydrology in large forest-dominated watersheds to date.  16  2.3  Impacts of forest disturbances on hydrology  The past century has witnessed increasing research interest in the relationships between streamflow and forest disturbances (e.g., harvesting, urbanization, land use change, wildfire, and insect infestation). A general understanding is that forest disturbances can significantly increase annual mean flow, and increase or decrease peak flow or low flow in small watersheds (Stednick, 1996; Neary et al., 2003; Bruijnzeel, 2004; Moore and Wondzell, 2005). However, the research on the impacts of forest disturbances on hydrology in large watersheds (>500 km2) is limited (Wei and Zhang, 2010a; Vose, et al., 2011), and the results are inconsistent (Ring and Fisher, 1985; Buttle and Metcalfe, 2000; Costa et al., 2003; Tuteja et al., 2007; Wei and Zhang, 2010b). Moreover, the majority of studies on forest disturbance-water relationships have focused on logging while the hydrological impacts of fire and insect infestation have been less documented. The following sections provide a concise overview on the hydrological impacts of different forest disturbance types (logging, fire, and insect infestation). 2.3.1  Impacts of logging on hydrology  2.3.1.1 Mean flow The impacts of logging on mean flows have been studied for a century since the first paired watershed was established in 1909 at Wagon Wheel Gap, Colorado (Bates and Henry, 1928). Numerous studies on the responses of mean flows have been subsequently conducted worldwide mainly in small watersheds. These studies have consistently shown that logging can increase annual mean flows. This is because the lack or loss of vegetation cover after logging can reduce interception and evapotranspiration, resulting in more water being available for streamflow generation.  The change magnitude of annual mean flows caused by logging in small watersheds has been well documented by many studies. Based on a worldwide database of hydrologic studies, Calder (1993) has estimated that 10% change in forest cover can, on average, lead to a 32.6 mm increment in annual mean flows. As a matter of fact, the change magnitude of annual mean flow caused by logging can be highly variable and unpredictable among watersheds due to differences in watershed conditions such as vegetation types. The annual mean flow responses to the logging  17  of coniferous forests can be higher than the responses of deciduous forests. It is estimated that a 10% change in the forest cover of a coniferous forest can cause a 40 mm change in annual mean flow while only 25 mm for deciduous hardwoods according to a review conducted by Bosch and Hewlett (1982). Similarly, Sahin and Hall (1996) have concluded that a 10% reduction of conifer forests can lead to a 20-25 mm increment in annual mean flows, slightly higher than the increment of deciduous hardwoods (17-19 mm). Moreover, the differences in mean flow responses among watersheds may also be associated with the types of dominated hydrological processes and watershed properties. Annual mean flows in rainfall-dominated watersheds tend to be more sensitive to logging than in snow dominated watersheds. For example, in the Pacific Northwest, on average, up to a 60 mm increment in annual mean flows for each 10% increase in harvested area was found in rainfall dominated watersheds, while only 2.5–30 mm was estimated for snow dominated watersheds (Moore and Wondzell, 2005).  Large watershed studies on logging-related changes in annual mean flows are rather limited and less consistent conclusions have been reached as compared to the small watershed studies. Some studies have demonstrated that the hydrological impacts of logging in large watersheds are insignificant or minor (Buttle and Metcalfe, 2000; Wilk et al., 2001; Robinson et al., 2003; Thanapakpawin et al., 2007). For example, in the Bowron River watershed (3420 km2), logging at a level of 30% CECA yielded no detectable changes in annual mean flows (Wei and Davidson, 1998). Another study in Canadian boreal forests (with watersheds area from 401 to 11,900 km2), with disturbance levels ranging from 5% to 25% of watershed areas, also failed to find definitive changes in annual mean flows (Buttle and Metcalfe, 2000). These studies suggest that large watersheds have great ability to buffer hydrological impacts caused by a low level of forest disturbances. More surprisingly, even with forest cover reduced by 53%, Wilk et al. (2001) did not detect any significant change in annual mean flows in the Nam Pong River Basin (12,100 km2) of Northeast Thailand. On the contrary, significant hydrological responses to deforestation have been identified in many other large watershed studies (Eschner and Satterlund, 1966; Ring and Fisher, 1985; Huff et al., 2000; Matheussen et al., 2000; VanShaar et al., 2002; Costa et al., 2003; Sun et al., 2005; Siriwardena et al., 2006; Li et al., 2007a; Lin and Wei, 2008). For example, the study of the Tocaintins River watershed from a tropical region showed that a 19% reduction in forest cover led to an average increment of 88 mm (or 24%) in  18  annual mean flows (Costa et al., 2003). An example from the upper Yangtze River also found that annual mean flows increased by 38 mm on average with only 15.5% of watershed area logged (Zhang et al., 2012a). These inconsistent results may be due to significant complexities in large watersheds where various components, processes, and their interactions are involved.  Since watersheds always have ability to buffer changes caused by disturbances, a theoretical threshold of forest disturbance level must exist, below which a significant change on hydrology may not be detected. The identification of forest disturbance thresholds is useful for guiding forest management practices and protecting water resources and public safety. Efforts have already been made to determine the thresholds of forest logging in small watersheds. Generally, it is believed that more than 20% of a watershed area must be changed or disturbed to detect a significant change in annual mean flows in small watersheds (Bosch and Hewlett, 1982; Hetherington, 1987). Such thresholds tend to be variable due to differences in topography, vegetation, geology, hydrological regime, and climate. For examples, in the Appalachian Mountains, USA, only a 10% reduction in forest cover can produce a detectable response in annual mean flows (Swank et al., 1988), while in the Central Plains of the Unite States, a 50% harvest might be required for a significant change on flows. In the Rocky Mountain/ Inland Intermountain region only 15% of a watershed area logged can lead to a measurable increase in annual mean flows. These different thresholds might be explained by differences in topography, vegetation, soils, climate conditions, and hydrologic regimes among study watersheds (Stednick, 1996). However, the identification of logging thresholds for large watersheds can be very difficult given their more complex components and processes. To my knowledge, no related research has been conducted on this subject in large watersheds. 2.3.1.2 Peak flow The impacts of logging on peak flows are of great concern given their significance in flood control (Andréassian, 2004). In spite of growing studies on the impacts of forest disturbances, especially the effects of logging on peak flows, consistent conclusions are rarely drawn even for small watersheds. Many studies have suggested that peak flows can be significantly increased by logging (Troendle and King, 1985; Cheng, 1989; Gottfried, 1991; Bari et al., 1993; Caissie et al., 2002; Whitaker et al., 2002; Macdonald et al., 2003; Iroumé et al., 2006; Lin and Wei, 2008;  19  Tonina et al., 2008), while others have found insignificant changes in peak flows after logging (Harr et al. 1975; Harris 1977; Harr 1983; Harr et al. 1982; Wei and Davidson, 1998; Buttle and Metcalfe, 2000; Rodriguez et al., 2010). Interestingly, a few studies even suggested a negative effect of logging on peak flows (Cheng et al., 1975; Brandt et al., 1988; Moore and Brett, 2010). Obviously, the responses of peak flows to logging are highly variable among watersheds.  It is generally believed that the effects of logging on peak flows in a given watershed are closely related to the severity of soil disturbances after logging (Bruijnzeel, 2004). Logging operations using heavy machinery can lead to increased soil compaction, and thus greatly reduced soil infiltration capacity or hydrological conductivity. The peak flow pattern will shift from subsurface flow dominated to overland flow dominated, resulting in dramatic increases in peak flows or stormflows after severe soil compaction. Interestingly, severe soil disturbances can sometimes lead to decreases in peak flows after logging. For example, a study conducted in the UBC Malcolm Watershed Knapp Research Forest, Canada found a 22 % decrease in peak flows followed logging (Cheng et al., 1975). This decrease may be due to extensive soil disturbances that diminished the function of soil macropores, resulting in slower delivery of water to the channel. On the contrary, for watersheds with minor soil disturbances, insignificant changes in peak flows after logging have been reported (Harris 1977; Wei and Davidson, 1998). For example, studies from H.J. Andrews Experimental Forest found no significant changes in peak flows after logging in catchments HJA6 and HJA7 (Harr et al., 1982; Harr, 1983). In general, it is expected that the better the soil drainage, the smaller the responses of peak flows to logging would be.  In addition, the effects of logging on peak flows are also associated with the generation mechanism of peak flows. In the catchments with peak flows caused by rainfall, the hydrological responses to logging often depend on factors including rainfall intensity and duration, soil hydraulic conductivity, and slope morphology. Normally, with decreased evapotranspiration during growing seasons after logging, soil becomes wetter than during pre-harvest conditions, which leads to early soil saturation, and then potentially higher peak flows when rainfall occurs (Harr, 1976). In the Oregon Coast Range and South Coastal British Columbia, the relative increase in peak flows due to logging can be 20-194% according to experimental watershed  20  studies. For example, in a small experimental watershed South Fork watershed (2.3 km2), USA, peak flows were increased by 64.7% after 52% of the total watershed area was logged (Gottfried, 1991). In two larger watersheds (152 and 133 km2) in Portugal, peak flows were increased by 121% and 195%, respectively (David et al., 1994). In tropical areas, the relative change of peak flows caused by logging can be even higher (Bruijnzeel, 2004). In contrast, in catchments with peak flows driven by snowmelt process, the hydrological responses to logging tend to be smaller than the responses of rainfall-driven peak flows. In small snow-dominated watersheds in the Pacific Northwest, peak flows can be increased, on average, by 23-87% after logging (Moore and Wondnell, 2005). But insignificant changes in peak flows have also been reported (Wei and Davidson, 1998; Buttle and Metcalfe, 2000). For example, in the Deadhorse Creek in the Rocky Mountains, Troendle and King (1987) found no significant changes in peak flows caused by logging. A study in two subbasins of Horse Creek in North-Central Idaho also generated similar results (King and Tennyson, 1984).  In snow mountainous watersheds, the responses of peak flows to logging may be closely related with the synchronization of snowmelt processes at a watershed scale. Snowmelt starts earlier at lower elevation areas and south-facing slopes, and then moves to upper elevation areas. The snowmelt water from lower elevation areas and south-face slopes normally fills up rivers, while the snowmelt water from higher elevation areas and north-face slopes contributes to the generation of peak flows or potential floods. Even though with limited watershed-scale studies, it is believed that logging at higher elevation areas or north-face slopes can potentially increase peak flows or floods in snow-dominated watersheds in BC, mainly because logging at these areas can increase snow accumulation (Toews and Gluns, 1986; Troendle and King, 1987; Storck et al., 2002; Winkler et al., 2005) and accelerate snowmelt processes due to increasing temperature and radiation (Berris and Harr, 1987; Adams et al., 1998; Toews and Gluns, 1986; Berris and Harr, 1987; Winkler et al., 2005). Thus, logging occurring at higher elevation areas or north-face slopes can lead to more synchronizations of snowmelt as compared to that at lower elevation areas or south-face slopes. This can consequently increase and advance peak flows (Verry et al., 1983; Brandt et al., 1988; Schnorbus et al., 2004). On the contrary, logging concentrated at lower elevation areas or south-face slopes can cause more de-synchronizations of snowmelt among different elevations, which eventually yield insignificant impacts on peak flows  21  or even reduce them. A watershed-scale modeling study in the Reddish Creek watershed in the BC interior suggests that logging in the bottom 20% of the catchment causes little or no change in peak flows (Whitaker et al., 2002). Similarly, a study in Sweden suggested that logging at lower elevation areas, for example, near the outlet of a watershed, produced less pronounced impacts on peak flows as compared to that at higher elevation areas (Brandt et al., 1988).  Logging can actually yield different impacts on peak flows with different sizes or return periods. There is a well-established perception in forest hydrology that logging can generate more pronounced effects on small to medium-sized peak flows while its effects on large peak flows are limited based on a century of experimental paired studies at a small watershed scale (<100 km2) (Brooks et al., 2003; Calder, 2005). For example, research in the Oregon Cascade Mountains found that treatment effects on peak flows decreased as flow event sizes increased and were not detectable for flows with 2-year return intervals or greater in the treated watersheds (Thomas and Megahan, 1998). In tropic regions, relative increases in peak flows after logging can be roughly 100-300% for small rainfall events but decline to 10% or less for large events (Bruijnzeel, 2004). Larger peak flows are insensitive to logging mainly because the amounts of rainfall during these large storms are generally much greater than the increased soil moisture due to logging, and the differences in peak flows between the cut and uncut areas are primarily determined by their interception differences. Since the interception capacity of canopies is much lower for large storm or snowfall events than that for small ones, this consequently causes small or little change in large peak flows after logging. However, the understanding that the effects of forest logging on peak flows decrease with their sizes has recently been challenged by a series of modeling studies, suggesting that forest harvesting-related changes in the magnitude and frequency of floods are positively related with return periods (Alila et al., 2009; Green and Alila, 2012). Unlike the traditional approach that compares chronologically (e.g., equal meteorology or storm event) paired peak flows in the control and treatment catchments, these new studies have adopted an innovative approach by comparing the frequency distributions of peak flows in control and treatment catchments. Unfortunately, the frequency pairing study fails to exclude confounding effects from storm intensity and duration, vegetation, topography, and subsurface drainage patterns, and lacks statistical power when it compares the observed and predicted frequency distributions of floods to show the effects of logging on large floods (Lewis et al.,  22  2010). In addition, the frequency pairing study relies on hydrological models which can be overparameterized to meet high accuracy levels, ignoring the equifinality problem that different parameter sets for a model might yield the same results during calibration (Beven, 1992; Kirchner, 2006). More importantly, the predictions can be misleading and ungrounded given that the hydrological model applied is actually built on the understanding that forests have limited effects on large storm events and their resultant large peak flows. Therefore, compared with the findings by a traditional chronological approach, the results from the frequency pairing studies appear less reliable. In comparison to tremendous small watershed studies, little research on large watersheds has been performed. Whether the responses of floods to forest disturbances increase or decrease with return periods at a large watershed scale (>500 km2) still lacks evidence. 2.3.1.3 Low flow Like the impacts of logging on peak flows, less consistent conclusions have been drawn on low flows (Bruijnzeel, 2004; Moore and Wondzell, 2005). The responses of low flows to logging can be positive, negative or even insignificant (Calder, 2005). The majority of studies from rainfall dominated watersheds (Bari et al., 1993; Bent, 2001; Cornish and Vertessy, 2001; Robinson and Dupeyrat, 2005; Webb et al., 2007) and some small-scale studies from snow-dominated watersheds (Van Haveren, 1988; Swanson et al., 1986; Gottfried, 1991) suggest that logging can increase low flows. For example, the study of the Wagon Wheel Gap, Colorado indicated that 100 % clear-cutting led to a 17% increase in low flows (Van Haveren, 1988). This increase is because the removal or death of forests can decrease evapotranspiration and interception in disturbed sites, which ultimately increases soil moisture and groundwater recharge (Bosch and Hewlett, 1982). However, some studies have reported no changes or decreases in dry season flows or low flows after forest disturbances (Bruijnzeel, 2004; Calder, 2005). For example, the study from the Oregon Cascades showed that low flows after logging were reduced by 20% for the 70–100 year timber rotation schedule (Hicks et al., 1991).  As a matter of fact, the impacts of forest logging on dry season flows or low flows are mainly determined by the degree of soil disturbances (e.g., soil compaction) associated with logging and resultant soil erosion. When soil disturbances are minor or insignificant after logging (e.g.,  23  logging occurs in winter seasons when soils are completely frozen in high-latitude regions), higher dry season flows are expected since the removal of vegetation reduces evapotranspiration and thus more soil water storage to promote soil infiltration and groundwater recharge (Calder, 2005; Stednick, 2008; Zhang and Wei, 2012b). In contrast, when the topsoil is seriously compacted by logging activities, soil infiltration capacity can be severely impaired, leading to more surface runoff and less recharge to deep soil and groundwater systems and consequently causing reductions in low flows (Bruijnzeel, 2004; Zhang et al., 2012a). Moreover, the removal of cloud forests in some coastal watersheds, where fog drips intercepted by forests serves as an important precipitation input, also leads to declines in low flows. For example, with decreased fog drips after logging, water input for streamflow was reduced, resulting in a decline of low flows in summer (Harr, 1982).  In general, low flows are expected to increase in the watersheds where logging operations produce limited soil disturbances and resultant soil erosions, while decreased low flows are likely detected after logging in the areas with severely impaired soil infiltration capacity and soil water storage. However, most low flow studies are from rainfall-dominated watersheds. The impacts of low flow responses to logging in small snow-dominated watersheds have been less examined because of difficulties in accurately measuring winter flows due to ice formation in channels and on weirs (Pike and Scherer, 2003). 2.3.2  Impacts of fire on hydrology  Fire can impact hydrology through the modification of vegetation and soils. First, it can cause the loss or even death of trees and understory vegetations, resulting in decreased canopy interception and evapotranspiration, as well as reduced water storage of litters. Secondly, fire can change the chemical and physical properties of the soils and accordingly alter their infiltration capacity or hydrological conductivity. Under some circumstances, the topsoil can become hydrophobic since volatile organic gases produced by fire can condense on mineral soil grains in cooler layers below the surface after severe fires (Letey, 2001). These hydrophobic layers impede infiltration and lead to the generation of Hortonian overland flow (Dyrness, 1976). Thus, with reduced interception, evapotranspiration, and infiltration, more precipitation becomes  24  available for surface runoff generation. Coupled with decreased surface roughness, flow movements tend to be accelerated after fires.  Compared with numerous studies on logging-related hydrological responses, research on the hydrological impacts of fire, wildfire in particular is rather limited. Most of them were conducted in rain-dominated areas at the hill-slope scale or small watershed scale, possibly due to the lack of pre-fire data (Shakesby and Doerr, 2006). Unlike scheduled logging, the occurrence of wildfire is likely unpredictable. This impedes the application of traditional experimental watershed study except in a few serendipitous cases, where watersheds that were being monitored for other purposes were burned by wildfires (Hoyt and Troxell, 1934; Helvey, 1980; Woodsmith, 2004). Alternatively, the effects of prescribed fire on hydrology have been investigated to understand the hydrological impacts of fire (Lindley et al., 1988; Scott, 1993; Soler et al., 1994; Soto et al., 1994; Bêche et al., 2005; Mayor et al., 2007).  According to small watershed studies, annual mean flows can either be increased or decreased after fires and the change magnitude ranges from -13% to 700%. Small watersheds generally tend to have greater changes. For example, in a small catchment in the San Dimas Experimental Forest in the southern California, annual mean flows were increased by 200-300% during the first year after a wildfire, with an average of 30% increment for the next 5 years (Hoyt and Troxell, 1934). Another small watershed study in the Entiat Experimental Forest in Washington also found a 200-300% increase in annual mean flows over 7 years after a wildfire (Helvey, 1980). Campbell et al. (1977) even found annual mean flows were increased by up to 800% in a pine forest area in Arizona. However, annual mean flows of large watersheds tend to be less sensitive to fire compared with those of small watersheds. The change magnitude of annual mean flows caused by fire normally varied from 11% to 300% in large watersheds (Anderson et al., 1976; Nasseri, 1988). Contrarily, in the highland eucalyptus forests of Australia, annual mean flows showed an unusual decrease by 13-30 % (Langford, 1976). This reduction is mainly due to higher transpiration of regenerated young eucalyptus forests as compared with that of the burned mature forests.  The impacts of fire on peak flows can be more pronounced. As well documented, post-fire  25  increments in peak flows range from 0.45 to 870 folds. Small watersheds tend to have more significant changes. For example, a 1.45-fold increase in peak flows was observed in a coastal Oregon forest by Anderson (1974). And a 10- to 100-fold increase in peak flows was common in Coon Creek (38.86 km2), USA (Neary et al., 2003). The largest increase in peak flows ranging from 20- to 870-fold was found at the San Dimas Experimental Forest in southern California (SanDimas, 1954; Sinclair and Hamilton; 1955; Krammes and Rice, 1963). The responses of peak flows in large watersheds are less sensitive. For example, a large watershed study showed that annual peak flows were increased by about 45% after the Tillamook Burn of ponderosa pine in Trask and Wilson River watersheds (about 350 and 400 km2 respectively) (Anderson et al., 1976). Generally, peak flow increments vary from 0.45-fold to 6-fold in large watersheds while the changes can go up to 870-fold in small watersheds (Anderson et al., 1976; Campbell et al., 1977; Woody and Martin, 2001).  In general, peak flows are more sensitive to fire disturbances than annual mean flows. Moreover, hydrology in large watersheds tends to be less sensitive to fire disturbances. The fire-induced changes in the magnitude of peak flows and annual mean flows are often far beyond the normal range of hydrological changes caused by logging (Neary et al., 2003). The magnitude of hydrological responses to fire can be highly variable, which is mainly determined by factors such as the degree of soil disturbances and vegetation damage, as well as the water consumption of regenerated vegetation. Severe soil disturbances and high death rate of vegetation after fires can always cause more intensive hydrological changes. However, if vigorous regeneration occurs after fires and the new vegetation has higher water consumption than the old vegetation, insignificant increases or even decreases in flows can be expected. 2.3.3  Impacts of insect infestation on hydrology  Insect infestation is regarded as an important disturbance agent for forests. Although it is believed that insect infestation might be crucial in the long run in maintaining forest ecosystems (Logan et al., 2003), large-scale outbreaks of insects may generate damaging or even catastrophic effects. This is particularly true in the BC interior of Canada, where forests have been attacked by MPB since the 1990s and had a large-scale outbreak around 2003. As of 2008, MPB in the BC interior has affected approximately 14.5 million hectares of forests. The large26  scale MPB infestation and subsequent salvage logging can potentially produce significant hydrological effects, for example, increasing the magnitude and changing the timing of floods or increasing the magnitude of annual mean flows (Forest Practices Board, 2007; Winkler et al. 2008).  After the death of trees due to MPB infestation, the evapotranspiration of trees stops, resulting in more available water for streamflow (Uunila et al., 2006). Moreover, reductions in canopy interception as the loss of needles and branches can increase ground snow accumulation in winter and accelerate snow ablation and snowmelt in early spring as compared with undisturbed forests. This possibly leads to advanced timing and higher magnitude of peak flows. Unlike logging with the removal of trees, dead trees after MPB infestation remain standing in the forests with their canopies lost over time, changing from green stands to red stands, and then grey stands. As reported by some studies, beetle-killed trees lose all needles within 3-5 years after death and most branches within 10-15 years (Lewis and Huggard, 2006 and 2010). Accordingly, the impacts of MPB infestation on snow accumulation and snow ablation are relatively lower at the beginning (the red stand period) when trees start to defoliate. But increases in snow accumulation become significant as trees continue to lose their needles and branches during the grey stand period, and then slow down as vegetation regeneration. An early stand-level study in the north of Tabernash, Colorado examined the effects of MPB infestation on snow accumulation and summer rainfall interception, showing that net precipitation (precipitation reaching the ground) in most infested sites resembled their respective noninfested sites during the first four years after the attack, mainly due to needles retention and the existence of understory (Schmid et al., 1991). Recent studies in the BC interior, Canada compared the hydrological effects of MPB infestation with those of the clear-cuts, and also found less pronounced hydrological response during the period of red stands. On average, snow accumulation in mature and older green/red stands was 25% less than that in the openings while this number was reduced to only 13% in mature and older grey stands (Winkler et al., 2010). However, the difference in snow ablation between the clear-cut and the MPB infested sites was insignificant. The snow ablation rate in the mature and older green/red stands and the mature and older grey stands were similar, on average, 38% and 31% less than that of the openings, respectively (Boon, 2009b). As to the average snow duration in green/red stands and grey stands, similar result was also found, 3-4 days longer than  27  that in the openings (Winkler and Boon, 2009).  Comparing with logging and fire that always cause more intensive hydrologic changes in early years after disturbances and then followed by decreased responses as vegetation recovery, the hydrological impacts of MPB infestation are minor at early years after the MPB attack and then reach the maximum, and finally decline as vegetation recovery. For example, Bethlahmy (1974 and 1975) found the smallest increases in annual mean flows during the first 5-year period and the largest increases 15 years later, and the increases lasted at least 25 years after the beetle epidemic. However, it is important to note that the hydrological impacts of MPB infestation can vary among study sites and years. The stand-level studies conducted by Winkler et al. (2010) suggested great variations in the effects of MPB infestation on snow accumulation and melt. These variations are often associated with site conditions such as aspect, elevation, stand age, and attack level, as well as weather conditions. MacDonald and Stednick (2003) also believed that similar to logging-related effects, the hydrological effects of beetle infestation are most significant during wet years and less pronounced during dry years.  Despite the significance of hydrological impacts of beetle infestation, less than 10 water-scale studies have been published to date. The most classic studies were conducted in the White and Yampa Rivers (1564 km2) of Colorado by Bethlahmy (1974 and 1975). Their results suggested that annual mean flows were about 10% greater 25 years after a beetle epidemic. Another study in the Jack Creek watershed (133 km2 ) of the southwestern Montana showed that with 35% death of forest stands due to the MPB epidemic from 1975 to1977, annual mean flows and low flows were increased by 15% and 10%, respectively, and annual hydrographs were advanced by 2-3 weeks while insignificant changes were found in peak flows (Potts, 1984). A modeling study performed in the North Platte River watershed of Wyoming and Colorado also found that annual mean flows were predicted to be increased by 56 mm by the 10th year after a spruce bark beetle epidemic with 30-50% tree mortality, and this increment can continue but slow down over 60-70 years (Troendle and Nankervis, 2000). In general, the hydrological impacts of the MPB infestation are less pronounced than that of clear-cuts and similar to partial clear-cuts (Boon, 2007 and 2009a). Although those studies improve the understanding regarding MPB infestation and hydrology, more studies are clearly needed to provide a comprehensive understanding in this  28  subject across various spatial scales. 2.3.4  Forest disturbances and flow regimes  Forest disturbances such as drought, insect infestation, and wildfire are part of natural processes, which contribute to landscape diversity, nutrient circulation, species evolution, forest succession, and thus more resilient ecosystems (Dale et al., 2000). However, due to climate change, more frequent droughts, wildfire, and insect outbreaks have been reported in the last 10 years and are expected to be even more severe in terms of intensity and size in the future (Logan et al., 2003; Allen et al., 2010; Haughian and Burton, 2012). These frequent, large-scale natural disturbances coupled with increasing anthropogenic disturbances such as logging, agricultural activities, and urbanization can directly cause significant impacts on forest and watershed ecosystems. Forest disturbances can also indirectly produce ecological impacts through modifying biological, hydrological, and morphological processes. In watershed ecosystems, the interactions among biological, hydrological, morphological, and other ecological processes are complex and dynamic (Figure 2.2). Because of these close linkages and interactions, the hydrological alterations caused by forest disturbances can generate important effects on many other biological and morphological processes and functions, and consequently affects the integrity of watershed ecosystems.  Forest disturbances: Logging, insect infestation, wildfire  Biological Process: Life cycle, species distribution/diversity/abundance  Flow regime: Magnitude, Frequency, duration, timing, variability  Morphology: Channel, floodplain, bank  Ecosystem: Function and integrity  Figure 2.2 Interactions between forest disturbances and key watershed ecological processes 29  In order to reveal the possible ecological consequences of the hydrological changes caused by forest disturbances, the hydrological metrics that are closely related to ecological functions must be identified first. From an ecological perspective, hydrological conditions can be comprehensively described by flow regimes that include five flow components (magnitude, duration, timing, frequency, and variability or change rate of flows). Flow regimes are believed to play a critical role in determining the structures and functions of aquatic, floodplain, and riparian ecosystems (Poff et al., 1997 and 2010a; Pettit et al., 2001). Flow regimes can affect aquatic ecosystems directly by regulating the life cycle, distribution, abundance, and diversity of aquatic and riparian species (Sparks, 1995; Greenberg et al., 1996; Marchetti et al., 2001). For example, the natural timing of high flows or low flows can serve as important environmental cues for fish to spawn or migrate upstream or downstream (Montgomery et al., 1983; Sparks, 1995). Similarly, frequent flushing flows or floods can wash out flood-intolerant exotic species and protect native flood-resistant species (Rood and Mahoney, 1990). Flow regimes can also indirectly influence aquatic ecosystems by shaping the geomorphologic features of channels and floodplains. These channels and floodplains constitute an array of different habitats required by many aquatic and riparian species to complete their life cycles within and along a river (Dunne and Leopold, 1978; Greenberg et al., 1996; Reeves et al., 1996). Thus, the alterations of flow regimes can negatively affect the diversity and abundance of aquatic and riparian species, channel morphological attributes, and many other ecosystem functions (Sparks, 1992; Poff and Allan, 1995; Stanford et al., 1996).  Existing studies on the impacts of forest disturbances on flow regimes have mainly targeted small watersheds (<100 km2), with limited components of flow regimes. Moreover, the majority of these studies have focused on one type of disturbance (Bethlahmy, 1974; Bruijnzeel, 2004; Woodsmith et al., 2004; Stednick, 2008) rather than various types of forest disturbances that accumulate over time and space in a watershed. Among those existing studies, there are large variations on flow regime responses to forest disturbances. For examples, forest disturbances such as logging are believed to increase the magnitude and frequency of small peak flows or floods while larger floods are unlikely altered (Jones and Grant, 1996; Jones, 2000; Jones and Post, 2004). Moreover, forest disturbances can produce less profound effects or even insignificant effects on the magnitude and frequency of floods in snow-dominated watersheds  30  than in rainfall-dominated ones (Moore and Wondnell, 2005). In addition, in snow-dominated watersheds, forest disturbances have also been reported to advance the timing of snowmelt peak flows (Troendle and King, 1985; Cheng, 1989). The impacts of forest disturbances on low flows are inconsistent, and mainly focus on the magnitude of low flows only. Positive, negative, and insignificant effects on the magnitude of low flows, mainly due to logging, have all been found (Bruijnzeel, 2004; Calder, 2005; Moore and Wondzell, 2005).  In contrast with a wealth of small watershed studies, the alteration of flow regimes resulting from forest disturbances in large watersheds has been less investigated. Among the limited large watershed studies, most of them have only examined limited flow metrics such as the magnitude of selected hydrological variables (e.g., the magnitude of annual water yield, peak flows, and low flows) since those variables are important for water supply and public safety. The results are generally inconsistent (Ring and Fisher, 1985; Buttle and Metcalfe, 2000; Costa et al., 2003; Tuteja et al., 2007; Wei and Zhang, 2010b; Zhang et al., 2012b), mainly due to the great complexity of large watersheds. The limited studies and inconsistent results clearly highlight a significant research gap on the effects of forest disturbances on flow regimes in large watersheds. However, a comprehensive understanding of flow regime alteration due to forest disturbances is essential for water and forest managers to design management strategies for long-term water sustainability and the protection of watershed ecosystem functions. This is particularly evident as forest disturbances (e.g., insect infestation and wildfire) become more frequent and catastrophic due to climate change and human activities. Therefore, more comprehensive studies on the effects of forest disturbances on all components of flow regimes are urgently needed, particularly in large watersheds.  31  3 Chapter: Research design and study areas 3.1 Research design This study involves six major tasks including watershed selection, the quantification of watershed-scale forest disturbance level, the identification of significant impacts of forest disturbances on hydrological variables of interest, the quantification of forest disturbance related changes in annual mean flows, high flows, and low flows, the comparison of the results from study watersheds across environmental gradients, and the discussion on possible implications for future watershed management and ecosystem protection. First, six large watersheds (>500 km2) along environmental gradients were selected in the BC interior, Canada. They include the Willow River watershed in the central-northern interior, the Cottonwood River, Baker Creek, and Moffat Creek watersheds in the central interior, and the Tulameen and Ashnola River watersheds in the southern interior(Figure 3.1). The Baker and Moffat Creek watersheds have experienced extremely severe forest disturbances including logging, MPB infestation, and fire in the last 10 years, with 62.2 % and 65.7% of CECA, respectively. The Willow and Tulameen River watersheds, with their lengthy disturbance histories, have progressively experienced substantial forest disturbances since the1950s, even though they have lower disturbance levels (with 35.4 % and 33.8% of CECA, respectively) as compared with the Baker and Moffat Creek watersheds. The Cottonwood and Ashnola River watersheds are the least disturbed with 12.2 % and 6.4% of CECA, respectively, and were used as “controls” of quasi-paired watersheds in this study (more details are provided in the next section).Table 3.1 provides a brief description of the watershed properties, and more detailed descriptions can be found in the next section.  Cumulative equivalent clear-cut area (CECA) was used to indicate cumulative forest disturbances over space and time at a large watershed scale. The overall methodology is that time series cross-correlation analysis was first conducted to test the statistical significance of the cause-effect relationships between forest disturbances and hydrological variables for each study watershed (statistical assessment). The single watershed approach or quasi-paired watershed approach or both were further applied to quantitatively assess the impacts of forest disturbances on annual mean flows that were significantly related to forest disturbances according to the statistical assessment. Similarly, the paired year approach was used to quantify the impacts of 32  forest disturbances on flow regimes that were significantly altered by forest disturbances according to the statistical assessment (Figure 3.2).  To quantify the impacts of forest disturbances on annual mean flows, the selected watersheds with significant cause-effect relationships between forest disturbances and annual mean flows were analyzed individually (a single-watershed approach) or were quasi-paired with their comparable “control” watersheds for analysis (a quasi-paired watershed approach) depending on data availability and watershed conditions (Figure 3.2). The single watershed approach that combined techniques including modified double mass curve (hereinafter referred to as “MDMC”), autoregression, and time series Autoregressive Integrated Moving Average (hereinafter referred to as “ARIMA”) intervention analysis, was employed to quantify the annual mean flow changes attributed to forest disturbances in the intensively disturbed watersheds including the Willow River, Baker Creek, and Moffat Creek watersheds. The quasi-paired watershed approach was applied to two pairs of watersheds only: Willow-Cottonwood and Tulameen-Ashnola. The Cottonwood and Ashnola River watersheds served as the “control” watersheds in their respective pairs. The Willow River watershed was analyzed by both the single watershed approach and quasi-paired watershed approach. This research strategy provided a useful way to compare the validity of these two approaches. The results from these four watersheds were further analyzed to explore how the forest disturbance-induced changes in annual mean flows varied along environmental gradients by use of statistical techniques such as Kendall tau correlation, boxplots, and Mann-Whitney U test. Moreover, the results from the Willow and Tulameen River watersheds, two watersheds with long disturbed periods, were also analyzed to explore the role of watershed resilience in the responses of annual mean flows to forest disturbances. Finally, the implications of these findings for watershed management and ecosystem protection were discussed.  A paired year approach was developed to quantify the changes in the flow regimes of high flows and low flows for each significantly impacted watershed (a single watershed study). This approach matched the years in the reference period with the years in the disturbed period according to their similarities in climate conditions in each watershed. Statistical techniques used in the analysis included the Kendall tau correlation, canonical correlation, and the Mann-  33  Whiteney U test. Finally, the results were compared to explore the role of environmental factors in determining the hydrological responses to forest disturbances in the study watersheds. Possible ecological implications for future watershed management were also discussed.  34  Figure 3.1 The distribution of study watersheds in the BC interior  35  Table 3.1 A summary of watershed characteristics Metrics  Willow  Cottonwood Baker  Moffat  Tulameen  Ashnola  Drainage area(km2)  3185  1910  1560  539  1780  1050  H60 Line(m)  970  1040  1040  1060  1300  1800  Elevation range(m)  570-2080  542-1936  475-1500  778-2155  629-2302  478-2627  Watershed slope (%)  0.3  0.5  0.6  0.7  1.0  2.2  Soil type  Luvisolic  Luvisolic  Luvisolic  Luvisolic  Brunisolic  Brunisolic  Annual Precipitation(mm) 824  840  543  603  942  727  Winter Snowpack(mm)  335  321  209  247  572  442  Mean Temperature(°C)  2.8  3.4  2.5  3.4  4.3  3.8  Annual ET(mm)  400  420  430  442  500  420  Hydrological regime  Snow  Snow  Snow  Snow  Hybrid  Hybrid  Annual mean flow(mm)  424  395  105  135  386  226  Runoff Coefficient  0.52  0.47  0.19  0.22  0.41  0.31  Peak flow(m3/s)  238  198  44.9  24.1  197.3  74.2  Timing of peak flow  Apr to Jun  Apr to Jun  Apr or May  Apr or May May or Jun  May or Jun  Low flow(m3/s)  7.2  3.0  0.5  0.4  1.8  0.8  BioGeoClimatic zone  SBS, ESSF  SBS, ESSF  SBPS, SBS  SBPS,SBS,  IDF, ESSF  PP,IDF,  ESSF  ESSF  Dominant tree species  SW,FS  SW,FS  SW,PL  SW,PL  PL,FID  Dominant Disturbance  Logging,  Logging,  Logging,  Logging,  Logging, MPB Logging,  Type  MPB  MPB  MPB  MPB  CECA (%)  35.4  12.2  62.2  65.7  33.8  6.4  Disturbance level  Moderate  Low  High  High  Moderate  Low  Disturbance above H60  59.2%  43.6%  66%  70.6%  52.8%  76.6%  PL,FID  MPB  Note: H60 Line referes to the elevation for which 60% of the watershed area is above; SW, FS, PL, and FID are white spruce, subalpine fir, lodgepole pine, and interior Douglas fir.  36  Watershed Selection  Quantification of forest disturbance level  Statistical analysis of forest disturbance impacts on hydrology  Frequency  Flow regimes of high lows and low flows  Variability  Annual mean flows  Quasi-paired Watershed Single Watershed  Timing Magnitude  Single Watershed  Single Watershed  Mean flows (annual, seasonal)  CECA  Duration  High flows Quantitative analysis of forest disturbance impacts on hydrology  Paired Year  Low flows  Comparisons of results across watersheds Figure 3.2 Research design  37  3.2 Study areas 3.2.1  Willow River watershed  The Willow River, about 223 km in length and with a drainage area of 3185 km2, locates in the north-central interior of BC, Canada. It originates in the northern Cariboo Plateau near Barkerville and flows northwest to enter the Fraser River near Prince George (Figure 3.3). The total length of stream network is 2729.6 km and the stream network density is 0.86 km/km2. This watershed is mainly comprised of a long broad valley with a low relief (20% of the total area flat). The average slope of the watershed is 0.3%. Elevations for the watershed range from 570 m at the river mouth to 2080 m at the headwaters, with 60% of the total watershed area above 970 m (H60 line, Figure 3.4) and less than 10% above 1500 m. Most areas in the watershed are characterized by Luvisolic soils with the headwaters dominated by Podzolic soils. The climate of the Willow River watershed is continental, characterized by long, severe, snowy winters and relatively warm, moist, and short summers, along with moderate annual precipitation. Mean annual precipitation is about 820 mm. Most areas in the watershed are under snow from November to March and snowfall accounts for about 41% of annual precipitation. Annual mean temperature is 2.8 °C. The climate is coldest in January (monthly mean temperature varying from -22.2 to -3.1°C with an average of -9.5°C) and warmest in July (monthly mean temperature varying from 11.5 to16.5°C with an average of 13.4°C). The hydrology of the Willow River watershed is snowmelt-dominated with peak flows occurring in late spring (April to June) due to snowmelt. Annual mean flows range from 230 to 682 mm with an average of 424 mm during the study period. Annual peak flows fluctuate from 98 to 536 m3/s with an average of 238 m3/s, and annual 7-day low flows vary from 3.2 to 13.3 m3/s with an average of 7.2 m3/s. The average runoff coefficient (the ratio of annual mean flow to annual precipitation) is 0.52.  38  Figure 3.3 Location of the Willow River watershed  39  2500  elevation(m)  2000 1500 1000 500 0 0  10  20  30  40  50  60  70  80  90  100  % of total watershed area over  Figure 3.4 Hypsometric curve in the Willow River watershed  According to the biogeoclimatic ecosystem classification (hereinafter referred to as “BEC”) system, this watershed mostly lies in the sub-boreal spruce (hereinafter referred to as “SBS”) zone, specifically SBS moist cool (SBSmk) and SBS moist hot (SBSmh) zones, and higher elevation areas are dominated by the Engelmann Spruce Subalpine Fir zone (hereinafter referred to as “ESSF”), specifically ESSF wet cold (ESSFwk) and ESSF wet cool (ESSFwc) zones. The upland forests are dominated by coniferous species such as white spruce (Picea glauca), hybrid white spruce (Picea engelmannii x glauca), and subalpine fir (Abies lasiocarpa) in moist areas, and lodgepole pine (Pinus contorta) and Douglas-fir (Pseudotsuga menziesii) in drier areas. Broad-leave species including trembling aspen (Populus tremuloides) and paper birch (Betula papyrifera) are frequent components of seral stands. Riparian vegetation consists of small alders (Alnus sp.) and willow (Salix sp.) (Beaudry and Nassey, 1994). The shrub layer includes species such as black huckleberry (Vaccinium membranaceum), thimbleberry (Rubus parviflorus), highbush-cranberry (Viburnumedule), and Sitka alder (Alnus crispa ssp. sinuata) (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). Logging is the leading disturbance type, and then followed by MPB infestation and to a less extent, fire in the Willow River watershed. Based on the Vegetation Resources Inventory (hereinafter referred to as “VRI”) data, a forest stand in this watershed was disturbed either by 40  one type of disturbance (logging, fire, or MPB) or two types of disturbances (logging + fire or logging +MPB) from 1953 to 2008 (Figure 3.5). Logging activities dated back to 1955 and intensive logging lasted for a long period between 1963 and 1996, followed by a steady decline from 1997 to 2002 and then a sharp increase since 2003. Annual logged area (including areas disturbed by logging, logging+fire, and logging+MPB) was, on average, 20.8 km2 (0.65% of the total watershed area) over the study period. In the most intensively disturbed year (1972), 1.87% of the watershed area (59.6 km2) was logged (Figure 3.6). Up to 2008, cumulative logged area reached 1174 km2 (36.87% of the total watershed area). The large-scale MPB outbreak started in 2003. The area infested by MPB (including MPB and MPB + logging) covered 5.8 % of the total watershed area (183 km2) during the period of 2003-2008. Fire is another active disturbance agent in the Willow River watershed. The cumulative area disturbed by fire was 79 km2 until 2008. The largest fire occurred in 1961 with about 63 km2 areas burned. Up to 2008, cumulative disturbed area (including all types of disturbances) was 1312.6 km2, accounting for 41.2% of the total watershed area. Moreover, most of the logging activities occurred at higher elevation areas with 59.2% of logging distributed above 970 m (H60 line). In general, the Willow River watershed has been substantially disturbed by forest disturbances for a long period, especially at higher elevation areas. The Willow River provides migration, spawning, incubation, and rearing habitats for fish species including rainbow trout (Oncorhynchus mykiss), largescale sucker (Catostomus macrocheilus), mountain whitefish (Prosopium williamsoni), pink salmon (Oncorhynchus gorbuscha), and chinook salmon (Oncorhynchus tshawytscha) (BC Ministry of Environment, 2012). Streams in this watershed are major spawning and rearing habitats for chinook salmon and pink salmon. The main commercial anadromous species in this watershed is chinook salmon, while the stock of pink salmon is relatively lower (Department of Fisheries and Oceans, 1990).  41  Figure 3.5 Cumulative forest disturbances in the Willow River watershed  42  2.5 All MPB Logging Logging and Fire MPB and Logging Fire  Annual Disturbed Area (%)  2  1.5  1  0.5  1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007  0  Year Figure 3.6 Annual disturbed area by different types of forest disturbances in the Willow River watershed  3.2.2  Cottonwood River watershed  The Cottonwood River, about 146 km in length and with a drainage area of 1910 km2, lies in the central interior of BC, Canada. It originates in the northern Cariboo Plateau and flows northwest and enter the Fraser River north of Quesnel (Figure 3.7). The total length of stream network is 1324 km and the stream network density is 0.74 km/km2. The watershed is located within the Quesnel highlands with a low to moderate relief. The average slope of the watershed is 0.5%. Elevations for the watershed range from 542 m at the river mouth to 1936 m at the headwaters, with 60% of the total area above 1040 m (H60 line, Figure 3.8) and less than 10% above 1500 m. Most of the watershed is characterized by Luvisolic soils while the headwaters are dominated by Podzolic soils with a small portion of Brunisolic soils.  43  Figure 3.7 Location of the Cottonwood River watershed  44  2500  elevation(m)  2000 1500 1000 500 0 0  10  20  30  40  50  60  70  80  90  100  % of total watershed area over  Figure 3.8 Hypsometric curve in the Cottonwood River watershed  The climate of the Cottonwood River watershed is continental, characterized by warm, moist summers and cool winters with a moderate growing season. The mean annual precipitation is about 840 mm. Most of the watershed is under snow from November to March, with snow accounting for 38% of annual precipitation. Annual mean temperature is 3.4 °C. The climate is coldest in January (monthly mean temperature from -22 to -1.7°C with an average of -9.1°C) and warmest in July (monthly mean temperature from 12.1 to16.8°C with an average of 14.5°C). The hydrology of the Cottonwood River is snowmelt-dominated with peak flows occurring in the late spring (April to June) due to snowmelt. During the study period (1965-1995), annual mean flows range from 252 to 624 mm with an average of 396 mm. Annual peak flows fluctuate between 73.1 and 371 m3/s with an average of 196 m3/s, and annual 7-day low flows vary from 0.71 to 7.7 m3/s with an average of 2.8 m3/s. The average runoff coefficient is 0.47. According to the BEC system, most of this watershed lies in the SBS zone, specifically SBS moist warm (SBSmw) and SBS dry warm (SBSdw) zones, and the higher elevation areas are dominated by ESSF zone, specifically ESSF wet cold (ESSFwk) and ESSF wet cool (ESSFwc) zones. The upland forests are dominated by coniferous species such as white spruce (Picea glauca), and subalpine fir (Abies lasiocarpa) in the moist slopes while lodgepole pine (Pinus contorta) and Douglas-fir (Pseudotsuga menziesii) in drier areas. Broad-leave species including trembling aspen (Populus tremuloides) and paper birch (Betula papyrifera) are frequent components of seral stands. Riparian vegetation consists of small alders (Alnus sp.) and willow 45  (Salix sp.) (Beaudry and Nassey,1994). The shrub layer includes species such as black huckleberry (Vaccinium membranaceum), thimbleberry (Rubus parviflorus), highbush-cranberry (Viburnumedule ), and Sitka alder (Alnus crispa ssp. sinuata) (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). There is a small portion of Alpine Tundra Zone in high elevation areas along the eastern boundary of the Cottonwood River watershed. During the study period from 1965 to 1995, logging is the leading disturbance type and to a lesser extent, fire in the Cottonwood River watershed. Based on the VRI data, a forest stand in this watershed was disturbed by either one type of disturbance (logging, fire, and MPB) or two types of disturbances (logging + fire) (Figure 3.9). Logging activities dated back to 1956 but only on a small-scale except in 1979 and 1994, with 1.25% (23.9 km2) and 1.05% (20.0 km2) of the total watershed area logged, respectively (Figure 3.10). Mean annual clear-cut area (including areas disturbed by logging and logging + fire) was 5.4 km2 (0.28% of the total watershed area) over the study period. Up to 1995, cumulative clear-cut area came up to 231.9 km2 (11.7% of the total watershed area). Fire disturbances are minor in the Cottonwood River watershed. The cumulative area burned by fire was 2.7 km2 until 1995. The largest fire occurred in 1978 with 2 km2 areas burned. The MPB infestation hardly occurred before 1995. Up to 1995, cumulative disturbed area (including logging and fire) was up to 233.6 km2, accounting for 12.2% of the total watershed area. Moreover, most of the logging activities concentrated at lower elevation areas with only 43.6% of clear-cuts above 1040 m (H60 line). In general, forest disturbances in the Cottonwood River watershed before 1995 are limited, especially at higher elevation areas. The Cottonwood River provides migration, spawning, incubation, and rearing habitats for fish species including chinook salmon (Oncorhynchus tshawytscha), coho salmon (Oncorhynchus kisutch), pink salmon (Oncorhynchus gorbuscha), bull trout (Salvelinus confluentus), and rainbow trout (Oncorhynchus mykiss) (BC Ministry of Environment, 2012). Streams in this watershed are major spawning and rearing habitats for chinook salmon, the main commercial anadromous species of this watershed. The stocks of pink salmon and coho salmon are relatively low (Department of Fisheries and Oceans, 1990).  46  Figure 3.9 Cumulative forest disturbances in the Cottonwood River watershed  47  1.4 All MPB Logging Fire Logging and Fire  Annual Disturbed Area (%)  1.2 1 0.8 0.6 0.4 0.2  1995  1992  1989  1986  1983  1980  1977  1974  1971  1968  1965  1962  1959  1956  1953  0  Year  Figure 3.10 Annual disturbed area by different types of forest disturbances in the Cottonwood River watershed  3.2.3  Baker Creek watershed  The Baker Creek, about 114 km in length and with a drainage area of 1570 km2, flows into the Fraser River in Quesnel in the central interior of BC, Canada (Figure 3.11). The total length of the stream network is 1400 km and the stream network density is 0.90 km/km2. The average slope of the watershed is 0.6%. Most of the watershed is a plateau with a low relief. Elevations for the watershed range from 475 m at the river mouth to 1500 m at the headwaters, with 60% of the total area above 1040 m (H60 line, Figure 3.12) and less than 1% above 1500 m. Areas at higher elevations and the valley bottom above the canyon section are characterized by volcanic bedrock. Unconsolidated sediments are dominant at middle elevation areas, while the middle or canyon section of the watershed is a complex of meta sedimentary and volcanic rock. The watershed is dominated by Luvisolic soils with a small portion of Podzolic soils at higher elevations.  48  Figure 3.11 Location of the Baker Creek watershed  49  1600 1400 elevation(m)  1200 1000 800 600 400 200 0 0  10  20  30  40  50  60  70  80  90  100  % of total watershed area over  Figure 3.12 Hypsometric curve in the Baker Creek watershed  The climate of the Baker Creek watershed is typical continental, characterized by long, severe winters and relatively cool, very dry, and short summers, along with low annual precipitation due to its interior location in the shadow of the Coast Mountains. The dry air and clear skies created by the rain shadow effect result in significant night-time radiation cooling and low overnight temperatures. Night-time frosts are common in all months. The mean annual precipitation is about 542 mm. Most of the watershed is under snow from November to March with snow accounting for about 39 % of annual precipitation. Annual mean temperature is 2.5 °C. The climate is coldest in January (monthly mean temperature from -22.7 to -2.4°C with an average of -9.3°C) and warmest in July (monthly mean temperature from 11.4 to16.0°C with an average of 13.4°C). The Baker Creek watershed is very dry with highly variable streamflow. The hydrology of this watershed is snowmelt-dominated with peak flows occurring in the late spring (April or May) due to snowmelt. During the study period, annual mean flows, with an average of 105 mm, can be as low as 24 mm while in the wettest year can be 182 mm. Annual peak flows fluctuate between 8.7 and 129 m3/s with an average of 44.8 m3/s and annual 7-day low flows vary from 0.1 to 1.4 m3/s with an average of 0.5 m3/s. The average runoff coefficient is only 0.19.  According to the BEC system, this watershed is primarily located within the Sub-Boreal Pine Spruce (hereinafter referred to as “SBPS”) biogeoclimatic zone, specifically SBPS dry cool (SBPSdc) and SBPS moist cold (SBPSmk). The SBS zone (mainly SBS moist cold (SBSmc) and  50  SBS dry warm (SBSdw)) and the Montane Spruce (hereinafter referred to as MS) biogeoclimatic zones (mainly MS very dry very cold (MSxv)) can also be found at middle and higher elevations, respectively. The dominant tree species include lodgepole pine (Pinus contorta) and white spruce (Picea glauca) (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). The upland coniferous forests are dominated by lodgepole pine. Large area of forests actually contains no tree species other than lodgepole pine. Due to an extensive fire history, the pine trees are generally young, even-aged, and often dense with low productivity limited by the harsh climate. White spruce is usually occurs in the understory and occasionally presents in the canopy of mature pine stands in wetter areas. Trembling aspen (Populus tremuloides) is a common seral species throughout the zone but the stands it dominates are usually small. Douglas fir (Pseudotsuga menziesii) and black cottonwood (Populus balsamifera subsp. Trichocarpa) can also be found (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). Logging, MPB infestation, and fire are recognized as three major forest disturbance types in the Baker Creek watershed. Based on the VRI data, a forest stand in this watershed was disturbed by either one type of disturbance (logging, fire or MPB) or two types of disturbances (logging + fire or logging +MPB) (Figure 3.13). Between the 1960s and 1970s, forest disturbances were rather limited except for a fire that occurred in 1961 in the Long John Creek-Wentworth Lake area, a tributary to the Baker Creek. As a result of this burn, about 0.3% of the watershed area was burned. The cumulative area burned by fire was less than 1% until 2009. MPB disturbances were rare before 2000, and then became the dominant disturbance type after the large-scale outbreak in 2003, with 17.3% of the watershed area affected in that year alone. 85% of forest stands are pine-leading and 83% of them have been attacked by MPB. Up to 2009, forests attacked by MPB comprised up to 70.2% (1095.1 km2) of the total watershed area. Logging has been the most dominant anthropogenic disturbance type since 1970. Large-scale logging activities occurred in two periods (1975-1980 and 1989-2009) (Figure 3.14). The most intensive logging took place between 2001 and 2009 as a result of salvage logging in response to the large-scale MPB outbreak. During that period, 23.8% (371.3 km2) of the watershed (14% salvage logged) was harvested with an average clear-cut rate of 2.6% (40.6 km2) per year. Up to 2009, cumulative logged area accounted for 41.4% (645.8 km2) of the total watershed area (Figure 5a). Cumulative disturbed area (including all disturbance types) is up to 1470.3 km2, accounting for 94.3% of the  51  total watershed area. Thus, the Baker Creek watershed has been severely disturbed by logging, MPB infestation and subsequent salvage logging, particularly in the last 10 years.  Figure 3.13 Cumulative forest disturbances in the Baker Creek watershed  52  25 All MPB Logging Fire Logging and Fire MPB and Logging  Annual Disturbed Area (%)  20  15  10  5  1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008  0 Year  Figure 3.14 Annual disturbed area by different types of forest disturbances in the Baker Creek watershed  The Baker Creek provides migration, spawning, incubation, and rearing habitats for fish species including chinook salmon (Oncorhynchus tshawytscha), coho salmon (Oncorhynchus kisutch), pink salmon (Oncorhynchus gorbuscha), chum salmon (Oncorhynchus keta), westslope cutthroat trout (Oncorhynchus clarkii lewisi), and rainbow trout (Oncorhynchus mykiss) (BC Ministry of Environment, 2012). Streams in this watershed are major spawning habitats for chinook salmon and pink salmon, the main commercial anadromous species of this watershed. The stocks of coho salmon and chum salmon are relatively low (Department of Fisheries and Oceans, 1990).  3.2.4  Moffat Creek watershed  The Moffat Creek, about 88.1 km in length and with a drainage area of 540 km2, flows into the Fraser River near Horsefly in the central interior of BC, Canada (Figure 3.15). The total length of the stream network is 552 km and the stream network density is 1.0 km/km2. The average slope of the watershed is 0.7%. Elevations for the watershed range from 778 m at the river mouth to  53  2155 m at the headwaters, with 60% of the total area above 1060 m (H60 line, Figure 3.16) and about 10% above 1500 m. Most of the watershed lies in the Fraser Plateau physiographic area which is characterized by flat and gently rolling country with large areas of undissected upland lying between 1200 and 1500 m in elevation (Holland 1976). Areas at higher elevations and the valley bottom above the canyon section are characterized by volcanic bedrock. Unconsolidated sediments are dominant at middle elevations, while the middle or canyon section of the watershed is a complex of meta sedimentary and volcanic rock. The watershed is dominated by Luvisolic soils with some Podzolic soils at higher elevations.  Figure 3.15 Location of the Moffat Creek watershed  54  2500  elevation(m)  2000 1500 1000 500 0 0  10  20  30  40  50  60  70  80  90  100  % of total watershed area over  Figure 3.16 Hypsometric curve in the Moffat Creek watershed  The climate of the Moffat Creek watershed is typical continental, characterized by long, severe winters and relatively cool, very dry, and short summers, along with low annual precipitation due to its interior location in the shadow of the Coast Mountains. But the headwaters in the west mountaineous area of the watershed tend to be wetter than the flat plateau areas in the east. Mean annual precipitation is about 603 mm and about 41% is snow. Most of the watershed is under snow from November to March. Annual mean temperature is 3.4 °C. The climate is coldest in January (monthly mean temperature varies from -21.0 to -1.5°C with an average of -8.5°C) and warmest in July (monthly mean temperature varies from 11.6 to17.0°C with an average of 14.4°C). The hydrology of this watershed is snowmelt-dominated with peak flows occurring in the late spring (April or May) due to snowmelt. During the study period, the annual mean flows fluctuate between 65mm and 246 mm with an average of 135mm. Annual peak flows vary between 11.4 and 62.8 m3/s with an average of 24.1 m3/s, and annual 7-day low flows vary from 0.1 to 0.7 m3/s with an average of 0.4 m3/s. The Moffat Creek watershed is relatively dry, similar to the Baker Creek watershed. The average runoff coefficient is only 0.22.  According to the BEC system, this watershed is primarily located within the SBPS and SBS biogeoclimatic zones, specifically SBPS moist cold (SBPSmk) and SBS dry warm (SBSdw). ESSF zones (mainly ESSF wet cool (ESSFwc) and ESSF wet cold (ESSFwk)) can also be found at middle and higher elevations, respectively. Similar to the Baker Creek watershed, the 55  dominant tree species include lodgepole pine (Pinus contorta) and white spruce (Picea glauca) (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). The upland coniferous forests are dominated by lodgepole pine. Large area of forests actually contains no tree species other than lodgepole pine. White spruce is usually occurs in the understory and occasionally presents in the canopy of mature pine stands in wetter areas while Douglas-fir (Pseudotsuga menziesii ) in drier slopes. Trembling aspen (Populus tremuloides) is a common seral species throughout the zone but the stands it dominates are usually small (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). Logging and MPB infestation are recognized as two major forest disturbance types in the Moffat Creek watershed while fire disturbances are minor. Based on the VRI data, a forest stand in this watershed was disturbed by either one type of disturbance (logging, fire or MPB) or two types of disturbances (logging + fire or logging +MPB)(Figure 3.17). Before 1990, forest disturbances were limited and the cumulative disturbed area was only 41.3 km2 (7.6% of the total watershed area). During the 1990s, forest disturbances started to increase (Figure 3.18). Up to 1998, cumulative disturbed area was 109.5 km2 (20.3% of the total watershed area). MPB disturbances were rare before 2000, and then became the dominant disturbance type after its large-scale outbreak in 2003, with 41.4% (223.3 km2) of the watershed area affected between 2003 and 2007. Up to 2009, forests attacked by MPB reached 57.0% (307.1 km2) of the total watershed area. Intensive logging activities occurred since 2003 in response to the large-scale outbreak of MPB. 24.1% (130 km2) of the watershed (9.5% salvage logged) was harvested during the period of 2003-2009 with an average clear-cut rate of 3.4% (18.6 km2) per year. Up to 2009, cumulative logged area accounted for 48.3% (260.1 km2) of the total watershed area. Until 2009, cumulative disturbed area is up to 483.5 km2, accounting for 89.7% of the total watershed area. Thus, the Moffat Creek watershed has been severely disturbed by logging, MPB infestation and subsequent salvage logging, particularly in the last 10 years. The Moffat Creek provides migration, spawning, incubation, and rearing habitats for fish species including sockeye salmon (Oncorhynchus nerka), chinook salmon (Oncorhynchus tshawytscha), coho salmon (Oncorhynchus kisutch), pink salmon (Oncorhynchus gorbuscha), rainbow trout (Oncorhynchus mykiss), redside shiner (Richardsonius balteatus), and northern pikeminnow (Ptychocheilus oregonensis) (BC Ministry of Environment, 2012). Streams in this watershed are 56  major spawning habitats for sockeye salmon, the main commercial anadromous species of this watershed. They also provide spawning habitats for kokanee salmon, another type of sockeye salmon. The stocks of chinook salmon, coho salmon, and pink salmon are relatively low (Department of Fisheries and Oceans, 1990).  Figure 3.17 Cumulative forest disturbances in the Moffat Creek watershed  57  20 All MPB Logging Fire Logging and Fire MPB and Logging  18  Annual Disturbed Area (%)  16 14 12 10 8 6 4 2  1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008  0  Year  Figure 3.18 Annual disturbed area by different types of forest disturbances in the Moffat Creek watershed  3.2.5  Tulameen River watershed  The Tulameen River, about 148.5 km in length and with a drainage area of 1780 km2, flows into the Similkameen River near Princeton in the southern interior of BC, Canada (Figure 3.19), between the Coastal Mountains and Okanagan Valley. It originates in the North Cascades, part of the Cascade Mountains. The total length of the stream network is 1281.6 km and the stream network density is 0.72 km/km2. The northern part of this watershed lies within the Southern Thompson Plateau physiographic area, featured with a gently undulating upland with a low relief while the headwaters in the south consist of flat-floored valleys, rugged mountain ranges, and plateau areas with dry land vegetation and forest (Holland, 1976). The average slope of the watershed is 1.0%. Elevations for the watershed range from 629 m at the river mouth to 2302 m at the headwaters, with 60% of the total area above 1300 m (H60 line, Figure 3.20) and about 30% above 1500 m. The watershed is featured with glacially-shaped bedrocks, glacial till, and meltwater channels and deposits. Brunisolic soils are the dominant soil type, occurring on welldrained deposits with forest cover. Chernozemic soils can also be found in grassland areas with organic matter accumulation while Luvisolic soils are in areas developed in surface deposits with 58  some clay.  Figure 3.19 Location of the Tulameen River watershed  59  2500  elevation(m)  2000 1500 1000 500 0 0  10  20  30  40  50  60  70  80  90  100  % of total watershed area over  Figure 3.20 Hypsometric Curve in the Tulameen River watershed  The climate of the Tulameen River watershed is continental, characterized by dry, warm summers (a fairly long growing season) and cool winters, along with moderate annual precipitation due to the rainshadow created in the lee of topographic barriers (the Coast, Cascade, and Columbia mountains) to prevail easterly flowing air. Areas in the west of the watershed tend to be wetter than areas in the east. Mean annual precipitation is about 942 mm and about 61% is snow, mostly from higher mountainous areas. Annual mean temperature is 4.3 °C. The climate is coldest in January (monthly mean temperature varies from -15.8 to -0.3°C with an average of 5.8°C) and warmest in July (monthly mean temperature varies from 12.8 to18.4°C with an average of 15.4°C). Substantial growing season moisture deficits are common and frosts can occur at any time. The hydrological regime of this watershed is hybrid with peak flows in the late spring (May or June) from snowmelt or in autumn by rainfall. During the study period, annual mean flows fluctuate between 251mm and 740 mm with an average of 386 mm. Annual peak flows vary between 71.7 and 374 m3/s with an average of 197.3 m3/s and annual 7-day low flows vary from 0.8 to 4.8 m3/s with an average of 1.8 m3/s. The average runoff coefficient is 0.41. According to the BEC system, this watershed is primarily located within the Interior Douglas Fir (hereinafter referred to as “IDF”) and Engelmann Spruce Subalpine Fir (hereinafter referred to as “ESSF”) biogeoclimatic zones. MS zone can also be found. The lower elevation areas are dominated by IDF dry cold (IDFdk) and IDF very dry hot (IDFxh) zones while higher elevation  60  areas are featured with ESSF dry cold (ESSFdc), ESSF moist warm (ESSFmw), and MS dry mild (MSdm) zones with a small portion of ESSF very dry cold (ESSFxc), MS moist warm (MSmw), and MS very dry very cold (MSxk) zones. The dominant tree species in this watershed include lodgepole pine (Pinus contorta) and interior Douglas fir (Pseudotsuga menziesiiinterior). Ponderosa pine persists as a climax species on drier sites at lower elevations. Mixed stands of interior Douglas fir and lodgepole pine are extensive on drier sites at moderate elevations, and lodgepole pine commonly dominates the landscape in the driest regions due to crown fires. Engelmann spruce, hybrid white spruce (Picea engelmannii x glauca), and subalpine fir (Abies lasiocarpa) are the dominant climax tree species on the wetter sites at higher elevations. Trembling aspen (Populus tremuloides) is also a widely distributed seral species (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). Logging, MPB infestation, and fire are recognized as three major forest disturbance types in the Tulameen River watershed. Based on the VRI data, a forest stand in this watershed was disturbed by either one type of disturbance (logging, fire or MPB) or two types of disturbances (logging + fire or logging +MPB)(Figure 3.21). Logging activities, the leading forest disturbance type, dated back to 1957 but on a small scale before 1961. Since 1962, logging activities were on a steady increase and the watershed was progressively logged with an average clear-cut rate of 0.42% per year (7.5 km2). Particularly, about 1% of the total watershed area (18 km2) was logged in 1976, 1977, 1991, and 1993 (Figure 3.22). Up to 2009, cumulative clear-cut area came up to 457.5 km2 (25.7% of the total watershed area). MPB infestation is the second leading disturbance type. MPB infestation was limited before 2003, except in 1986 with a small area infested. Between 2003 and 2007, forests attacked by MPB were on an increase with 356 km2 forests infested (20% of the total watershed area) during that period. Fires occurred occasionally on a small scale in 1985, 1986, 2001, 2004, and 2006. Up to 2009, cumulative disturbed area was up to 817 km2 (45.9% of the total watershed area) in the Tulameen River watershed. The Tulameen River watershed has experienced substantial forest disturbances, particularly logging since 1957. The Tulameen River provides migration, spawning, incubation, and rearing habitats for fish species including rainbow trout (Oncorhynchus mykiss), longnose dace (Rhinichthys cataractae), and chinook salmon (Oncorhynchus tshawytscha) (U.S. DEBPA Division of Fish and Wildlife, 1984; BC Ministry of Environment, 2012). 61  Figure 3.21 Cumulative forest disturbances in the Tulameen River watershed  62  14 All MPB Logging Fire Logging and Fire MPB and Logging  Annual Disturbed Area (%)  12 10 8 6 4 2  1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008  0  Year  Figure 3.22 Annual disturbed area by different types of forest disturbances in the Tulameen River watershed  3.2.6  Ashnola River watershed  The Ashnola River, about 51.2 km in length and with a drainage area of 1050 km2 (the area in Canada is 872 km2) flows into the Similkameen River near Keremeos in the southern interior of BC, Canada-US border (Figure 3.23), between the Coastal Mountains and Okanagan Valley. It originates in the North Cascades, part of the Cascade Mountains. One third of the watershed (334.5 km2) lies within Cathedral Provincial Park. The total length of the stream network is 1045.9 km and the stream network density is 1.2 km/km2. This watershed lies within the Southern Thompson Plateau physiographic area, consisting of flat-floored valleys, rugged mountain ranges, and plateau areas with dry land vegetation and forest (Holland 1976). The valley sides are steep and the average slope of the watershed is 2.2%. Elevations for the watershed range from 478 m at the river mouth to 2627 m at the headwaters, with 60% of the total area above 1800 m (H60 line, Figure 3.24) and about 80% above 1500 m. The watershed is featured with glacially-shaped bedrocks, glacial till, and meltwater channels and deposits. Brunisolic soils are the dominant soil type, occurring on well-drained deposits with forest cover. Chernozemic soils can also be found in grassland areas with organic matter accumulation while 63  Luvisolic soils are in areas developed in surface deposits with some clay.  Figure 3.23 Location of the Ashnola River watershed (The map only shows the area of the Ashnola River watershed falls within Canada)  64  3000  elevation(m)  2500 2000 1500 1000 500 0 0  10  20  30  40  50  60  70  80  90  100  % of total watershed area over  Figure 3.24 Hypsometric curve in the Ashnola River watershed  The valley bottoms are dominated by a semi-arid steppe climate with extremely hot and dry summers and largely overcast and cold winters. Precipitation increases with elevation resulting in a humid continental climate at higher elevations with dense stands of conifer forests. Occasionally larger moist pacific weather systems spill over the Cascade Mountains, providing much of the regions precipitation in the form of short duration rain or snowstorms or periodic summer thunderstorms. Mean annual precipitation is about 727 mm and about 61% is snow mostly from higher mountainous areas. Annual mean temperature is 3.8°C. The climate is coldest in January (monthly mean temperature from -15.1 to -0.3°C with an average of -5.8°C) and warmest in July (monthly mean temperature from 11.5 to17.1°C with an average of 14.0°C). The hydrological regime of this watershed is hybrid with peak flows in the late spring (May or June) from snowmelt or in autumn by rainfall. During the study period, annual mean flows fluctuate between 110 mm and 427 mm with an average of 226 mm. Annual peak flows vary between 25.8 and 181.0 m3/s with an average of 73.5 m3/s, and annual 7-day low flows vary from 0.3 to 1.6 m3/s with an average of 0.9 m3/s. The average runoff coefficient is 0.31. Most of the study watershed falls within the IDF, ESSF, and Ponderosa Pine (PP) zones. The very hot dry PP zone dominates the lower valley bottom and the very hot dry IDF zone occupies areas at moderate elevations. Areas at higher elevation are mainly within the very dry cold and the dry cold ESSF zones. The very hot dry Bunchgrass zone can also be found in the lower 65  valley bottom in the south of the study area. The dominant tree species in this watershed include lodgepole pine (Pinus contorta), ponderosa pine (Pinus ponderosa), and interior Douglas fir (Pseudotsuga menziesii-interior). Ponderosa pine persists as a climax species on drier sites at lower elevations. Mixed stands of subalpine fir, Douglas fir, and lodgepole pine are extensive on drier sites at moderate elevations and lodgepole pine commonly dominates the landscape especially in the driest regions due to crown fires. Engelmann spruce, hybrid white spruce (Picea engelmannii x glauca), and subalpine fir (Abies lasiocarpa) are the (Picea engelmannii) dominant climax tree species on the wetter sites at higher elevations (BC Ministry of Forests, Lands and Natural Resource Operations, 2012). Logging, MPB infestation, and fire are recognized as three major forest disturbance types in the Ashnola River watershed. Based on the VRI data, a forest stand in this watershed was disturbed by either one type of disturbance (logging, fire, or MPB) or two types of disturbances (logging + fire or logging +MPB) (Figure 3.25). Logging activities were limited before 1980 and slightly on an increase but with low levels (Figure 3.26). Up to 2009, cumulative clear-cut area was only 50 km2 (4.8% of the total watershed area). MPB infestation was limited before 2003. Between 2003 and 2007, forests attacked by MPB were on an increase with 190.4 km2 infested (18.1% of the total watershed area) during that period. Fires occurred occasionally in 1975, 1982, 1985, and 1986. The largest fire in 1985 burned an area of about 3.2 km2. Up to 2009, cumulative disturbed area was 247 km2 (23.5% of the total watershed area). The Ashnola River watershed has experienced limited forest disturbances over the study period. The Ashnola River provides migration, spawning, incubation, and rearing habitats for fish species including rainbow trout (Oncorhynchus mykiss), sculpin (Cottus spp.) and chinook salmon (Oncorhynchus tshawytscha) (U.S. DEBPA Division of Fish and Wildlife, 1984; BC Ministry of Environment, 2012) However, snow pack and precipitation at higher elevations are major sources for streamflow in the arid valley bottom, and rivers tend to be relatively cold that results in slow growth rates of fish.  66  Figure 3.25 Cumulative forest disturbances in the Ashnola River watershed (The map only shows the area of the Ashnola River watershed falls within Canada)  The study area of the Ashnola River watershed only includes the area within Canada (83% of the total watershed area). Since the area of this watershed falling within the U.S. is mostly located in the Cathedral Provincial Park with limited forest disturbances, the use of area within Canada to represent disturbance conditions for the whole watershed is acceptable.  67  4 Chapter: Impacts of forest disturbances on mean flows 4.1 Background The impacts of forest disturbances on mean flows have been studied for a century, either by the traditional experimental paired watershed approach or hydrological modeling. A general conclusion drawn from numerous small watershed studies (<100 km2) is that forest disturbances such as logging, fire, and insect infestation can increase annual mean flows due to the reduction of interception and evapotranspiration after the removal or death of trees (Stednick, 1996; Neary et al., 2003; Bruijnzeel, 2004; Moore and Wondzell, 2005). However, the effects of forest disturbances on seasonal mean flows lack consistency, and can be either positive or negative in small watersheds. Both increases and decreases in dry season flows have been reported worldwide (Bruijnzeel, 2004; Calder, 2005; Moore and Wondzell, 2005). In contrast to a wealth of studies conducted at a small watershed scale (<100 km2), the impacts of forest disturbances on mean flows at a large watershed scale (>500 km2) have been less investigated. Most studies have merely examined one type of forest disturbance (Wei and Zhang, 2010a; Vose et al., 2011). Moreover, inconsistent results have been found even for the effects on annual mean flows (Ring and Fisher, 1985; Buttle and Metcalfe, 2000; Costa et al., 2003; Tuteja et al., 2007; Wei and Zhang, 2010b). The large watershed studies is mainly constrained by the lack of a suitable and efficient methodology, the lack of a comprehensive indicator for cumulative forest disturbances over space and time, and the availability of long-term data on hydrology, climate, and forest disturbances (See Chapter 1 and Chapter 2 for more details). However, scientific information on the impacts of forest disturbances on hydrology at large watersheds is critical since watershed planning and management are increasingly operated at this spatial scale.  Watersheds in the BC interior have been substantially disturbed by various types of forest disturbances (e.g., logging, wildfire, and mountain pine beetle (MPB) infestation). There are a number of watersheds that have been intensively disturbed by forest disturbances due to the large-scale outbreak of MPB in 2003 and subsequent salvage logging. A comprehensive understanding of the hydrological impacts of forest disturbances in these intensively disturbed 68  watersheds is in critical need to guide forest practices and water resources management and to protect watershed ecosystems. In this study, in order to investigate the impacts of forest disturbances on mean flows at large watersheds, six large watersheds (Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River) in the BC interior were selected along environmental gradients. The cumulative equivalent clear-cut area (CECA) was selected as an integrated indicator for forest disturbance level.  The overall research design was that the statistical assessment using the time series crosscorrelation analysis was conducted first to identify the statistical significance of cause-effect relationships between forest disturbances and mean flows. For those significantly impacted annual mean flows, the single watershed and quasi-paired watershed approaches were further used to separate the effects of forest disturbances with those of climatic variability on annual mean flows so that the impacts of forest disturbances were quantified. The single watershed approach that involves techniques including modified double mass curve, ARIMA intervention analysis, regression or autoregression were applied to quantify the impacts of forest disturbances on annual mean flows in a moderately disturbed watershed (the Willow River watershed) and two severely disturbed watersheds (the Baker Creek and Moffat Creek watersheds). The quasipaired watershed approach was applied to two pairs of watersheds: Willow-Cottonwood and Tulameen-Ashnola. The Cottonwood and Ashnola River watersheds are the “control” watersheds in their respective pairs. The Willow River watershed was analyzed by both the single watershed and quasi-paired watershed approaches in order to test the validity and robustness of the two approaches. The results from these four watersheds were further compared to explore how the forest disturbance-induced changes in annual mean flows varied along environmental gradients by use of statistical techniques such as Kendall tau correlation, boxplots, and Mann-Whitney U test. In addition, the results from the Willow and Tulameen River watersheds, two watersheds with lengthy disturbed periods, were analyzed further to explore the role of watershed resilience in the responses of annual mean flows to forest disturbances. Finally, the implications of these findings for watershed management and ecosystem protection were discussed.  69  4.2 Data 4.2.1  Hydrological data  Hydrological data including the daily flows for six large watersheds (the Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River watersheds) were obtained from the Water Survey Canada, which were used to calculate annual and seasonal mean flows.  The Willow River watershed, located in the northern-central interior of BC, Canada, has one active hydrometric station (Station ID: 08KD006), which was installed in 1976 as a relocation of the original one constructed in 1953 (Station ID: 08KD003). As a result of this relocation, the drainage area was reduced from 3110 to 2860 km2 (an 8% reduction). Hence, the hydrological records from 1953 to 1976 observed at the original site were adjusted slightly by multiplying by a factor of 0.92 (equal to the ratio of the relocated drainage area to the original drainage area) to reflect the change in drainage area (Lin and Wei, 2008). Given the low relief and relatively uniform topography of the Willow River watershed, the potential error associated with this adjustment would be minor (Wei and Zhang, 2010a). With the adjustment of flow records, the daily flow records in the Willow River watershed can cover the period between 1953 and 2008.  The Cottonwood River watershed, adjacent to the Willow River watershed, has one discontinued hydrometric station with a drainage area of 1910 km2. The daily flow records in the Cottonwood River watershed are available for the years between 1965 and 1995 when this watershed was less disturbed. The hydrological regimes of these two watersheds are typically snow-dominated. The average annual mean flows of the Willow and Cottonwood River watersheds are 424 mm and 395 mm, respectively. The annual peak flows driven by snowmelt process usually occur in late April or May in both watersheds. The Willow River watershed, with a larger drainage size, often has higher peak flows than the Cottonwood River watershed.  The Baker Creek watershed has one active hydrometric station at the outlet (Station ID: 08KE016), with hydrological records for the period between 1964 and 2009. The Moffat Creek watershed, near the Baker Creek watershed, also has one active hydrometric station at the outlet 70  (Station ID: 08KH019), with daily flow records available from 1967 to 2009. Located in the dry central interior of BC, these two snow-dominated watersheds are characterized by similar hydrological regimes. The average annual mean flows of the Baker and Moffat Creek watersheds are 105 mm and 135 mm, respectively. The annual peak flows driven by snowmelt process usually occur in late April or May in both watersheds. The Baker Creek watershed, with its larger drainage size, often has higher peak flows than the Moffat Creek watershed.  The Tulameen and Ashnola Rivers are the tributaries of the Similkameen River, located in the southern interior of BC. The Tulameen River watershed has one active hydrometric station at the outlet (Station ID: 08NL024), with hydrological records available from 1954 to 2009. The Ashnola River watershed has one active hydrometric station at the outlet (Station ID: 08NL004), with daily flow records available from 1947 to 2009. Located in the dry southern interior of BC, these two watersheds belong to hybrid hydrological regimes. The average annual mean flows of the Tulameen and Ashnola River watersheds are 386 mm and 226 mm, respectively. Annual peak flows can be driven by snowmelt in May or June or by rainfall in September in both watersheds. The Tulameen River watershed, with a larger drainage size and wetter climate, often has higher peak flows than the Ashnola River watershed.  According to the long-term hydrological records, for a water year (November to October) , the annual stream flow hydrograph in these watersheds was divided into four seasons: spring (April– June), summer (July–August), autumn (September to October), and winter (November–March). All seasonal as well as annual mean flows were calculated for water years in the study watersheds accordingly. 4.2.2  Climatic data  The climate data used in this study are from two sources: the historical data from climate stations and ClimateWNA dataset. The historical data, such as minimum, average, maximum temperature and precipitation on monthly and daily scales for the climate stations within or near the study watersheds were collected from Weather Canada. The availability of weather stations at higher elevations in a large watershed is essential for hydrological studies. However, climate stations at  71  higher elevations are always deficient, especially the ones with long-term records. Such a deficiency may be one of the greatest challenges in hydrological studies at a watershed scale.  In the study area, the only long-term climate station at a high elevation is the Barkerville station (Climate ID: 1090660, elevation: 1283 m), which has records since 1888. This climate station, located in the headwaters of the Willow River watershed can represent parts of the Willow and Cottonwood River watersheds.  There are two active climate stations in or near the Willow River watershed. The Prince George Airport climate station (Climate ID: 1096450, elevation: 691 m), with records dating back to 1947, is located near the outlet of the Willow watershed, while the Barkerville climate station (Climate ID: 1090660) is situated in the southern mountainous area of the watershed, a relatively more humid area. The average values of the climate data from the Barkerville climate station and the Prince George Airport climate station were used in the hydrological analysis of the Willow River watershed. There are also two climate stations near the Cottonwood River watershed. The Barkerville climate station is close to the eastern mountainous area of the Cottonwood River watershed, and the Quesnel Airport climate station, with records from 1947, is adjacent to the western plateau of this watershed. The data from the Barkerville climate station and the Quesnel Airport climate station were averaged for the hydrological analysis of the Cottonwood River watershed.  The other four watersheds lack long-term data from climate stations at high elevations. In the Baker Creek watershed, there is only one long-term climate station near its outlet, the Quesnel Airport climate station. Similarly, the Tulameen River watershed has only one long-term active climate station, the Princeton Airport climate station (Climate ID: 1126510, elevation: 701.7 m) near its outlet with records since 1953. The Ashnola River watershed has two long-term discontinued weather stations, the Keremeos (Climate ID: 1124110, elevation: 415 m, with records from 1891 to 1995) and Keremeos 2 (Climate ID: 1124112, elevation: 434.9 m, with records from 1924 to 2000) near its outlet. A long-term climate station is unavailable within or near the Moffat Creek watershed. For these four watersheds that lack long-term climate stations at higher elevations, climate data such as monthly mean, maximum and minimum temperature,  72  and precipitation were obtained from the ClimateWNA dataset. The ClimateWNA is a gridded climate dataset for the Western North America, which downscales and integrates monthly and annual historical climate data (1901-2009) (Mitchell and Jones, 2005; Mbogga et al., 2009). Given large spatial variations in climate and topography, the gridded monthly climate data of the ClimateWNA were derived with a resolution of 10 km*10 km and then aggregated to generate the monthly climate data series at a watershed scale. 4.2.3  Forest disturbance data  The Geographic Information System (GIS) based data on the histories of forest disturbances for the study watersheds were derived by use of ArcGIS 9.2 from two provincial databases: Cutblocks 2010 and Vegetation Resources Inventory (VRI) 2010, developed and maintained by BC Ministry of Forests, Lands and Natural Resources Operations. The Cutblocks 2010 database combines logging information up to 2010 from BC Ministry of Forests, Lands and Natural Resources Operations and the forestry companies. It contains complete records of cutblock sizes and logged-years but detailed vegetation information has not been included. The VRI 2010 database records information on various disturbances (i.e., fire, infestation, and logging) and provides a detailed vegetation description up to 2009. However, its records on logging are incomplete due to delayed submissions from the forestry companies. Thus, both datasets are complementary and were used in this study. Data from these two databases were combined and analyzed to generate complete long-term records on forest disturbance histories for the study watersheds by use of ArcGIS 9.2. 4.2.4  Geographic data  Geographic data regarding watershed boundaries, stream networks, lakes, rivers, provincial parks, and topography were used to generate base maps of the study watersheds by use of ArcGIS 9.2. The gridded digital elevation model dataset for BC was applied in the topographic analysis for the determination of H60 line in every study watershed. This dataset is created from the 1:20 000 scale Terrain Resource Information Management (TRIM) Digital Elevation Model (DEM) with a resolution of 25 meters. All the geographic data were collected from GeoBC.  73  4.3 Methods 4.3.1  Quantification of forest disturbance levels  Logging, fire, and MPB infestation are recognized as three major forest disturbance types in the study watersheds. Since many forest disturbances are cumulative over both space and time, cumulative equivalent clear-cut area (CECA) was used in this study as an integrated indicator that combines all types of forest disturbances spatially and temporally with a consideration of vegetation and hydrological recovery following disturbances. The CECA is the sum of annual equivalent clear-cut area (ECA). ECA is defined as the area that has been clear-cut, fire-killed or infested by MPB, with a reduction factor (ECA coefficient) to account for the hydrological recovery due to forest regeneration (BC Ministry of Forests and Rangeland, 1999). For example, an ECA coefficient of 100% means no hydrological recovery in a disturbed forest stand, while an ECA coefficient of 0% indicates a full hydrological recovery. However, the generation of ECA coefficients for each type disturbance can be challenging because the hydrological recovery of a forest stand is determined by various factors, mainly including disturbance type, climate, and tree species (Hudson, 2000; Talbot and Plamondon, 2002).  The relationship between vegetation growth (expressed by ages and tree heights) and hydrological recovery rate (ECA coefficients: 0-100%) was generally used to estimate ECA after logging for different tree species, mainly spruce, lodgepole pine, and interior Douglas fir forests in the watershed assessment (BC Ministry of Forests and Rangeland, 1999). The relationship, however, varies among watersheds due to the differences in climate and tree species.  The six study watersheds were classified into three groups according to climate conditions. The first group consists of the Willow and Cottonwood River watersheds, featured with moist cool summers and severely cold winters, and dominated by tree species including white spruce, subalpine fir, and lodgepole pine. The second group includes the Baker and Moffat Creek watersheds, characterized with very dry cool summers and severely cold winters, and predominated by lodgepole pine. The third group is composed of the Tulameen and Ashnola River watersheds, where summers are very dry and hot, and winters are cold but relatively mild  74  compared to the other two groups. With relatively warmer climate, tree species, such as interior Douglas fir, are common in these two watersheds.  The relationships between the age and height for different tree species were established according to their growth curves and dominant site indexes in each group. Then, the ECA coefficients time series for different tree species since logging or fire disturbances were estimated based on the Interior Watershed Assessment Procedure (IWAP) guidelines and professional judgment from local forest hydrologists (Figure 4.1) (IWAP, 2006). With similar growth rates, the interior Douglas fir and lodgepole pine generally have similar ECA coefficients, while spruce forests have a lower ECA coefficient due to their relatively lower growth rates (Thrower and Goudie, 1992; Thrower et al., 1994). For MPB infested stands, Lewis and Huggard (2010) have developed a model to quantify the effects of MPB infestation on ECA calculations based on the monitoring in different biogeoclimatic zones. Based on their studies and the input from local forest hydrologists, we also estimated the time series of ECA coefficients for the MPB infested forest stands in the SBPS, SBS, and MS biogeoclimatic zones.  The hydrological impacts of MPB infestation on forests are different from those of logging. Since dead trees remain in stands, the hydrological function of dead trees is incompletely damaged unlike that caused by the removal of trees by logging (Winkler et al., 2008). Moreover, the understory beneath the MPB attacked stands and other trees at overstorey without MPB infestation can also intercept precipitation and transpire water. Thus, the hydrological changes due to MPB infestation are much lower than those from the logging, especially within 1-2 years after MPB attacks. However, as dead trees lose their canopy over time, the hydrological effects of MPB attack increase, and then decrease with the regeneration of young trees. For example, the ECA coefficient for the forest stands in SBS/SBPS zone is only about 15% one year after MPB attacks, and then reaches the maximum of 75% 18-20 years later, and finally drops to 10% after 60 years (Lewis and Huggard, 2010).  Figure 4.1 provides the estimated time series of ECA coefficients for logging, fire, and MPB in different watersheds, which were used to estimate ECA data series for each forest stand based on their disturbed area (e.g., annual clear-cut area) derived from historic disturbance records. Forest  75  disturbances that occur in areas above the H60 line are believed to have more pronounced hydrological impacts. The disturbed areas above the H60 line are multiplied by a weighted factor 1.5 as suggested in the IWAP guidelines. The reason to introduce H60 line is that in most areas of the BC interior, streamflow is normally comprised of snowmelt water at lower elevations in early spring when snow typically covers the upper 60% of a watershed. As the temperature rises, snowmelt water from high elevations (above H60 line) contributes to high flows in the late spring. Forest disturbances at higher elevations are, therefore, considered to be more influential  ECA coefficient (%)  on hydrology, particularly on high flows, than disturbances at lower elevations.  120 110 100 90 80 70 60 50 40 30 20 10 0  a)  ECA coefficient (%)  0  120 110 100 90 80 70 60 50 40 30 20 10 0  Logging/Fire-S(spruce) Logging/Fire-PL(pine) MPB-MS MPB-SBS  5  10  15  20 25 30 35 40 45 Time since disturbances(yrs)  5  55  60  65  Logging/Fire-S(Spruce) Logging/Fire-PL(Pine) MPB-MS MPB-SBPS/SBS  b)  0  50  10  15  20  25  30  35  40  45  50  55  60  Time since disturbances(yrs)  76  ECA coefficient (%)  120 110 100 90 80 70 60 50 40 30 20 10 0  Logging/Fire-S(Spruce) Logging/Fire-PL(Pine) MPB-MS MPB-SBPS/SBS Logging/Fire-FD(Fir)  c)  0  5  10  15  20  25  30  35  40  45  50  55  60  65  Time since disturbances Figure 4.1 ECA coefficients of different forest disturbance types for a) the Willow and Cottonwood River watersheds; b) the Baker and Moffat Creek watersheds; c) the Tulameen and Ashnola River watersheds  A forest stand in the study watersheds may actually be disturbed by a single disturbance agent or mixed types of disturbances chronologically or simultaneously. In order to calculate the cumulative equivalent clear-cut area (CECA) at a watershed-scale, the disturbed forest stands in the study watersheds were classified into five groups according to the disturbance records from two datasets, Cutblocks 2010 and VRI 2010. The five groups are listed below:  a) Forest stands disturbed by logging; b) Forest stands disturbed by MPB; c) Forest stands disturbed by fire; d) Forest stands disturbed by both logging and fire; e) Forest stands disturbed by both logging and MPB.  Annual ECA data series for each group was calculated individually and then summed to estimate annual ECA data for all types of disturbances in each watershed. Finally, annual ECA data for the whole watershed were added to generate time series of CECA and forest disturbance levels in terms of CECA for a given watershed were quantified. A program named “ECA Calculator” was developed to calculate cumulative equivalent clear-cut area in this study (Appendix B). Figure 4.2 displays the calculated long-term CECA data series for the study watersheds. The most 77  severely disturbed watersheds are the Baker and Moffat Creek watersheds, with the CECA of 62.2% and 65.6% in 2009, respectively (Figure 4.2c and d). The average annual ECA of the Barker and Moffat Creek watersheds was 1.2% and 1.3%, respectively. Forest disturbances in these two watersheds were limited before 1990 and remarkably increased due to the large-scale MPB outbreak in 2003 and subsequent salvage logging. The CECA of the Baker Creek watershed was about 1% in 1975, and increased slowly before 2000, and then increased from 22.4% in 2002 to 62.2% in 2009. The Moffat Creek watershed followed a similar pattern. During the most intensively disturbed period, the CECA of the Moffat Creek watershed increased from 27.1% in 2002 to 65.7% in 2009.  The Willow River watershed, despite a relatively lower CECA (35.4%) in 2008 compared with the Baker and Moffat Creek watersheds, experienced long-term substantial forest disturbances (mainly logging) lasting from the early 1960s to the mid 1990s, and then declined to some extent, but increased again after 2002. The CECA of the Willow River watershed was less than 5% in 1969, with the average annual ECA of 0.4% during the period of 1960 to 1969. Then, an intensive and steady increase from 5.4% in 1970 to 33.1% in 1996 was observed, with the average annual ECA more than doubling (1.1%) (Figure 4.2a). After a short stable period (19972002), the CECA of the Willow River watershed increased again to reach a high level of 35.4% in 2008.  The Tulameen River watershed had a similar disturbance level with that of the Willow River watershed. The CECA of the Tulameen River watershed was less than 5% in 1975 and then was followed by a steady increase from 6% in 1976 to 20% in 2005 (Figure 4.2e). A sharp increase with an average annual ECA of 3.5% occurred from 2005 to 2009, resulting in 33.8% of CECA in 2009.  The Cottonwood River watershed, adjacent to the Willow River watershed, was less disturbed. The CECA of Cottonwood River watershed was about 12% in 1995, with the average annual ECA of less than 0.4% (Figure 4.2b). Forest disturbances in the Ashnola River watershed near the Tulameen River watershed were the least, with the CECA of 6.8% in 2009, and most of  78  which was attributed to disturbances since 2003 (Figure 4.2f). The average annual ECA was only 0.2% between 1975 and 2009 in the Ashnola River watershed.  45  a) 40 35  b) 12 10 CECA(%)  CECA(%)  30  14  All MPB Logging Logging and Fire MPB and Logging Fire  25 20 15  All MPB Logging Fire Logging and Fire  8 6 4  10  2  5  0 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005 2009  1953 1956 1959 1962 1965 1968 1971 1974 1977 1980 1983 1986 1989 1992 1995  0  Year  Year  70  40  All MPB Logging Fire Logging and Fire MPB and Logging  30  60 50 40  d)  All MPB Logging Fire Logging and Fire MPB and Logging  30 20  10  10  0  0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  20  1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  CECA(%)  50  c)  CECA(%)  60  70  Year  Year  79  40  8 e)  35  6 CECA(%)  25  7  20 15  5  f)  All MPB Logging Fire Logging and Fire MPB and Logging  4 3 2  5  1  0  0 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  10  1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005 2009  CECA(%)  30  All MPB Logging Fire Logging and Fire MPB and Logging  Year  Year  Figure 4.2 Cumulative equivalent clear-cut area (CECA) in a) the Willow River watershed from 1953 to 2008; b) the Cottonwood River watershed from 1953 to 1995; c) the Baker Creek watershed from 1960 to 2009; d) the Moffat Creek watershed from 1960 to 2009; e) the Tulameen River watershed from 1957 to 2009; and f) the Ashnola River watershed from 1960 to 2009  4.3.2  Time series cross-correlation analysis  The time series cross-correlation analysis was performed to determine the statistical significance of the cuase-effect relationships between forest disturbance levels (CECA) and mean flows (annual, spring, summer, fall and winter mean flows) for each watershed. The time series crosscorrelation analysis is found to be an effective technique to investigate the cause-effect relationships among environmental variables because it can address the autocorrelation issue in data series and can identify the lagged causality between the two data series (Chatefield, 1989; Jassby and Powell, 1990; Lin and Wei, 2008; Zhou et al., 2010). All the hydrological data series along with CECA data series were pre-whitened to remove autocorrelations by fitting Autoregressive Integrated Moving Average (ARIMA) models. The white noises or model residuals from the ARIMA models with best performance in terms of their achievements in model stationarity and coefficient of determination (R2) were selected for cross-correlations (Lin and Wei, 2008).  80  4.3.3  Quantification of annual mean flow responses  Watersheds where annual mean flows were significantly changed by forest disturbances were further analyzed to quantify their change magnitude of annual mean flows attributed to forest disturbances. Both the single and quasi-paired watershed approaches were used in this study. The single watershed approach tends to be more flexible and widely applicable due to fewer constraints in watershed selection, while the application of the paired watershed approach is restricted in many regions in light of its strict requirements (e.g., long-term data, comparable disturbance history, and similar watershed characteristics) for both control and disturbed watersheds. 4.3.3.1 Single watershed approach 4.3.3.1.1 Removal of the effect of climate variability on annual mean flow by the modified double mass curve (MDMC) For a large forested watershed, climatic variability and forest disturbances are two primary drivers for hydrological variations. Climatic variability is typically more dominant, and it can always obscure the effects of other changes on hydrology (Cong et al., 2009). The challenge for us, therefore, is how to exclude the climatic effects on annual mean flow variation so that the impacts of forest disturbances on annual mean flows can be quantified.  In order to separate the effects of climate variability and forest disturbances on annual mean flows, the “modified double mass curve” developed by Wei and Zhang (2010b) was used to eliminate the influence of climatic variability on annual mean flows. According to watershed water balance, streamflow is determined by the difference between precipitation, evapotranspiration (evaporation and transpiration), and change in soil water storage. Over an annual scale, because change in soil water storage can be generally assumed to be a constant and minor term in the water balance equation and the transpiration of vegetation can also be assumed constant for an undisturbed watershed (Zhang et al., 2001). Therefore, streamflow variations are mainly affected by precipitation and evaporation under undisturbed conditions. Here, we defined the difference between precipitation and evaporation as effective precipitation (Pe), an integrated climatic indicator for streamflow generation (Wei and Zhang, 2010b). The evaporation (E) was  81  estimated by the Budyko equation using precipitation and potential evaporation (E0) (Budyko, 1974). Given the limited long-term data in these large watersheds, the Hargreaves equation (Hargreaves and Samani, 1985), a temperature-based potential evaporation estimation method, was applied to compute potential evaporation. The Hargreaves equation has been recognized as a good method to estimate potential evaporation for data-limited watersheds (Shuttleworth, 1993; Sankarasubramanian et al., 2001). It requires only data on mean, minimum and maximum air temperature, and extraterrestrial radiation, which are all available in the study watersheds.  Table 4.1 Equations for evaporation estimation Type  Expression E0,n=0.0023*Ra*[(Tmax,n+Tmin,n) /2+17.8)]*(Tmax,n-  Hargreaves equation (Hargreaves and Samani,1985) Budyko equation  Tmin,n)0.5 E0 = E0,1+ E0,2 +E0,3 +…+E0,n (n=1,2,3…,12) E={P[1-exp(-E0/P)]* E0*tanh(P/ E0)}0.5  (Budyko and Miller,1974) Note: Ra, extraterrestrial radiation; Tmax,n, monthly mean maximum temperature in ºC; Tmin,n, monthly mean minimum temperature in ºC; E0,n, monthly potential evaporation; P, annual precipitation; E: annual evaporation; E0, annual potential evaporation.  Zheng et al. (2009) indicated that annual mean flow variations were linearly associated with climate variability. Similarly, we also assumed a linear relation between annual mean flow variation and effective precipitation, but a different approach was developed to remove climatic effects. The double mass curve (DMC), frequently used in paired watershed studies to detect hydrological changes caused by forest disturbances, was modified in this study (Beschta, 1978; Ziemer, 1981; Trimble et al., 1987; Buttle and Metcalfe, 2000; Huang et al., 2003; Troendle, 2007). Unlike the traditional DMC where the accumulated annual mean flows from the disturbed watershed are plotted against the accumulated annual mean flows from the undisturbed one, modified double mass curve (MDMC) plots accumulated annual mean flows versus accumulated annual effective precipitation for each study watershed (Figure 4.3). In this way, climatic effects  82  on annual mean flows can be eliminated. In the period without or with minor forest disturbances (namely the reference period), a straight line is expected. This serves as a baseline that describes the linear relation between annual mean flows and annual effective precipitation, and a break in this curve indicates the change in annual mean flows caused by the factors other than climatic variability, for example, forest disturbances or land use change. In other words, a step change or regime shift occurs in the slope of the modified double mass curve, and the slope before the breakpoint is different from that afterwards.  ARIMA Intervention analysis was applied to detect the change points with statistical significance in the slope of MDMC (Jassby and Powell, 1990; Wei, 1990; Francis and Hare, 1994). Unlike most tests for change detection, this analysis can identify multiple step interventions with statistical significance while accounting for the autocorrelation in the data (Rodionov, 2005). The intervention type was set as abrupt and permanent. Once the significant breakpoints in the slope of MDMC are identified, the study period can be divided into the reference and disturbed periods by the first significant breakpoints. The disturbed period can be further divided into several phases by the rest of the identified significant breakpoints.  Finally, a regression model based on the relationship between the accumulated annual mean flows and accumulated annual effective precipitation during the reference period was used to predict the accumulated annual mean flows without forest disturbances during the disturbed period. The differences between the observed accumulated annual mean flows and predicted accumulated annual mean flows without forest disturbances after the first breakpoint can be viewed as the cumulative effects of forest disturbances on annual mean flows (referred to as ∆Qaf) as compared with undisturbed conditions (Figure 4.3). In addition to the traditional linear regression model, the AUTOREG procedure in SAS 9.2 was also applied in some of the study watersheds since it provides regression analysis of linear models with autocorrelated or conditional heteroscedastic errors. In the development of regression models, log-transformed values of variables were used instead of their original values. This allows for meeting the assumptions of regression models, for example constant variance and normal distribution of errors, as well as better model performance. Only modified double mass curves plotted using original values were presented in the main text to illustrate the change points and disturbance  83  effects with a better expression of time. Similar graphs plotted using log-transformed values were presented in Appendix A.  ∆Qaf(t) = Qa(t)-Qa0 (t)  (4.1)  Where, Qa(t) and Qa0(t) are the observed accumulated annual mean flow and predicted accumulated annual mean flow without forest disturbances at the tth year, respectively; ∆Qaf (t) stands for the accumulated annual mean flow variation attributed to forest disturbances at the tth year, respectively.  25000 20000  Observed line  15000 Qa  Predicted line  change point  10000 5000 0 0  5000  10000 15000 Pae Pae: accumulated annual effective precipitation (mm) Qa: accumulated annual mean flow (mm) Figure 4.3 A hypothesized example of MDMC for single watershed study  4.3.3.1.2 Calculation of the relative contributions of forest disturbances and climatic variability to streamflow variations We assumed that during the disturbed period, the streamflow variations from the annual mean flows of the reference period in the study watersheds are caused by forest disturbances and climatic variability. During the disturbed period, the annual mean flow variation attributed to forest disturbances and climatic variability can be calculated as follows:  ∆Q(t)= Q(t)- Qr  (4.2)  84  ∆Qa(t)=∆Q(1)+∆Q(2)+······+∆Q(t-1)+∆Q(t)  (4.3)  ∆Qf(t)=∆Qaf(t)-∆Qaf(t-1)  (4.4)  ∆Qc(t)=∆Qa(t)-∆Qf(t)  (4.5)  ∆Qac(t)=∆Qc(1)+∆Qc(2)+······+∆Qc(t-1)+∆Qc(t)  (4.6)  Where, Q(t), ∆Q(t), ∆Qa(t), ∆Qf(t), and ∆Qc(t) are annual mean flow, annual mean flow variation from Qr, accumulated ∆Q(t), annual mean flow variation due to forest disturbances, and annual mean flow variation due to climate variability in the tth year, respectively; Qr is average annual mean flow during the reference period; ∆Qaf(t) and ∆Qaf(t-1) are accumulated ∆Qf(t) for the tth year and (t-1)th year, respectively; and∆Qac(t) is accumulated ∆Qc(t) in the tth year Then, the relative contributions of forest disturbances and climate variability, reflecting the relative impact strength of forest disturbances and climate variability on annual mean flow, can be estimated by the equations below:  Rf(t)=100*|∆Qf(t)|/(|∆Qf(t)|+|∆Qc(t)|)  (4.7)  Rc(t)=100*|∆Qc(t)|/(|∆Qf(t)|+|∆Qc(t)|)  (4.8)  Where Rf(t) and Rc(t) are the relative contributions of forest disturbances and climate variability on annual mean flow variation for the tth year, respectively.  We also calculated the ratios of annual mean flow variation and its components to long-term annual mean flow, indicating the relative change of annual mean flow as well as the relative change of annual mean flow attributed to forest disturbances and climate variability compared with the long-term annual mean flow.  ∆Q%(t)= 100*∆Q(t)/Q  (4.9)  ∆Qf%(t)= 100*∆Qf(t)/Q  (4.10)  ∆0c%(t)= 100*∆Qc(t)/Q  (4.11)  Where, ∆Q%(t), ∆Qf%(t), and ∆Qc%(t) are relative annual mean flow variation from Qr, relative annual mean flow variation attributed to forest disturbances, relative annual mean flow variation attributed to climatic variability for the tth year, respectively; Q is the long-term annual mean flow over the whole study period.  85  The single watershed approach has been applied not only in the disturbed watersheds (the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds), but also in the controlled watersheds (the Cottonwood River and Ashnola River watersheds). This allows for validating the single watershed approach and also supports the application of the Cottonwood River and Ashnola River watersheds as the controls in the quasi-paired watershed approach.  4.3.3.2 Quasi-paired watershed study approach The application of the quasi-paired watershed study can be less feasible given the great difficulty to find a suitable pair of study watersheds. To be eligible for a quasi-paired watershed study, the paired watersheds must meet the requirements below:  1) be geographically close to ensure their climatic patterns are similar and comparable; 2) the control watershed is undisturbed or less disturbed while the disturbed watershed must be severely disturbed; 3) have similar hydrological regime, forest types, soil types, topography, and land-use patterns; 4) have long-term hydrological and forest disturbance records (at least 30 years) to meet the requirements of the selected statistical methods.  In the study area, two pairs of watersheds were identified. The first pair includes the Cottonwood (the “controlled watershed”) and Willow (the “disturbed watershed”) River watersheds, two neighbouring watersheds in the central interior of BC, Canada. Since the hydrological records of the Cottonwood River only covered the period from 1965 to 1995, the study period for Cottonwood-Willow pair was set to be between 1965 and 1995. Until 1995, the CECA of the Willow River watershed was 32.5%, nearly three times that of the Cottonwood River watershed (12%). The Willow River watershed was analyzed by both single and quasi-paired watershed approaches, which enabled us to compare their validity and robustness between these two approaches. The other pair of watersheds includes the  86  Ashnola (the “controlled watershed”) and Tulameen (the “disturbed watershed”) River watersheds, within the Similkameen River watershed in the southern interior of BC, Canada. Forest disturbances in the Ashnola River watershed were minor, and remained undisturbed for a long period, with only 6.8% of CECA in 2009. The CECA of the Tulameen River watershed was 33.8%, five times that of the Ashnola River watershed.  In the experimental paired watershed study, the effects of climate variability on streamflow are assumedly excluded by pairing, which is acceptable in small watershed studies. In large watersheds, a simple pairing, though with the ability to remove the effects of regional-scale climate oscillation, are unlikely to exclude the effect of climate variability completely in view of the great spatial variation in climate, especially in precipitation. The integrated climatic indicator “effective precipitation” was also used, but in a different way in the quasipaired watershed study. A new hydrological indicator namely “annual effective runoff coefficient,” referring to the ratio of annual mean flow to annual effective precipitation was introduced here (Equation 4.13). Since the integrated climatic indicator “effective precipitation” was used as a denominator, this hydrological indicator is capable of indicating the actual efficiency of a watershed to convert effective precipitation into streamflow, which is only sensitive to non-climatic factors such as forest changes, rather than to climate variability. Unlike the traditional DMC, the accumulated annual effective runoff coefficient of the disturbed watershed was plotted against the accumulated annual effective runoff coefficient of the controlled watershed. If both watersheds were undisturbed or less disturbed, a straight line should be observed. However, with the controlled watershed less disturbed and the disturbed one severely disturbed, a bias from the original line could occur (Figure 4.4). Similarly, the ARIMA Intervention analysis was also adopted to determine the significant change points in the slope of a modified double mass curve. With significant change points identified, the study period was divided into the reference period and disturbed period. The differences between the observed accumulated annual effective runoff coefficient of the disturbed watershed and its predicted values by the original linear regression during the disturbed period were regarded as forest disturbance effect on annual effective runoff coefficient. Similar to the single watershed approach, log-transformed values of variables were used for the development of regression models. However, the modified double mass  87  curves shown in the text were plotted using original data for better illustration of the change points and disturbance effects. The modified double mass curves plotted using logtransformed values can be found in the appendices.  C0(t)=Q(t)/Pe (t)  (4.12)  ∆Caf(t) = Ca(t)-Ca0 (t)  (4.13)  ∆Caf(t) = (C(1)+ C(2) +…+ C(t))- (C0(1)+ C0 (2) +…+ C0 (t)) (4.14) = (C(1)- C0(1))+ (C(2) -C0(2))+…+ (C(t) -C0(t))  (4.15)  Where C(t), Q(t), and Pe(t) are observed annual effective runoff coefficient, annual mean flow, annual effective precipitation at tth year in the disturbed watershed , respectively; ∆Caf(t) and Ca(t) stands for the accumulated variation of annual effective runoff coefficient attributed to forest disturbances and observed accumulated annual effective runoff coefficient at tth year in the disturbed watershed, respectively; Ca0 (t) is predicted accumulated annual effective runoff coefficient without forest disturbances for the tth year in the disturbed watershed (derived by the linear regression model in the reference period); and C0(t) are predicted annual effective runoff coefficient without disturbances at tth year in the disturbed watershed.  Since annual effective runoff coefficient is a hydrological indicator insensitive to climate variability, the differences between C(t) and C0(t) are considered as the effects of forest disturbances on annual effective runoff coefficient in the disturbed watershed. Then the annual mean flow variation attributed to forest disturbances was calculated accordingly by the following equations:  ∆Caf (t)= ∆Cf (1)+ ∆Cf (2) +…+ ∆Cf (t)  (4.16)  ∆Cf (t)= ∆Qf (t)/Pe(t)  (4.17)  ∆Qf (t)= ∆Cf (t) * Pe(t)  (4.18)  ∆Qaf (t)= ∆Qf (1) + ∆Qf (2) +…+ ∆Qf (t)  (4.19)  Where ∆Cf (t), ∆Qf (t), Pe(t) and ∆Qaf (t) are the variation of annual effective runoff coefficient attributed to forest disturbances, annual mean flow variation attributed to forest  88  disturbances, annual effective precipitation, and accumulated annual mean flow variation attributed to forest disturbances for the tth year in the disturbed watershed, respectively.  The calculations of ∆Cc(t), ∆Qf %(t), ∆Qc %(t), Rf (t), and Rc(t) followed the same procedures in the single watershed study.  16 14  Observed line  12  Cad  10  Change point  8 6  Predicted line  4 2 0 0  2  4  6  8  10  12  14  Car Car: accumulated annual effective runoff coefficient in the controlled watershed Cad: accumulated annual effective runoff coefficient in the disturbed watershed Figure 4.4 A hypothesized example of MDMC for quasi-paired watershed study  4.3.3.3 Integrated analysis The change magnitude of annual mean flows caused by forest disturbances varied from watershed to watershed, as well as from year to year. Differences in climate, such as precipitation and temperature among watersheds or years are believed to be one of the key factors that control the responses of annual mean flows to forest disturbances in small watershed studies (Stednick, 1996). However, this belief lacks evidence from large watershed studies. In order to address this issue, specifically, we synthesized the results from this study watersheds to explore the roles of precipitation (i.e., annual precipitation) and temperature (i.e., annual mean temperature) in the responses of annual mean flows to forest disturbances. To avoid the effect of differences in forest disturbance levels, the normalized annual mean flow variation attributed to forest disturbances, defined as the ratio of annual mean flow variation attributed to forest disturbances (∆Qf) to cumulative equivalent clear-cut area (CECA), was generated. This variable  89  can allow better comparisons among watersheds with different climate conditions. Both linear regression and Kendall tau correlation were applied to investigate the correlations between the normalized annual mean flow variation to forest disturbances and annual precipitation/annual mean temperature. Boxplots that grouped the normalized annual mean flow variations attributed to forest disturbances in different watersheds in the order of their annual mean temperature or annual precipitation were used to show if there are any patterns of the hydrological responses to forest disturbances in watersheds along climatic gradients. Mann-Whitney U test was performed to confirm whether the differences in annual mean flow responses among these watersheds were statistically significant.  90  4.4 Results 4.4.1  Statistical test of cause-effect relationships between forest disturbances and annual and seasonal mean flows  As suggested by the time series cross-correlation analysis in Table 4.2, forest disturbances had significant positive impact on mean flows in the watersheds with high CECA (>30%), such as in the Baker Creek, Moffat Creek, Willow River, and Tulameen River watersheds, while in the watersheds with low CECA (<15%), we hardly detected any significant cause-effect relationship between forest disturbances and mean flows. The impacts of forest disturbances on annual mean flows were consistent in the Baker Creek, Moffat Creek, Willow River, and Tulameen River watersheds, where annual mean flows were significantly and positively correlated with CECA. The effects of forest disturbances on seasonal mean flows, however, varied among the watersheds. In the Baker Creek and Tulameen River watersheds, both fall and winter mean flows showed a significant positive relationship with forest disturbances, while in the Moffat Creek watershed, a significant positive correlation between forest disturbances on both summer and winter mean flows were detected. In the Willow River watershed, however, only spring mean flows were significantly increased by forest disturbances while the effects on other seasonal mean flows lacked statistical significance.  Table 4.2 Time series cross-correlation between CECA and mean flows (annual and seasonal scales) Watersheds  Annual Q  Spring Q  Summer Q  Fall Q  Winter Q  Baker  0.34*(lag=1)  0.22(lag=1)  0.17(lag=0)  0.47*(lag=1)  0.38*(lag=1)  Moffat  0.35*(lag=6)  0.27(lag=5)  0.40*(lag=6)  -0.22(lag=4)  0.32*(lag=6)  Cottonwood  -0.25(lag=1)  -0.12(lag=1)  -0.07(lag=1)  -0.04(lag=1)  0.23(lag=1)  Willow  0.32*(lag=2)  0.32*(lag=2)  0.22(lag=2)  0.13(lag=2)  0.09(lag=2)  Tulameen  0.28*(lag=2)  -0.02(lag=2)  0.07(lag=2)  0.31*(lag=2)  0.36*(lag=2)  Ashnola  -0.13(lag=1)  -0.15(lag=1)  -0.11(lag=1)  0.07(lag=1)  0.04(lag=1)  Note: *, significant correlation at α=0.05  91  4.4.2  Quantification of the impact of forest disturbances on annual mean flow  As suggersted by the time series cross-correlation analysis, forest disturbances yielded significant impacts on annual mean flows in the Baker Creek, Moffat Creek, Willow River, and Tulameen River watersheds (the disturbed watersheds), while insignificant forest disturbanceinduced impacts on annual mean flows were identified in the Cottonwood River and Ashnola River watersheds (the controlled watersheds). Moreover, the modified double mass curves of the Cottonwood River and Ashnola River watersheds using a single watershed approach also showed insignificant changes in annual mean flows caused by forest disturbances. This also supports the use of those two watersheds as the controlled watersheds. Therefore, in the final analysis, only the Baker Creek, Moffat Creek, Willow River, and Tulameen River watersheds were further analyzed to quantify the change magnitude of annual mean flows caused by forest disturbances.  4.4.2.1 Willow River watershed 4.4.2.1.1 Single watershed approach Figure 4.5 displays the modified double mass curve for the Willow River watershed using original values of variables, where accumulated annual mean flows are plotted against accumulated annual effective precipitation. According to the ARIMA Intervention model of the slope in Figure 4.5, three significant breakpoints (1968, 1985, and 1991) were detected at α=0.05 (Table 4.3). Thus, we defined the reference period as between 1953 and 1967, while the disturbed period was from 1968 to 2008. The disturbed period was further divided into three phases: phase 1 from 1968 to 1984, phase 2 from 1985 to1990, and phase 3 from 1991 to 2008. As shown in Figure 4.5, a straight line (linear relationship) was observed between accumulated annual mean flows and accumulated annual effective precipitation in the period from1953 to 1967. After 1968, the observed line started to deviate from the original line (predicted line), suggesting that more annual mean flows were generated than predicted. By use of the logtransformed data from the reference period, a linear regression model with a significance level of 0.05 was established between accumulated annual mean flows and accumulated annual effective precipitation (Table 4.4), which was then used to predict accumulated annual mean flows without disturbances in the disturbed period. The differences between observed accumulated  92  annual mean flows and predicted values from 1968 to 2008 were referred to as accumulated annual mean flow variations attributed to forest disturbances. The accumulated annual mean flow variations attributed to climate variability was calculated accordingly.  24000 21000  1991  18000 Qa (mm)  1985 15000 12000  1968  9000 6000 3000 0 0  3000  6000  9000  12000 Pae (mm)  15000  18000  21000  24000  Figure 4.5 The modified double mass curve for the Willow River watershed (Qa: accumulated annual mean flow; Pae: accumulated annual effective precipitation) Table 4.3 The ARIMA intervention model of the slope in the MDMC for the Willow River watershed  AR part  p(1)  p(2)  -0.88  -0.37  (p<0.001) (p=0.01)  Int  MA  part part d(1)  q(1)  1  0  Intervention Part CP1  CP2  CP3  (1968)  (1985)  (1991)  ω(1)  ω(2)  ω(3)  0.24*  0.43*  -0.28*  (p=0.05) (p=0.001) (p=0.02)  Change Type  AP  Model Structure  Ln(x) (2,1,0)  MS  0.03  Note: *, significant change point at α=0.05; p(1), d(1), and q(1) are parameters for autoregression, differencing, and moving average; ω(1), ω(2), and ω(3) are parameters for intervention; AR part, Int part, and MA part refer to autoregressive part, integrated part, and moving average part, respectively; CP1, CP2, CP3, AP, and MS refer to the first change point, the second change point, the third change point, abrupt permanent change, and model residual, respectively; The slope is the ratio of annual mean flow to annual effective precipitation.  93  Table 4.4 The regression model of the MDMC for the Willow River watershed PE  Parameter estimates  method  β0(constant)  OLS  0  β1(ln(Pae)) 0.988 (p<0.001)  DW  AC(1)  statistics  coefficient  1.50  0.10  R2  RMSE  0.999  0.029  Note: PE method is parameter estimation method; OLS is ordinary least square; RMSE is rooted mean squared error; Qa(t) is accumulated annual mean flow; Pae(t) is accumulated annual effective precipitation; γ(t) is regression model error; DW is Durbin-Watson statistics; AC(1) coefficient is lag 1 autocorrelation coefficient of model errors; Regression model is ln(Qa(t))=0.988 ln(Pae(t))+γ(t) (See Figure A-1 for its regression plot)  Figure 4.6 shows the accumulated annual mean flow variations and their components during the disturbed period. Up to 2008, the accumulated annual mean flow variation attributed to forest disturbances was 2503.6 mm, while its variation attributed to climate variability was -3824.6 mm. This cumulatively resulted in a net reduction of 1321.0 mm in annual mean flow by the end of the study period as compared to the reference period. The annual mean flow variations attributed to forest disturbances were then computed. As shown in Figure 4.7, the annual mean flow variations attributed to forest disturbances ranged from -53.2 mm (12.5% of the average annual mean flow from 1953-2008) to 181.2 mm (42.6% of the average annual mean flow from 19532008), with an average of 61.1 mm (14.4% of the average annual mean flow from 1953-2008). The annual mean flow variations attributed to climate variability varied from -235.5 mm to 128.2 mm with an average of -93.3 mm (Figure 4.8). Meanwhile, during the disturbed period, CECA experienced a substantial increase from 4.5% in 1968 to 35.4% in 2008.  The estimated annual mean flow variations attributed to forest disturbances were normalized by CECA (∆Qf /CECA) for an easier interpretation of the results as well as a better comparison from year to year and from watershed to watershed. As suggested in Figure 4.9, the normalized annual mean flow variations attributed to forest disturbances varied from -2.1 mm/% to 36.7 mm/%, with an average of 4.5 mm/%. That means, during the disturbed period, per 10% CECA caused a -21 mm (0.5% of the average annual mean flow from 1953-2008) to 367 mm (86.4% of the average annual mean flow from 1953-2008) change in annual mean flows and on average,  94  per 10% CECA led to a 45 mm (10.6% of the average annual mean flow from 1953-2008) change in annual mean flows.  40 a)  ∆Qaf(mm)  ∆Qa ∆Qaf 95CI CECA  35 30 25 20 15  CECA(%) ∆Qac(mm)  3500 3000 2500 2000 1500 1000 500 0 -500 -1000 -1500 -2000  10 5  b)  ∆Qa ∆Qac 95CI  1968 1973 1978 1983 1988 1993 1998 2003 2008  1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  0  500 0 -500 -1000 -1500 -2000 -2500 -3000 -3500 -4000 -4500  Year  Year  Figure 4.6 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf); and b) the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Willow River watershed estimated by single watershed approach  50  35  40  150  30  30  30  100  25  20  25  50  20  10  20  0  15  0  15  -100 -150  ∆Qf 95CI CECA  Year  10  -10  5  -20  0  -30  40 b)  ∆Qf% 95CI CECA  CECA%  35  10 5 0  1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  -50  1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  ∆Qf(mm)  200  a)  CECA% ∆Qf%(%)  40  250  Year  Figure 4.7 a) The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances (∆Qf%) in the Willow River watershed  95  ∆Qc 95CI  1968 1971 1974 1977 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007  ∆Qc(mm)  200 150 100 50 0 -50 -100 -150 -200 -250 -300  Year  40  b)  35  30  10  Year  25 ∆Qf%/CECA 95CI CECA  4 2  20 15  CECA%  15  ∆Qf%/CECA  6  25 20  35  8  30 ∆Qf/CECA 95CI CECA  40  10  5  0  0  -2  5 0 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  a)  10  CECA%  45 40 35 30 25 20 15 10 5 0 -5 -10  1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008  ∆Qf/CECA(mm/%)  Figure 4.8 The annual mean flow variation attributed to climatic variability (∆Qc) in the Willow River watershed  Year  Figure 4.9 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); b) The normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA) in the Willow River watershed  Table 4.5 summarizes forest disturbance effects on annual mean flows in different phases, which helps us explore the temporal dynamics of hydrological impacts of forest disturbances. In phase  96  1, with average CECA of 13.5%, the average annual mean flow variation attributed to forest disturbances was 60.4 mm/yr (equivalent to 14.2% of the average annual mean flow from 19532008), which rose to 101.3 mm/yr (equivalent to 23.8% of the average annual mean flow from 1953-2008) in phase 2, the most intensively disturbed phase. It then slightly decreased to 48.3 mm (equivalent to 11.4% of the average annual mean flow from 1953-2008) in phase 3. The normalized annual mean flow variation attributed to forest disturbances, however, displayed a contrary pattern. In phase 1, per 10% CECA resulted in a 142 mm increment in annual mean flow, while in phase 2, this value declined to 32 mm and was even lower in phase 3 (24 mm).  Unlike forest disturbances, climate variability impacted annual mean flows in a negative way. The average annual mean flow variation attributed to climate variability was -73.0, -162.2, and 89.4 mm in phase 1, phase 2, and phase 3, respectively (Table 4.6).  Table 4.5 Forest disturbance effects on annual mean flows in the Willow River watershed in different periods (1968-2008) ∆Q  ∆Qf  ∆Qf/CECA  ∆Q%  ∆Qf%  (mm)  (mm)  (mm%)  (%)  (%)  1:1968-1984  -12.7  60.4±28.2  14.2±6.6  -3.0  7.9±5.1  1.9±1.2  13.5  2:1985-1990  -60.9  101.3±6.9  3.2±0.3  -14.3  3.8±1.8  0.9±0.4  26.8  3:1991-2008  -41.1  48.3±5.3  2.4±0.2  -9.7  1.5±1.0  0.3±0.2  32.5  Average  -32.2  61.1±20.0  4.5±2.3  -7.6  14.4±4.7  1.1±0.6  23.8  Phase  ∆Qf%/CECA  CECA (%)  Note: CECA is cumulative equivalent clear-cut area; ∆Q and ∆Q% are annual mean flow variation and relative annual mean flow variation (∆Q%= ∆Q/Q, and Q is average annual mean flow from 1953 to 2008 (425mm)), respectively; ∆Qf and ∆Qf % are annual mean flow variation attributed to forest disturbances and relative annual mean flow variation attributed to forest disturbances(∆Qf%= ∆Qf/Q), respectively; ∆Qf/CECA and ∆Qf%/CECA are normalized annual mean flow variation attributed to forest disturbances and normalized relative annual mean flow variation attributed to forest disturbances by CECA, respectively.  97  Table 4.6 Climate variability effects on annual mean flows in the Willow River watershed in different phases (1968-2008) Phase  ∆Q (mm)  ∆Qc(mm)  ∆Qc%  1: 1968-1984  -12.7  -73.0±44.7  -17.2±10.5  2: 1985-1990  -60.9  -162.2±43.0  -38.2±10.1  3: 1991-2008  -41.1  -89.4±30.1  -21.0±7.1  Average  -32.2  -93.3±24.8  -21.9±5.8  Note: ∆Q is annual mean flow variation; ∆Qc and ∆Qc% are annual mean flow variation attributed to climate variability and relative annual mean flow variation attributed to climate variability (∆Qc%= ∆Qc/Q, and Q is average annual mean flow from 1953 to 2008 (425mm))  Table 4.7 The relative contributions of forest disturbances and climate variability on annual mean flow variations in the Willow River watershed Phase  ∆Q (mm)  Rf (%)  Rc (%)  CECA (%)  1: 1968-1984  -12.7  42.9±12.2  57.1±12.2  13.5  2: 1985-1990  -60.9  38.0±16.3  62±16.3  26.8  3: 1991-2008  -41.1  37.1±12.7  62.9±12.7  32.5  Average  -32.2  39.7±7.8  60.3±7.8  23.8  Note: CECA is cumulative equivalent clear cut area; ∆Q is annual mean flow variation; Rf and Rc are relative contributions of forest disturbances and climate variability, respectively (Rf=100*|∆Qf|/(|∆Qf|+|∆Qc|); Rc=100*|∆Qc|/(|∆Qf|+|∆Qc|)).  Table 4.7 demonstrates the relative contributions of forest disturbances and climatic variability on annual mean flow variation. Forest disturbances and climate variability produced opposite effects on annual mean flows, and the strength of climate variability related impact on annual mean flow was greater than that of forest disturbances. The relative contribution of forest disturbances on annual mean flows was, on average, 39.7% while the relative contribution of climate variability was 60.3%. In phase 1, 42.9% of the variation in annual mean flows was explained by forest disturbances, while 57.1% of that was accounted by climate variability. However, in phase 2, the impact of forest disturbances tended to be less prounouced, and the relative contribution of forest disturbances on annual mean flow variations (Rf) declined to 98  38.0%, compared with 62.0% of the variation explained by climate variability. In phase 3, the relative contribution of forest disturbances further declined to 37.1%, while that of climate variability was 62.9%.  4.4.2.1.2 Quasi-paired watershed approach Figure 4.10 presents the modified double mass curve for the Cottonwood-Willow quasi-paired watershed study using original values of variables, where the accumulated annual effective runoff coefficient of the Willow River watershed is plotted against the accumulated annual effective runoff coefficient of the Cottonwood River watershed. According to the ARIMA Intervention model of the slope in Figure 4.10, one significant breakpoint at 1986 was identified at α=0.05 (Table 4.8). Thus, we defined the reference period as between 1965 and 1985 and the disturbance period from 1986 to 1995. As described in the single watershed study, the differences between the observed line and the predicted line were also ascribed to forest disturbances. By use of log-transformed data from the reference period, a linear model with autoregressive errors at α=0.05, was established between the accumulated annual effective runoff coefficient of the Willow River watershed and the accumulated annual effective runoff coefficient of the Cottonwood River watershed (Table 4.9). This was then used to predict the accumulated annual effective runoff coefficient of the Willow River watershed without disturbances in the disturbed period. Once the differences between observed accumulated annual effective runoff coefficient of the Willow River watershed and their predicted values from 1986 to 1995 were calculated, the annual mean flow variations attributed to forest disturbances and accumulated annual mean flow variations attributed to forest disturbances were estimated. The accumulated annual mean flow variations attributed to climate variability were calculated accordingly. As shown in Figure 4.11, up to 1995, the accumulated annual mean flow variation attributed to forest disturbances was 616.0 mm while its variation attributed to climate variability was -1233.8 mm by the end of the study period. This cumulatively resulted in a 617.8 mm decrease in annual mean flows by 1995. Meanwhile, during the disturbed period, the CECA of the Willow River watershed experienced a substantial increase from 24.6% in 1986 to 32.5% in 1995, whereas the CECA of the Cottonwood River increased slightly from 7.4% to 12.0%. 99  35 30  Caw  25 20  1986  15 10 5 0 0  5  10  15  20  25  30  35  Cac  Figure 4.10 The modified double mass curve for the Cottonwood-Willow quasi-paired watershed study  Caw and Cac are the accumulated annual effective runoff coefficient of the Willow River watershed and the Cottonwood River watershed, respectively.  Table 4.8 The ARIMA Intervention model of slope in the MDMC for the Cottonwood-Willow quasi-paired watershed study  AR part  p(1) -037 (p=0.04)  Int part  Intervention Part MA part  CP1 (1999)  d(1)  q(1)  1  0  Change Type  Model Structure  MS  ω(1) 0.6* (p=0.05)  AP  (1,1,0)  0.14  Note: *, significant change point at α=0.05; p(1), d(1), q(1) and ω(1) are parameters for autoregression, differencing, moving average and intervention parts; AR part, Int part, and MA part refer to autoregressive part, integrated part, and moving average part, respectively; CP1, AP, and MS refer to the first change point, abrupt permanent change, and model residual, respectively; The slope is the ratio of annual mean flow to annual effective precipitation.  100  Table 4.9 Regression model with autoregressive errors of MDMC for the Cottonwood-Willow quasi-paired watershed study Parameter estimates  PE method  ML  β0  β1  φ1  (constant)  (ln(Cac))  (AR1)  1.069  -0.582  (p<0.001)  (p=0.006)  0  DW  AC(1)  statistics  coefficient  1.90  -0.03  R2  0.99 9  RMSE  0.025  Note: PE method is parameter estimation method; ML is maximum likelihood; RMSE is rooted mean squared error; Caw (t) and Cac (t) are the accumulated annual effective runoff coefficient of the Willow River watershed and the Cottonwood River watershed at year t, respectively; γ(t), regression model error; ε(t), autoregressive model(AR1) error; DW: Durbin-Watson statistics; AC(1) coefficient: Lag 1 Autocorrelation coefficient of model errors; Model: ln(Caw(t))=1.069 ln(Cac(t))+1.069,γ(t)=0.582γ(t1)+ε(t) ( See Figure A-2 for its regression plot).  40 38 36 34 32  600  30 400  28  200  26 24  0  22 20 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995  -200  200 0 -200 -400 -600 -800 -1000 -1200 -1400 -1600 -1800  b)  ∆Qa Qac 95CI 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995  Qaf(mm)  800  ∆Qa Qaf 95CI CECA  Qac(mm)  1000  a)  CECA(%)  1200  Year Year Figure 4.11 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf) and b) the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Willow River watershed estimated by single watershed approach  Table 4.10 summarizes forest disturbance effects on annual mean flows in the Willow River watershed during the disturbed period. With average CECA of 29.4%, the average annual mean flow variation attributed to forest disturbances was 61.6 mm/yr (equivalent to 14.5% of the average annual mean flow from 1965 to 1995). The normalized annual mean flow variation attributed to forest disturbances was 2.3 mm/%, that is, per 10% CECA resulted in a 23 mm 101  increment in annual mean flow (equivalent to 5% of the average annual mean flow from 1965 to 1995). On the contrary, climate variability impacted annual mean flows in a negative way. The average annual mean flow variation attributed to climate variability was -123.4 mm.  Table 4.10 Forest disturbance effects on annual mean flows from 1986 to 1995 estimated by the Cottonwood-Willow quasi-paired watershed study ∆Q  ∆Qf  Qf/CECA  ∆Q%  ∆Qf%  (mm)  (mm)  (mm/%)  (%)  (%)  -61.8  61.6±52.8  2.3±1.9  -14.5  14.4±12.4  ∆Qf%/CECA 0.5±0.4  ∆Qc  ∆Qc%  CECA  (mm)  (%)  (%)  -123.4±63.6 -28.9±14.9  29.4  Note: Refer to the denotes in Table 4.5. Q here is average annual mean flow from 1965 to 1995 (427mm) in the Willow River watershed.  Table 4.11 demonstrates the relative contributions of forest disturbances and climatic variability on annual mean flow variation in the Willow River watershed during the disturbed period. Forest disturbances and climate variability have opposite effects on annual mean flow, and with a high level of CECA, the influence of forest disturbances on annual mean flow variation was lower than that of climate variability. The relative contribution of forest disturbances on annual mean flow, on average, was 41.8% and the relative contribution of climate variability was 58.2%.  Table 4.11 The relative contributions of forest disturbances and climate variability on annual mean flow variations from 1986 to 1995 estimated by the Cottonwood-Willow quasi-paired watershed study ∆Q (mm)  Rf (%)  Rc (%)  -61.8  41.8±13.4  58.2±13.4  CECA (%) 20.1  Note: Refer to the denotes in Table 4.7.  4.4.2.2 Baker Creek watershed Figure 4.12 shows the modified double mass curve for the Baker Creek watershed using original values of variables, where the accumulated annual mean flows are plotted against the  102  accumulated annual effective precipitation. According to the ARIMA intervention analysis of the slope in Figure 4.12, a significant breakpoint in 1999 was detected at α=0.05 (Table 4.12). Thus, we defined the reference period as between 1964 and 1998, while the disturbance period was from 1999 to 2009. As shown in Figure 4.12, a straight line (linear relationship) was observed between the accumulated annual mean flows and the accumulated annual effective precipitation in the period from 1964 to 1998. After 1999, the observed line started to deviate from the original line (predicted line), suggesting that more annual mean flows were generated than the predicted value. Table 4.13 provides detailed information about the linear model with autoregressive errors developed by use of log-transformed values of variables, which was used to predict the accumulated annual mean flows of the Baker Creek watershed without disturbances by their corresponding accumulated annual effective precipitation in the disturbed period. The differences between the observed accumulated annual mean flows and the predicted values from 1999 to 2009 are referred to as accumulated annual mean flow variations attributed to forest disturbances. The accumulated annual mean flow variations attributed to climate variability and annual mean flow variations attributed to forest disturbances were then calculated accordingly.  5000 4500 4000 3500 Qa(mm)  3000  1999  2500 2000 1500 1000 500 0 0  1000  2000  3000 Pae(mm)  4000  5000  6000  Figure 4.12 The modified double mass curve for the Baker Creek watershed  103  Table 4.12 The ARIMA Intervention model of the slope in the MDMC for the Baker Creek watershed  AR part  Intervention Part  Int part  p(1)  d(1)  0  1  MA part  CP1  Change  (1999)  Type  q(1)  ω(1)  0.90  0.36*  (p<0.001)  (p=0.004)  AP  Model Structure  Ln(x), (0,1,1)  MS  0.07  Note: *, Significant change point at α=0.05; p(1), d(1), q(1) and ω(1) are parameters for autoregression, differencing, moving average and intervention parts; AR part, Int part, and MA part refer to autoregressive part, integrated part, and moving average part, respectively; CP1, AP, and MS refer to the first change point, abrupt permanent change, and model residual, respectively; The slope is the ratio of annual mean flow to annual effective precipitation. Table 4.13 The autoregressive model of the MDMC for the Baker Creek watershed Parameter estimates  PE method  ML  β0  β1  φ1  (constant)  (ln(Pae))  (AR1)  -1.154  1.120  -0.833  (p<0.001)  (p<0.001)  (p<0.001)  DW  AC(1)  statistics  coefficient  2.20  -0.1  R2  RMSE  0.997  0.05  Note: PE method: parameter estimation method; ML: Maximum Likelihood; RMSE: rooted mean squared error; Qa(t), accumulated annual mean flow; Pae(t), accumulated annual effective precipitation; γ(t), regression model error; ε(t), autoregressive model(AR1) error; DW: Durbin-Watson statistics; AC(1) coefficient: Lag 1 Autocorrelation coefficient of model errors; Regression model: ln(Qa(t))= 1.154+1.120*ln(Pae(t))+γ(t), γ(t)=0.833γ(t-1)+ε(t) (See Figure A-3 for its regression plot).  As we can seen in Figure 4.13, up to 2009, the accumulated annual mean flow variation attributed to forest disturbances was up to 236.5 mm while its variation attributed to climate variability was 57.0 mm. This cumulatively resulted in a 293.5 mm increment in annual mean flow by the end of the study period. The annual mean flow variations attributed to forest disturbances ranged from -52.5 mm (51.0% of the average annual mean flow from 1964 to 2009) to 70.5 mm (68.3% of the average annual mean flow from 1964 to 2009), with an average of 21.5 mm (20.9% of the average annual mean flow from 1964 to 2009) ( Figure 4.14, Table 4.14). The annual mean flow variations attributed to climate variability ranged from -52.1 mm to 65.1 mm (Figure 4.15), with an average of 5.2 mm (5.0% of the average annual mean flow from 1964 104  to 2009) (Table 4.14). Meanwhile, the CECA experienced a significant increase from 19.6% in 1999 to 62.2% in 2009 with an average of 35%. The normalized annual mean flow variations attributed to forest disturbances varied from -2.1 mm/% to 2.8mm/%, with an average of 0.7 mm/% (Figure 4.16). That means, during the disturbed period, per 10% CECA caused -21 mm (20.4% of the average annual mean flow from 1964 to 2009) to 28 mm (27.2% of the average annual mean flow from 1964 to 2009) change in annual mean flows and on average, per 10% CECA led to 7 mm (6.8% of the average annual mean flow from 1964 to 2009) change in annual mean flow.  Table 4.15 demonstrates the relative contributions of forest disturbances and climatic variability on annual mean flow variations. Both forest disturbanceas and climate variability produced positive effects on annual mean flows. The influence of forest disturbances on annual mean flows was stronger during the disturbed period. On average, 57.5% of the variation in annual mean flows was explained by forest disturbances and 42.5% of that was accounted by climate variablity.  50  200  40  100  30  0  20  -100  10  -200  0 Year  ∆Qac(mm)  60  300  1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  ∆Qaf(mm)  400  ∆Qa ∆Qaf 95CI  350 300 250 200 150 100 50 0 -50 -100 -150 -200  b)  ∆Qa ∆Qac 95CI  1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  70 a)  CECA(%)  500  Year  Figure 4.13 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf); and b) the accumulated annual mean flow variation attributed to climate variability (∆Qaf) in the Baker Creek watershed  105  120  40  50  20  40  0  30  -20  20 ∆Qf CECA  -40  100  ∆Qf% CECA  60 50  80 60  40  40  30  20  20  0  10  -20  10  0  -40  0  1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  -60  b)  CECA(%)  60  70  1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  a)  140  ∆Qf%(%)  ∆Qf(mm)  60  70  CECA(%)  80  Year  Year  Figure 4.14 a) The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances (∆Qf%) in the Baker Creek watershed  80 60  ∆Qc(mm)  40 20 0 -20 -40 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  -60  Year  Figure 4.15 The annual mean flow variation attributed to climate variability (∆Qc) in the Baker Creek watershed  106  70  6  60  5  2  50  4  1  40  a)  3  ∆Qf/CECA  -1  20  -2  10  -3  0  Qf%/CECA  30  CECA(%)  0  b)  60  ∆Qf%/CECA CECA  50  3  40  2 30  1  20  0  -1  10  -2  0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  ∆Qf/CECA(mm/%)  CECA  70  CECA(%)  4  Year  Year  Figure 4.16 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); and b) the normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA)in the Baker Creek watershed  Table 4.14 Annual mean flow variation and its components in the Baker Creek watershed (19992009) ∆Q  ∆Qf  ∆Qf/CECA  ∆Q%  ∆Qf%  (mm)  (mm)  (mm/%)  (%)  (%)  25.5  21.5±23.3  0.7±0.8  24.7  61.8±24.7  ∆Qf%/CECA 1.9±0.9  ∆Qc  ∆Qc%  CECA  (mm)  (%)  (%)  5.2±8.7  5.0±8.7  35%  Note: Refer to the denotes in Table 4.5. Q is average annual mean flow from 1964 to 2009 (103 mm) in the Baker Creek watershed.  Table 4.15 The relative contributions of forest disturbances and climate variability on annual mean flow variation in the Baker Creek watershed (1999-2009) ∆Q (mm)  Rf (%)  Rc (%)  CECA(%)  25.5  57.5±19.0  42.5±19.0  35  Note: Refer to the denotes in Table 4.7  107  4.4.2.3 Moffat Creek watershed Figure 4.17 presents the modified double mass for the Moffat Creek watershed by use of original values of variables, where the accumulated annual mean flows are plotted against the accumulated annual effective precipitation. According to the ARIMA intervention analysis of the slope in Figure 4.17, a significant breakpoint in 1998 was detected at α=0.05 (Table 4.16). Thus, we defined the reference period as between 1967 and 1997, while the disturbed period was from 1998 to 2009. As shown in Figure 4.17, a straight line (linear relationship) was observed between the accumulated annual mean flows and accumulated annual effective precipitation in the period from1967 to 1997. After 1998, the observed line started to deviate from the original line (predicted line), suggesting that more annual mean flows were generated than the predicted. Table 4.17 provides detailed information about the linear model with autoregressive errors using log-transformed values of variables, which was used to predict the accumulated annual mean flows of the Moffat Creek watershed without disturbances by its corresponding accumulated annual effective precipitation in the disturbed period. The differences between the observed accumulated annual mean flows and the predicted values from 1998 to 2009 were referred to as the accumulated annual mean flow variations attributed to forest disturbances. The accumulated annual mean flow variations attributed to climate variability and annual mean flow variations attributed to forest disturbances were then calculated accordingly.  7000 6000  Qa(mm)  5000 4000  1998  3000 2000 1000 0 0  1000  2000  3000  4000  5000  6000  7000  8000  9000  Pae(mm)  Figure 4.17 The modified double mass curve for the Moffat Creek watershed  108  Table 4.16 The ARIMA Intervention model of slope in the MDMC for the Moffat Creek watershed Intervention Part AR part  Int part  MA part  CP1 (1998)  p(1)  d(1)  0  1  q(1)  ω(1)  0.92  0.39  (p<0.001)  (p=0.08)  Change Type  AP  Model  MS  Structure  Ln(x),(0,1,1)  0.1  Note: *, significant change point at α=0.05; p(1), d(1), q(1) and ω(1) are parameters for autoregression, differencing, moving average and intervention parts; AR part, Int part, and MA part refer to autoregressive part, integrated part, and moving average part, respectively; CP1, AP, and MS refer to the first change point, abrupt permanent change, and model residual, respectively; The slope is the ratio of annual mean flow to annual effective precipitation.  Table 4.17 The autoregressive model of MDMC for the Moffat Creek watershed Parameter estimates  PE method  ML  β0  β1  φ1  (constant)  (ln(Pae))  (AR1)  0.278  0.948  -0.591  (p=0.001)  (p<0.001)  (p=0.001)  DW  AC(1)  statistics  coefficient  2.26  -0.13  R2  RMSE  0.997  0.03  Note: PE method is parameter estimation method; MLis maximum likelihood; RMSE is rooted mean squared error; Qa(t) is accumulated annual mean flow; Pae(t) is accumulated annual effective precipitation; γ(t) is regression model error; ε(t) is autoregressive model (AR1) error; DW is DurbinWatson statistics; AC(1) coefficient is lag 1 autocorrelation coefficient of model errors; Regression model: ln(Qa(t))= 0.278+0.948*ln(Pae(t))+γ(t), γ(t)=0.591*γ(t-1)+ε(t) (See Figure A-4 for its regression plot)  As we can seen in Figure 4.18, up to 2009, the accumulated annual mean flow variation attributed to forest disturbances was up to 299.2 mm while its variation attributed to climate variability was -285.9 mm. This cumulatively resulted in a 13.3 mm increment in annual mean flow by the end of the study period. The annual mean flow variations attributed to forest disturbances ranged from 50.9 mm (37.7% of the average annual mean flow from 1967 to 2009) to 80.2 mm (59.5% of the average annual mean flow from 1967 to 2009), with an average of 21.6 mm (16.0% of the average annual mean flow from 1967 to 2009) ( Figure 4.19, Table 4.18). 109  The annual mean flow variations attributed to climate variability ranged from -80.1 mm to 34.0 mm (Figure 4.20), with an average of -19.7 mm (14.6% of the average annual mean flow from 1967 to 2009) ( Table 4.18). Meanwhile, CECA experienced a significant increase from 24.7% in 1999 to 65.7% in 2009 with an average of 38%. The normalized annual mean flow variations attributed to forest disturbances varied from -1.3 mm/% to 2.9 mm/%, with an average of 0.7 mm/% (Figure 4.21, Table 4.19). That means, during the disturbed period, per 10% CECA caused a -13 mm (9.6% of the average annual mean flow from 1967 to 2009) to 29 mm (21.5% of the average annual mean flow from 1967 to 2009) change in annual mean flow, with an average of 7 mm (5.2% of the average annual mean flow from 1967 to 2009).  Table 4.19 demonstrates the relative contributions of forest disturbances and climatic variability to annual mean flow variations. The influence of forest disturbances on annual mean flows was stronger during the disturbed period. On average, only 55.1% of the variation in annual mean flows was explained by forest disturbances and 44.9% of that was accounted by climate variability.  500.0  a)  30.0  0.0  20.0  1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  ∆Qa ∆Qaf 95CI CECA  Year  50.0  -100.0 -200.0  40.0  -300.0  30.0  -400.0 -500.0  10.0  -600.0  0.0  -700.0  ∆Qa ∆Qac 95CI CECA  20.0 10.0 0.0  1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  ∆Qaf (mm)  40.0  100.0  -300.0  60.0  CECA%  50.0  200.0  -200.0  70.0  b)  0.0  300.0  -100.0  100.0  60.0  CECA% ∆Qac(mm)  400.0  200.0  70.0  Year  Figure 4.18 a) The accumulated annual mean flow variation and accumulated annual mean flow variation attributed to forest disturbances (∆Qaf); and b) the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Moffat Creek watershed  110  ∆Qf(mm)  40  60  60  50  40  50  40  20  40  0  30  -20  20  20 30  0  20  -20 -40  ∆Qf CECA  -40  0  -60  1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  -60  10  b)  60  ∆Qf% CECA  CECA(%)  60  70  10 0  1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  a)  80  ∆Qf%(%)  80  70  CECA(%)  100  Year  Year  Figure 4.19 a) The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances(∆Qf%) in the Moffat Creek watershed  40 20  ∆Qc(mm)  0 -20 -40 -60 -80 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  -100  Year  Figure 4.20 The annual mean flow variation attributed to climate variability (∆Qc) in the Moffat Creek watershed  111  2.5  60  2  20 10 CECA  ∆Qf/CECA  0  ∆Qf%/CECA  CECA(%)  30  b)  60  1.5  50 40  70  50  1  40  0.5 30  0  20  -0.5  CECA(%)  a)  70  10  -1 ∆Qf%/CECA  -1.5  CECA 0  1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009  ∆Qf/CECA(mm/%)  3.5 3 2.5 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2  Year  Year  Figure 4.21 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); b) the normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA)in the Moffat Creek watershed  Table 4.18 Annual mean flow variation and its components in the Moffat Creek watershed (19982009) ∆Q  ∆Qf  Qf/CECA  ∆Q%  ∆Qf%  (mm)  (mm)  (mm/%)  (%)  (%)  1.9  21.6±25.1  0.7±0.8  -1.4  16.0±18.6  ∆Qf%/CECA 0.5±0.6  ∆Qc  ∆Qc%  CECA  (mm)  (%)  (%)  -19.7±23.7 -14.6±17.6  40%  Note: Refer to the denotes in Table 4.5. Q is average annual mean flow from 1967 to 2009 (135 mm) in the Moffat Creek watershed.  Table 4.19 The relative contributions of forest disturbances and climate variability on annual mean flow variation in the Moffat Creek watershed (1998-2009) ∆Q (mm)  Rf (%)  Rc (%)  CECA(%)  1.9  55.1.3±7.5  44.9±7.5  40  Note: Refer to the denotes in Table 4.7  112  4.4.2.4 Tulameen River watershed Figure 4.22 presents the modified double mass curve for the Ashnola-Tulameen quasi-paired watershed study using the original values of variables, where the accumulated annual effective runoff coefficient of the Tulameen River watershed is plotted against the accumulated annual effective runoff coefficient of the Ashnola River watershed. According to the ARIMA Intervention model of the slope in Figure 4.22, two significant breakpoints at 1984 and 1999 were identified at α=0.05 (Table 4.20). Thus, we defined the reference period as between 1954 and 1983 and the disturbed period from 1984 to 2009. The disturbed period was further divided into two phases: phase 1 from 1984 to 1998 and phase 2 from 1999 to 2009. The differences between the observed line and the predicted line were ascribed to forest disturbances. By use of log-transformed values of variables, the linear model with autoregressive errors that relates the accumulated annual effective runoff coefficient of the Tulameen River watershed with the accumulated annual effective runoff coefficient of the Ashnola River watershed was used to predict the accumulated annual effective runoff coefficient of the Tulameen River watershed without disturbances in the disturbed period (Table 4.21). Then the annual mean flow variations attributed to forest disturbances were estimated accordingly. The accumulated annual mean flow  Cat  variations attributed to climate variability were also calculated.  55 50 45 40 35 30 25 20 15 10 5 0  1999 1984  Cat and Caa are the accumulated annual effective runoff coefficient of the Tulameen River watershed and the  0  5  10  15  20  25 30 Caa  35  40  45  50  55  Figure 4.22 The modified double mass curve for the Ashnola-Tulameen quasi-paired watershed study  113  Table 4.20 The ARIMA Intervention model of the slope in the MDMC for the Ashnola-Tulameen quasi-paired watershed study Int  AR part  part  p(1)  d(1)  -0.76  3  (p<0.001)  Intervention Part MA part  CP1  CP2  (1984)  (1999)  q(1)  ω(1)  ω(2)  0.89  0.86*  -0.61*  (p<0.001)  (p=0.009)  (p=0.05)  Change Type  AP  Model Structure  Ln(x),(1,3,1)  MS  0.2  Note: *, Significant change point at α=0.05; p(1), d(1), and q(1) are parameters for autoregression, differencing, and moving average; ω(1) and ω(2) are parameters for intervention; AR part, Int part, and MA part refer to autoregressive part, integrated part, and moving average part, respectively; CP1, CP2, AP, and MS refer to the first change point, the second change point, abrupt permanent change, and model residual, respectively; The slope is the ratio of annual mean flow to annual effective precipitation.  Table 4.21 The autoregressive model of MDMC for the Tulameen River watershed Parameter estimates  PE method  ML  β0  β1  φ1  (constant)  (ln(Cat))  (AR1)  0.322  0.945  -0.555  (p<0.001)  (p<0.001)  (p=0.006)  DW  AC(1)  statistics  coefficient  2.06  -0.03  R2  RMSE  0.999  0.02  Note: PE is parameter estimation method; ML is maximum likelihood; RMSE is rooted mean squared error; Cat (t) and Caa (t) are the accumulated annual effective runoff coefficient of the Tulameen River watershed and the Ashnola River watershed at year t, respectively; γ(t) is regression model error; ε(t) is autoregressive model(AR1) error; DW is Durbin-Watson statistics; AC(1) coefficient is lag 1 autocorrelation coefficient of model errors; Regression model: ln(Cat (t))=0.322+0.935ln(Caa (t))+γ(t), γ(t)=0.555γ(t-1)+ε(t) (See Figure A-5 for its regression plot).  As seen in Figure 4.23, up to 2009, the accumulated annual mean flow variation attributed to forest disturbances was 1566.6 mm while its variation attributed to climate variability was 1668.4 mm. This cumulatively resulted in a 101.8 mm decrease in annual mean flow by 2009. The annual mean flow variations attributed to forest disturbances ranged from -86.7 mm (22.5% of the average annual mean flow from 1954 to 2009) to 202.9 mm (52.6% of the average annual mean flow from 1954 to 2009), with an average of 60.3 mm (15.6% of the average annual mean flow from 1954 to 2009) (Figure 4.24, Table 4.22). The annual mean flow variations attributed to climate variability varied from -292.7 mm to 181.1 mm with an average of 112.5 mm (Figure 114  4.25, Table 4.22). Meanwhile, during the disturbed period, the CECA of the Tulameen River watershed experienced a substantial increase from 10.6% in 1985 to 33.8% in 2009, whereas the CECA of the Ashnola River increased slightly from 2.0% to 6.8%.  As suggested in Figure 4.26 (a,b), the normalized annual mean flow variations attributed to forest disturbances varied from -5.3 mm/% to 15.7 mm/%, with an average of 4.6 mm/%. That means, during the disturbed period, per 10% CECA caused a -53 mm (13.7% of the average annual mean flow from 1954 to 2009) to 157 mm (40.7% of the average annual mean flow from 1954 to 2009) change in annual mean flows, with an average of 44 mm (11.4% of the average annual mean flow from 1954 to 2009).  3000 2000  40  ∆Qa ∆Qaf ∆Qac 95CI CECA  35 30  1000 0  %  mm  25 20  -1000  15 10  -3000  5 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008  -2000  Year  Figure 4.23 The accumulated annual mean flow variation (∆Qa), accumulated annual mean flow variation attributed to forest disturbances (∆Qaf), and the accumulated annual mean flow variation attributed to climate variability (∆Qac) in the Tulameen River watershed  115  40.0  a)  150.0  30.0  100.0  25.0  50.0  20.0  0.0  15.0  35.0  60.0  30.0 40.0  25.0  20.0  20.0 15.0  0.0  10.0  10.0  5.0 2008  2005  2002  0.0 1999  1996  1993  1984  2008  2005  2002  1990  -40.0  0.0 1999  ∆Qf% 95CI CECA  -20.0  5.0  1996  1987  1984  -150.0  1993  -100.0  1990  ∆Qf 95CI CECA  1987  -50.0  40.0  b)  35.0  CECA(%) ∆Qf%(%)  ∆Qf(mm)  200.0  80.0  CECA(%)  250.0  Year  Year  Figure 4.24 The annual mean flow variation attributed to forest disturbances (∆Qf); and b) the relative annual mean flow variation attributed to forest disturbances (∆Qf%) in the Tulameen River watershed  300.0  40.0  200.0  35.0  ∆Qc(mm)  25.0  0.0  20.0 -100.0  15.0  -200.0 ∆Qc 95CI CECA  -300.0  10.0 5.0 0.0  1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008  -400.0  CECA(%)  30.0  100.0  Year  Figure 4.25 The annual mean flow variation attributed to climate variability (∆Qc) in the Tulameen River watershed  116  35.0  4.0  30.0  3.0  30.0  2.0  25.0  1.0  20.0  0.0  15.0  25.0  5.0  20.0 15.0  0.0  10.0 -5.0  1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008  -10.0  ∆Qf/CECA 95CI CECA  Year  40.0  b)  35.0  -1.0  5.0  -2.0  0.0  -3.0  CECA(%)  ∆Qf/CECA(mm/%)  10.0  5.0  10.0 ∆Qf%/CECA 95CI CECA  5.0 0.0  1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008  a) 15.0  40.0  CECA(%) ∆Qf%/CECA  20.0  Year  Figure 4.26 a) The normalized annual mean flow variation attributed to forest disturbances (∆Qf/CECA); b) The normalized relative annual mean flow variation attributed to forest disturbances (∆Qf%/CECA) in the Tulameen River watershed  Table 4.22 summarizes forest disturbance effects on annual mean flows in different phases, which helps us explore the temporal dynamic of hydrological impacts of forest disturbances. In phase 1, with an average CECA of 12.8%, the average annual mean flow variation attributed to forest disturbances was 90.0 mm/yr (equivalent to 23.3% of the average annual mean flow from 1954 to 2009), but it greatly declined to 19.7 mm/yr (equivalent to 5.1% of the average annual mean flow from 1954 to 2009) in phase 2, the most intensively disturbed phase. The normalized annual mean flow variations attributed to forest disturbances displayed a similar pattern. In phase 1, per 10% CECA resulted in a 72 mm increment in annual mean flow while in phase 2, this number declined to 9 mm. Unlike forest disturbances, climate variability impacted annual mean flows in a negative way. The average annual mean flow variations attributed to climate variability in phase 1 and phase 2 were -129.9 mm and -88.8mm, respectively.  117  Table 4.22 Forest disturbance effects on annual mean flows in the Tulameen River watershed in different phases (1984-2009) ∆Q  ∆Qf  ∆Qf/CECA  ∆Qf%  (mm)  (mm)  (mm/%)  (%)  1:1984-1998  -40.0  90.0±33.6  7.2±2.6  23.3±8.7  1.9±0.7  -129.9±63.6 -33.7±16.0  12.8  2:1999-2009  -69.1  19.7±43  0.9±2.4  5.1±11.1  0.2±0.6  -88.8±67.2 -23.0±16.5  21.7  Average  -129.9  60.3±29.8  4.6±2.2  15.6±7.7  1.2±0.6  -112.5±45.4 -29.2±11.8  16.6  Phase  ∆Qf%/CECA  ∆Qc  ∆Qc%  CECA  (mm)  (%)  (%)  Note: Refer to the denotes in Table 4.5 and 4.6. Q is average annual mean flow from 1954 to 2009 (386 mm) in the Tulameen River watershed.  Table 4.23 demonstrates the relative contributions of forest disturbances and climatic variability to annual mean flow variations. Forest disturbances and climate variability produced offsetting effects on annual mean flows, and climate variability was more influecial. The relative contribution of forest disturbances on annual mean flows was averaged 33.0% while the relative contribution of climate variability was 67.0%. In phase 1, 34.3% of the variation in annual mean flows was explained by forest disturbances and 65.7% of that was accounted by climate variability. During phase 2, the relative contribution of forest disturbances on annual mean flow variations slightly dropped to 31.2%, compared with 68.4% of variation explained by climate variability.  Table 4.23 The relative contributions of forest disturbances and climate variability on annual mean flow variations in the Tulameen River watershed (1984-2009) Phase  ∆Q (mm)  Rf (%)  Rc (%)  CECA(%)  1: 1984-1998  -40.0  34.3±7.2  65.7±7.2  12.8  2: 1999-2009  -69.1  31.2±8.6  68.8±8.6  21.7  Average  -129.9  33.0±5.6  67.0±5.6  16.6  Refer to the denotes in Table 4.7.  118  4.4.3  Integrated analysis  The results from the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds were combined to explore how the estimated change magnitude of annual mean flows varied along environmental gradients, climatic gradients (precipitation and temperature) in particular.  Firstly, we examined how annual mean flow responses to forest disturbances varied with precipitation gradient. As shown in Figure 4.27, a significant positive correlation between the normalized annual mean flow variations attributed to forest disturbances (∆Qf/CECA) and annual precipitation (P) was detected by both linear regression and Kendall tau correlation. In other words, there is a significant upward tendency on annual mean flow variations attributed to forest disturbances (∆Qf) with increased annual precipitation. Figure 4.28 presents the normalized annual mean flow responses to forest disturbances grouped by watersheds in the order of their long-term average annual precipitation. The watersheds with higher long-term average annual precipitation such as the Tulameen and Willow River watersheds are characterized by greater annual mean flow responses to forest disturbances. The Mann-Whitney U test further confirmed that the normalized annual mean flow responses to forest disturbances in the wetter watersheds (Tulameen and Willow) were significantly higher than those in the drier watersheds (Baker and Moffat), and there were no significant differences in the annual mean flow responses between the two drier watersheds (Table 4.24).  Table 4.24 Mann Whitney U test on differences in annual mean flow responses to forest disturbances (∆Qf/CECA) between different watersheds Pairs Statistics Z statistics  Tulameen-  Tulameen-  Tulameen-  Willow-  Willow-  Baker-  Willow  Baker  Moffat  Baker  Moffat  Moffat  -1.95*  3.5*  3.6*  2.9*  2.8*  0.07  (p=0.05)  (p<0.001)  (p<0.001)  (p=0.003)  (p=0.004)  (p=0.94)  *, significant at α=0.05.  119  20  ∆Qf/CECA=0.0095P -4.24 (R2=0.27, p<0.0001) 15 ∆Qf/CECA (mm/%)  Kendall tau=0.35(p<0.05) 10  5  0  -5 400  500  600  700  800 900 1000 1100 1200 P (mm)  Figure 4.27 The correlation between annual precipitation (P) and annual mean flow responses to forest disturbances (∆Qf/CECA)  18 16  Median  25%-75%  Non-Outlier Range  14 ∆Qf/CECA (mm/%)  12 10 8 6 4 2 0 -2 -4 Baker  Moffat  Willow Tulameen  Precipitation: Low to High  Figure 4.28 The comparison of annual mean flow response to forest disturbances in the watersheds with different long-term average annual precipitations  We also investigated the role of temperature in annual mean flow responses to forest disturbances. Figure 4.29 relates the normalized annual mean flow responses to forest disturbances with annual mean temperatures, while Figure 4.30 shows the normalized annual 120  mean flow responses to forest disturbances grouped by watersheds in the order of their long-term annual mean temperatures. Both graphs failed to suggest any significant differences in annual mean flow responses to forest disturbances across the temperature gradient.  20 ∆Qf/CECA = 0.17T + 2.32 R² = 0.003(p=0.49) Kendall tau=-0.01(p>0.05)  ∆Qf/CECA (mm/%)  15  10  5  0  -5 0  1  2  3  4  5  6  7  T (°C)  Figure 4.29 The correlation between annual mean temperature (T) and annual mean flow responses to forest disturbances (∆Qf/CECA)  18 16  Median  25%-75%  Non-Outlier Range  14  ∆Qf/CECA (mm/%)  12 10 8 6 4 2 0 -2 -4 Baker  Moffat Willow  Tulameen  Temperature: Low to High  Figure 4.30 The comparison of annual mean flow response to to forest disturbances in the watersheds with different long-term annual mean temperatures  121  4.5 Discussion 4.5.1  Determination of forest disturbance thresholds for significant changes in annual mean flows  A watershed is a system underpinned by complex processes and interactions involving geology, hydrology, vegetation, climate, topography, and biology. These complexities and interactions make any watershed system resilient to both internal and external changes. The watershed ecosystems can be altered by disturbances from time to time, and may be collapsed by severe catastrophic disturbances. It is commonly believed that a theoretical threshold of forest disturbances must exist, below which a watershed ecosystem remains stable, and conversely, collapses or is significantly affected.  From a watershed management perspective, the identification of the forest disturbance threshold for a given watershed is needed because it can guide the design of forest management practices to avoid significant and negative effects on watershed ecosystem functions and services. Efforts have already been made to determine possible thresholds of forest logging in small watersheds. Such thresholds tend to be variable due to the differences in topography, vegetation, geology, hydrological regime, and climate among watersheds. For examples, in the Appalachian Mountains, USA, only a 10% reduction in forest cover can yield a detectable response in annual mean flows (Swank et al., 1988), while in the Central Plains of the United States, 50% harvest might be required for a significant change in flows (Stednick, 1996). The general belief is that more than 20% of the watershed area must be changed or disturbed to detect a significant change in annual mean flows in small watersheds (Bosch and Hewlett, 1982; Hetherington, 1987). In large watersheds, information on the forest disturbance thresholds for significant hydrological changes is rare. The general perception is that the thresholds for large watersheds should be larger than those for small watersheds due to their greater buffering capacities in large watersheds. For example, the study in several large watersheds (from 401 to11900 km2) in Canadian boreal forests (Buttle and Metcalfe, 2000) showed that disturbance levels ranging from 5 to 25% of watershed areas yielded insignificant changes in annual mean flows.  122  Determining the forest disturbance thresholds above which hydrological variables are significantly changed is a challenging task as it requires long-term experiments of assessing the effects of various levels of forest disturbances on hydrological variables of interest. While this experimental approach may be feasible for small watersheds, it is difficult to be applied in large watersheds. It should be noted that many published studies on the impacts of forest changes or disturbances on hydrology may not be used to define the forest disturbance thresholds mainly because the reported forest disturbance levels in the published studies are just the levels when the investigation took place. Without information on the gradients of forest disturbance levels and their comparable effects, it is impossible to define reliable thresholds. In this study, the application of modified double mass curves and the statistical identification of change points on those modified double mass curves provide a meaningful way to estimate forest disturbance thresholds. In the Willow River watershed, three significant change points in annual mean flows due to forest disturbances were identified (Figure 4.5). The first significant change point in annual mean flows due to forest disturbances was in 1968 and the second significant change point was in 1985. This suggests that annual mean flows in the period of 1968 to 1984 were significantly altered by forest disturbances, compared with those in the reference period before 1968. In other words, the forest disturbance levels must be equal to or greater than the CECA (23%) in 1984 to detect significant hydrological responses. However, due to the 2-year lagged responses between annual mean flows and forest disturbances (as shown in Table 4.2), the above threshold was adjusted to 20.6% (CECA) in 1982 (Table 4.25). Thus, 20.6 % CECA was recognized as the forest disturbance threshold for a significant change in annual mean flows in the Willow River watershed. Similarly, the forest disturbance threshold in terms of CECA in the Tulameen River watershed was identified to be 15% (Table 4.25). Unlike the Willow and Tulameen River watersheds with long disturbed periods and multiple significant change points in their modified double mass curves, the Baker Creek watershed has a very short disturbed period (about 10 years) and has only one change point in the modified double mass curve. During that short period, there was a sharp increase in CECA from about 19.2% to 62.2%. This impedes us from identifying such a threshold. Therefore, we estimate that the forest disturbance threshold must fall in the range between 19.2% (1999) and 62.2% (2009). After considering the one-year lagged effect, this range was adjusted to be 18.9% (1998) to 123  55.3% (2008) (Table 4.25). Similarly, the forest disturbance threshold in the Moffat Creek watershed was estimated to be in the range of 19.4-29.2% (Table 4.25). With longer disturbed periods in the Baker and Moffat Creek watersheds in the future, we should be able to determine the exact values of the forest disturbance thresholds in these two watersheds.  Table 4.25 The estimated forest disturbance thresholds for significant changes in annual mean flows in the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds Watershed  CECA  Annual mean  Thresholds  size (km2)  (%)  flow change  (CECAf) (%)  Willow  3185  35.4  Significant  21  Baker  1560  62.2  Significant  ∈ [19, 55]  Moffat  539  65.7  Significant  ∈ [20, 29]  Tulameen  1780  33.8  Significant  15  Watersheds  Note: CECA is cumulative equivalent clear-cut area  The estimates (Table 4.25) suggest that even in the BC interior, the forest disturbance thresholds for significant changes in annual mean flows in large watersheds can be highly variable. The large variations of forest disturbance thresholds in the study watersheds may be due to the differences in climate, which governs how forest and water interact. Forests in the southern interior of BC, with warmer weather and resultant longer growing seasons, are likely to have high evapotranspiration. As a result, annual mean flows in these forested watersheds (e.g., Tulameen) in the southern interior of BC tend to be more sensitive to forest changes or disturbances. Therefore, a relatively low forest disturbance threshold (e.g., 15% CECA in Tulameen) is expected. In comparison, the watersheds in the central interior of BC, characterized by long, cold winter and short, cool summer, have a shorter growing season. Forests in these watersheds tend to consume less water, which consequently make annual mean flows less sensitive to forest changes. This may help partially explain a higher forest disturbance threshold in the Willow River watershed. In the dry and cold Baker and Moffat Creek watersheds, runoff yield is low (the runoff coefficient is 0.19-0.22), forests grow slowly with their 124  evapotranspiration mainly constrained by water availability. As a result, the hydrological responses to forest disturbances are also likely less sensitive, and thus higher forest disturbance thresholds (CECA>20%) are expected. Watershed properties and configurations can also play an important role in forest disturbance thresholds in large watersheds. As compared with small watersheds, there are greater complexities and spatial heterogeneities in topography, vegetation, soils, climate conditions, and hydrologic regime in large watersheds. It can be expected that the watersheds with more complexities have greater buffering capacities so that higher forest disturbance thresholds are needed to cause significant change on hydrology. For example, about 21% of CECA in the Willow River watershed caused significant change in annual mean flows. In contrast, the similar level of forest disturbances in the neighboring Bowron River watershed caused no detectable changes in annual mean flows (Wei and Davison, 1998). The different hydrological responses between two watersheds are due to the difference in watershed properties (Zhang and Wei, 2013b). The Bowron River watershed is lake headed in the high mountainous area with larger variations in elevation; the drainage area above 1500 m accounts for 18.8% of the total watershed area. On the contrary, the Willow River watershed lies mostly in the low elevation areas characterized by flat to gently sloping topography with only 8% of the watershed area above 1500 m. Such differences in watershed properties suggest that the Bowron River watershed has greater water storage, higher buffering capacities, and is more resilient to forest disturbances. Thus, the Bowron River watershed requires a higher forest disturbance threshold for a detectable change in annual mean flows. This example clearly demonstrates that forest disturbance thresholds in large watersheds also depend on watershed properties. The findings may provide some useful implications for designing forest management practices in the BC interior. However, more case studies in large watersheds across different environmental gradients are needed to draw a more generalized conclusion on forest disturbance thresholds.  125  4.5.2  Forest disturbances impact on annual mean flow  As suggested in this analysis, forest disturbances significantly increased annual mean flows in the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds over their respective study periods. This is in accordance with the findings from small watershed studies (Stednick, 1996; Moore and Wondzell, 2005). The reduction of vegetation cover after forest disturbances reduces interception and evapotranspiration, resulting in more water available for streamflow generation. Annual mean flow variations attributed to forest disturbances in the Willow River and Tulameen River watersheds rank the highest among the four disturbed watersheds, with an average of 61.1 and 60.3 mm per year, respectively. Forest disturbances produced similar impacts on annual mean flows in these two watersheds, mainly due to their close disturbance levels (35.4% and 33.8% CECA in the Willow River and Tulameen River watersheds, respectively). In spite of the moderate levels of forest disturbances, the Willow River and Tulameen River watersheds were substantially disturbed for over 40 years. In a watershed, particularly a large watershed, there are always time-lagged hydrological responses to forest disturbances, and the delayed responses can be cumulative over time and space, resulting in hysteretic processes (Blöschl et al., 2007). Given the great heterogeneity in topography, landscape, climate, and their resultant great buffering ability, this hysteretic effect can be particularly noticeable in large watersheds. Thus, the long-lasting intensive forest disturbances in the Willow River and Tulameen River watersheds caused not only an early presence of detectable responses of annual mean flows, but also provided long impact durations that allow for the detection of more hysteretic effects during their study periods. Consequently, more pronounced changes in annual mean flows due to forest disturbances are expected in the Willow River and Tulameen River watersheds.  The Baker Creek and Moffat Creek watersheds, with similar cumulative forest disturbance levels (Table 4.26), were estimated to have similar annual mean flow variations attributed to forest disturbances, with an average annual mean flow change of 21.5 mm and 21.6 mm, respectively. Despite over 60% of CECA, annual mean flow changes due to forest disturbances in the Baker Creek and Moffat Creek watersheds were lower than those in the Willow River and Tulameen River watersheds. This is partly related to the fact that most of the forest disturbances in the Baker Creek and Moffat Creek watersheds occurred in the last 10 years, as a result of the large126  scale outbreak of MPB and subsequent salvage logging since 2003. The short-term intense disturbances in both the Baker and Moffat Creek watersheds probably failed to provide sufficient time to detect the hysteretic processes of annual mean flow changes, thus resulting in less annual mean flow variations attributed to forest disturbances during the study period. However, more delayed and cumulative impacts of forest disturbances on annual mean flows may be expected in the future. Another reason for lower disturbance impacts of annual mean flows in the Baker and Moffat Creek watersheds could be that trees in these two watersheds, situated in extremely dry and cold areas, have relatively lower evapotranspiration, as compared with those in the Willow and Tulameen River watersheds. Therefore, in these water-limited systems, less reduction in evapotranspiration due to vegetation removal is expected, and their resultant increments in annual mean flows after forest disturbances in those two watersheds tend to be less. These two reasons can support the finding that higher magnitude of forest disturbances induced change in annual mean flows in the wetter Willow and Tulameen watersheds, while lower annual mean flow changes in the drier Baker and Moffat Creek watersheds.  Although the annual mean flow variations attributed to forest disturbances in these four watersheds are different, their relative annual mean flow variations attributed to forest disturbances are similar. The annual mean flow variations attributed to forest disturbances in the Willow River, Tulameen River, Baker Creek, and Moffat Creek watersheds account for 14.2% 15.6%, 20.9%, and 16.6% of their long-term average annual mean flows, respectively.  Table 4.26 The quantification of forest disturbance effect on annual mean flow ∆Qc  ∆Qf  ∆Qf/CECA  ∆Qf%  (mm)  (mm)  (mm/%)  (%)  Willow  -93.3±24.8  61.1±20.0  4.5±2.3  15.2±6.3  1.1±0.7  23.6  Baker  5.2±8.7  21.5±23.3  0.7±0.8  61.8±24.7  1.9±0.9  35.0  Moffat  -19.7±23.7  21.6±25.1  0.7±0.8  16.0±18.6  0.5±0.6  40.0  Tulameen  -112.5±45.4  60.3±29.8  4.6±2.2  15.6±7.7  1.2±0.6  16.6  Watersheds  ∆Qf%/CECA  CECA (%)  Note: Refer to the denotes in Table 4.5 and 4.6.  127  4.5.3  Forest disturbances impact on seasonal mean flows  The time series cross-correlation analysis suggested that forest disturbances produced significant impacts on seasonal mean flows in the study watersheds. The specific seasonal variables impacted by forest disturbances, however, varied among watersheds. Only in the Willow River watershed was a significant positive correlation between the spring mean flows and CECA detected. Forest disturbances impacted the spring mean flows mainly by altering winter snow accumulation and snow ablation. Forest disturbances such as logging increase snow accumulation in winters due to the reduction in snow interception and evaporation by canopies, resulting in more snowpack available for snowmelt in springs. In addtion, forest disturbances can increase the incoming solar radiation and allow for more turbulent transfer of sensible heat to the snowpack surface and more transfer of latent heat by condensation and freezing, resulting in accelerated snowmelt process (Boon, 2009, 2011; Winkler et al., 2010). Therefore, faster and greater amounts of snowmelt water are normally expected after forest disturbances in snowdominated small watersheds. However, for large watersheds, many other complex watershed processes can mask these increments in snowmelt water, often leading to insignificant changes in spring mean flows.  The Willow River watershed, mostly characterized by more uniform topography and a lower relief, is favourable for a faster snowmelt process. More importantly, most of the logging activities occurred at higher elevations with 59.2% of the clear-cuts above 970 m (H60 line), resulting in a synchronized process of snowmelt water from both lower and higher elevations and eventually more snowmelt water flowing into river channels in spring. In a snow-dominated watershed like the Willow River watershed, snow melting normally starts at lower elevation areas in early spring and gradually spreads upwards to the higher elevation areas as the temperature rises. Melting water from lower elevations normally fills up stream channels, while the melting water from higher elevations causes peak flows or high flows. Thus, forest disturbances at higher elevations can increase snow accumulation and accelerate snowmelt rates, leading to an earlier and greater amount of snowmelt water from higher elevations. This can be  128  synchronized with melt water from low elevations and consequently increase spring mean flows in the Willow River watershed.  Forest disturbances have generated significant positive impacts on summer mean flows in the Moffat Creek watershed and on fall mean flows in the Baker Creek and Tulameen River watersheds. Since summer and fall are growing seasons with vigorous vegetation growth, the increments in summer and fall mean flows are closely related to the changes in soil moisture and evapotranspiration. After forest disturbances, due to an increase in snow accumulation and melting, more recharges to soils and groundwater from snowmelt water can lead to higher soil moisture in spring, the beginning of the growing season. Accordingly, less water is needed for soil moisture recharge during the growing season (summer and fall). Hence, more water is available in summer and fall for streamflow generation. The significant increase in summer mean flows in the Moffat Creek watershed may be associated with earlier high soil moisture as a result of less evapotranspiration and more soil water recharge as compared with that in the Baker and Tulameen River watersheds.  A significant positive response of winter mean flows to forest disturbances has also been detected in the Baker Creek, Moffat Creek, and Tulameen River watersheds. Streamflow in winter is mainly maintained by groundwater discharging which normally depends upon groundwater recharge processes through soil infiltration, mountain system recharging mechanisms (Wahi 2005; Wahi et al. 2008) and regional aquifer characteristics. After forest disturbances in these watersheds, more water is available for soil moisture and consequently more groundwater recharge from spring to fall, which leads to more streamflow in winter. 4.5.4  Climatic gradients and mean flow response to forest disturbances  The change magnitude of annual mean flows caused by forest disturbances varied from watershed to watershed, as well as from year to year. Differences in climate, especially in precipitation and temperature, among watersheds or years are believed to be one of the key factors controlling the responses of annual mean flows to forest disturbances. The following sections discuss the role of climate, mainly precipitation and temperature in annual mean flow responses to forest disturbances. 129  As suggested by Figure 4.27, annual mean flow responses to forest disturbances in the drier years tend to be less than those in the wetter years. Similarly, the responses in the drier watersheds appear less intensive that those in the wetter watersheds (Figure 4.28). The watersheds with higher long-term average annual precipitation such as the Tulameen and Willow River watersheds are characterized by greater annual mean flow responses to forest disturbances than the two dry watersheds (Baker Creek and Moffat Creek). This finding is consistent with small watershed studies (Baker, 1986; Troendle and Kaufmann, 1987). In the drier years or drier watersheds, the increment of available water due to the reduction in interception and evapotranspiration after forest disturbances is proportioned to recharge the soil moisture first. Once the dry soils are saturated, the excess water can be transformed into streamflow. In the wetter years or wetter watersheds, on the contrary, with higher soil moisture, less water is needed for soil moisture recharge and thus more water is converted into streamflow. Hence, more pronounced responses of annual mean flows to forest disturbances in the wetter years or wetter watersheds are expected. However, we failed to find any differences in annual mean flow responses to forest disturbances across the temperature gradient as suggested by Figure 4.29 and 4.30.  This analysis clearly demonstrates that precipitation can have greater impact on annual mean flow responses to forest disturbances, regardless of forest disturbance levels. However, no definitive conclusion can be drawn on the effects of temperature on annual mean flow responses to forest disturbances. 4.5.5  Watershed resilience and the response of annual mean flows to forest disturbances  The normalized annual mean flow variation attributed to forest disturbances, referred as the annual mean flow variation attributed to per unit CECA, reveals the intensity of forest disturbance impacts on annual mean flows. Both the Willow and Tulameen River watersheds had long disturbance periods, which provide us an opportunity to investigate how the strength of forest disturbance impacts on annual mean flows might change over time. As suggested by the analysis (Table 4.6), the normalized annual mean flow variation attributed to forest disturbances in the Willow River watershed during the period of 1968 to 1984 was 14.2 mm/%, which greatly 130  declined to 3.2 mm/% during the period of 1985 to 1990 and further dropped to 2.4 mm/% during the period of 1991 to 2008. Similarly, in the Tulameen River watershed, the normalized annual mean flow variation attributed to forest disturbances was 7.2 mm/% during the period of 1984 to1998 and then greatly reduced to 0.9 mm/% in the period of 1999 to 2009 (Table 4.21). Those results suggest that the strength of forest disturbance related impacts on annual mean flows were higher at the early stage of the disturbed period, and then declined with the progression of forest disturbances (Figure 4.31).  A system perspective may help us understand the above-mentioned phenomenon. The properties of hydrological system behavior in a large watershed generally include stability, hysteresis, resilience, reversibility or irreversibility. In the reference period with no or fewer forest disturbances, forest disturbance impacts could be minor or hysteretic due to the buffering ability of a large watershed. This stage can be characterized by stability (Blöschl et al., 2007). As forest disturbances progress and reach the forest disturbance threshold, the hysteretic and cumulative responses of annual mean flows emerge, leading to an abrupt hydrological change. This initial high strength of forest disturbance impact on hydrology may then decline in spite of increased cumulative forest disturbances. Such a decline may be due to the strengthened resilience of a watershed system to continuous forest disturbances. This strengthened resilience normally enables the altered hydrological system to gradually recover towards the initial state if the system is not collapsed (Figure 4.31a). However, if the cumulative forest disturbances increase to an extremely high level, resulting in severe damage of watershed resilience, the changes in hydrological system become irreversible and may never return to the initial state (Figure 4.31b). For the Willow and Tulameen River watersheds, there was a downward tendency in the intensity of annual mean flow response to forest disturbances during their respective disturbed periods, which indicates that these two watersheds are in the periods with positive hydrological recovery due to watershed resilience. But when the altered hydrological system will return to the initial state is difficult to be determined in this study. The answer to this question must await the availability of more data in the future. Also, long-term continuous data will allow us to determine if the watershed resilience is damaged in the other two extremely disturbed watersheds (the Baker and Moffat Creek watersheds) in the future. Moreover, this finding that hydrological responses of forest disturbances decreased as the progression of forest disturbances over time,  131  possibly due to watershed resilience, can have an implication for future hydrological modeling. Current hydrological modeling, normally based on small-scale processes, always assume a constant relationship between forest disturbances and hydrology, which may produce misleading conclusions. As suggested in this study, there were hardly any simple linear relationships between cumulative forest disturbances and hydrological changes. In future hydrological studies, watershed resilience to watershed disturbances must be taken into account, which will provide a better modeling framework that captures some non-linear properties in large watersheds.  Reference period  Disturbed period  ∆Qf/CECA(mm/%)  ∆Qf/CECA(mm/%)  b) a)  Reference period  Disturbed period  Time  Time Figure 4.31 a) The conceptualized graph of watershed resilience to the forest disturbance-induced change in annual mean flow with a moderate to high level of disturbances; and b) with extreme high level of disturbances  4.5.6  Joint effects and relative contributions of forest disturbances and climate variability  In forest dominant watersheds, forest change and climatic variability are commonly recognized as two major drivers for hydrological changes. Understanding their interactive, dynamic effects is important for sustainable water management and for the protection of ecosystem functions and public safety. Forest disturbances and climate variability can affect annual mean flows in the same or opposite (offsetting) directions with different strengths (Figure 4.32). According to the analysis in the Willow River, Moffat Creek, and Tulameen River watersheds, forest disturbances and climatic variability produced opposite impacts on annual mean flows. Forest disturbances increased annual mean flows in these watersheds. The accumulated annual mean flow variations attributed to forest disturbances during the disturbed period in the Willow 132  River, Moffat Creek, and Tulameen River watersheds were 2503.6 mm, 299.2 mm, and 1566.6 mm, respectively. In contrast, climatic variability decreased annual mean flows. The accumulated annual mean flow variations attributed to climate variability during the disturbed period in the Willow River, Moffat Creek, and Tulameen River watersheds were -3824.6 mm , 285.9 mm, and -1668.4 mm, respectively. These counteracting or cancelling effects of forest disturbances and climate variability eventually led to a stable trend of annual mean flows over the study period in the study watersheds. The counteracting effects of forest disturbances and climate variability have also been reported in a similar study based on the long-term records in 35 small headwater basins in the United States and Canada (Jones et al., 2012).  However, in the Baker Creek watershed, both forest disturbances and climate variability produced positive effects on annual mean flows during the disturbed period from 1999 to 2009. Up to 2009, the accumulated annual mean flow variation attributed to forest disturbances was up to 236.5 mm and accumulated annual mean flow variation attributed to climate variability was 57.0 mm. This resulted in a 293.5 mm increment in annual mean flows in total, which may have positive implications for water supply. On the other hand, it can also lead to the potential risks of higher floods and threaten pubic safety. Moreover, there is another possibility that both forest disturbances and climate variability can cause the reduction of streamflow (e.g., forest recovery and climate change decrease streamflow), which may have significant negative impacts on water supply. As forests and climate continue to change in the future, their combined effects (offsetting or additive) on annual mean flows will have significant implications for watershed management and protection.  133  Positive effect Change magnitude  Change magnitude  a)  Offsetting effect ∆Qaf  ∆Qa  b)  ∆Qa ∆Qaf ∆Qac Time  Time ∆Qaf  ∆Qac ∆Qac ∆Qa Negative effect Figure 4.32 The conceptualization of possible joint effects of forest change and climate variability on annual mean flows Note: Offsetting effect means the positive effect of forest disturbances and the negative effect of climate variability; positive effect means the positive effect of forest disturbances and the positive effect of climate variability; negative effect means the negative effect of forest regeneration and the negative effect of climate variability; ota, ∆Qac, and ∆nf represent total accumulated annual mean flow variation, accumulated annual mean flow variation attributed to climate variability, and accumulated annual mean flow variation attributed to forest disturbances.  In addition to the impact directions of climate variability and forest disturbances, their strength or relative contributions to annual mean flow variations are also important and meaningful. In the Tulameen River and Willow River watersheds with moderated levels of forest disturbances (CECA 33.8% and 35.4%, respectively), forest disturbances are less influencial than climate variability. The relative contribution of forest disturbances to annual mean flow variations was about 30- 40% while the relative contribution of climate variability was about 60-70%. However, in the Baker Creek and Moffat Creek watersheds, with a high level of forest disturbances, the  134  influence of forest disturbances was slightly greater than that of climate variability. The relative contribution of forest disturbances to annual mean flow variations was about 55-58%, while the contributions of climate variability accounted for only about 42-45% (Table 4.27). Although climate variability is generally believed to be the most influential driver for streamflow variation, the findings in this study suggest that the relative contribution of forest disturbances to annual mean flows can exceed that of climate variability if forest disturbances are severe such as in the Baker Creek and Moffat Creek watersheds.  Studies from small watersheds generated similar results that the impacts of forest disturbances on streamflow can sometimes override those of climate variability. For example, in three longterm experimental forests (Andrews, Coweeta, and Hubbard Brook), increments of daily streamflow in the late summer and early fall caused by forest harvesting can be up to 300% in the early years after disturbances while climate induced changes in streamflow can be 10-50% (Jones and Post, 2004). Although similar forest disturbance-related studies in large watersheds have not been conducted to my knowledge, research on the hydrological impact of land use changes also provides similar findings. In large watersheds that experienced significant land use changes, their influence on streamflow appeared to be greater than those from climatic variability. For example, in the headwaters of the Yellow River Basin, China, only 30% of the streamflow reduction in the 1990s was caused by climate variability while land use change was responsible for 70% of the reduction (Zheng et al., 2009). The similar results were also reported in the Chaobai River watershed of China by Wang et al. (2009) and Zhang et al. (2008).  Moreover, the relative contribution of forest disturbances tended to decline over time inspite of increased forest disturbance levels in the Willow River and Tulameen River watersheds. For example, during the period of 1968-1984, the relative contribution of forest disturbances to annual mean flow variation was about 43%, which declined to 38% during the period of 19851990 and further declined to 37% in the period of 1991-2008 in the Willow River watershed. Similar result was found in the Tulameen River. The downward tendency in the relative contribution of forest disturbances over time further support the finding that the impact strength of forest disturbances can decline over time with continueous forest disturbances due to watershed resilience.  135  These studies clearly highlight the importance of understanding the effects of forest disturbances and climate variability, as well as their relative contributions in the context of climate change and global warming. Those results can support us to predict available water resources in a changing environment in the future. Accordingly, forest practices or land use planning can be designed to mitigate and adapt to the effects of climate change. In this way, the negative impacts of climate change and forest disturbances or land use change on water and aquatic functions can be effectively minimized.  Table 4.27 The relative contributions of forest disturbances and climate variability to annual mean flow variation  4.5.7  CECA  Watersheds  Rf (%)  Rc (%)  Willow  39.7±7.8  60.3±7.8  23.8  Baker  57.5±19.0  42.5±19.0  35.0  Moffat  55.1±7.5  44.9±7.5  40.0  Tulameen  33.0±5.6  67.0±5.6  16.6  (%)  Single watershed study V.S. quasi-paired watershed study  In this study, both the single watershed and quasi-paired watershed approaches were used to quantify the impacts of forest disturbances on annual mean flows. The quasi-paired watershed study has an advantage due to its ability to exclude regional-scale confounding noises (e.g., regional climate oscillation), as compared with the single watershed approach. The application of quasi-paired watershed study, however, under most circumstances, can be challenging in large watershed studies given the great difficulty to locate a suitable pair of study watersheds. This is because an eligible pair of watersheds must meet a list of requirements such as being geographically close to each other, with one less disturbed by forest disturbances and the other severely disturbed, and both watersheds with long-term hydrological and forest disturbance  136  records. In contrast, the single watershed approach, with fewer restrictions, demonstrates greater flexibility and wider applicability (Wei and Zhang 2010b). Both approaches were applied to the Willow River watershed. As seen in Table 4.28, the two approaches yielded similar estimations the average annual mean flow variations attributed to forest disturbances estimated by the quasipaired watershed study and the single watershed study were 61.6 mm and 60.7 mm, respectively. This suggests that the single watershed study is as powerful as the quasi-paired watershed study. Thus, for the regions that have difficulties in locating suitable pairs of watersheds, the single watershed approach is a sound alternative.  Table 4.28 Comparison of the quasi-paired watershed and single watershed studies on forest disturbance effect on annual mean flow (1986-1995) Approach Quasi-paired watershed Single watershed  4.5.8  ∆Q  ∆Qf  (mm)  (mm)  ∆Qf/CECA ∆Q%  ∆Qf% (%)  ∆Qf%/CECA  CECA  (mm/%)  (%)  (%)  -61.8 61.6±52.8  2.3±1.9  -14.5 14.4±12.4  0.5±0.4  20.1  -61.8 60.7±38.3  2.2±1.5  -14.5 14.3±9.0  0.5±0.3  29.4  Implication for watershed management  This study shows that forest disturbances have produced significant, positive impacts on annual mean flows in the Willow River, Tulameen River, Baker Creek and Moffat Creek watersheds. In addition, forest disturbances have significantly increased spring mean flows in the Willow River watershed, summer/winter mean flows in the Moffat Creek watershed, fall/winter mean flows in the Baker Creek and Tulameen River watersheds. From a water supply perspective, these increases can be positive and substantial for the more populated watersheds- the Baker Creek, Moffat Creek, and Tulameen River. They can be particularly important for the very dry watersheds like the Baker Creek. The average annual mean flow in the Baker Creek watershed is only 103.3 mm with great inter-annual variability, suggesting that water supply is likely constrained or stressed, especially in the dry seasons from late summer to winter. The positive  137  effects of forest disturbances on streamflow will certainly help alleviate the water supply stress within the watershed and the downstream of the watershed. But these positive effects will gradually diminish with forest regeneration over time. Resource managers must recognize these dynamic, positive effects and incorporate them into the design of sustainable water management.  Despite the possibility to reduce drought risks and enhance water supply due to increases in mean flows caused by forest disturbances, forest disturbances can produce negative effects on aquatic habitat and other watershed functions. Streams in the Willow and Baker watersheds are major spawning and rearing habitats for chinook salmon and pink salmon, while the streams in the Moffat Creek watershed are major spawning habitats for sockeye salmon. Significantly increased summer and fall mean flows may affect salmon migration and spawning due to the alteration of flow magnitude and associated water quality. More research is needed to further explore the potential impacts of mean flow changes on aquatic ecosystems.  Forest disturbances and climate variability have yielded counteracting effects on streamflow in the Willow River, Tulameen River, and Moffat Creek watersheds, which helps maintain a stable water supply in the study watersheds. However, both forest disturbances and climate variability have produced positive effects on streamflow in the Baker Creek watershed, which may potentially lead to catastrophic floods. Moreover, there is another possibility that both climate variability and forest change (e.g., forest recovery after disturbances) have negative effects on streamflow, which consequently lead to severe droughts and water shortages. Understanding these effects is particularly important in the context of climate change and global warming. The simulations predicted that an increase in annual mean flows and decrease in summer flows are expected to occur due to global warming in the interior of BC (Shrestha et al., 2012). This indicates that if the study watersheds experience on-going forest disturbances, annual mean flows are very likely to be dramatically augmented by both climate change and forest disturbances under future climate conditions. Therefore, an opportunity exists on managing forest disturbances to minimize its negative effects on water resources and watershed functions as well as to mitigate the negative impacts of climate change.  138  We have also identified the forest disturbance thresholds for detectable changes in annual mean flows in the four study watersheds that cover typical watershed types in the southern and central interior of BC. These thresholds can be used to guide forestry practices in the region as well as in other watersheds where physical properties and climate are similar. To avoid significant annual mean flow changes, the forest disturbance levels must be restricted below the forest disturbance thresholds for any given watersheds.  139  4.6 Summary Forest disturbances have produced significant impacts on annual mean flows in the Willow River, Baker River, Moffat Creek, and Tulameen River watersheds. The forest disturbance threshold in terms of CECA for detectable changes in annual mean flows varies from 15% to 55%. Forest disturbances have also yielded significant impacts on seasonal mean flows in these four watersheds. The specific seasonal flow variables altered by forest disturbances vary among watersheds.  The change magnitude of annual mean flows caused by forest disturbances vary from watershed to watershed, as well as from year to year mainly due to the differences in climate conditions. Generally, in the drier years or the drier watersheds, annual mean flow responses to forest disturbances tend to be less that those in the wetter years or wetter watersheds. However, no definitive conclusion can be drawn on the effects of temperature on annual mean flow responses to forest disturbances.  For the Willow and Tulameen River watersheds, there was a downward tendency in the strength of annual mean flow responses to forest disturbances during their respective disturbed periods. This suggests that the hydrological changes caused by current levels of forest disturbances are in the recovery process probably due to the resilience, stability, and hysteresis of watershed systems.  In the Willow River, Moffat Creek, and Tulameen River watersheds, forest disturbances and climate variability have produced counteracting or cancelling effects on streamflow, eventually leading to stable trends of annual mean flows over the study periods in these watersheds. However, both forest disturbances and climate variability have yielded positive effects on annual mean flows in the Baker Creek watershed. There is also another possibility that future forest changes (e.g., forest recovery) and climate variability can affect streamflow in a negative depending on their impact directions and strength.  140  The findings mentioned above have important implications for forest and watershed management regarding water resource allocation, flood control, forestry practice design, and ecosystem protection in the context of intensified forest disturbances and climate change. Moreover, the results on hydrological responses to forest disturbances along climate gradients can support future modeling.  141  5 Chapter: Impacts of forest disturbances on flow regimes 5.1 Background Flow regimes including five hydrological components (magnitude, duration, timing, frequency, and variability or change rate of flows) play a critical role in determining the structures and functions of aquatic, floodplain, and riparian ecosystems (Poff et al., 1997 and 2010a; Pettit et al., 2001). Flow regimes can affect aquatic ecosystems directly by regulating the life cycle, distribution, abundance, and diversity of aquatic and riparian species (Sparks, 1995; Greenberg et al., 1996; Marchetti et al., 2001), and can indirectly influence aquatic habitat and ecosystem functions by shaping the geomorphologic features of channels and floodplains (Dunne and Leopold, 1978; Greenberg et al., 1996; Reeves et al., 1996). Thus, the alteration of flow regimes can negatively affect the diversity and abundance of aquatic and riparian species, channel morphological attributes, and many other ecosystem functions (Poff and Allan, 1995; Sparks, 1992; Stanford et al., 1996).  Flow regimes in forested watersheds can be altered by forest disturbances in both direct and indirect ways. Forest disturbances such as logging can have direct impacts on hydrology through the modifications of forest hydrological processes such as canopy interception, soil infiltration, evapotranspiration or interactions between groundwater and surface water (Stednick, 1996; Bruijnzeel, 2004). Forest disturbances can also influence channel morphology via sediment transportation and debris flows, which further lead to the changes of flow regimes in an indirect way (Benda et al., 2003, 2004; Moore and Wondzell, 2005). Thus, forest disturbances may affect flow regimes and geomorphology, and consequently lead to degraded aquatic habitat, biodiversity, and other aquatic functions.  Existing studies on the impacts of forest disturbances on flow regimes have mainly targeted small watersheds (<100 km2), with rare examinations of large watersheds. Among those existing studies, they mainly focused on one type of forest disturbance (Bethlahmy, 1974; Bruijnzeel, 2004; Woodsmith et al., 2004; Stednick, 2008), with limited flow regime components. These limitations may lead to incomplete understanding of forest disturbances and their effects on flow regimes. This is particularly true in large watersheds where various types of forest disturbances 142  exist and they are accumulative over space and time. Thus, to fully understand the relationship between forest disturbances and flow regimes at large watersheds, it is critical to consider various types of forest disturbances and their impacts on all flow regime components. To my knowlege, there is no study examining the quantitative impacts of cumulative forest disturbances on a full spectrum of flow regimes in large watersheds to date. However, a comprehensive understanding of forest disturbance-induced alteration of flow regime is essential for water resource and forest managers to design management strategies for long-term water supply and the protection of watershed ecosystem functions. This is particularly important as forest disturbances (e.g., insect infestation and wildfire) are becoming more frequent and catastrophic due to climate and land use changes. Clearly, insufficient research, together with the growing demand on this type of information highlights a significant research gap on the relationship between forest disturbances and flow regimes in large watersheds. In this study, six large watersheds (>500 km2) along environmental gradients were selected in the BC interior to examine the impacts of forest disturbances on flow regimes. These watersheds are the Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River watersheds. The Baker and Moffat Creek watersheds have experienced extremely severe forest disturbances including logging, mountain pine beetle (MPB), and fire in the last 10 years, with CECA of 62.2 % and 65.7%, respectively. The Willow and Tulameen River watersheds, with long disturbance histories, have progressively experienced substantial forest disturbances since the 1950s, even though they had lower disturbance levels (with CECA of 35.4 % and 33.8%, respectively) than those from the Baker and Moffat Creek watersheds. The Cottonwood and Ashnola River watersheds were less disturbed with CECA accounting for 12.2 % and 6.4%, respectively. The availability of long-term data on hydrology, climate, and forest disturbances in those selected watersheds offers an excellent opportunity of examining how forest disturbances affect flow regimes in the BC interior.  The methodology that combined advanced statistical analysis (e.g., time series cross-correlation analysis) with graphical methods (e.g., flow duration curve) was firstly developed to test the statistical significance of the cause-effect relationships between cumulative forest disturbances and flow regimes in six study watersheds. Then for those watersheds with significant correlations  143  between forest disturbances and flow regimes, the paired-year approach was further applied to quantitatively assess the impacts of forest disturbances on flow regimes.  The objectives of this study were (1) to assess the impacts of cumulative forest disturbances on various components of high flow regimes (magnitude, frequency, timing, duration, and variability) (see the definitions in the Method section); (2) to assess the impacts of cumulative forest disturbances on all five components of low flow regimes; and (3) to discuss the potential ecological implications of the altered flow regimes.  144  5.2 Data 5.2.1  Hydrological data  As mentioned in Chapter 3, daily flow data for six large watersheds (Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River) were obtained from Water Survey Canada. Those daily flow data were used to develop yearly flow duration curves (FDCs). The time series of high and low flows for each watershed were then derived and the metrics of flow regimes (the magnitude, duration, timing, frequency, and variability of both high and low flows) were calculated accordingly. Detailed information on hydrometric stations and their data descriptions for the study watersheds can be found in Chapter 4. 5.2.2  Climatic data  Climate data such as mean, maximum, and minimum temperatures and precipitation at daily and monthly scales are required in the quantification of forest disturbance-induced impacts on flow regimes. In consideration of data availability from active climate stations, both statistical and quantitative assessments were conducted only in the Willow River and Baker Creek watersheds, while only statistical assessments on the impacts of forest disturbances on flow regimes for other study watersheds were conducted. The Prince George airport climate station (Climate ID: 1096450, elevation: 691 m), with records dating back to 1947, is located near the outlet of the Willow River watershed, while the Barkerville station (Climate ID: 1090660, elevation: 1283 m), with records since 1888, is in the southern mountainous areas. Data including daily mean, maximum, and minimum temperatures, and daily precipitation from these two climate stations were collected and averaged to represent the climate in the Willow River watershed. Daily and monthly climate data used in the Baker Creek watershed were obtained from the Quesnel airport climate station (Climate ID: 1096630, Elevation: 545 m). This station, located approximately 2 km north of the outlet of the Baker Creek watershed, was constructed in 1953 and then replaced by Quesnel airport auto station (Climate ID: 1096631) after 2006. 5.2.3  Forest disturbances and geographic data  Forest disturbances and geographic data used in this chapter are the same as in the mean flow analysis of Chapter 4. 145  5.3 Methods 5.3.1  Quantification of forest disturbance level  See Chapter 4 for more details. 5.3.2  Definitions of flow regime components and studied hydrological variables  From an ecological perspective, each flow regime component has different implications for ecological processes in aquatic ecosystems (Poff et al., 1997). The specific definitions of these five components of flow regimes are given in Table 5.1. In this study, we focused on high flows and low flows (see the next paragraph for their definitions) since they are important hydrologic variables that affect the integrity of river ecosystems (Wolman and Miller, 1960; Stromberg et al., 2007; Poff et al., 1997, 2010a,b).  The flow duration curve (FDC), a cumulative distribution function of daily flows over a time interval of interest, for example, monthly and annually, shows the percentage of time that streamflow equals or exceeds a given amount. As a frequently used hydrograph, FDC can be used to describe flow regimes of a watershed at a given point of time (Brown et al., 2005). In this study, flow duration curves for each year were generated using daily flows, while flows at a given percentile (denoted as Qp%) were derived. Here, high flows are defined as the flows that equal to or are greater than Q5% or Qh (Q5%: flows exceeded at 5% of the time in a given year), while low flows refer to the flows that are equal to or less than Q95% or Ql (Q95%: flows exceeded at 95% of the time in a given year). Based on those definitions, annual data series of the five components of flow regimes for both high and low flows were generated.  A flood frequency analysis using maximum daily flow data during the low disturbance period when the effects of forest disturbances on streamflow were assumed minor or ignorable (with CECA≤10%) was conducted for each study watershed. According to the forest disturbance levels in terms of CECA, the periods used to estimate the flood frequency distributions for the Willow River, Cottonwood River, Baker Creek, Moffat Creek, Tulameen River, and Ashnola River were 1953-1972, 1955-1992,1964-1989,1967-1989, 1955-1981, and 1954-2009, respectively. The best-fitted distribution adopted in this study is Log-Pearson Type III. Once the flood frequency 146  distribution was generated for each watershed, the frequencies and return periods of high flows for the study period were estimated accordingly. Similarly, by use of minimum daily flow data from the same period as in the flood frequency analysis, low flow frequency analysis was performed. The frequencies and return periods of low flows for study period were also calculated.  Table 5.1 Definitions for flow regime components Components Magnitude  Definitions The amount of water moving past the watershed outlet per day (m3/s); High flows: Daily flows≥Q5% ; Low flows: Daily flows ≤ Q95%  Timing  The time for a given flow event (e.g. annual peak flow) that occurs regularly. It is also referred to as the predictability of a given flow event; High flows: average date for the occurrence of high flows in a water year (e.g., the 200th day in a water year); Low flows: average date for the occurrence of low flows in a water year.  Duration  The period of time for given flow events (e.g. high flows) last; High flows: the number of days with daily flows equal to or greater than high flow threshold (High flow threshold: the median of high flows during the less disturbed period); Low flows: the number of days with daily flows equal to or less than high flow threshold (low flow threshold: the median of low flows during the less disturbed period).  Frequency  How often a flow above a given magnitude recurs above or below a specified time interval, for example, a 50-yr flood is equaled to or exceeded once every 50 yrs on average and a 50-yr low flow is equaled to or below once every 50 yrs on average; High flows: the return period of high flows; Low flows: the return period of low flows;  Variability  The variations of flows, describing how spread out or closely clustered of flow magnitude; Coefficient of variation is often used to measure variability for a set of data; High flows: Coefficient of variation for high flows at each year (the coefficient of variation measures variability in relation to the mean (or average) and is used to compare the relative dispersion in one type of data with the relative dispersion in another type of data, a measure of dispersion); Low flows: Coefficient of variation for low flows at each year. 147  5.3.3  Statistical test of cause-effect relationships between forest disturbances and flow regimes  For a large forested watershed, climatic variability and forest disturbances are two primary drivers for hydrological variations. A direct examination of trends in hydrological variables unlikely distinguishes the impacts of forest disturbances. A method with the ability to detect how hydrological variables vary with changes of forest disturbance levels is needed. In this study, time series cross-correlation was performed to detect the statistical significance of the causeeffect relationships between CECA data series and hydrological variables (magnitude, timing, duration, frequency, and variability of high flows and low flows on an annual scale) in each study watershed. The time series cross-correlation is found to be an effective approach to investigate the causality among environmental variables since it can not only address the autocorrelation issue in data series but also identify the lagged causality between two data series with a plausible causal relation (Chatfield, 1989; Jassby and Powell, 1990; Lin and Wei, 2008; Zhou et al., 2010). All the hydrological data series along with CECA data series were prewhitened first to remove the autocorrelations by fitting the Autoregressive Integrated Moving Average (ARIMA) models (Box and Jenkins, 1976). The white noises or model residuals of selected ARIMA models with best performance in terms of their achievements of model stationarity and coefficient of determination (R2) were used in the time series cross-correlation analysis to test the statistical significance of the causal relationships between the CECA data series and each hydrological variable.  5.3.4  Quantitative assessment on the impact of forest disturbances on high flow and low flow regimes  The watersheds with significant changes in flow regimes as a result of forest disturbances were further analyzed to quantitatively assess the impacts of forest disturbances on flow regimes. In a large forested watershed, climatic influence is typically dominant, and always obscure the possible hydrological effects of other variables (Cong et al., 2009). To quantify the impacts of forest disturbances on high and low flows, the effects of climate variability must be excluded. To accomplish this objective, the paired-year approach was developed. In the paired-year approach,  148  a year in the reference period (namely the reference year) was paired with its comparable year in the disturbed period (namely the disturbed year) according to their similarities in climate conditions. To achieve this, the first step was to determine the climate variables that were mostly related to both high and low flows in each watershed. The Kendall tau correlation analysis was applied first to test the correlations between the components of high flows (Mh) and possible relevant climatic variables (annual precipitation P, winter snowpack Pw, March accumulated snowmelt temperature Ts3, April accumulated snowmelt temperature Ts4, antecedent accumulated snowmelt temperature Ts, and annual mean/maximum/minimum temperature (Tave, Tmax, Tmin)). Climate variables that were significantly correlated with high flows were identified. The climatic variables of significant correlations with the components of low flows (Ml) were also identified in the same way.  The above climatic variables with the significant correlations with high flows were then combined with those identified for low flows in each watershed. However, for climatic variables having significant correlations with each other, only the one mostly related to high flows or low flows was retained each time in order to address the multicollinearity problem. This resulted in several sets of climatic variables relevant to high flows and low flows for each study watershed (Table 5.2 and 5.3). These sets of climatic variables were then used in the canonical correlation analysis (Table 5.2 and 5.3). Unlike the Kendall tau correlation analysis that tests the correlation between two variables, the canonical correlation analysis is able to test the correlation between two sets of variables. Thus, the canonical correlation analysis was conducted here to detect the overall correlation between the set of hydrological variables (high flows and low flows) and each set of climatic variables (Table 5.2 and 5.3). The set of climatic variables that were mostly correlated with the set of hydrological variables was finally found. Using this process, the identified climatic variables were determined and then were used as the controlling climatic factors in the pairing.  According to the Kendall tau correlation analysis in the Willow River watershed, low flows were significantly correlated with annual precipitation (P) and annual winter snowpack (Pw), while high flows were significantly correlated with annual precipitation (P), winter snowpack (Pw), March accumulated snowmelt temperature (Ts3), April accumulated snowmelt temperature (Ts4),  149  antecedent accumulated snowmelt temperature (Ts), and annual mean/maximum/minimum temperatures (Tave, Tmax, Tmin). These variables formed 12 sets of climatic variables for canonical correlation analysis. As suggested by the canonical correlation analysis, climatic variables including P, Tave, and Ts were selected as the controlling climatic variables for the pairing in the Willow River watershed. Similarly, climatic variables including Pw, Tmax, and Ts were identified as the controlling climatic variables for the pairing in the Baker Creek watershed.  Table 5.2 Canonical correlation analysis between the set of hydrological variables and the set of climatic variables in the Willow River watershed Canonical R  Climate Variable Sets  Hydrological Variable Set (Mh, Ml)  Set 1(P, Tave, Ts)  0.80(p<0.001)  Set 2(P, Tmax, Ts)  0.79(p=0.001)  Set 3(P, Tmin, Ts)  0.78(p<0.001)  Set 4(P, Tave, Ts3, Ts4)  0.78(p=0.004)  Set 5(P, Tmax, Ts3, Ts4)  0.78(p=0.008)  Set 6(P, Tmin, Ts3, Ts4)  0.78(p=0.003)  Set 7(Pw, Tave, Ts)  0.74(p<0.001)  Set 8(Pw, Tmax, Ts)  0.74(p<0.001)  Set 9 (Pw, Tmin, Ts)  0.75(p<0.001)  Set 10(Pw, Tave, Ts3, Ts4)  0.74(p=0.002)  Set 11(Pw, Tmax, Ts3, Ts4)  0.75(p=0.001)  Set 12(Pw, Tmin, Ts3, Ts4)  0.74(p=0.001)  Note: P and Pw are annual precipitation and annual winter snowpack (the total precipitation from November to April), respectively.Ts3, Ts4, and Ts are March accumulated snowmelt temperature, April accumulated snowmelt temperature, and antecedent accumulated snowmelt temperature (snowmelt temperature refers to the temperature over 0°C), respectively. Tave, Tmax, and Tmin are annual mean/maximum/minimum temperatures, respectively.  150  Table 5.3 Canonical correlation analysis between the set of hydrological variables and the set of climatic variables in the Baker Creek watershed Canonical R  Climate Variable Sets  Hydrological Variable Set (Mh, Ml)  Set 1(P, Tave, Ts)  0.60(p<0.001)  Set 2(P, Tmax, Ts)  0.60(p=0.001)  Set 3(P, Tmin, Ts)  0.59(p<0.001)  Set 4(P, Tave, Ts3, Ts4)  0.59(p=0.004)  Set 5(P, Tmax, Ts3, Ts4)  0.59(p=0.008)  Set 6(P, Tmin, Ts3, Ts4)  0.59(p=0.003)  Set 7(Pw, Tave, Ts)  0.68(p<0.001)  Set 8(Pw, Tmax, Ts)  0.69(p<0.001)  Set 9 (Pw, Tmin, Ts)  0.67(p<0.001)  Set 10(Pw, Tave, Ts3, Ts4)  0.67(p=0.002)  Set 11(Pw, Tmax, Ts3, Ts4)  0.67(p=0.001)  Set 12(Pw, Tmin, Ts3, Ts4)  0.67(p=0.001)  Note: Refer to the denotes in Table 5.2.  Therefore, for the eligibility of the pairing in the Willow River watershed, each pair of years must have similar annual precipitation, antecedent snowmelt temperature, and mean temperature, where 10% biases are allowed. In the Baker Creek watershed, each pair of years must have similar annual winter snowpack, antecedent snowmelt temperature, and maximum temperature. In addition, monthly mean/maximum /minimum temperatures and monthly precipitation in any reference year in each pair must be insignificantly different from those in its counterpart according to a nonparametric test (Mann-Whitney U test). This ensures a more reliable pairing. According to those criteria, twelve and nine pairs were identified in the Willow River and Baker Creek watersheds, respectively. Detailed information regarding the selected pairs is provided in Tables 5.4 and 5.5. In this way, the climatic effects were excluded, and the reference year and the disturbed year in each pair were comparable. Hence, the differences in high flows or low flows between the reference year and disturbed year in each pair were accounted as the effects of forest disturbances on flow regimes of high or low flows. The Mann-Whitney U test was performed to  151  further confirm the statistical significance of the disturbance effects by comparing hydrological variables in the reference years with those in the disturbed years.  Table 5.4 Selected pairs for the Willow River watershed Reference  Disturbed  P  Ts  Tave  CECA  Year  Year  (mm)  (°C)  (°C)  (%)  1957  Reference  923.1  161.8  2.2  0.5  1  1972  Disturbed  855.2  145.5  1.9  9.0  2  1974  Disturbed  926.1  151.8  3.0  12.0  1958  Reference  777.1  162.7  5.4  0.5  1970  Disturbed  715.0  184.2  4.5  5.4  1959  Reference  995.1  168.9  3.1  0.5  1974  Disturbed  926.1  151.8  3.0  12.0  1960  Reference  882.0  205.2  3.8  1.0  2007  Disturbed  873.5  192.1  4.6  33.8  1961  Reference  765.2  195.3  4.7  1.8  6  1970  Disturbed  715.0  184.2  4.5  5.4  7  1977  Disturbed  807.3  231.2  5.2  15.1  8  1991  Disturbed  839.7  204.0  3.5  29.9  9  1993  Disturbed  790.4  194.0  3.7  31.8  1964  Reference  1104.7  118.8  3.9  3.1  1982  Disturbed  1108.8  124.5  2.9  20.6  1966  Reference  885.2  156.4  3.2  3.5  1974  Disturbed  926.1  151.8  3.0  12.0  1967  Reference  849.8  99.1  4.0  3.7  2008  Disturbed  926.6  100.0  3.8  35.4  Pair#  3  4  5  10  11  12  Note: Refer to the denotes in Table 5.2.  152  Table 5.5 Selected pairs for the Baker Creek watershed Pair# 1  2  3  4  5  6  7  8  9  Reference  Disturbed  Pw  Tmax  Ts  CECA  Year  Year  (mm)  (°C)  (°C)  (%)  1964  Reference  218.1  8.2  267.3  0.5  2009  Disturbed  222.6  8.7  250.2  62.2  1968  Reference  249.6  8.3  348.6  0.5  2007  Disturbed  256.7  8.6  370.0  48.0  1971  Reference  236.7  7.8  248.6  0.4  2009  Disturbed  222.6  8.7  250.2  62.2  1975  Reference  205.5  7.7  194.6  1.1  2008  Disturbed  195.1  8.3  218.9  55.3  1979  Reference  221.9  7.6  263.8  6.4  2009  Disturbed  222.6  8.7  250.2  62.2  1988  Reference  182.6  9.2  436.2  10.0  2003  Disturbed  216.1  9.7  444.3  25.5  1989  Reference  176.1  8.9  368.2  10.4  1999  Disturbed  193.4  8.9  393.0  19.2  1991  Reference  242.7  8.2  351.3  12.6  2007  Disturbed  256.7  8.6  370.0  48.0  1993  Reference  223.4  8.2  374.2  15.2  2007  Disturbed  256.7  8.6  370.0  48.0  Note: Refer to the denotes in Table 5.2.  153  5.4 Results 5.4.1  Statistical relationships between forest disturbances and high flow regimes  As suggested by the time series cross-correlation analysis (Table 5.6), with extremely severe forest disturbances, significant impacts on most components of high flow regimes were detected in the Baker and Moffat Creek watersheds. In comparison, the significant impacts on the high flow regimes in the Willow River watershed were only found on two regime components, while significant impacts on any flow regime components in the Tulameen River watershed were not identified. In the two least disturbed watersheds (the Cottonwood and Ashnola River watersheds), one flow regime component was significantly related to forest disturbances (CECA): timing (TMh) in the Cottonwood River watershed and variability (CVh) in the Ashnola River watershed (Table 5.6). These results generally suggest that the more severe forest disturbances, the greater alternations to the flow regimes of high flows in the study region.  Table 5.6 also shows that the CECA date series were significantly and positively correlated with the magnitude (Mh) of high flows in the Baker Creek, Moffat Creek, and Willow River watersheds, suggesting that the magnitude of high flows in these watersheds can be significantly augmented by forest disturbances. Significant negative correlations between the timing (TMh) of high flows and the CECA time series were found in the Baker Creek, Moffat Creek, and Cottonwood River watersheds (Table 5.6). In other words, forest disturbances can significantly advance the timings of high flows in these watersheds. The durations (DRh) of high flows in the Moffat Creek and Willow River watersheds were significantly and positively related to the CECA data series, showing that forest disturbances can result in a longer period of high flows in these watersheds. Significant positive correlations between the return periods (RPh) of high flows and the CECA time series were detected in the Baker Creek and Moffat Creek watersheds, indicating that the occurrence of large floods can become more frequent as a result of severe forest disturbances in these watersheds. The variability (CVh) of high flows was significantly and positively correlated with the CECA data series in the Baker Creek, Moffat Creek, and Ashnola River watersheds (Table 5.6). This suggests that high flows can become more variable and dispersed as a result of forest disturbances in these watersheds.  154  Table 5.6 Cross-correlations between forest disturbance levels (CECA) and high flow regimes Watersheds  Baker  Moffat  Cottonwood  Willow  Tulameen  Ashnola  ARIMA  Mh  TMh  DRh  RPh  CVh  0.32*  -0.35*  0.29  0.41*  0.35*  (1, 1, 1) non-  (lag=2)  (lag=2)  (lag=1)  (lag=1)  (lag=0)  constant  0.33*  -0.33*  0.35*  0.40*  0.42*  (lag=3)  (lag=0)  (lag=3)  (lag=2)  (lag=3)  -0.18  -0.31*  -0.17  -0.23  -0.16  (lag=2)  (lag=1)  (lag=2)  (lag=1)  (lag=2)  0.28*  -0.15  0.29*  -0.11  -0.21  (lag=2)  (lag=0)  (lag=2)  (lag=0)  (lag=0)  -0.19  -0.12  -0.15  0.12  0.18  (lag=0)  (lag=0)  (lag=0)  (lag=0)  (lag=0)  -0.15  -0.06  -0.22  -0.08  0.30*  (lag=0)  (lag=0)  (lag=0)  (lag=1)  (lag=0)  model  Ln(x),(4,2,0)  (1,1,0)  (2,1,0)  (2,2,0)  (1,1,0)  Note: *, Significant at α=0.05; Mh, TMh, DRh, RPh, and CVh refer to the magnitude, timing, duration, return period, and variability of high flows, respectively.  5.4.2  Statistical relationships between forest disturbances and low flow regimes  According to the time series cross-correlation analysis, the impacts of forest disturbances on the components of low flow regimes in the study watersheds were less pronounced than the impacts on high flow regimes. As shown in Table 5.7, only in the Baker Creek watershed was a significant and positive correlation between the magnitude (Ml) of low flows and the CECA detected, indicating forest disturbances can increase low flows in this watershed. Forest disturbances yielded insignificant impacts on the timing (TMl) of low flows in all watersheds. The duration (DRl) of low flows in the Cottonwood River watershed was significantly and positively correlated with the CECA, suggesting forest disturbances can cause a longer period of low flows below the threshold in this watershed. Interestingly, significant and positive correlations between the return periods (RPl) of low flows and forest disturbance levels were identified in the Moffat Creek and Cottonwood River watersheds, while a significant and negative relationship was found in the Willow River watershed. The variability (CVl) of low  155  flows was significantly and negatively related to the forest disturbance levels in the Baker Creek, while this relationship was positive in the Tulameen River watershed. Clearly, compared with the impacts of forest disturbances on high flows, the impacts on low flows were less pronounced and inconsistent, with some changes even being on opposite directions for some regime components. This suggests that changes of low flow regimes may be more complicated as they are influenced by not only forest disturbances but also soils and aquifers.  Table 5.7 Cross-correlations between forest disturbance levels (CECA) and low flow regimes Watersheds Baker  Moffat  Cottonwood  Willow  Tulameen  Ashnola  Ml  TMl  DRl  RPl  CVl  0.32*  0.15  -0.14  -0.22  -0.33*  (lag=1)  (lag=2)  (lag=1)  (lag=4)  (lag=2)  -0.22  0.21  0.13  0.31*  -0.11  (lag=0)  (lag=3)  (lag=3)  (lag=0)  (lag=0)  -0.21  -0.25  0.33*  0.47*  -0.22  (lag=3)  (lag=0)  (lag=2)  (lag=2)  (lag=2)  0.26  0.26  -0.25  -0.36*  -0.07  (lag=0)  (lag=1)  (lag=0)  (lag=0)  (lag=0)  -0.13  0.27  0.12  0.14  0.30*  (lag=1)  (lag=2)  (lag=1)  (lag=0)  (lag=1)  -0.08  -0.19  -0.16  -0.14  0.25  (lag=0)  (lag=2)  (lag=1)  (lag=2)  (lag=1)  ARIMA model (1, 1, 1) non-constant  Ln(x),(4,2,0)  (1,1,0)  (2,1,0)  (2,2,0)  (1,1,0)  Note: *, Significant at α=0.05; Ml, TMl, DRl, RPl, and CVl are the magnitude, timing, duration, return period, and variability of low flows, respectively.  156  5.4.3  Quantitative assessment on the changes in high and low flow regimes caused by forest disturbances  5.4.3.1 Baker Creek watershed 5.4.3.1.1 High flows  Magnitude As illustrated in Figure 5.1 and Table 5.8, the magnitude of high flows in the disturbed years was significantly higher than those in the reference years. For example, in the reference year of 1968, the averaged magnitude of high flows was 29.4 m3/s (ranging from 26.8 to 32.6 m3/s), while in its comparable year of 2007, with 48% CECA, this value was greatly increased by about 116% (63.4 m3/s) (Figure 5.2). Similarly, for the 1964-2009 pair, the averaged magnitude of high flows was augmented by 136% due to forest disturbances at the level of 62.2% CECA. In addition, the smaller-sized high flows had greater responses. For example, in the reference year of 1989, the average magnitude of high flows was 7.4 m3/s while in its paired year of 1999, with 19.2% CECA, this value was greatly increased by about 530% (Figure 5.2). Based on the calculation of all selected pairs, the average magnitude of high flows was increased by 154.3% due to the average difference of 41.1% CECA (Table 5.9). That is, per 10% CECA can averagely lead to a 92.7% increment in the magnitude of high flows.  In order to explore how the responses of high flows to forest disturbances changed with the return periods, we summarized the results by the return periods of high flows. As shown in Table 5.9, the high flows with the return periods equal to or less than 1 year were increased by up to 285.4% due to the average difference of 12.1% CECA or a 311.5% increment in high flows per 10% CECA. However, the high flows with the return periods of 1-2 years were increased by 143.7%, attributable to the average difference of 49.5% CECA (a 29.7% increment in high flows per 10% CECA). The increment was greatly reduced in the high flows with the return periods of 5-10 years, with an 8.2% increment in high flows per 10% CECA. In short, the responses of high flows to forest disturbances decreased with increasing return periods of high flows, suggesting that forest disturbances can produce more pronounced impacts on the magnitude of small-sized high flows, while its impacts on the magnitude of large-sized high flows are lower. 157  120 Median  25%-75%  Min-Max  Magnitude-h (m3 /s)  100  80  60  40  20  0  Reference Year  Disturbed Year  Figure 5.1 Comparison on the magnitude of high flows in the reference years and the disturbed years  Table 5.8 The Mann-Whitney U test on the differences in flow regimes between the paired reference and disturbed years in the Baker Creek watershed Variables  Z statistics  Mh  -11.8*(p<0.001)  TMh  9.2*(p<0.001)  RPh  -11.8*(p<0.001)  CVh  -7.6*(p<0.001)  Ml  -13.4 *(p<0.001)  CVl  12.5*(p<0.001)  CECA  -15.2*(p<0.001)  Note: *, Significant at α=0.05; Mh, TMh, RPh, and CVh are the magnitude, timing, return period, and variability of high flows, respectively, while Mh and CVl are the magnitude and variability of low flows, respectively. CECA is the cumulative equivalent clear-cut area %.  158  1  a)  1989 1999  Discharge (m3/s)  a)  0.4 0.2 99.5  98.9  99.5  97.8  97.3  96.7  96.2  98.4  2007  98.9  0 10 20 30 40 50 60 70 80 90 100  98.4  95.1  0  97.8  20  97.3  40  96.7  60  1968  b)  96.2  80  1968 2007  Discharge (m3/s)  b)  1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0  95.6  95.1  0  120 Discharge (m3/s)  1999  0.6  0 10 20 30 40 50 60 70 80 90 100 % of time over  100  1989  0.8  95.6  Discharge (m3/s)  90 80 70 60 50 40 30 20 10 0  1  c)  1964 2009  c)  0.8 0.6 0.4 1964  0.2  2009 99.5  98.9  98.4  97.8  97.3  96.7  96.2  95.6  0  0 10 20 30 40 50 60 70 80 90 100 % of time over  95.1  90 80 70 60 50 40 30 20 10 0  Discharge (m3/s)  Discharge (m3/s)  % of time over  Figure 5.2 Flow duration curves for the selected pairs in the Baker Creek watershed: a) 1989-1999; b) 1968-2007; and c) 1964-2009  159  Table 5.9 The disturbance effects on the magnitude of high flows (Mh) in the Baker Creek watershed ∆ Mh  ∆ M h%  ∆CECA  Normalized ∆Mh%  (m3/s)  (%)  (%)  (%/10%CECA)  ≤1  21.9  285.4  12.1  311.5  1-2  35.4  143.7  49.5  29.7  2-5  24.8  73.3  50.6  15.6  5-10  24.0  47.0  57.8  8.2  Average  28.8  154.3  41.1  92.7  Return Period  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆Mh is the absolute difference in the magnitude of high flows between the paired disturbed and reference years; ∆Mh% is the relative difference in high flows between the paired disturbed and reference years (∆Mh%=(Mhd - Mhr)*100%/Mhr; Mhd and Mhr are the magnitude of high flows in the paired disturbed and reference years, respectively)  Timing As illustrated in Figure 5.4 and Table 5.8, the timing of high flows in the disturbed years was significantly lower than that in the reference years. For example, in the reference year of 1964, the average timing of high flows was the 192nd day, while in its paired year of 2009, with 48% CECA, the timing (174th day) was greatly advanced by about 18 days. Similarly, for the 19891999 pair, the average timing of high flows was advanced by 5 days due to forest disturbances at the level of 19.2% CECA. Based on the calculation of all selected pairs, the timing of high flows was, on average, advanced by about 9 days or 2 days per 10% CECA due to the average difference of 41.1% CECA (Table 5.10). Table 5.10 also demonstrated how the responses of high flows to forest disturbances changed with the return periods. For example, the high flows with the return periods equal to or less than 2 years were advanced by about 2 days per 10% CECA, while the high flows with the return periods of 2-5 years were advanced by 2.4 days per 10% CECA. Similar results were found for the high flows with the return periods of 5-10 years.  160  220 Median  25%-75%  Min-Max  Timing-h (day)  210  200  190  180  170  160  Reference Year  Disturbed Year  Discharge (m3/s)  Figure 5.3 Comparison on the timing of high flows in the paired reference and disturbed years in the Baker Creek watershed  100  18 days  a)  1964 2009  80 60 40 20 0  01-Oct  01-Sep  01-Aug  01-Jul  01-Jun  01-May  01-Apr  01-Mar  01-Feb  01-Jan  01-Dec  01-Nov  Date  Discharge (m3/s)  100  5 days  b)  80  1989 1999  60 40 20 0  01-Oct  01-Sep  01-Aug  01-Jul  01-Jun  01-May  01-Apr  01-Mar  01-Feb  01-Jan  01-Dec  01-Nov  Date  Figure 5.4 Hydrographs for the paired years with similar climates to show advancing of high flows after forest disturbances in the Baker Creek watershed: a) 1989-1999; and b) 1964-2009 161  Table 5.10 The disturbance effects on the timing of high flows (TMh) in the Baker Creek watershed Return  Normalized ∆TMh  ∆TMh(days)  ∆CECA(%)  ≤1  -1.1  12.1  -2.0  1-2  -9.8  49.5  -2.0  2-5  -12.8  50.6  -2.4  5-10  -13.0  57.8  -2.3  Average  -8.7  41.1  -2.1  Period  (days/10%CECA)  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆TMh is the difference between the timing of high flows in the disturbed year and that in the reference year.  Frequency As illustrated in Figure 5.5 and Table 5.8, the return periods of high flows in the disturbed years were significantly greater than those in the reference years. For example, in the reference year of 1964, the average return period of high flows was 1.6 years while in its paired year of 2009, with 62.2% CECA, the average return period was greatly increased to 9.3 years. Similarly, for the 1989-1999 pair, the average return period of high flows was from 0.8 year to 5.0 years with CECA increasing to 19.2%. Based on the calculation of all selected pairs, the averaged return period of high flows was increased by 11.1 years due to the average difference of 41.1% CECA or a 3-year increase per 10% CECA (Table 5.11). As shown in Table 5.11, however, the increments in the return periods of high flows per 10% CECA did not change with increasing the return periods, ranging from 2.7 to 3.4 years.  162  120 Median  25%-75%  Min-Max  Return Period-h (years)  100  80  60  40  20  0  Reference Year  Disturbed Year  Figure 5.5 The comparison on the return periods of high flows in the paired reference and disturbed years in the Baker Creek watershed  Table 5.11 The disturbance effects on the return periods of high flows (RPh) in the Baker Creek watershed Return  Normalized ∆RPh  ∆RPh(yrs)  ∆CECA(%)  ≤1  3.0  12.1  3.4  1-2  13.1  49.5  2.7  2-5  13.8  50.6  3.1  5-10  17.6  57.8  3.0  Average  11.1  41.1  3.0  Period  (years/10%CECA)  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆RPh is the difference between the return periods of high flows in the disturbed year and that in the reference year.  163  Variability As illustrated by Figure 5.6 and Table 5.8, the averaged CV of high flows in the disturbed years was significantly higher than that in the reference years, suggesting greater variations in the high flows during the disturbed years. For example, in the reference year of 1964, the CV of high flows was 7% while in its paired year of 2009, with 62.2% CECA, the CV greatly increased to 34%. Similarly, for the 1968-2007 pair, the CV of high flows increased from 5% to 51% due to the forest disturbances difference of 48% CECA. In addition, their changing patterns were different (Figure 5.7). The pattern of the high flows in the reference years was stable, while more fluctuated in the disturbed years. A similar pattern was found in the flow duration curves of these two pairs in Figure 5.2. Based on the calculation of all selected pairs, the averaged CV of high flows was increased by 324.2% due to the average difference of 41.1% CECA or an 84.4% increase per 10% CECA (Table 5.12). Clearly, forest disturbances significantly increased the variability of high flows in the study watershed.  100 Median  25%-75%  Min-Max  CV-h (%)  80  60  40  20  0  Reference Year  Disturbed Year  Figure 5.6 The comparison of the variability (CV) of high flows in the paired reference and disturbed years in the Baker Creek watershed  164  120  120  a)  b) 100 Discharge (m3/s)  Discharge (m3/s)  100  1968 2007  80 60 40 20  1964 2009  80 60 40 20  0  0  1 2 3 4 5 6 7 8 9 101112131415161718  1 2 3 4 5 6 7 8 9 101112131415161718  Figure 5.7 High flows for the paired years in the Baker River watershed: a) 1976-2007 and b) 19642009 (According to the definitions in Table 5.1; high flows contain 18 data points each year)  As shown in Table 5.12, the responses of the CV of high flows to forest disturbances did not show a consistent tendency with the increasing return periods. The greatest response of variability (CV) was found in the high flows with the return periods of 1-2 years, which was a 132.9% increase in the CV per 10% CECA. However, the high flows with the return periods of 5-10 years produced the smallest response, which was only a 1.7% increase in the CV per 10% CECA.  Table 5.12 The disturbance effects on the variability of high flows (CVh) in the Baker Creek watershed Return Period  ∆CVh% (%)  ∆CECA  Normalized ∆CVh% (%/10%CECA)  (%) 76.3 12.1 43.1 ≤1 536.5 49.5 132.9 1-2 245.6 50.6 54.8 2-5 10.9 57.8 1.7 5-10 324.2 41.1 84.4 Average Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆CVh% is the relative difference in CV of high flows between the disturbed and reference years (∆CVh%=(CVhd CVhr)*100%/CVhr; CVhd and CVhr are the CV of high flows in the disturbed and reference years, respectively).  165  5.4.3.1.2 Low flows  Magnitude As illustrated in Figure 5.8 and Table 5.8, the magnitude of low flows in the disturbed years was significantly higher than those in the reference years. For example, in the reference year of 1968, the average magnitude of low flows was 0.40 m3/s, while in its paired year of 2007, with 48% CECA, this value was greatly increased by about 200% (1.2 m3/s). Similarly, for the 1989-1999 pair, the average magnitude of low flows was augmented by 159% due to forest disturbances at the level of 62.2% CECA. Based on all selected pairs, the averaged magnitude of low flows was increased by 163.4% due to the average difference of 41.1% CECA or 60.5% increment per 10% CECA (Table 5.13).  Table 5.13 The disturbance effects on the magnitude and CV of low flows in the Baker Creek watershed ∆ Ml  ∆ Ml%  Normalized ∆Ml %  ∆CVl%  Normalized ∆CVl%  3  (m /s)  (%)  (%/10%CECA)  (%)  (%/10%CECA)  0.5  163.4  60.5  -57.0  -14.1  ∆CECA (%) 41.1  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆Ml is the absolute difference in the magnitude of low flows between the paired disturbed and reference years; ∆Ml% is the relative difference in low flows between the paired disturbed and reference years (∆Ml%=(Mld – Mlr)*100%/Mlr; Mld and Mlr are the magnitude of low flows in the disturbed and reference years, respectively); ∆CVl% is the relative difference in CV of low flows between the disturbed and reference years, respectively (∆CVl%=(CVld – CVlr)*100%/CVlr; CVld and CVlr are the CV of high flows in the disturbed and reference years, respectively)  166  2.0 1.8  Median  25%-75%  Min-Max  Magnitude-l(m3 /s)  1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0  Reference Year  Disturbed Year  Figure 5.8 The comparison of the magnitude of low flows in the paired reference and disturbed years in the Baker Creek watershed  Variability The CV of low flows in the disturbed years was significantly lower than those in the reference years (Figure 5.9 and Table 5.8), suggesting smaller variations in the low flows during the disturbed years. For example, in the reference year of 1964, the CV of low flows was 3% while in its paired year of 2009, with 62.2% CECA, the CV declined to 2%. Based on the calculation of all selected pairs, the CV of low flows was averagely reduced by 57% due to the average difference of 41.1% CECA (a 14.1% reduction per 10% CECA) (Table 5.13). The decreased CV in the low flows from forest disturbances may be associated with increasing magnitude of low flows as shown in Figure 5.8.  167  35 Median  25%-75%  Min-Max  30  CV-l (%)  25 20 15 10 5 0  Reference Year  Disturbed Year  Figure 5.9 The comparison of the CV of low flows in the paired reference and disturbed years in the Baker River watershed  5.4.3.2 Willow River watershed 5.4.3.2.1 High flows  Magnitude As shown in Figure 5.10 and Table 5.14, the magnitude of high flows in the disturbed years was significantly higher than those in the reference years in the study watershed. For example, in the reference year of 1957, the averaged magnitude of high flows was 203.7 m3/s while in its paired year of 1974, with 12% CECA, this value was increased by about 5.4% (214.7 m3/s). For the pair of 1960-2007, the averaged magnitude of high flows was augmented by 77% due to increasing of forest disturbances at the level of 33.8% CECA. Based on all selected pairs, the averaged magnitude of high flows was increased by 36.2% due to the average difference of 17.6% CECA (a 22.6% increment per 10% CECA change).  To demonstrate how the responses of high flows to forest disturbances changed with increasing return periods, we summarized the results by the return periods of high flows. Unexpectedly, forest disturbances produced the greatest impact on the magnitude of the largest high flows (the return periods of 5-10 years) with the average increment of 41.9% in high flows per 10% CECA 168  (Table 5.15). For other return periods, the averaged increments of high flows per 10% CECA were 27.2%, 20.7%, and 16.8% for the return periods of ≤1 year, 1-2 years, and 2-5 years, respectively, following a general decreasing pattern with increasing return periods.  600 550  Median  25%-75%  Min-Max  Magnitude-h(m3 /s)  500 450 400 350 300 250 200 150 100  Reference Year  Disturbed Year  Figure 5.10 The comparison on the magnitude of high flows in the paired reference and disturbed years in the Willow River watershed  250 200  a)  1957 1974  150 100 50 0 0 10 20 30 40 50 60 70 80 90 100 % of time over  Discharge (m3/s)  Discharge (m3/s)  300  400 350 300 250 200 150 100 50 0  b)  1960 2007  0 10 20 30 40 50 60 70 80 90 100 % of time over  Figure 5.11 Flow duration curves for the selected pairs in the Willow River watershed: a) 19571974 and b) 1960-2007  169  Table 5.14 The Mann-Whitney U test on the differences of flow regimes in the paired reference and disturbed years in the Willow River watershed Variables  Z statistics  Mh  -7.1*(p<0.001)  DRh  -4.3*(p<0.001)  RPl  1.9*(p=0.05)  CECA  -12.2*(p<0.001)  Note: *, Significant at α=0.05; Mh and DRh are the magnitude and duration of high flows, respectively; and RPl is the return period of low flows.  Table 5.15 The disturbance effects on the magnitude of high flows in the Willow River watershed Return Period  ∆ Mh (m3/s)  ∆Mh%(%)  ∆CECA (%)  Normalized ∆Mh% (%/10%CECA)  ≤1  62.2  73.2  19.2  27.2  1-2  33.7  46.7  10.8  20.7  2-5  98.4  18.9  23.8  16.8  5-10  229.3  39.4  17.5  41.9  Average  65.1  36.2  17.6  22.6  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆Mh is the absolute difference in the magnitude of high flows between the paired disturbed and reference years; ∆Mh% is the relative difference in high flows between the paired disturbed and reference years (∆Mh%=(Mhd - Mhr)*100%/Mhr; Mhd and Mhr are the magnitude of high flows in the paired disturbed and reference years, respectively)  Duration The durations of high flows over the thresholds in the disturbed years were significantly longer than those in the reference years (Figure 12 and Table 5.15). For example, in the reference year of 1957, the duration of high flows was 15 days, while in its paired year of 1974, with 12% CECA, this value was increased to 19 days. For the 1960-2007 pair, the duration of high flows was increased from 0 days to 19 days due to forest disturbances at the level of 33.8% CECA. Based on all selected pairs, the averaged duration of high flows was increased by 4.2 days due to the averaged difference of 17.6% CECA or a 2.3 day increment per 10% CECA (Table 5.16).  170  To examine how the responses of the duration of high flows to forest disturbances changed with the increasing return periods, we summarized the results by the return periods of high flows. As shown in Table 5.16, forest disturbances produced the greatest impact on the duration of the smaller-sized high flows. Per 10% CECA averagely led to an increment of 4.3 days in the duration of the high flows with return periods ≤1 year. In comparison, the responses of the duration of the largest-sized high flows to forest disturbances were minor. Per 10% CECA yielded no change in the duration of the high flows with the return periods of 5-10 years. 25 Median  25%-75%  Min-Max  Duration-h(days)  20  15  10  5  0  -5  Reference Year  Disturbed Year  Figure 5.12 The comparison on the durations of high flows in the paired reference and disturbed years in the Willow River watershed Table 5.16 The disturbance effects on the durations of high flows in the Willow River watershed Return Period  ∆DRh  ∆CECA (%)  (days)  Normalized ∆DRh (years/10%CECA)  ≤1  8.3  19.2  4.3  1-2  1.5  10.8  0.6  2-5  1.4  23.8  1.5  5-10  0.0  17.5  0.0  Average  4.2  17.6  2.3  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆DRh is the difference between the durations of high flows in the disturbed years and those in the reference years.  171  Discharge (m3/s)  350  a)  High flows over the threshold  300 250  1957 1974  Threshold: 182.7m3/s  200 150 100 50 0 31-May  26-May  21-May  16-May  11-May  06-May  01-May  26-Apr  21-Apr  16-Apr  11-Apr  06-Apr  01-Apr  Date  Discharge (m3/s)  400 350  b)  300  High flows over the threshold  250  1960 2007  Threshold: 182.7 m3/s  200 150 100 50 0  31-May  26-May  21-May  16-May  11-May  06-May  01-May  26-Apr  21-Apr  16-Apr  11-Apr  06-Apr  01-Apr  Date  Figure 5.13 The comparison on the durations of high flows in the pairs of 1957-1974 (a) and 19602007 (b) in the Willow River watershed  5.4.3.2.2 Low flows  Return period As shown in Figure 5.14 and Table 5.17, the return periods of low flows in the disturbed years were significantly smaller than those in the reference years. For example, in the reference year of 1954, the average return period of low flows was 6.2 years while in its paired year of 1970, with  172  12% CECA, the average return period was reduced to 2.5 years. Similarly, for the 1960-2007 pair, the average return period of low flows was reduced from 3.9 years to 1.2 years due to increasing forest disturbances of 33.8% CECA. Based on all selected pairs, the averaged return period of low flows was reduced by 1.4 years due to the average difference of 17.6% CECA (a 2.2 year reduction per 10% CECA) (Table 5.17).  40 Median  25%-75%  Min-Max  35  Return Period-l(years)  30 25 20 15 10 5 0  Reference Year  Disturbed Year  Figure 5.14 The comparison on the return periods of low flows in the paired reference and disturbed years in the Willow River watershed  Table 5.17 The disturbance effects on the return periods of low flows in the Willow River watershed ∆RPl (yrs) -1.4  ∆CECA (%) 17.6  Normalized ∆RPl (yrs/10%CECA) -2.2  Note: ∆CECA is the difference in the CECA between the paired disturbed and reference years; ∆RPl is the difference between the return periods of low flows in the disturbed year and that in the reference year.  173  5.5 Discussion 5.5.1  Forest disturbances and high flow regimes  The findings (Table 5.18) show that forest disturbances have caused significant alterations to most components of high flow regimes in the two most severely disturbed watersheds: the Baker and Moffat Creek watersheds. In the Willow River watershed, a moderately disturbed watershed, less pronounced effects on high flow regimes have been found, with only two flow regime components (magnitude and duration) being significantly related to forest disturbances. However, in the Tulameen River watershed with similar forest disturbances as in the Willow River watershed, there were insignificant effects on all five components of high flow regimes. Surprisingly, in the Cottonwood and Ashnola River watersheds, two least disturbed watersheds, we have found significant impacts on one component of high flow regimes in each watershed: timing in Cottonwood and variability in Ashnola.  It is important to note that the directions of forest disturbance-induced impacts on each component of high flow regimes are consistent in the study watersheds (Table 5.18). In the significantly impacted watersheds, forest disturbances can have the following directional effects on high flow regimes: increasing the magnitude, advancing the timing, prolonging the duration, increasing the return period, and increasing the variability of high flows. These findings are consistent with the general conclusions from small snow-dominated watershed studies (King and Tennyson, 1984; Gottfried, 1991; Moore and Wondnell, 2005; Alila et al., 2009; Green and Alila, 2012), as well as in line with some studies on the responses of peak flows or floods to forest disturbances in large watersheds (Troendle and King, 1985; Brandt et al., 1988; Cheng, 1989; Wei and Davidson, 1998; Buttle and Metcalfe, 2000; Lin and Wei, 2008).  174  Table 5.18 Cross-correlations between forest disturbances (CECA) and hydrological variables in the six study watersheds Hydrological  Baker  Moffat  Mh  +  +  TMh  -  -  variables*  DRh  Cottonwood  +  +  CVh  +  +  Ml  +  Ashnola  Tulameen  + -  +  RPh  Willow  +  +  TMl DRl  +  RPl CVl  + -  +  +  Note: Refer to the denotes in Table 5.6 and 5.7  The study watersheds are snow-dominated where high flows or floods are always driven by snowmelt processes in spring. The high flow regime for a given year is mainly determined by factors such as winter snowpack (total precipitation from November to March) and weather conditions (e.g., temperature and radiation) during the snowmelt season. In the mountainous watersheds, the snowmelt starts at lower elevation areas in early spring and then proceeds to upper elevation areas as the temperature rises. Under undisturbed conditions, the snowmelt water from lower elevations normally fills up rivers in late March, while the melt water from higher elevations contributes to the generation of high flows or potential floods in late April or May. After forest disturbances, such as logging, fire, and MPB infestation, the snow accumulation is expected to increase due to the reductions in canopy snow interception, and provide more water available for the generation of high flows.  175  As suggested by the stand level studies in the BC interior, clear-cuts can accumulate 5-70% more snow than the forest sites, depending on the size of snowfall events and canopy densities (Boon, 2009, 2011; Winkler and Boon, 2010). Moreover, snowmelt process is expected to be faster in the openings than in the forested sites due to the increased solar radiation available in the openings. Although snow evaporation can also be increased in the openings, field studies have indicated that evaporation losses from the snowpack during the snowmelt season can be negligible, and snowmelt dominates snow evaporation in the snow ablation (Winkler et al., 2008). A long-term study in the BC interior has shown that snow ablation is 15% lower and snow can last up to 8 days longer in mature lodgepole pine stands than that in the clear-cuts (Winkler and Moore, 2006; Winkler et al., 2008). Another study in boreal forest stands has also found a similar result: the snowmelt rate increased dramatically with increasing canopy gap fraction (the fraction of sky visible to a sensor under the canopy) (Metcalfe and Buttle, 1998). Similarly, snowmelt rates in the MPB attacked forest stands are also higher and faster than undisturbed forests (Winkler and Boon, 2010). Thus, more snowmelt water and faster snowmelt after forest disturbances can potentially lead to an increase in the magnitude of high flows, and can advance the timing of high flows during the snowmelt season at a watershed scale.  In addition, the elevation of a disturbed forest stand in snow-dominated watersheds is believed to be another important factor that determines the responses of high flows to forest disturbances. Forest disturbances that distribute at higher elevations can lead to more synchronizations of snowmelt water between the higher and lower elevations. These synchronizations can consequently cause stronger responses of high flow regimes to forest disturbances (the increased magnitude and advanced timing of high flows) (Verry et al., 1983; Brandt et al., 1988; Schnorbus et al., 2004). On the contrary, forest disturbances that concentrate at lower elevations may cause more de-synchronizations of snowmelt processes at different elevations, which eventually reduces the impacts of forest disturbances on high flows or peak flows. A watershedscale modeling study in the Reddish Creek watershed in the BC interior has shown that logging 20% of the area at the bottom of the catchment can cause little or no change in peak flows (Whitaker et al., 2002). A similar result has also been found in Sweden by Brandt et al. (1988). In order to take the elevation factor into consideration for designing forest harvesting at a watershed scale, the H60 concept, referring to the elevation of the snowline when the upper 60%  176  of a watershed is cover with snow, has been applied in the BC interior (IWAP, 2006). Harvesting above H60 is generally believed to yield more pronounced impacts on peak flows or high flows in the BC interior (Gluns, 2001; Whitaker et al., 2002).  In the Moffat Creek watershed, the most severely disturbed one among the six study watersheds, forest disturbances at the extremely high level (65.7% CECA) can greatly increase the snow accumulation and snowmelt rate. In addition, about 71% of logging in the Moffat Creek watershed was conducted above H60, promoting more synchronizations of snowmelt from low elevations and higher elevations. It is not surprising that such an extremely high level of forest disturbances, along with a great proportion of logging above H60 has caused the most pronounced impacts on high flow regimes, with all the components of high flow regimes being significantly altered. Similarly, in the Baker Creek watershed, the severe forest disturbances (62.2% CECA), coupled with the fact that 64% of logging distributes above H60 (Table 3.1), has also led to strong impacts on the high flow regimes, with the four components of high flow regimes significantly related to forest disturbances. According to the paired year analysis, per 10% CECA has averagely led to a 92.7% increment in the magnitude, a 2-day advance in the timing, a 3-year increase in the return period, and an 84.4% increment in the variability of high flows in the Baker Creek watershed.  In the Willow River watershed, with a moderate level of forest disturbances (35.4% CECA) and a relatively lower proportion of logging above H60 (59.2%, Table 3.1) than those in the Baker and Moffat Creek watersheds, less pronounced impacts on the high flow regimes were found with only two components (magnitude and duration) of the high flow regimes being significantly related to forest disturbances. On average, per 10% CECA can lead to a 22.6% increment in the magnitude and a 2.3-day increase in the duration of high flows in the Willow River watershed. In addition, the higher magnitude of high flows may also contribute to the lower value of relative change in the Willow River watershed as compared with Baker and Moffat Creek watersheds. The average magnitude of high flows in the Willow River watershed is 238 m3/s, 10 times as much as that in the Baker Creek watershed. Therefore, a moderate level of forest disturbances together with a relative high magnitude of high flows in the Willow River watershed can result in less pronounced impacts on high flow regimes.  177  To my surprise, there were no significant impacts of forest disturbances on any components of high flow regimes in the Tulameen River watershed, even though its forest disturbance level is similar to that in the Willow River watershed. This insignificant response may be partly explained by the reduced synchronizations of snowmelt water between higher elevations and lower elevations due to the relatively even distribution of logging in the Tulameen River watershed, with only 52.8% of logging above H60 (Table 3.1). In addition, the differences in landforms and topography in these two watersheds can also affect their responses of high flows to forest disturbances. The Tulameen River watershed, with elevations ranging from 629 m to 2302 m, about 60% of the total area above 1300 m (Figure 3.20), and with an average slope of 1%, is featured with a gently undulating upland of a low relief in the northern areas and rugged mountain ranges in the headwaters. On the contrary, the Willow River watershed is mainly comprised of a long, narrow broad valley with about 20% of the total watershed area above 1300 m (Figure 3.4) and with an average slope of 0.3%. The differences in topography and landscape patterns between the two watersheds suggest that the Tulameen River watershed has greater spatial heterogeneity, which can lead to more complex hydrological processes including flow routing and surface water-groundwater interactions, and consequently afford this watershed a greater ability to buffer forest disturbance effects. Therefore, the relatively even distribution of forest disturbances at different elevations and more complex topography are judged to be the key factors that account for the insignificant change in high flow regimes in the Tulameen River watershed.  In the Cottonwood River watershed, adjacent to the Willow River watershed, due to a low level of forest disturbances, the increment in snow accumulation and ablation is expected to be lower than that in the Willow River watershed, and consequently insignificant change in the magnitude of high flows. However, the timing of high flows in the Cottonwood River watershed has been significantly advanced by forest disturbances. Since forest disturbances are mostly concentrated at lower elevations in this watershed, with about 56% of logging below H60 (Table 3.1), more snowmelt water at lower elevations is expected to flow into the channels earlier, which consequently cause an advance of high flows as compared to that in pre-disturbance conditions. In the Ashnola River watershed, even with a very low level of forest disturbances (CECA 6.4%),  178  a significant positive impact on the variability (CV) of high flows has been identified. This unusual response may be due to the spatial distribution of forest disturbances: about 77% of logging located over H60 (Table 3.1).  In summary, this study has clearly demonstrated that the severe forest disturbances can lead to significant impacts on high flow regimes. This study also confirmed that the responses of high flows to forest disturbances are highly variable at large watersheds. The various responses may be related to the factors such as the levels and types of forest disturbances, watershed properties (e.g., landscape pattern and landform) and their interactions in large watersheds. In addition, my case studies show that the spatial distribution of forest disturbances also plays an important role in determining the responses of high flows to forest disturbances in large snow-dominated watersheds. 5.5.2  Forest disturbances and low flow regimes  According to the time series cross-correlation analysis, forest disturbance-related effects on low flow regimes include increased magnitude, prolonged duration, increased or decreased return period, and elevated or declined variability of low flows depending on the individual watershed under investigation. Those results suggest that the overall impact of forest disturbances on low flow regimes appeared less pronounced than that on the high flow regimes, and only limited flow regime components were significantly related to forest disturbances even in the severely disturbed watersheds. Moreover, the directions of effects were not consistent for some components of the low flow regimes.  As shown in Table 5.18, a significant positive correlation between the magnitude of low flows and forest disturbance levels has been identified in the Baker Creek watershed, a severely disturbed one with the CECA of 62.2%, suggesting that forest disturbances can significantly increase the magnitude of low flows. In addition, the variability of low flows in this watershed was significantly decreased by forest disturbances, indicating more stable low flows after forest disturbances. As suggested by the paired year study in the Baker Creek watershed, per 10% CECA, on average led to up to a 60.5% increment in the magnitude of low flows and a 14.1% reduction in the variability of low flows. The higher and more stable low flows as a result of 179  forest disturbances in the Baker Creek watershed may be beneficial to water supply in the watershed and downstream communities from a water quantity perspective. Forest disturbanceinduced changes, however, in other components (timing, duration, and return period) of low flows in the Baker Creek watershed were insignificant.  In the Willow River watershed, with a moderate level of forest disturbances (35.4% CECA), the significant response of low flow regimes appeared less pronounced. Only a significant but negative relationship between the return periods of low flows and forest disturbance levels was detected in the watershed (Table 5.18), suggesting that low flows with shorter return periods can be more frequent while the severe low flows or droughts had lower chance of occurrence after forest disturbances in the Willow River watershed. In summary, forest disturbances in the Baker Creek and Willow River watersheds can help reduce the risk of droughts by increasing the magnitude of low flows or reducing the responses of extreme low flow events, thus resulting in a more stable water supply.  On the contrary, in the Moffat Creek watershed (CECA 65.7%), with the forest disturbance level similar to that in the Baker Creek watershed, forest disturbances have caused significant positive effects on the return periods of low flows, suggesting more frequent low flows with greater return periods and consequently more severe droughts after forest disturbances in the Moffat Creek watershed. Similarly, in the Cottonwood River watershed, a low level of forest disturbances (CECA 12.2%) has also caused significant positive effects on both the return periods and the duration of low flows, suggesting lower and longer low flows and thus possibly more severe droughts after forest disturbances. In other words, forest disturbances in the Moffat Creek and the Cottonwood River watersheds can increase the risk of drought and potentially exacerbate water shortage in dry seasons when the water demands from both human beings and aquatic needs are high. In the Tulameen River watershed, forest disturbances (CECA 33.8%) have produced a significant positive effect on the variability of low flows (Table 5.18), suggesting larger variations in low flows after forest disturbances and consequently a less stable water supply in the low flow seasons. In the Ashnola River watershed, adjacent to the Tulameen River watershed, with very limited disturbances (CECA 6.4%), none of the components of low flow regimes has been significantly altered by forest disturbances.  180  The above inconsistent results on low flow regimes found in these study watersheds clearly show that the responses of low flows to forest disturbances are highly variable and watershed-specific, which generally agree with the common understanding of the impacts of forest disturbances on low flows (Bruijnzeel, 2004; Moore and Wondzell, 2005). According to the published scientific literature, both negative and positive impacts of forest disturbances on low flows have been reported (Swanson et al., 1986; Gottfried, 1991; Bent, 2001; Calder, 2005; Webb et al., 2007). It is commonly believed that the hydrological responses of low flows to forest disturbances depend on not only the levels of vegetation removal but also the levels of soil damage. With a severe topsoil disturbance (e.g., soil compaction caused by logging activities and soil loss due to soil erosion), the capacity of soil infiltration can be greatly impaired, resulting in more surface runoff and reduced recharge to deep soils and groundwater systems, and consequently leading to the reduction of low flows (Zhang et al., 2012b). On the contrary, with minor soil damage after forest disturbances, higher low flows may be expected since the removal or death of trees decreases evapotranspiration and interception, which increase the amount of water available in the soils to promote soil infiltration and recharge to groundwater, and accordingly to yield more discharges to streams in the dry seasons (Bosch and Hewlett, 1982).  In the Baker Creek and Willow River watersheds, forest disturbances have produced positive impacts on low flow regimes, resulting in higher low flows after forest disturbances. Logging activities in the BC interior are normally conducted in winter seasons when soils are completely frozen, which can cause minor or insignificant damage to soils. Moreover, since both the Baker Creek and Willow River watersheds are characterized by relieves with gentle slopes or even flat lands (Figure 3.11 and Figure 3.3), soil loss due to soil erosion after forest disturbances could be minor. Therefore, with the severe forest disturbances in those two watersheds, more recharges to soils and groundwater systems and consequently higher low flows are expected due to the reduction in evapotranspiration and interception.  As described earlier, the positive impacts of forest disturbances on low flow regimes in the Baker Creek watershed were identified, while the negative effects were found in the Moffat Creek watershed. The contrasting hydrological responses of low flows in those two most severely  181  disturbed watersheds may be due to the following factors. The first factor is the difference in topography. The Moffat Creek watershed is situated in the transition zone from the Fraser Plateau to the Quesnel Highlands, with an average slope of 0.7%. The western portion of the Moffat Creek watershed is dominated by a flat plateau while its eastern portion is occupied by high mountains, with the elevations ranging from 778 to 2155 m (Figure 3.16). In comparison, the Baker Creek watershed is mostly a plateau with the elevations ranging from 475 to 1500 m (Figure 3.12). The difference in topography may make the soils in the Moffat Creek more erodible and consequently greater soil loss during the spring high flow seasons or summer storms, which may, in turn, greatly impair soil infiltration, reduce soil storage, and thus, exacerbate soil moisture deficit and decrease low flows in the dry seasons. The second possible factor is the difference in the growth of vigorous new vegetations after forest disturbances. Compared with the Baker Creek watershed, the climate in the Moffat Creek watershed (higher temperature and precipitation) is more favorable for vegetation regeneration. Therefore, new vegetation at the disturbed sites in the Moffat Creek watershed can consume more water than those in the Baker Creek watershed during the growing seasons, resulting in higher soil water deficit in the former watershed. The third possible factor is that forest disturbances may greatly alter the interactions between surface water and groundwater in a different way. In short, the differences in the potentials of soil loss, the vigor of vegetation growth, and surface water and groundwater interactions are judged to contribute to the contrasting responses of low flows to forest disturbances between the Moffat Creek and Baker Creek watersheds. More research is clearly needed to study the mechanisms governing the relationship between forest disturbances and low flows in the BC interior.  The difference in the hydrological responses of low flows to forest disturbances between the Cottonwood and Willow River watersheds is noticeable. In the Cottonwood River watershed, a low level of forest disturbances (12.2% CECA) has caused significant positive effects on both the duration and the return period of low flows, suggesting potentially higher risks for more severe, long-lasting droughts after forest disturbances. However, in the Willow River watershed, a higher level of forest disturbances (35.3% CECA) has yielded a negative effect on the return periods of low flows, indicating a reduced risk of severe droughts after forest disturbances. The contrasting findings in these two neighbouring watersheds may be due to the differences in the  182  characteristics of low flows (e.g., the magnitude and the occurring season of low flows) in addition to their different levels of forest disturbances. The low flows in the Cottonwood River watershed mostly occur in late summer and fall (August to October) when precipitation is very low. However, during this low flow period, vegetation growth is normally fast and the interactions between vegetation and water, especially the interactions between riparian vegetation and streamflow can be very active. With a low level of disturbances, vigorous new vegetation can consume more water than the “saved water” due to the reduction of evaporation and interception after the removal of old vegetation in the Cottonwood River watershed. This makes the low flows become very sensitive to vegetation regrowth in the Cottonwood River watershed. In contrast, the low flows in the Willow River watershed are mostly observed in winter (November to March) when the topsoil is frozen and interactions between vegetation and water are virtually limited or inactive. The winter low flows are mainly maintained by groundwater discharge, normally depending upon groundwater recharge processes through soil infiltration and mountain system recharging mechanisms and regional aquifer characteristics (Wahi 2005; Wahi et al. 2008). With a higher level of forest disturbances, more water can be saved for soil moisture and groundwater recharge during growing seasons and contributes to a higher level of winter low flows after forest disturbances in this watershed. Therefore, the winter low flows tend to be less sensitive to vegetation regrowth in the Willow River watershed. Moreover, the average magnitude of low flows in the Cottonwood River watershed (3.0 m3/s) is much lower than that in the Willow River watershed (7 m3/s). This difference may further cause higher sensitivity of low flows to forest disturbances in the Cottonwood River watershed. Thus, the occurrence of low flows in the season of late summer to fall, a lower level of forest disturbances together with the much lower magnitude of low flows in the Cottonwood River watershed may suggest that the low flows in this watershed are much more sensitive to vegetation growth, which may account for the contrasting results between the two watersheds. The above results further indicate that the responses of low flows to forest change or disturbances are complicated, and data on vegetation changes alone are unlikely sufficient to explain low flow changes.  In summary, this study highlights the fact that the responses of low flows to forest disturbances are complex and watershed-specific. Forest disturbance level, soil disturbances and vegetation  183  regeneration after forest disturbances, topographic and geologic features, and low flow characteristic (e.g., the magnitude or the occurring season of low flows) are believed to be the major factors that determine the responses of low flow to forest disturbances. 5.5.3  Return periods and the responses of high flows to forest disturbances  According to the quantitative assessments by use of the paired year approach in the Baker Creek watershed, we have found that the responses of high flows (or floods) to forest disturbances were either decreasing or increasing with the return periods (for the high flows with the return periods ≤10 years). As presented in Table 5.19, in the Baker Creek watershed, an averaged 311.5% increment per 10% change in CECA in the magnitude of high flows with the return periods ≤ 1years were found. This increment dramatically declined to less than 30% for high flows with the return periods of 1-2 years and to only 8.2% for high flows with the return periods of 5-10 years. This finding is in accordance with the well-established concept in forest hydrology that the effects of forest disturbances on floods decrease with increasing return periods according to a century of paired-watershed experimental studies on a small watershed scale (<100 km2) (Brooks et al., 2003; Calder, 2005; Moore and Wondnell, 2005). This result can be explained by the reasons below. First, any given increment in snowmelt water due to forest disturbances is expected to transfer into lower percentages of changes in large-sized high flows since we use the relative difference between the controlled and disturbed years as a metric to indicate the response of high flows. Second, the relative increase in snow accumulation due to forest disturbances tends to decrease with the sizes of snowfall events. The results from Winkler and Moore (2006) suggest that in the year of heaviest snowfall, the relative differences in snow water equivalent between the mature forest and clear-cut are not the greatest, possibly due to the existence of an upper limit for the interception capacity of canopies. Such a relationship can be translated to the relationship between the forest disturbance-related responses of high flows and the sizes (return periods) of high flows.  In the Willow River watershed, this negative relationship is valid for high flows with the return periods ≤ 5years. However, the response of high flows with the return periods of 5-10 years was greater than that with the smaller-sized (or shorter return periods) high flows in the Willow River watershed, which is different from that in the Baker Creek watershed. This finding in the Willow 184  River watershed is in line with the recent research (Alila et al., 2009; Green and Alila, 2012) showing that forest disturbances can potentially yield greater effects on the magnitude and frequency of large floods. Unlike the traditional approach that compares the chronologicallypaired peak flows in the control and treatment catchments, Alila et al. (2009) have developed an approach that compares the frequency distributions of peak flows in the control and treatment catchments.  The results in this study using the paired year approach with the Baker Creek and Willow River watershed have presented different relationships between the return periods and high flow responses. This difference is likely due to the dissimilarities in watershed characteristics (e.g., landscape pattern, slope gradient, drainage density, and travel time) that control the generation mechanism of high flows between these two watersheds. The Willow River watershed is mainly comprised of a long, narrow valley where melt water can be transferred quickly from the upslope lands to the main watercourse due to shorter distances from main watercourse (Figure 3.3), resulting in a faster hydrological response, especially for large floods. In comparison, the shape of the Baker Creek watershed is circular (Figure 3.11), where the distances between the upslope lands and the main watercourse are longer than those in the Willow River watershed, leading to a slower generation of high flows. Therefore, it is possible that the effects of forest disturbances on larger-sized (longer return periods) high flows is more significant than the smaller-sized (shorter return periods) high flows in the Willow River watershed as compared with the Baker Creek watershed. Moreover, this difference can be partly due to the small sample size of high flows with the return period of 5-10 years. High flows with the return period of 5-10 years account for only about 5% of the total data in the Baker and Willow watersheds. Such a small sample size impedes us from drawing a more reliable inference on high flows with the return period of 5-10 years. Nevertheless, due to data limitations, the responses of high flows with the return periods >10 years have not been examined. Whether these findings are valid for the high flows with the return periods >10 years remain uncertain. This study highlights the fact that the relationships between the return periods and forest disturbance-induced responses of floods are varied and watershed specific.  185  Table 5.19 The responses of high flows with the different return periods in the Baker Creek and Willow River watersheds Return Period  Willow-Normalized  Baker-Normalized  ∆Mh% (%/10%CECA)  ∆Mh% (%/10%CECA)  ≤1  27.2  311.5  1-2  20.7  29.7  2-5  16.8  15.6  5-10  41.9  8.2  Average  22.6  92.7  Note: ∆Mh% is the relative difference in high flows between the paired disturbed and reference years.  5.5.4  Ecological implications and watershed management  Forest disturbances have produced significant impacts on high flow regimes such as increased magnitude, variability, duration, frequency, and advanced timing of high flows in the Baker Creek, Moffat Creek, and Willow River watersheds. These changes can potentially increase floods and thus threaten the public safety of downstream cities or communities. This is particularly evident in the two severely disturbed, dry watersheds (Baker Creek and Moffat Creek), where forest disturbances have increased high flows dramatically. Due to the recent wide-spread forest disturbances such as MPB infestation and subsequent salvage logging in the upper reaches of Fraser River (mainly in the central interior of BC), the significant alternations to high flow regimes can be added and amplified, and consequently lead to a chance of more catastrophic floods to threaten downstream populated cities of the Fraser River. In the Baker Creek watershed, forest disturbances have increased the magnitude of low flows and decreased the variability, which indicates the reduction of the drought risks and ensures a more stable water supply in this dry watershed. On the contrary, in the Moffat Creek and Cottonwood River watersheds, forest disturbances tend to increase the risk of droughts, exacerbating the water shortage in dry summers and falls. Apparently, the forest disturbance-induced changes in flow 186  regimes in those watersheds can have significant negative impacts on human beings, although the increase in low flows in some watersheds may benefit human beings from a water supply perspective.  The alterations of flow regimes due to forest disturbances can potentially affect aquatic ecosystems since a watershed ecosystem is featured with complex and dynamic interactions among biological, hydrological, morphological, and other ecological processes (Poff et al., 1997). Forest disturbances such as logging, insect infestation, and fire can either directly or indirectly cause significant impacts on forest ecosystems through modifying biological, hydrological, and morphological processes (Figure 2.1). Because of the close linkages and interactions among those processes, the hydrological alteration caused by forest disturbances can, in turn, generate important effects on many other biological and morphological processes and functions, and consequently on the integrity of watershed ecosystems (Poff et al., 1997).  As summarized in Table 5.20, high flow regimes have been significantly impacted by forest disturbances in the Baker Creek, Moffat Creek, and Willow River watersheds. Forest disturbances can increase the magnitude, duration, and return period of high flows, which may lead to unstable and widened stream banks, increased erosion, and more sediment transportation and deposition, and consequently degraded aquatic habitats. These hydrological and morphological changes can also cause downward incision and floodplain disconnection, resulting in the damage of floodplain habitats. In addition, more frequent high flows can cause more large wood debris jams and the modification of the streambed morphology and substrate compositions that have implications for aquatic habitat such as riffle and pool sequences. The change of the timing of high flows can also have negative ecological impacts. Since many native fish species use certain levels of flows as cues for spawning, egg hatching, and migration, the advanced high flows can disrupt the life cycle of these species and may reduce the growth rates of juvenile fish species and their population sizes, and eventually modify entire food webs (Nasje et al., 1995, Wootton et al., 1996).  An advance in high flows may also harm native riparian plant species whose successful germination requires the synchronized supplies of water and heat. For example, in the Baker  187  Creek watershed, the high flows were advanced to early April with the temperatures lower in the disturbed periods than those in the undisturbed high flow periods (late April and early May). This is unfavourable for native species with specific requirements for thermal input to germination. Moreover, the increased variability of high flows can also negatively affect aquatic and riparian species. Many small fish species and the young of large species can be severely impaired by frequent fluctuations of flows, which may eventually cause the native fish species to be replaced by those species tolerant to high flows with greater variations (Moore and Gregory, 1988; Stanford, 1994).  Forest disturbance-related alterations of low flow regimes may yield profound effects on aquatic ecosystems. However, those alterations were less consistent as compared to those of high flow regimes based on the findings from this study. Forest disturbances can lead to the alterations of low flows including the increased magnitude, decreased return period, or reduced variability of low flows in the Baker Creek or Willow River watersheds. These changes in low flows can yield negative effects on ecosystems, such as disrupting the life cycle of aquatic and riparian species and causing the reduction of native species, resulting in potentially higher risks to aquatic ecosystems. On the other hand, in the Moffat Creek, Cottonwood River, and Tulameen River watersheds, forest disturbances have produced negative effects on low flow regimes such as the prolonged duration, increased return period, and/or elevated variability of low flows, indicating higher potential risks of droughts. The severe, long-lasting droughts can cause the widespread mortality of riparian and aquatic species, resulting in significant damage to the health of watershed ecosystems (Carnicer et al., 2011).  The Willow River, Cottonwood River, Baker Creek, and Moffat Creek watersheds are critical ecosystems for salmons, such as chinook, pink, and sockeye, the most valuable commercial anadromous species in BC, Canada. Streams and lakes in these watersheds provide important habitats for their spawning, hatching, rearing, and migrating. The mature salmons always migrate upstream and spawn during the low flow seasons from August to October. The hatching takes place in the winter seasons from December to February, while the juveniles emerge in the spring seasons, and then migrate downstream (Healey, 1991; Heard, 1991). Given the fact that low flows in these watersheds occur in late summer, fall or winter and high flows always occur in  188  spring driven by snowmelt process, not surprisingly, the alterations of both high and low flow regimes in these watersheds can inevitably affect the life cycle of salmon species.  The increased high flows with greater return periods can transport more sediment from the upslope to streams, which will significantly degrade water quality and substrate composition, and reduce the growth rate of the juveniles. Moreover, the decreased low flows with greater variability and prolonged duration in late summer or fall can reduce the amount and quality of habitats available for spawning, and even dewater redds, resulting in drying out of salmon eggs (Rood and Hamilton, 1995). Research has demonstrated that forest disturbances such as logging can cause great degradation of stream habitat for salmons and reduce the population size of native fish species in the Pacific Northwest, mainly via effects on streamflow and sediment transportation (NRC, 1996). A recent study by Nelitz et al. (2012) has also reported that the decline in sockeye salmon in the Fraser River watershed is partly attributed to forest disturbances in this watershed.  In the study watersheds, due to the differences in the responses of flow regimes to forest disturbances, the corresponding potential impacts on critical fish species are expected to be variable (Table 5.20). In the Baker Creek and Moffat Creek watersheds, due to the very high levels of forest disturbances, most components of the flow regimes have been significantly modified, which may have already produced severe impacts on critical salmon fish species such chinook, pink, and sockeye in these two watersheds. In the Willow and Cottonwood River watersheds, three components of high and low flow regimes have been significantly altered by forest disturbances, which can lead to a moderate level of the impact on those important fish species. In the Tulameen and Ashnola River watersheds, only one component of flow regimes was found to be significantly related to forest disturbances, which may produce a low level of the impact on aquatic habitat and species in these two watersheds.  In a large watershed, it often takes several years to detect any significant hydrological changes after forest disturbances. And it may take an even longer time to identify any significant changes in geomorphic features and their associated biological or ecological effects due to the complexity of ecosystems and other confounding factors such as climate change (Knox, 1972). Therefore,  189  resource managers must pay immediate and special attention to the possible ecological risks due to the alterations of flow regimes as a result of forest disturbances in the disturbed watersheds in the BC interior. Moreover, given the dynamic interactions among biological, hydrological, morphological, and ecological processes, any watershed management strategy must be designed in a broad context with the consideration of interactions among these important watershed processes and their responses to forest disturbances.  Table 5.20 Forest disturbance-induced changes in the high flow and low flow regimes and their possible effects on critical fish species Watershed  Changes in high flow regime  Changes in low flow regime  Critical fish species  Impact  Baker Creek  Increased magnitude, return period and variability, and advanced timing  Increased magnitude and decreased variability  Chinook salmon and pink salmon  High  Moffat Creek  Increased magnitude, duration, return period and variability, and advanced timing  Increased return period  Sockeye salmon and kokanee salmon  High  Willow River  Increased magnitude and duration  Decreased variability  Pink salmon and chinook salmon  Moderate  Cottonwood River  Advanced timing  Increased duration and return period  Chinook salmon  Moderate  Tulameen River  No significant impact  Increased variablity  Chinook salmon, Rainbow trout, and longnose dace  Low  Ashnola River  Increased variability  No significant impact  Chinook salmon, Rainbow trout, and sculpin  Low  Note: The impact level is determined by the number of components of flow regimes being significantly altered by forest disturbances. 190  5.6 Summary Forest disturbances have significantly altered the magnitude, variability, duration, and frequency of high flows, and advanced the timing of high flows in the Baker Creek, Moffat Creek, or Willow River watersheds. These changes can potentially increase flood potentials and thus threaten the public safety of downstream cities or communities. The impacts of forest disturbances on low flow regimes can be either negative or positive. The positive effects, from a water quantity perspective, can help alleviate droughts and ensure a more stable water supply in dry seasons, while the negative effects can exacerbate the water shortage. Apparently, forest disturbance-induced changes in flow regimes mostly have significant negative impacts on human beings as well as aquatic habitat and watershed functions. We expect that these significant and negative ecological impacts in the severely disturbed watersheds such as the Baker and Moffat Creek watersheds may lead to an increased chance of catastrophic and irreversible ecological consequences. Therefore, immediate attention should be paid to those possible ecological risks due to the alterations of flow regimes after forest disturbances in these severely disturbed watersheds.  This study also demonstrates inconsistent and uncertain relationships between forest disturbances and low flow regimes. This is mainly due to the complicated interactions or mechanisms governing the generation of low flows at large watersheds. More research is needed to examine this relationship in a broader context for a better understanding on forest disturbances and its effects on low flow regimes.  191  6 Chapter: Conclusions and future studies 6.1 Conclusions Despite numerous studies on the relations between forest disturbances and water, the majority of them are conducted at hillslope and small watershed scales, with limited studies that focus on large watersheds and a broad forest disturbance context. Moreover, most existing studies have focused on hydrological variables such as mean, high and low flows with rare examinations of a full spectrum of flow regimes (magnitude, frequency, duration, variability, and timing of flow). Maintaining natural flow regimes is crucial for the conservation of riverine habitat and aquatic ecosystem integrity (Poff et al., 1997). Understanding how flow regimes are altered by forest disturbances is prerequisite for conserving and restoring aquatic ecosystems in forested watersheds. This study took advantage of long-term data availability on climate, hydrology, and forest disturbances in the BC interior to examine the effects of cumulative forest disturbances on annual and seasonal mean flows and flow regimes. The followings are key conclusions drawn from this study. 6.1.1  Cumulative forest disturbances and mean flows  Forest disturbances have produced significant impacts on annual mean flows in the four severely disturbed watersheds (the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds) with the forest disturbance levels ranging from 33.8% to 65.7% CECA. Annual mean flow variations attributed to forest disturbances were 21.5- 61.1 mm (5-19 mm per 10% change in CECA). Forest disturbances have also yielded significant impacts on some of seasonal mean flows in these four watersheds. The specific seasonal flows altered by forest disturbances vary among the study watersheds. 6.1.2  Cumulative forest disturbances and high flow regimes  In the significantly impacted watersheds (the Baker Creek, Moffat Creek, and Willow River watersheds), forest disturbances have yielded the following consistent and directional effects on high flow regimes: increasing the magnitude, advancing the timing, prolonging the duration, increasing the return period, and increasing the variability of high flows. These effects clearly  192  indicate that forest disturbances can increase the chance of floods. The extent of alterations and the altered components of high flow regimes vary among these four watersheds, which highlights that the impacts of forest disturbances on high flow regimes are watershed specific 6.1.3  Cumulative forest disturbances and low flow regimes  Forest disturbance-related impacts on low flow regimes have included increasing the magnitude, prolonging the duration, increasing or decreasing the return period, and increasing or decreasing the variability of low flows. The overall impact of forest disturbances on low flow regimes appeared less pronounced than that on the high flow regimes, and only limited components of low flow regimes were significantly related to forest disturbances even in the severely disturbed watersheds. Moreover, the impact directions were opposite for some components of the low flow regimes. These impacts clearly demonstrate that the relationships between forest disturbances and low flow regimes are more complicated as it involves soil conditions and groundwater processes in addition to forest changes. 6.1.4  The offsetting effects of forest disturbances and climate variability on annual mean flows  In the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds, forest disturbances and climate variability have produced counteracting or joint positive effects on annual mean flows. Forest disturbances increased streamflow while climatic variability decreased it, resulting in a relatively stable trend of annual mean flows over the study period in the Willow, Tulameen, and Moffat watersheds. However, both forest disturbances and climate variability produced positive effects on annual mean flows in the Baker Creek watershed. Annual mean flow variations caused by climate variability range from -112.5 mm to 5.2 mm in these four watersheds. 6.1.5  Forest disturbance thresholds for detectable changes in annual mean flows  The modified double mass curves and statistical multiple change points analysis on those curves provide a meaningful way to estimate forest disturbance thresholds. The forest disturbance level thresholds for detectable changes in annual mean flows in the four study watersheds (the Willow River, Baker Creek, Moffat Creek, and Tulameen River watersheds) are from 15% to 55%. The 193  Tulameen River watershed in the southern interior of BC has a lower forest disturbance threshold than those from the other three watersheds located in the central interior of BC. The large variations of the forest disturbance thresholds in those four study watersheds may be due to the differences in watershed properties (e.g., climate, vegetation, soil, and topography) that govern how forests and water interact. 6.1.6  Climatic gradients and hydrological responses  The change magnitude of annual mean flows caused by forest disturbances vary along climatic gradients. The comparisons show that in the drier years or the drier watersheds (the Baker Creek and Moffat Creek watersheds), annual mean flow responses to forest disturbances are less than those in the wetter years or wetter watersheds, suggesting that precipitation is a critical factor influencing the hydrological responses to forest disturbances. However, no definitive conclusions can be drawn on temperature gradients. 6.1.7  Hydrological responses and watershed resilience  In the Willow and Tulameen River watersheds, the strength of forest disturbance impacts on annual mean flows is higher at the early stage of the disturbed period, and then on a decline with the progression of forest disturbances. This suggests that the hydrological changes caused by the current levels of forest disturbances in these two watersheds are in the recovery process probably due to the resilience, stability, and hysteresis of watershed systems. The findings mentioned above are of great implications for designing forest and watershed management strategies to protect watershed ecosystem functions and public safety in the context of future forest and climate changes.  194  6.2 Limitations and future studies 6.2.1  Limitations  Although this study successfully assessed the hydrological impacts of cumulative forest disturbances in large watersheds by use of a new methodology and long-term data, some limitations may still exist. The first limitation is weak prediction ability. This methodology combing statistical and graphic approaches has advantages of providing robust inferences or empirical relationships based on measured data. However, since it treats watersheds as “black boxes”, this methodology lacks the ability to explore mechanisms or hydrological processes, and thus, the prediction of hydrological responses under future forest change and climate change scenarios can be constrained. Another limitation is its high dependency on data. The availability of long-term data on hydrology, vegetation, and climate in these study watersheds in the BC interior provided us a unique opportunity to investigate the hydrological responses to forest disturbances. Although those long-term data were collected by reliable government agencies such as Water Survey Canada, Enivironment Canada, and BC Ministry of Forests, Lands and Natural Resources Operations, they may not be perfect and may contain some errors and inconsistency. Thus, some uncertainties asscociated with data quality cannot be completely ignored in this study. 6.2.2  Future studies  The research assessing forest disturbances and hydrology in large watersheds is constrained by the lack of commonly accepted methodology and insufficient data. In the past, researchers have generally applied statistical and modeling approaches to study this topic in large watersheds. The statistical approach provides robust inferences or empirical relationships based on measured data, but it treats watersheds as “black boxes.” In comparison, the modeling approach can predict hydrological responses to future forest and climate changes, but reliable simulations depend on careful model calibration and validation, as well as various empirical relationships on various watershed processes and components. These empirical relationships are normally not available in large watersheds. Thus, both approaches are complementary. With this philosophy and on the basis of this study, I propose the following topics for future research.  195  6.2.2.1 Hydrological modeling Hydrological simulations are useful for predicting the effects of future forest changes (e.g., disturbances or reforestation) on hydrology. The results from this study can greatly support future modeling studies. Various empirical relationships between cumulative forest disturbances and hydrological variables derived from this study can be used to support calibration and validation processes in any selected models for improved simulations (e.g., VIC, DHSVM, and MIKE-SHE). Based on the results, the Willow River watershed may be one of the best watershed candidates to conduct hydrological modeling due to its long and high-level forest disturbances and its significant hydrological responses. Another opportunity is to include watershed resilience into future hydrological modeling. As suggested in this study, there are hardly any simple linear relationships between cumulative forest disturbances and hydrological changes. We have also found the consistent, dynamic, and reduced hydrological responses to the progression of forest disturbances in the Willow and Tulameen River watersheds, which is likely due to watershed resilience. However, current hydrological modeling, normally based on small-scale processes, always assume a constant relationship between forest disturbances and hydrology, which may produce misleading conclusions. In future hydrological studies, watershed resilience to watershed disturbances must be taken into account, which will provide a better modeling framework that captures some nonlinear properties in large watersheds. 6.2.2.2 More case studies are needed Modern watershed resource managers are eagerly seeking scientific information on large watersheds to support resources management for the protection of water resources. This is particularly true when climate change and anthropogenic activities are dramatically altering watershed processes at large spatial scales. However, research on evaluating the hydrological impacts of cumulative forest disturbances in large watersheds systems is challenging and limited, mainly because of insufficient data and complexity. Moreover, the results from this study as well as from other studies show that research results on forest disturbances and hydrological effects in large watersheds are less consistent when compared with small watershed studies. Clearly, the forest-water relationship in large watersheds is likely watershed specific, and thus more case  196  studies are needed to draw general conclusions on this subject. Future studies should target a broader range of environmental gradients, including large watersheds from rainfall-dominated systems. 6.2.2.3 Further studies in the Baker Creek and Moffat Creek watersheds Both the Baker and Moffat Creek watersheds have experienced the highest levels of forest disturbances due to the large-scale outbreak of mountain pine beetle and subsequent salvage logging in the last 10 years. Such a short disturbed period provides a snap-shot assessment, but it generally fails to evaluate the long-term dynamic responses of hydrology to forest disturbances. A longer disturbed period in the future will allow the determination of how long the impacts would last, and if watershed resilience was or will be damaged in those two extremely disturbed watersheds. In addition, longer-term data will help the determination of exact forest disturbance thresholds. 6.2.2.4 Application of landscape ecology This study shows that the hydrological responses to forest disturbances are watershed specific. In addition to climate, watershed properties such as topography, watershed shape, lakes and wetlands, and channel connectivity can also affect the hydrological responses to forest disturbances. Landscape ecology studies spatial heterogeneity and connectivity, scales, and interactions between spatial patterns and processes or functions. The theory and approach in landscape ecology can be introduced to forest hydrology to explore how watershed properties such as watershed shape, channel connectivity, spatial heterogeneity, and land use and cover patterns affect the hydrological responses to forest disturbances. 6.2.2.5 More studies on forest disturbances and flow regimes This study demonstrates that forest disturbances have caused significant impacts on high and low flow regimes. The alterations of flow regimes due to forest disturbances can potentially affect aquatic ecosystems. Most rivers in the BC interior are critical ecosystems for salmons, the most valuable commercial anadromous species. How forest disturbances affect flow regimes and consequently fish habitat, is highly significant in BC. Because the Fraser River basin is home to many salmon fish species and is currently facing serious pressures from the MPB infestation and  197  subsequent salvage logging, more studies are urgently needed to evaluate the interplays among forest disturbances, flow regimes, and fish habitats. 6.2.2.6 Application of geochemical approach Information on water geochemistry (major ions) and isotopes is useful for understanding flow paths and sources. It can be used to investigate surface water and groundwater integration. While the statistical approach treats watershed as “black boxes”, the geochemical approach will lead to more insights on water fluxes, paths, and water source separations. Those process-level data are important for understanding the mechanisms that govern the hydrological responses to forest disturbances at a watershed scale, and can be also used to support the model calibration and validation. 6.2.2.7 Risk assessment for water resource management The significant impacts of severe forest disturbances in large watersheds can inevitably impose stresses on water supply and public safety in downstream communities. A critical need is to assess potential risks as to how forest disturbance-induced hydrological changes in upstream watersheds impact downstream water resource sustainability, particularly in the context of future climate change. Such a study will provide a solid basis to support implementation of the “sourceto-tap” wholistic approach.  198  References Adams, R.S., Spittlehouse, D.L., Winkler, R.D. 1998. The snowmelt energy balance of a clearcut, forest and juvenile stand. In: Proceedings of 23rd Conference on Agricultural and Forest Meteorology, Albuquerque, N. Mex., American Meteorological Society, Boston Mass., pp. 54–57. Ager, A., Clifton, C., 2005. Software for calculating vegetation disturbance and recovery by using the equivalent clearcut area model. General technical report PNW-GTR-637. USDA Forest Service, Portland. http://www.fs.fed.us/pnw/pubs/pnw_gtr637.pdf. (Accessed 11 Feb 2009) Alila, Y., 2001. Using numerical modelling to address hydrologic forest management issues in British Columbia. Hydrol. Process. 15, 3371-3387. Alila, Y., Kuras, P.K., Schnorbus, M., Hudson, R., 2009. Forests and floods: A new paradigm sheds light on age-old controversies. Water Resour. Res. 45(8), W08416. Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H., Gonzalez, P., Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.H., Allard, G., Running, S.W., Semerci, A., Cobb, N., 2010. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. Forest Ecol. Manage. 259, 660-684. Anderson, H.W., 1974. Fire effects on water supply, floods, and sedimentation. In: Annual Proceedings Tall Timber Fire Ecology Conference, Pacific Northwest, October 16–17, 1974, 15: 249–260 Anderson, H.W., Marvin, D.H., Kenneth, G.R., 1976. Forest and water: effects of forest management on floods, sedimentation, and water supply. General Technical Report PSW-18, USDA (Forest Service). Berkeley, California. Andréassian, V., 2004. Waters and forests: from historical controversy to scientific debate. J. Hydrol. 291, 1-27. Arnold, J.G, Srinivasan, R., Muttiah, R.S., Williams, J.R.,1998. Large area hydrologic modeling and assessment part 1: model development. J. Am. Water Resour. As. 34(1), 73-89. Baker, M.B. 1986. Effects of ponderosa pine treatments on water yields in Arizona. Water Resour. Res. 22, 67-73. Bari, M.A., Smith, N., Ruprecht, J.K., Boyd, B.W.,1996. Changes in streamflow components following logging and regeneration in the southern forest of Western Australia. Hydrol. Process.10, 447–461. Bates, C.G., Henry, A.J., 1928. Second phase of streamflow experiment at Wagon Wheel Gap, Colo. Mon. Weather Rev. 56, 79-81. Beschta, R. L., 1978. Long-term patterns of sediment production following road construction and logging in the Oregon Coast Range. Water Resour. Res., 14(6), 1011-1016. BC Ministry of Forests, Lands and Natural Resource Operations, 2012. Biogeoclimatic Ecosystem Classification Program-BEC zone and subzone descriptions. 199  http://www.for.gov.bc.ca/hre/becweb/resources/classificationreports/subzones/index.html (Accesed in September, 2012) BC Ministry of Environment, 2012. Fisheries Information Summary System(FISS)-Fisheries Inventory Data Queries (FIDQ). (Accesed in September, 2012) BC Ministry of Forests and Rangeland,1999. Interior Watershed Assessment Guidebook (Appendix 8) 2nd edition, Victoria, British Columbia, Canada. Beaudry, P., Nassey, J., 1994. Watershed Assessment-Middle Reach Bowron River, Prince George Forest Region, BC Ministry of Forests, p38. Bêche, L.A., Stephens, S.L., Resh, V.H., 2005. Effects of prescribed fire on a Sierra Nevada (California, USA) stream and its riparian zone. For. Ecol. Manag. 218, 37–59. Benda, L., Miller, D., Bigelow, P., Andrus, K., 2003. Effects of post-wildfire erosion on channel environments, Boise River, Idaho. J. Forest Ecol. Manage. 178, 105–119. Benda, L., Poff, N.L., Miller, D., Dunne, T., Reeves, G., Pess, G., Pollock, M., 2004. The network dynamics hypothesis: how channel networks structure riverine habitats. BioScience 54(5), 413-427. Bent, G.C., 2001. Effects of forest-management activities on runoff components and groundwater recharge to Quabbin Reservoir, central Massachusetts. Forest Ecol. Manag. 143, 115– 129. Bent, G.C., 2001. Effects of forest-management activities on runoff components and groundwater recharge to Quabbin Reservoir, central Massachusetts, Forest Ecol. Manag.143, 115– 129. Berris, S.N., Harr, R.D., 1987. Comparative snow accumulation and melt during rainfall in forested and clear-cut plots in the Western Cascades of Oregon. Water Resour. Res. 23(1), 135–142. Bethlahmy, N., 1974. More streamflow after a bark beetle epidemic. J. Hydrol. 23(3-4), 185-189. Bethlahmy, N., 1975. A Colorado episode: beetle epidemic, ghost forests, more stream flow. Northwest Sci. 49 (2), 95-105. Beven, K., 1992. The future of distributed models: Model calibration and uncertainty prediction. Hydrol. Process. 6, 279-298. Blöschl, G., Ardoin-Bardin, S., Bonell, M., Dorninger, M., Goodrich, D., Gutknecht, D., Matamoros, D., Merz, B., Shand, P., and Szolgay, J., 2007. At what scales do climate variability and land cover change impact on flooding and low flows?. Hydrol. Process. 21(9), 1241-1247. Boon, S., 2007. Snow accumulation and ablation in a beetle-killed pine stand in Northern Interior British Columbia. BC J. Ecosyst. Manage. 8(3),1–13. Boon, S., 2009a. Impact of mountain pine beetle infestation and salvage harvesting on seasonal snow melt and runoff. Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria, BC. Mountain Pine Beetle Working Paper 2008-24. 28p. Boon, S., 2009b. Snow ablation energy balance in a dead forest stand. Hydrol. Process. 23, 26002610. 200  Boon, S., 2011. Snow accumulation following forest disturbance. Ecohydrology. DOI:10.1002/ eco.212. Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. J. Hydrol. 55(1-4), 3-23. Bouchard, M., Kneeshaw, D., Bergeron, Y., 2008. Ecosystem management based on large-scale disturbance pulses: A case study from sub-boreal forests of western Quebec (Canada). Forest Ecol. Manag. 256, 1734-1742. Bowling, L.C., Storck, P., Lettenmaier, D.P., 2000. Hydrologic effects of logging in western Washington, United States. Water Resour. Res. 36, 3223–3240. Box, G., Jenkins, G.M., 1976. Time series analysis: forecasting and control (revised ed). Holden Day, San Francisco, USA Brandt, M., Bergström, S., Gardelin, M., 1988. Modelling the effects of clearcutting on runoffExamples from Central Sweden. AMBIO 17, 307–313. Brooks, K.N., Folliott, P.F., Gregersen, H.M., DeBano, L.F., 2003. Hydrology and the Management of Watersheds. Blackwell, Ames, IA. Brown, A.E., Zhang, L., McMahon, T.A., Western, A.W., Vertessy, R.A., 2005. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. J. Hydrol. 310(1-4), 28-61. Bruijnzeel, L., 2004. Hydrological functions of tropical forests: not seeing the soil for the trees? Agric. Ecosyst. Environ. 104(1), 185-228. Budyko, M., Miller, D. H., 1974. Climate and life. Academic Press. New York. Buttle, J.M., 1996. Identifying hydrological responses to basin restoration: an example from southern Ontario. In: McDonnell, J.J., Stribling, J.B., Neville, L.R. et al (ed), Watershed restoration management: physical, chemical, and biological considerations. American Water Resources Association, Herndon, pp 5–13. Buttle, J., Metcalfe, R., 2000. Boreal forest disturbance and streamflow response, northeastern Ontario. Can. J. Fish. Aquat. Sci. 57(S2), 5-18. Caissie, D., Jolicoeur, S., Bouchard, M., Poncet, E., 2002. Comparison of streamflow between pre and post timber harvesting in Catamaran Brook (Canada). J. Hydrol. 258, 232-248. Calder, I.R., 1993. Hydrologic effects of land-use change. In: Maidment, D.R. (Ed.), Handbook of Hydrology. McGraw-Hill, New York, 13.11-13.50. Calder, I.R., 2005. Blue revolution – integrated land and water resources management, 2nd ed. Earthscan, London, UK. Campbell, R.E., Baker, M.B., Folliott, P.E., Larson F.R., Avery, C.C., 1977. Wildfire effects on a ponderosa pine ecosystem: an Arizona case study. USDA Forest Service Research Paper RM-191, 1-12. Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G., Peñuela, J., 2011. Widespread crown condition decline, food web disruption, and amplified tree mortality with increased climate change-type drought. Proc. Natl. Acad. Sci 108,1474-1478.  201  Chatfield, C., 1989. The analysis of time series: an introduction. Chapman & Hall, London, UK Chen, W., Wei, X., 2008. Assessing the relations between aquatic habitat indicators and forest harvesting at watershed scale in the interior of British Columbia. Forest Ecol. Manag. 256, 152-160. Cheng, J.D., Black, T.A., DeVries, J., Willington, R.P., Goodell, B.C., 1975. The evaluation or initial changes in peak streamflow following logging or a watershed on the West Coast or Canada. Int. Assoc. Hydrol. Sci. Publ. 117:475-486. Cheng, J., 1989. Streamflow changes after clear-cut logging of a pine beetle-infested watershed in southern British Columbia, Canada. Water Resour. Res. 25(3), 449-456. Choi, W., Deal, B.M., 2008. Assessing hydrological impact of potential land use change through hydrological and land use change modeling for the Kishwaukee River basin (USA). J. Environ. Manage. 88, 1119-1130 Cobourn, J., 1989. Application of cumulative watershed effects (CWE) analysis on the Eldorado National Forest in California. In: the symposium on headwaters hydrology, American Water Resources Association, Bethesda Maryland, p 449-460. Cong, Z., Yang, D., Gao, B., Yang, H., Hu, H., 2009. Hydrological trend analysis in the Yellow River basin using a distributed hydrological model. Water Resour. Res. 45, W00A13. Cornish, P.M., Vertessy, R.A., 2001. Forest age-induced changes in evapotranspiration and water yield in a eucalypt forest. J. Hydrol. 242, 43-63. Costa, M.H., Botta, A., Cardille, J.A., 2003. Effects of large-scale changes in land cover on the discharge of the Tocantins River, Southeastern Amazonia. J. Hydrol. 283(1-4), 206-217. Dale, V.H., Joyce, L.A., McNulty, S., Neilson, R.P., 2000. The interplay between climate change, forests, and disturbances. Sci. Total Environ. 262, 201-204. David, J.S., Henriques, M.O., David, T.S., Tomé, J., Ledger, D.C., 1994. Clearcutting effects on streamflow in coppiced Eucalyptus globulus stands in Portugal. J. Hydrol. 162, 143–154. Department of Fisheries and Oceans. 1990. Fish Habitat Inventory and Information Program. Stream Summary Catalogue. Sub-district 29H Quesnel. Vancouver, B.C. DHI (Danish Hydraulic Institute), 2004. MIKE SHE-an integrated hydrological modeling system-user guide, edition2004, Denmark. Dunne,T., Leopold, L.B., 1978. Water in Environmental Planning. W. H. Freeman and Co, San Francisco. Dyrness, C.T., 1976. Effects of wildfi re on soil wettability in the high Cascades of Oregon. USDA Forest Service Research Paper PNW-202, Corvallis, OR, 18 pp. Eaton, B.C., Moore, R.D., Giles, T., 2010. Forest fire, root strength and channel instability: the “unusual” response of Fishtrap Creek, British Columbia. Earth Surf. Proc. Land. 35(10) 1167-1183. Edwards, P.J., Troendle, C.A., 2012. Water yield and hydrology. In: LaFayette, R., Brooks, M.T., Potyondy, J.P., Audin, L., Krieger, S.L., Trettin, C. (Eds.), Cumulative watershed effects of fuel management in the Eastern United States. Gen. Tech. Rep. SRS-161, 2012.  202  Asheville, NC: U.S. Department of Agriculture Forest Service, Southern Research Station. pp 229-281. Eschner, A.R., Satterlund, D.R., 1966. Forest protection and streamflow from an Adirondack watershed. Water Resour. Res. 2, 765-783. Fohrer, N., Haverkamp, S., Frede, H.G., 2005. Assessment of the effects of land use patterns on hydrologic landscape functions: development of sustainable land use concepts for low mountain range areas. Hydrol. Process. 19, 659-672. Forest Practices Board, 2007. The effect of mountain pine beetle attack and salvage harvesting on streamflows. Victoria, B.C. Spec. Invest. Rep. No. FPB/SIR/16. www.for.gov.bc.ca/hfd/library/ documents/bib106689.pdf (Accessed May2012). Francis, R.C., Hare, S.R., 1994. Decadal scale regime shifts in the large marine ecosystems of the Northeast Pacific: a case for historical science. Fish. Oceanogr. 3(4), 279-291. Franczyk, J., 2008. The effects of climate change and urbanization on the runoff of the Rock Creek basin in the Portland metropolitan area, Oregon, USA. Hydrol. Process. 23, 805-815. Gluns, D.R., 2001. Snowline pattern during the melt season: evaluation of the H60 concept. In: Toews, D.A., Chatwin, S. (Eds.),Watershed assessment in the Southern Interior of British Columbia: Workshop Proceedings, Pencticton, March 2000. BC Ministry of Forests Research Program. pp. 68-80 Gottfried, G.J., 1991. Moderate timber harvesting increases water yields from an Arizona mixed conifer watershed. Water Resour. Bull. 27, 537-547. Green, K.C., Alila, Y., 2012. A paradigm shift in understanding and quantifying the effects of forest harvesting on floods in snow environments, Water Resour. Res. 48(10), W10503. Greenberg, L., Svendsen, P., Harby, A., 1996. Availability of microhabitats and their use by brown trout (Salmo trutta) and grayling (Thymallus thymallus) in the River Vojman, Sweden. Regul. River. Res. Manage.12, 287-303. Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from temperature. Appl. Eng. Agric. 1(2), 96-99. Harr, D.R., 1983. Potential for augmenting water yield through forest practices in western Washington and western Oregon. Water Resour. Bull. 19, 383–393. Harr, R.D., Levno, A., Mersereau, R., 1982. Streamflow changes after logging 130-year-old Douglas fir in two small watersheds, Water Resour. Res.18, 637-644. Harr, R.D.,1976. Forest practices and streamflow in western Oregon, Gen. Tech. Rep. PNW-49, 18 pp., For. Serv., U.S. Dep. of Agric., Portland, Oregon. Harr, R.D., Harper, W.C., Krygier, J.T., Hsieh, F.S., 1975. Changes in storm hydrographs after road building and clear-cutting in the Oregon Coast Range. Water Resour. Res. 11, 436-444. Harr, R.D.,1979. Effects of timber harvest on streamflow in the rain-dominated portion of the Pacific Northwest. In: Proceedings of workshop on scheduling timber harvest for hydrologic concerns, Pacific Northwest Forest and Range Experiment Station, Portland, USDA, Oregon, pp1-45  203  Harr, R.D.,1982. Fog Drip in the Bull Run Municipal Watershed, Oregon. Water Resour. Bull.18, 785-789. Harris, D.D., 1977. Hydrologic changes after logging in two small Oregon coastal watersheds, U.S. Geol. Surv. Water Supply Paper 2037, pp31. Haughian, S.R., Burton, P.J., Taylor, S.W., Curry, C.L., 2012. Expected effects of climate change on forest disturbance regimes in British Columbia. BC J. Ecosyst. Manage. 13(1), 1– 24. Healey, M.C., 1991. Life History of Chinook Salmon (Oncorhynchus tshawytscha). In: Groot, C. and Margolis, L.(Eds.), Pacific salmon life histories. UBC Press Vancouver, Canada. pp313 393. Heard, W.R. 1991. Life History of Pink Salmon (Oncorhynchus gorbuscha). In: Groot, C., Margolis, L. (Eds.), Pacific salmon life histories. UBC Press Vancouver, Canada. pp121 230. Helvey, J.D., 1980. Effects of a north central Washington wildfire on runoff and sediment production. Water Resour. Bull.16, 627-634. Hetherington, E.D., 1987. The importance of forests in the hydrological regime. In: Healy, M.D. and Wallace, R.R. (Eds.), Canadian Bulletin Fisheries and Aquatic Sciences, Ottawa, Ont., No. 215, pp 533. Hicks, B.J., Beschta, R.L., Harr, R.D., 1991. Long-term changes in streamflow following logging in western Oregon and associated fisheries implications. Water Resour. Bull. 27, 217-226. Holland, S., 1976. Landforms of British Columbia. The Government of British Columbia, Victoria, Canada. Hoyt, W.G., Troxell, H.C., 1934. Forest and stream flow. American Society of Civil Engineers, Paper No. 1858, 1–111. Huang, M., Zhang, L., Gallichand, J., 2003. Runoff responses to afforestation in a watershed of the Loess Plateau, China. Hydrol. Process.17(13), 2599-2609. Hudson, R., 2000. Snowpack recovery in regenerating coastal British Columbia clearcuts. Can. J. For. Res. 30(4), 548-556. Huff, D., Hargrove, B., Tharp, M., Graham, R., 2000. Managing Forests for Water Yield: The Importance of Scale. J. Forest. 98(12),15-19. Huggard, D., Lewis, D., 2006. Effects of salvage options for beetle-killed pine stands on ECA: A synthesis of currently available data and uncertainties. B.C. Ministry of Environment, Kamloops, B.C., draft report. Iroumé, A., Mayen, O., Huber, A., 2006. Runoff and peak flow responses to timber harvest and forest age in southern Chile. Hydrol. Process. 20, p. 37–50. IWAP (Interior Watershed Assessment Procedure), 2006. Interior Watershed Assessment Guidebook, BC Ministry of Forests and Range, Victoria, BC Canada Jassby, A.D., Powell, T.M., 1990. Detecting changes in ecological time series. Ecology 71(6), 2044-2052. 204  Jones, J.A., Grant, G.E., 1996. Peak flow responses to clear-cutting and roads in small and large basins, western Cascades, Oregon. Water Resour. Res. 32, 959-974. Jones, J.A. et al., 2012. Ecosystem processes and human influences regulate streamflow response to climate change at long-term ecological research sites. Bioscience 62(4). Jones, J.A., 2000. Hydrologic processes and peak discharge response to forest removal, regrowth, and roads in 10 small experimental basins, Western Cascades, Oregon, Water Resour. Res. 36(9), 2621–2642. Jones, J.A., Post, D.A., 2004. Seasonal and successional streamflow response to forest cutting and regrowth in the northwest and eastern United States. Water Resour. Res. 40(5), W05203. Jost, G, Moore, R.D., Weiler, M., Alila, Y., 2008. A combined statistical and mechanistic modeling approach to study the effect of clear-cutting in a snow dominated watershed. In: Proceedings of the American Geophysical Union Fall Meeting, San Francisco, 15-19 Dec, 2008. Juckem, P.F., Hunt, R.J., Anderson, M.P., Robertson, D.M., 2008. Effects of climate and land management change on streamflow in the driftless area of Wisconsin. J. Hydrol. 355(1-4), 123-130. King, J.G., Tennyson, L.C., 1984. Alteration of streamflow characteristics following road construction in north central Idaho, Water Resour. Res. 20(8), 1159-1163. King, J., 1989. Streamflow responses to road building and harvesting: a comparison with the equivalent clearcut area procedure. Research Paper INT-401. Intermountain Research Station, USDA Forest Service, Ogden. Kirchner, J.W., 2006. Getting the right answers for the right reasons: Linking measurements, analyses, and models to advance the science of hydrology. Water Resour. Res. 42, W03S04. Kirchner, J.W., 2009. Catchments as simple dynamical systems: Catchment characterization, rainfall-runoff modeling, and doing hydrology backward. Water Resour. Res., 45, W02429. Knox, J.C., 1972. Valley alluviation in southwestern Wisconsin. Ann Assoc Am Geogr. 62, 401410. Krammes, J.S., Rice, R.M., 1963. Effect of fire on the San Dimas experimental forest. In: Arizona Watershed Symposium, Proceedings of 7th Annual Meeting, Phoenix, Arizona, pp31–34. Kuraś, P.K., Alila, Y., Weiler, M., 2012. Forest harvesting effects on the magnitude and frequency of peak flows can increase with return period. Water Resour. Res. 48, W01544. La Marche, J.L., Lettenmaier, D.P., 2001. Effects of forest roads on flood flows in the Deschutes River, Washington. Earth Surf. Proc. Land 26,115-134. Langford, K.J., 1976. Change in yield of water following a bushfire in a forest of Eucalyptus regnans. J. Hydrol. 29, 87–114. Letey, J., 2001. Causes and consequences of fire-induced soil water repellency. Hydrol. Process.15, 2867–2875. Lewis, D., Huggard, D., 2010. A model to quantify effects of mountain pine beetle on equivalent clearcut area. Streamline Watershed Manag. Bull. 13(2) , 42-51. 205  Lewis, J., Reid, L.M., Thomas, R.B., 2010. Comment on ‘‘Forests and floods: A new paradigm sheds light on age-old controversies’’ by Younes Alila et al..Water Resour. Res. 46, W05801. Li, K., Coe, M., Ramankutty, N., Jong, R.D., 2007a. Modeling the hydrological impact of landuse change in West Africa. J. Hydrol. 337(3-4), 258-268. Li, L., Zhang, L., Wang, H., Wang, J., Yang, J., Jiang, D., Li, J., Qin, D., 2007b. Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China. Hydrol. Process. 21(25), 3485-3491. Lin, Y., Wei, X., 2008. The impact of large-scale forest harvesting on hydrology in the Willow watershed of Central British Columbia. J. Hydrol. 359(1-2), 141-149. Lindley, A.J., Bosch, J.M., van Wyk, D.B., 1988. Changes in water yield after fire in fynbos catchments. Water South Af. 14(1), 7–12. Logan, J.A., Régnière, J., Powell, J.A., 2003. Assessing the impacts of global warming on forest pest dynamics. Front. Ecol. Environ. 1,130-137. Ma, X., Xu, J., van Noordwijk, M., 2010. Sensitivity of streamflow from a Himalayan catchment to plausible changes in land cover and climate. Hydrol. Process. 24(11), 1379-1390. Ma, X., Xu, J.C., Luo, Y., Shiv, P.A., Li, J.T., 2009. Response of hydrological processes to landcover and climate changes in Kejie watershed, south-west China. Hydrol. Process. 23, 11791191 Macdonald, J.S., Beaudry, P.G., Maclsaac, E.A., Herunter, H.E., 2003. The effects of forest harvesting and best management practices on streamflow and suspended sediment concentrations during snowmelt in headwater streams in sub-boreal forests of British Columbia, Canada. Can. J. Forest Res. 33, 1397-1407. Marchetti, M.P., Moyle, P.B., 2001. Effects of flow regime on fish assemblages in a regulated California stream. Ecol. Appl. 11, 530–539. Matheussen, B., Kirschbaum, R.L., Goodman, I.A., O'Donnell, G.M., Lettenmaier, D.P., 2000. Effects of land cover change on streamflow in the interior Columbia River Basin (USA and Canada). Hydrol. Process. 14(5), 867-885. Mayor, A.G., Bautista, S., Lovet, J., Bellot, J., 2007. Post-fire hydrological and erosional responses of a Mediterranean landscape: Seven years of catchment-scale dynamics. Catena 71, 68–75. Mbogga, M., Hamann, A., Wang, T., 2009. Historical and projected climate data for natural resource management in western Canada. Agr. Forest Meteorol.149, 881-890. McCormick, B.C., Eshleman, K.N., Griffith, J.L., Townsend, P.A., 2009. Detection of flooding responses at the river basin scale enhanced by land use change. Water Resour. Res. 45(8), W08401. McGurk, B.J., Fong, D.R., 1995. Equivalent roaded area as a measure of cumulative effect of logging. Environ. Manage.19, 609-621.  206  Menning, K., Erman, D., Johnson, K., Sessions, J., 1996. Modeling aquatic and riparian systems, assessing cumulative watershed effects, and limiting watershed disturbance. Addendum, chapter 2, Sierra Nevada Aquatic Research Lab, Mammoth Lakes, USA, pp 33-51 Milly, P., Dunne, K., 2002. Macroscale water fluxes 2: water and energy supply control of their interannual variability. Water Resour. Res. 38, 1206. Mitchell, T.D., Jones, P.D., 2005. An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int. J. Climatol. 25, 693-712. Montgomery, W.L., McCormick, S.D., Naiman, R.J., Whoriskey, F.G., Black, G.A., 1983. Spring migratory synchrony of salmonid, catostomid, and cyprinid fishes in Rivière á la Truite, Québec. Can. J. Zool. 61, 2495-2502. Moody, J.A., Martin, D.A., 2001. Post-fire, rainfall intensity-peak discharge relations for three mountainous watersheds in the western USA. Hydrol. Process.15, 2981-2993. Moore, K.M.S., Gregory, S.V., 1988. Response of young-of-the-year cutthroat trout to manipulations of habitat structure in a small stream. T. Am. Fish. Soc. 117, 162-170. Moore, R., Wondzell, S., 2005. Physical hydrology and the effects of forest harvesting in the Pacific Northwest: A review. J. Am. Water Reour. Assoc. 41(4),763-784. Nasseri, I., 1988. Frequency of floods from a burned chaparral watershed. In: Proceeding of the symposium on fire and watershed management. General Technical Report PSW-109, USDA (Forest Service). Berkeley, California. National Research Council (NRC), 1996. Upstream: salmon and society in the Pacific Northwest. National Academy Press: Washington (DC). Neary, D.G., Gottfried, G.J., Folliott, P.E., 2003. Post-wildfire watershed flood response. In: Proceedings of the 2nd International Fire Ecology Conference, Orlando, Florida, 16-20 November, Paper 1 B.7. Nelitz, M., Porter, M., Parkinson, E., Wieckowski, K., Marmorek, D., Bryan, K., Hall, A., Abraham, D., 2011. Evaluating the status of Fraser River sockeye salmon and role of freshwater ecology in their decline. ESSA Technologies Ltd. Cohen Commission Tech. Rept. 3: 222p. Vancouver, B.C. www.cohencommission.ca/en/pdf/TR/Project3-Report.pdf (Accessed on October 2012) Nzsje, T., Jonsson, B., Skurdal, J., 1995. Spring flood: a primary cue for hatching of river spawning Coregoninae. Can. J. Fish. Aquat. Sci. 52, 2190-2196. Pettit, N.E., Froend R.H., Davies, P.M., 2001. Identifying the natural flow regime and the relationship with riparian vegetation for two contrasting western Australian rivers. Regul. River. Res. Manag. 17, 201–215. Pike, R., Scherer, R., 2003. Low flows in snowmelt dominated watersheds. Streamline Watershed Manage. Bull. 8(1), 24-28. Pike, R.G., Redding, T., Wilford, D., Moore, R.D., Ice, G., Reiter, M., Toews, D., 2007. Chapter 14 – Detecting and predicting changes in watersheds [Draft]. In: Compendium of Forest Hydrology and Geomorphology in British Columbia [In Prep.] Pike, R.G. et al. (Eds). Land  207  Management Handbook (TBD). B.C. Ministry of Forests and Range Research Branch, Victoria, B.C. and FORREX Forest Research Extension Partnership, Kamloops, Canada. Pizarro, R., Araya, S., Jordan, C., Farías, C., Flores, J., Bro, P., 2006. The effects of changes in vegetative cover on river flows in the Purapel river basin of central Chile. J. Hydrol. 327 (1– 2), 249–257. Poff, N.L. et al., 2010a. The ecological limits of hydrologic alteration (ELOHA): a new framework for developing regional environmental flow standards. Freshwater Biol. 55, 147– 170. Poff, N.L., Allan, J.D., Bain, M.B., Karr J.R., Prestegaard, K.L., Richter, B.D., Sparks, R.E., Stromberg, J.C., 1997. The natural flow regime: a paradigm for river conservation and restoration. BioScience 47,769–784. Poff, N.L., Zimmerman, J.K.H., 2010b. Ecological responses to altered flow regimes: a literature review to inform the science and management of environmental flows. Freshwater Biol. 55,194–205. Potts, D.F., 1984. Hydrologic impacts of a large scale mountain pine beetle (Dendroctonus ponderosae Hopkins) epidemic. Water Resour. Bull., Paper No. 83122: 373-377. Putz, G., Burke, J., Smith, D.W., Chanasyk, D.S., Prepas, E.E., Mapfumo, E., 2003. Modelling the effects of boreal forest landscape management upon streamflow and water quality: Basic concepts and considerations. J. Environ. Eng. Sci. S2, 87-101. Reeves, G.H., Benda, L.E., Burnett, K.M., Bisson, P.A., Sedell, J.R.,1996. A disturbance-based ecosystem approach to maintaining and restoring freshwater habitats of evolutionarily significant units of anadromous salmonids in the Pacific Northwest. Am. Fish. Soc. Symp. 17, 334-349. Ring, P.J., Fisher, I.H., 1985. The effects of changes in land use on runoff from large catchments in the upper Macintyre Valley, NSW. In: Hydrology and Water Resources Symposium, Sydney, 14 -16 May, 1985, the Institution of Engineers, Australia, National Conference Publication 85/2:153-158. Roberts, M.R., 2007. A conceptual model to characterize disturbance severity in forest harvests. Forest Ecol. Manag. 242, 58-64. Robinson, M., Dupeyrat, A., 2005. Effects of commercial timber harvesting on streamflow regimes in the Plynlimon catchments, mid-Wales. Hydrol. Process. 19, 1213–1226. Rodionov, S. N., 2005. A brief overview of the regime shift detection methods. In: Proceedings of UNESCO-ROSTE/BAS Workshop on Regime Shifts, Varna, Bulgaria. Rodriguez, D.A., Tomasella, J., Linhares, C., 2010. Is the forest conversion to pasture affecting the hydrological response of Amazonian catchments? Signals in the Ji-Paraná Basin. Hydrol. Process. 24, 1254–1269. Rood, K.M., Hamilton, R.E., 1995. Hydrology and water use for salmon streams in the Quesnel Habitat Management Area, British Columbia. Canadian Manuscript Report of Fisheries and Aquatic Sciences 2296. http://dsp-psd.pwgsc.gc.ca/collection_2007/dfompo/ Fs97-42296E.pdf. (Accessed in October, 2012)  208  Rood, S.B., Mahoney, J.M., 1990. Collapse of riparian poplar forests downstream fromdams in western prairies: probable causes and prospects for mitigation. Environ. Manage.14, 451464. Sahin, V., Hall, M.J., 1996. The effects of afforestation and deforestation on water yield. J. Hydrol. 178, 293-309. San Dimas Technical Staff. 1954. Fire-flood sequences on the San Dimas experimental forest. US Department of Agriculture, California Forest and Range Experimental Station, Technical Paper No. 6; 29. Sankarasubramanian, A., Vogel, R.M., Limbrunner, J.F., 2001. Climate elasticity of streamflow in the United States. Water Resour. Res. 37(6), 1771-1781. Schindler, D., 2001.The cumulative effects of climate warming and other human stresses on Canadian freshwaters in the new millennium. Can. J. Fish. Aquat. Sci., 58, 18-29. Schmid, J.M., Mata, S.A., Martinez, M.H., Troendle, C.A., 1991. Net precipitation within small group infestations of the mountain pine beetle. U.S. Dep. Agric. For. Serv., Rocky Mtn. Forest and Range Ex. Stn., Fort Colllins, Colo. Res. Note RM-508. Schnorbus, M., Alila, Y., 2004. Forest harvesting impacts on the peak flow regime in the Columbia Mountains of southeastern British Columbia: An investigation using long-term numerical modeling. Water Resour. Res. 40, W05205. Scott, D.F., 1993. The hydrological effects of fire in South African mountain catchments. J. Hydrol. 150, 409-432. Scott, D.F., Prinsloo, F., 2008. Longer-term effects of pine and eucalypt plantations on streamflow. Water Resour. Res. 44(7), W00A08. Shakesby, R.A., Doerr, S.H. 2006. Wildfire as a hydrological and geomorphological agent. Earth Sci. Rev. 74, 269-307. Sharma, K.P., Vorosmarty, C.J., Moore, B., 2000. Sensitivity of the Himalayan hydrology to land-use and climatic changes. Climatic Change 47(1), 117-139. Shrestha, R.J., Schnorbus, M.A., Werner, A.T., Berland, A.J., 2012. Modelling spatial and temporal variability of hydrologic impacts of climate change in the Fraser River basin, British Columbia, Canada. Hydrol. Process. 26, 1840–1860. Shuttleworth, W.J., 1993. Evaporation. In: Maidment, D.R. (Eds.), Handbook of Hydrology, McGraw-Hill, New York, p418. Silins, U., 2003. An integrated forest-watershed planning and assessing model: ECA-Alberta. http://www.cefm.rr.ualberta.ca/Research_Notes/Research%20Note%2007-03.pdf (Accessed 11 Nov 2008) Sinclair, J.D., Hamilton, E.L., 1955. Streamflow reactions of a fire-damaged watershed. American Society of Civil Engineers, Hydraulics Division, 81, Separate No. 629. Siriwardena, L., Finlayson, B., McMahon, T., 2006. The impact of land use change on catchment hydrology in large catchments: The Comet River, Central Queensland, Australia. J. Hydrol. 326(1-4), 199-214.  209  Soler, M., Sala, M., Gallart, F., 1994. Post fire evolution of runoff and erosion during an eighteen month period. In: Sala, M., Rubio, J.L. (Eds.), Soil Erosion and Degradation as a Consequence of Forest Fires. Geoforma Ediciones, Logroño, pp. 149– 162. Soto, B., Basanta, R., Benito, E., Pérez, R., Diaz-Fierros, F., 1994. Runoff and erosion from burnt soils in Northwest Spain. In: Sala, M., Rubio, J.L. (Eds.), Soil Erosion as a Consequence of Forest Fires. Geoforma Ediciones, Logroño, Spain, pp. 91– 98. Sparks, R.E., 1992. Risks of altering the hydrologic regime of large rivers. In: Cairns, J., Niederlehner, B.R., Orvos, D.R.(Eds.), Predicting ecosystem risk-Advances in modern environmental toxicology.. Princeton Scientific Publishing Co., Princeton (NJ). pp119-152. Sparks, R.E., 1995. Need for ecosystem management of large rivers and their floodplains. BioScience 45, 168-182. Stanford, J.A., 1994. Instream flows to assist the recovery of endangered fishes of the Upper Colorado River Basin. US Department of the Interior, National Biological Survey. Biological Report nr 24,Washington (DC). Stanford, J.A., Ward, J.V., Liss, W.J., Frissell, C.A., Williams, R.N., Lichatowich, J.A., Coutant, C.C., 1996. A general protocol for restoration of regulated rivers. Regul. River. Res. Manage.12, 391-414. Stednick, J.D., 1996. Monitoring the effects of timber harvest on annual water yield. J. Hydrol. 176(1-4), 79-95. Stednick, J. D., 2008. Long-term streamflow changes following timber harvesting. In: Stednick, J. D.(Eds.), Hydrological and Biological Responses to Forest Practices: The Alsea Watershed Study. Springer, New York, pp139– 155. Stonesifer, C.S., 2007. Modeling the Cumulative Effects of Forest Fire on Watershed Hydrology: A Post-fire Application of the Distributed Hydrology-Soil-Vegetation Model (DHSVM). J. Hydrol.10, 282-290. Storck, P., Bowling, L., Wetherbee, P., Lettenmaier, D., 1998. Application of a GIS-based distributed hydrology model for prediction of forest harvest effects on peak stream flow in the Pacific Northwest. Hydrol. Process. 12, 889-904. Storck, P., Bowling, L., Wetherbee, P., Lettenmaier, D., 1998. Application of a GIS-based distributed hydrology model for prediction of forest harvest effects on peak stream flow in the Pacific Northwest. Hydrol. Process.12, 889-904. Stromberg, J.C., Beauchamp, V.B., Dixon, M.D., Lite, S.J., Paradzick, C., 2007. Importance of low-flow and high-flow characteristics to restoration of riparian vegetation along rivers in arid south-western United States. Freshwater Biol. 52, 651–679. Sun, G., McNulty, S.G., Lu, J., Amatya, D.M., Liang, Y., Kolka, R.K., 2005. Regional annual water yield from forest lands and its response to potential deforestation across the southeastern United States. J. Hydrol. 308, 258-268. Swank, W.T., Swift, J.R., Douglas, J.E., 1988. Streamflow changes associated with forest cutting, species conversion and natural disturbance. In: Forest Hydrology and Ecology at Coweeta, Ecological Studies 66, Swank, W.T., Crossley, D.A. (Eds.), Springer, New York, 297–312, 1988. 210  Swanson, R.H., Golding, D.L., Rothwell, R.L., Bernier, P.Y., 1986. Hydrologic effects of clearcutting at marmot creek and streeter watersheds, Alberta. In: Northern Forestry Centre Information Report NOR-X-278, pp27, Canadian Forestry Service, Edmonton, Alberta, Canada. Tague, C.L., Band, L.E., 2004. RHESSys: Regional hydro-ecologic simulation system-an objectoriented approach to spatially distributed modeling of carbon, water, and nutrient cycling. Earth Interact. 8, 1-42. Talbot, J., Plamondon, A.P., 2002. The diminution of snowmelt rate with forest regrowth as an index of peak flow hydrologic recovery, Montmorency Forest, Quebec. In: Proceedings of the 59th Eastern Snow Conference, Stowe, Vermont, USA. Thanapakpawin P., Richey, J., Thomas, D., Rodda, S., Campbell, B., Logsdon, M., 2007. Effects of landuse change on the hydrologic regime of the Mae Chaem river basin, NW Thailand. J. Hydrol. 334, 215-230. Thomas, R.B., Megahan, W.F., 1998. Peak flow responses to clear-cutting and roads in small and large basins, western Cascades, Oregon: A second opinion. Water Resour. Res. 34, 33933403. Thrower, J.S., Goudie, J.W., 1992. Development of height-age and site index functions for evenaged Interior Douglas-fir in British Columbia. B.C. Ministry of Forests Forest Science Research Branch, Research Note 109. Thrower, J.S., Nussbaum, A.F., Di Lucca, C.M., 1994. Site index curves and tables for British Columbia: interior species (2nd edition). B.C. Ministry of Forests Forest Science Research Branch Land Management Handbook, Field Guide Insert 6. Thyer, M., Beckers, J., Spittlehouse, D., Alila, Y., Winkler, R., 2004. Diagnosing a distributed hydrologic model for two high-elevation forested catchments based on detailed stand- and basin-scale data. Water Resour. Res. 40, W01103. Toews, D.A., Gluns, D.R., 1986. Snow accumulation and ablation on adjacent forested and clearcut sites in southeastern British Columbia. In: Proceedings of the 54th Annual Western Snow Conference, pp101–111, West. Snow Conf., Phoenix, Ariz. Tonina, D., Luce, C., Rieman, B., Buffington, J.M., Goodwin, P., Clayton, S.R., Ali, S. M., Barry, J.J., Berenbrock, C., 2008. Hydrological response to timber harvest in northern Idaho: Implications for channel scour and persistence of salmonids. Hydrol. Process. 22, 3223– 3235. Trimble, S.W., Weirich, F.H., Hoag, B.L. 1987. Reforestation and the reduction of water yield on the Southern Piedmont since circa 1940. Water Resour. Res., 23(3), 425-437. Troendle, C.A., King, R.M., 1987. The effect of partial and clearcutting on streamflow at Deadhorse Creek, Colorado, J. Hydrol. 90, 145–157. Troendle, C.A., Kaufmann, M., 1987. Influence of forests on the hydrology of the subalpine forest. In: Troendle, C.A., Kaufmann, M.R., Hamre, R.H., Winokur, R.P. (Eds.), Management of subalpine forests: building on 50 years of research , pp 68-78. U.S. Forest Service, Rocky Mountain Forest and Range Experiment Station, Gen. Tech. Report, RM149. 211  Troendle, C.A., King, R.M., 1985. The effect of timber harvest on the Fool Creek watershed, 30 years later. Water Resour. Res. 21(12), 1915–1922. Troendle, C.A., Nankervis, J.E., 2000. Estimating additional water yield from changes in management of national forests in the North Platte Basin. Report submitted to the U.S. Bureau of Reclamation, Lakewood, CO. p51. Troendle, C.A., 2007. The potential for water yield augmentation from forest management in the Rocky Mountain region. J. Am. Water Resour. Assoc. 19(3), 359-373. Tuteja, N.K., Vaze, J., Teng, J., Mutendeudzi, M., 2007. Partitioning the effects of pine plantations and climate variability on runoff from a large catchment in southeastern Australia. Water Resour. Res. 43(8), W08415. USDEBPA Division of Fish and Wildlife, 1984. Natural propagation and habitat inventory – Washington volume iiB -Similkameen river habitat inventory-final report. Portland, Oregon. https://pisces.bpa.gov/release/documents/documentviewer.aspx?doc=247 (Acessed in September 2012) USDA Forest Service, 1974. Forest hydrology. Part II: Hydrologic effects of vegetation manipulation, Missoula, Montana, USA, pp229. Uunila, L., Guy, B., Pike, R., 2006. Hydrologic effects of mountain pine beetle in the interior pine forests of British Columbia: Key questions and current knowledge. Streamline Watershed Manage. Bull. 9, 1–6. Van Haveren, B.P., 1988. A reevaluation of the Wagon Wheel Gap forest watershed experiment. Forest Sci. 34, 208-214. VanShaar, J.R., Haddeland, I., Lettenmaier, D.P., 2002. Effects of land cover changes on the hydrological response of interior Columbia River basin forested catchments. Hydrol. Process. 16(13), 2499-2520. Verry, E.S., Lewis, J.R., Brooks, K.N., 1983. Aspen clearcutting increases snowmelt and storm flow peaks in north central Minnesota. Water Resour. Bull. 19, 59–67. Verry, E.S., Lewis, J.R., Brooks, K.N., 1983. Aspen Clearcutting Increases Snowmelt and Storm Flow Peaks in North Central Minnesota. Water Resour. Bull. 19, 59-67. Voinov, A., Fitz, C., Boumans, R., Costanza, R., 2004. Modular ecosystem modeling. Environ. Model Softw. 19, 285-304. Vose, J.M., Sun, G., Ford, C.R., Bredemeier, M., Otsuki, K., Wei, X., Zhang, Z., Zhang, L., 2011. Forest ecohydrological research in the 21st century: what are the critical needs? Ecohydrology 4(2), 146-158. Wahi, A.K., 2005. Quantifying mountain system recharge in the Upper San Pedro Basin, Arizona, using geochemical tracers. M.S. thesis, Department of Hydrology and Water Resources, University of Arizona, Tucson, Arizona. Wahi, A.K., Hogan, J.F., Ekwurzel, B., Baillie, M.N., Eastoe, C.J., 2008. Geochemical quantification of semiarid mountain recharge. Ground Water 46, 414-425.  212  Wang, G., Xia, J., Chen, J., 2009. Quantification of effects of climate variations and human activities on runoff by a monthly water balance model: A case study of the Chaobai River basin in northern China. Water Resour. Res. 45(7), W00A11. Webb, A., Jarrett, B., Turner, L., 2007. Effects of plantation forest harvesting on water quality and quantity: Canobolas State Forest, NSW. In: Proceedings of the 5th Australian Stream Management Conference, Australian Rivers: Making a Difference, Wilson, A., Deehan, R., Watts, R., Page, K., Bownan, K., Curtis, A. (Eds.), Charles Sturt University, Thurgoona, Australia. Wei, A., Davidson, G., 1998. Impacts of large-scale timber harvesting on the hydrology of the Bowron River Watershed. In: Proceedings of the 51st Annual CWRA Conference, Mountains to Sea: Human Interaction with the Hydrologic Cycle, Victoria, Canada, June 10 12, pp45-52. Wei, W., 1990. Time Series Analysis: Univariate and Multivariates Methods. Addison-Wesley, Redwood City, CA. Wei, X., Zhang, M., 2010a. Research methods for assessing the impacts of forest disturbance on hydrology at large-scale watersheds. In: Chen, J., Chao, L., Lafortezza, R.(Eds.) , Landscape Ecology and Forest Management: Challenges and Solutions in a Changing Globe. Higher Educ. Press, Beijing & Springer-Verlag, Berlin Heidelberg, pp119–147. Wei, X., Zhang, M., 2010b. Quantifying streamflow change caused by forest disturbance at a large spatial scale: A single watershed study. Water Resour. Res. 46(12), W12525. Whitaker, A., Alila, Y., Beckers, J., Toews, D., 2002. Evaluating peak flow sensitivity to clearcutting in different elevation bands of a snowmelt-dominated mountainous catchment. Water Resour. Res. 38(9), 1172. Wilk, J., Andersson, L., Plermkamon, V., 2001. Hydrological impacts of forest conversion to agriculture in a large river basin in northeast Thailand. Hydrol. Process. 15(14), 2729-2748. Winkler, R., Boon, S., Zimonick, B., Baleshta, K., 2010. Assessing the effects of post-pine beetle forest litter on snow albedo. Hydrol. Process. 24, 803–812. Winkler, R., Boon, S., 2010. The effects of mountain pine beetle attack on snow accumulation and ablation: a synthesis of ongoing research in British Columbia. Streamline Watershed Manag. Bull. 13(2), 25-31. www.forrex.org/publications/streamline/ISS42/Streamline_Vol13_No2_art3. pdf (Accessed in May 2012). Winkler, R., Boon, S., 2009. A summary of research into the effects of mountain pine beetle related stand mortality on snow accumulation and melt in BC. In: Proceedings of the 77th Western Snow Conference, Canmore, Alberta. pp83–93. Winkler, R., Moore, R.D., 2006. Variability in snow accumulation patterns within forest stands on the interior plateau of British Columbia, Canada. Hydrol. Process. 20:3683–3695 Winkler, R., Rex, J.F., Teti, P., Maloney, D.A., Redding, T., 2008. Mountain pine beetle, forest practices, and watershed management. B.C. Min. For. Range, Res. Br., Victoria, B.C. Exten. Note No. 88, 2008. www.for.gov.bc.ca/hfd/pubs/Docs/En/En88.pdf (Accessed in March 2010). 213  Winkler, R., Spittlehouse, D.L., Golding, D.L., 2005. Measured differences in snow accumulation and melt among clearcut, juvenile, and mature forests in southern British Columbia. Hydrol. Process.19, 51–62. Wolman, M.G., Miller, J.P., 1960. Magnitude and frequency of forces in geomorphic processes. J. Hydrol.69, 54-74. Woodsmith, R.D., Vache, K.B., McDonnell, J.J., Helvey, J.D., 2004. Entiat experimental forest: catchment-scale runoff data before and after a 1970 wildfire. Water Resour. Res. 40,W11701. Wootton, J.T., Parker, M.S., Power, M.E., 1996. Effects of disturbance on river food webs. Science 273: 1558-1561. Zégre, N., Skaugset, A.E., Som, N.A., McDonnell, J.J., Ganio, L.M., 2010. In lieu of the paired catchment approach: Hydrologic model change detection at the catchment scale. Water Resour. Res. 46(11), W11544. Zhang, L., Dawes, W., Walker, G., 2001. Response of mean annual evapotranspiration to vegetation changes at catchment scale. Water Resour. Res. 37(3), 701-708. Zhang, M., Wei, X., 2012b. The cumulative effects of forest disturbance on streamflow in a large watershed in the central interior of British Columbia, Canada. Hydrol. Earth Syst. Sci. 16, 2021–2034. Zhang, M., Wei, X., 2013a. Alteration of hydrological regimes caused by large-scale forest disturbances. Ecohydrology, DOI: 10.1002/eco.1374. Zhang, M., Wei, X., 2013b. Contrasted hydrological responses to forest harvesting in two large neighboring watersheds in snow hydrology dominant environment: implications for forest management and future forest hydrology studies. Hydro. Process. (Submitted) Zhang, M., Wei, X., Sun, P., Liu, S., 2012a. The effect of forest harvesting and climatic variability on runoff in a large watershed: the case study in the Upper Minjiang River of Yangtze River basin. J. Hydrol. 464-465,1-11. Zhang, X., Zhang, L., Zhao, J., Rustomji, P., Hairsine, P., 2008. Responses of streamflow to changes in climate and land use/cover in the Loess Plateau, China. Water Resour. Res. 44(7), W00A07. Zhao, F., Zhang, L., Xu, Z., Scott, D.F., 2010. Evaluation of methods for estimating the effects of vegetation change and climate variability on streamflow. Water Resour. Res. 46(3), W03505. Zheng, H., Zhang, L., Zhu, R., Liu, C., Sato, Y., Fukushima, Y., 2009. Responses of streamflow to climate and land surface change in the headwaters of the Yellow River Basin. Water Resour. Res. 45(7), W00A19. Zhou, G., Wei, X., Luo, Y., Zhang, M., Li, Y., Qiao, Y., Liu, H., Wang, C., 2010. Forest recovery and river discharge at the regional scale of Guangdong Province, China. Water Resour. Res. 46,W09503. Ziemer, R. R. 1981. Storm Flow Response to Road Building and Partial Cutting. Water Resour. Res. 17(4), 907-917. 214  Appendices Appendix A: Modified double mass curves with log-transformed data 11 10.5 10  1991 1985  9.5  1968  LnQa  9 8.5 8 7.5 7 6.5 6 5.5 5.5  6  6.5  7  7.5  8  8.5  9  9.5 10 10.5 11  LnPae  Figure A-1 The modified double mass curve for the Willow River watershed study with logtransformed data  4 3.5  1986  3  LnCaw  2.5 2 1.5 1 0.5 0 -0.5 -0.5  0  0.5  1  1.5 2 LnCac  2.5  3  3.5  4  Figure A-2 The modified double mass curve for the Cottonwood-Willow quasi-paired watershed study with log- transformed data 215  9 8.5  1999  8  LnQa  7.5 7 6.5 6 5.5 5 4.5 5  5.5  6  6.5 7 LnPae  7.5  8  8.5  9  Figure A-3 The modified double mass curve for the Baker Creek watershed study with log transformed data  9 8.5  1998  8  LnQa  7.5 7 6.5 6 5.5 5 4.5 5  5.5  6  6.5  7 7.5 LnPae  8  8.5  9  9.5  Figure A-4 The modified double mass curve for the Moffat Creek watershed study with log transformed data  216  4.5  1999  4 3.5  1984  LnCat  3 2.5 2 1.5 1 0.5 0 0  0.5  1  1.5  2  2.5  3  3.5  4  4.5  LnCaa  Figure A-5 The modified double mass curve for the Ashnola-Tulameen quasi-paired watershed study with log transformed data  217  Appendix B: An example of ECA calculator  Figure B-1 The selection of data files  Figure B-2 The display of data for calculation  218  Figure B-3 The time series of calculated annual equivalent clear-area  219  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0071984/manifest

Comment

Related Items