UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Comparison of techniques for measuring forest carbon in British Columbia Skrivanos, Pano Manolis 2012

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

Item Metadata

Download

Media
24-ubc_2012_fall_skrivanos_pano.pdf [ 1.56MB ]
Metadata
JSON: 24-1.0072837.json
JSON-LD: 24-1.0072837-ld.json
RDF/XML (Pretty): 24-1.0072837-rdf.xml
RDF/JSON: 24-1.0072837-rdf.json
Turtle: 24-1.0072837-turtle.txt
N-Triples: 24-1.0072837-rdf-ntriples.txt
Original Record: 24-1.0072837-source.json
Full Text
24-1.0072837-fulltext.txt
Citation
24-1.0072837.ris

Full Text

  Comparison of techniques for measuring forest carbon  in British Columbia    by    Pano Skrivanos  B.Sc., The University of Victoria, 2002     A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE  in  The Faculty of Graduate Studies  (Forestry)     THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  June, 2012   © Pano Skrivanos, 2012        ii   Abstract  The Earth is currently in a period of rapid climate change and the anthropogenic release of carbon dioxide (CO2) is widely considered to be the primary cause. Forests are an important part of the global carbon cycle and their ability to mitigate atmospheric CO2 levels is increasingly being recognized. Forest carbon projects generate carbon offsets either through the application of specialized forest management practices, or through the protection or restoration of deforested or degraded land. The ability to quantify accurately the amount of carbon that would be sequestered by forest carbon project activities is critical to their success. A variety of forest carbon modeling techniques exist and use a variety of methods for data acquisition including forest inventory data, remotely sensed data, or ground measurements. However the accuracy of these modeling techniques varies, making their standard application difficult.  This thesis contributes to the understanding of forest carbon quantification by comparing forest carbon estimates derived from a ground-based technique with forest carbon estimates derived from three forest carbon modeling techniques: Canadian Forest Service Carbon Budget Model (CBM-CFS3), Vegetative Resource Inventory Biomass Equations, and Private Woodland Planner. Both differential and least squares mean statistical analyses were conducted to determine which modeling technique estimated forest carbon closest to estimates derived from the ground-based technique.  The hypothesis that framed this research was that CBM-CFS3 would be the most accurate modeling technique. However results indicated that forest carbon estimates from PWP are closest to those derived from the ground-based technique. Results derived from CBM-CFS3 are farthest from the ground-based technique.  The results from this project suggest that an ideal forest carbon quantification technique incorporates field sampling with broad-based model estimates. Current research in forest carbon modeling techniques shows a trend towards more accurate and efficient estimates, which will allow project developers to better measure forest carbon stocks, improve forest conservation, and generate greater economic opportunities. These improvements will also increase the effectiveness of forest carbon projects and their role in mitigating the effects of climate change. iii  Table of Contents  Abstract ............................................................................................................................... ii Table of Contents ………………………………………………………………………………………...iii List of Tables ...................................................................................................................... v List of Figures .................................................................................................................... vi Acknowledgements ........................................................................................................... vii 1. INTRODUCTION .......................................................................................................... 1 1.1 Climate Change and Forests ..................................................................................... 1 1.2 Forest Carbon Economy & Forest Carbon Projects .................................................. 3 1.3 Measuring Forest Carbon .......................................................................................... 4 1.3.1 Forest Carbon Pools ............................................................................................6 1.3.2 Carbon Estimation Models ...............................................................................10 1.3.3 Available Data Sources .....................................................................................11 1.3.4 Challenges .........................................................................................................11 1.4 Study Area .............................................................................................................. 13 1.5 Literature Review ................................................................................................... 15 1.5.1 Carbon Management in Forests ........................................................................15 1.5.2 Forest Carbon Quantification Techniques ........................................................17 1.6 Research Objective ................................................................................................. 19 2. METHODS ................................................................................................................... 20 2.1 Introduction ............................................................................................................. 20 2.2 Establishing Forest Carbon Plots ............................................................................ 20 2.2.1 Sampling Technique .........................................................................................20 2.2.2 Spatial Location of Forest Carbon Plots ...........................................................22 2.2.3 Forest Carbon Plots ...........................................................................................26 2.2.4 Field Data ..........................................................................................................26 2.3 Techniques and Processes ....................................................................................... 33 2.3.1 Calculating Carbon from Field Data .................................................................33 2.3.2 Calculating Carbon from Forest Carbon Estimation Techniques .....................34 2.3.2.1 Canadian Forest Service Carbon Budget Model ........................................34 2.3.2.2 Vegetative Resource Inventory Biomass Equations ..................................36 2.3.2.3 Private Woodland Planner .........................................................................37 2.4 Analysis .................................................................................................................. 40 2.4.1 Data Distribution ...............................................................................................40 2.4.2 Comparison Analysis ........................................................................................41 3. RESULTS ..................................................................................................................... 42 3.1 Control Data ............................................................................................................ 42 iv  3.1.1 Inventory Results ..............................................................................................42 3.1.2 Carbon Results ..................................................................................................43 3.2 Treatment Data ....................................................................................................... 48 3.2.1 Canadian Forest Service Carbon Budget Model (CBM-CFS3) ........................48 3.2.2 Vegetative Resource Inventory (VRI) Biomass Equations ..............................49 3.2.3 Private Woodland Planner (PWP) ....................................................................50 3.3 Control and Treatment Data Analysis .................................................................... 51 3.3.1 Data Distribution ...............................................................................................51 3.3.2 Comparison Analysis ........................................................................................54 3.3.2.1 Differential Analysis ..................................................................................55 3.3.2.2 Statistical Analysis .....................................................................................59 4. DISCUSSION ............................................................................................................... 61 4.1 Interpretation of the Results .................................................................................... 61 4.2 Guidelines, Recommendations and Considerations ................................................ 65 4.3 Conclusion .............................................................................................................. 68 References ......................................................................................................................... 70 Appendices ........................................................................................................................ 81 Appendix 1: Volume to Biomass Calculations ............................................................. 81 Appendix 2: Forest Carbon Plot Summary of Results .................................................. 84 Appendix 3: Canadian Forest Service Carbon Budget Model Summary of Results .... 85 Appendix 4: VRI Biomass Equations Summary of Results ......................................... 86 Appendix 5: Private Woodland Planner Summary of Results ...................................... 87 Appendix 6: Summary of Control and Treatment Results ........................................... 88 Appendix 7: Differences Between Control and Treatment Estimates by Plot ............. 89 Appendix 8: Closest Estimates to the Control by Plot .................................................. 90 Appendix 9: Differences Between Control and Treatments by Strata .......................... 91  v  List of Tables  Table 1. Stratifications in order of decreasing area for the study area. ........................................ 22 Table 2. Forest carbon plot frequency by strata. ........................................................................... 26 Table 3. Forest carbon plot UTM Easting and Northing coordinates. .......................................... 28 Table 4. Summary of carbon pools and associated measured parameters. ................................... 33 Table 5. Summary of ground-measured and corresponding measured model carbon pools. ....... 39 Table 6. Forest carbon plot stand characteristics. ......................................................................... 43 Table 7. Summary of estimated forest carbon pool size by forest carbon sample plot ................ 46 Table 8. Normality tests based on all observations....................................................................... 51 Table 9. P-Value tests based on observations without outliers. .................................................... 53 Table 10. f and Pr>F values. ......................................................................................................... 54 Table 11. Summary of differences between control and treatment estimates. ............................. 57 Table 12. Least Square Mean values for the control and each treatment. .................................... 60         vi  List of Figures  Figure 1. Terrestrial Ecozones of British Columbia. ...................................................................... 7 Figure 2. Map showing the Indian River Watershed (Study Area) in relation to the Tsleil- Waututh Nation Traditional Territory and the Metro Vancouver region. .................................... 14 Figure 3. Spatial distribution of stratifications. ............................................................................ 21 Figure 4. Identified candidate forest stands intersecting or adjacent to 200m road buffer. .......... 23 Figure 5. Randomly ranked candidate forest stands. .................................................................... 24 Figure 6. Forest carbon plot locations........................................................................................... 25 Figure 7. Sample field notes. ........................................................................................................ 29 Figure 8. Measuring the height of broken stumps. ....................................................................... 31 Figure 9. Decay class classification used for dead wood identification. ...................................... 32 Figure 10. Total forest carbon percentage by forest carbon pool for all 19 plots. ........................ 44 Figure 11. Carbon pool size percentage by tree species within the Total live tree carbon pool. .. 45 Figure 12. Total average forest carbon by forest carbon pool for all 19 plots. ............................. 47 Figure 13. Total forest carbon by plot, estimated using the OSU methodology. ......................... 48 Figure 14. Total forest carbon by plot, estimated using CBM-CFS3. .......................................... 49 Figure 15. Total forest carbon by plot, estimated using VRI Biomass Equations. ....................... 50 Figure 16. Total forest carbon by plot, estimated using Private Woodland Planner. ................... 51 Figure 17. Normal Probability Plot.. ............................................................................................. 53 Figure 18. A comparison of total forest carbon by plot for the control and all treatments. .......... 54 Figure 19. A comparison of mean values by plot. The net difference between the means of each treatment are shown relative to the means of the control. ............................................................ 56 Figure 20. Scatterplot showing the relationship between forest carbon estimates of the control and the CBM-CFS3 treatment for 19 plots. .................................................................................. 57 Figure 21. Scatterplot showing the relationship between forest carbon estimates of the control and the VRI Biomass Equations treatment for 17 plots………………………………………...….. 58 Figure 22. Scatterplot showing the relationship between forest carbon estimates of the control and the PWP treatment for 19 plots…………………………………………………...………………58           vii  Acknowledgements  I would like to thank all of the staff of the Tsleil-Waututh Nation Treaty, Lands and Resources Department for their technical and moral support, including special thanks to Leah George-Wilson, Evan Stewart, Ernie George, Micheal George, Jason Forsyth, Sophie Middleton and Lauren Taylor.  I would like to thank my committee members, Dr. Nicholas Coops, Dr. Ronald Trosper, Dr. Howard Harshaw, and faculty member Dr. John Nelson, for their guidance and support. I want to thank Dr. Tony Kozak, who provided assistance with statistical analysis, as well as my fellow students within the Sustainable Forest Management (SFM) lab. I would also like to extend a big thank you to each of my field assistants; Raelene Esteban, Josh George, Charles George, Sophie Middleton, Lianzhen Xu, Ed Thomas, Dave Thomas, Lindsay Skrivanos, and Andy Spence, who without their help this work would not be possible.  Finally I will like to thank my supervisor, Dr. John Innes, who provided his time, insight and guidance along this journey.    1  1. INTRODUCTION  1.1 Climate Change and Forests  Climate change, as defined by the Intergovernmental Panel on Climate Change (IPCC), is the statistically significant variation in the mean state of the climate, and is attributed directly or indirectly to the anthropogenic release of greenhouse gasses, primarily through fossil fuel use, land-use change and agriculture (IPCC, 2007a). The primary greenhouse gases in the atmosphere that affect the Earth's radiative budget include carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), chlorofluorocarbons (CFCs), carbon tetrachloride (CCI4), ozone (O3), carbon monoxide (CO) and water vapor (Watson, Meira Filho, Sanhueza & Janetos, 1992). However CO2 is the most significant anthropogenic greenhouse gas due to its large volume, long life-span and high radiative forcing properties (IPCC, 2007a). Concentrations of CO2 have increased from pre-industrial levels of 280 ppm to 389 ppm in 2011 (Tans and Keeling, 2011). Average global temperature has risen by an average of 0.74°C between 1906 and 2006 and the IPCC predicts a further doubling of CO2 concentration is likely to result in an average global temperature increase in the range of 2°C to 4.5°C (IPCC, 2007a). This temperature increase is predicted to result in a further rising of sea levels, decreases in snow and ice extent, increases in precipitation, and increases in the occurrence of severe weather phenomenon such as droughts, storms and heat waves (IPCC, 2007a).  There is evidence of recent climate change in British Columbia (BC). The past 100 years have seen temperatures rise by an average 0.6°C on the coast and 1.7°C in the north. These higher temperatures have resulted in decreases in snow pack, retreating mountain glaciers, greater proportions of lakes and rivers remaining ice-free in winter, increases in coastal ocean temperatures, increases in average fresh water lake and river temperatures and increasing sea level (LiveSmart BC, 2011; Gayton, 2008; Hamann and Wang, 2006; Spittlehouse and Stewartz, 2003). As a result scientists and researchers have documented, and predict, increasing frequency, intensity, and uncertainty of heat waves, droughts, heavy rain events, and forest fires (Natural Resources Canada, 2007, BC Ministry of Forests, Mines and Lands, 2010).  2  Carbon sequestration and storage within forests has the potential to play an important role in the mitigation of climate change. Deforestation and forest degradation account for 12% of total anthropogenic emissions worldwide (IPCC, 2007a; IPCC, 2007b; van der Werf, 2009). Forests sequester and store carbon within their biomass as they grow, and release carbon and other greenhouse gases as they burn or decompose (BC Ministry of Forests, Mines and Lands, 2010). This carbon is not “locked” in, rather factors that influence forest conditions such as temperature, precipitation, fire and disease, can alter forest carbon levels by either increasing carbon stores, thereby turning the forest into a sink for atmospheric carbon emissions, or decreasing carbon stores, thereby turning the forest into a source of carbon emissions.  BC’s forests play an important role within the global carbon cycle. BC contains 55 million hectares of forest, or 1.7% of the world’s total forest area (4 billion hectares) (BC Ministry of Forests, Mine and Lands, 2010; Food and Agriculture Organization, 2011). Within BC approximately 60% of the land area is classified as forest land; 41% of this forested land (23 million hectares) is classified as old growth (140 years and older) (BC Ministry of Forests, Mine and Lands, 2010). BC contains some of the largest forest carbon pools not only within Canada (Stinson et al., 2011; Penner et al., 1997; Power and Gillis, 2006; Kurz et al., 1996) but also the world (Dixon et al., 1994). Estimates from a 1996 Ministry of Forests Forest Science Program study indicate that the total aboveground and belowground biomass carbon pool in BC forests in 1989 was 5.48 Pg C (billion tonnes carbon) (FRDA, 1996). In this same study soil and other non- living organic material were found to contain another estimated 12.6 Pg C (FRDA, 1996).  From 1990 to 2003 BC forests were a net carbon sink, removing an average of 30 metric tonnes carbon per hectare (MtCO2e) from the atmosphere each year (BC Ministry of Forests, Mines and Lands, 2010). Since 2003, however, BC forests have become carbon sources, emitting an average 45 MtCO2e each year. This is due to a combination of the consequences of the mountain pine beetle outbreak, wildfires and harvesting-related activities such as slash burning and salvage logging (BC Ministry of Forests, Mines and Lands, 2010). The current deforestation rate is approximately 6,200 hectares per year, roughly 60% of which occurs on Crown land (BC Ministry of Forests, Mines and Lands, 2010). 3  1.2 Forest Carbon Economy & Forest Carbon Projects  Forests play a key role for their ability to mitigate atmospheric CO2. The global forest carbon market is growing significantly, with a total value of $37 million (US) in 2008, and a projected value of $60+ million for 2011 (Hamilton et al., 2010). By 2020, carbon is expected to be one of the largest traded commodities in the world and have a market value of almost $2 trillion (Tvinnereim et al., 2011). This has led to the development of a diverse international forest carbon economy, characterized by the establishment of specialized voluntary and regulatory markets, legal and political frameworks, project standards and methodologies, and expert resources. However, there are many potential obstacles that challenge further growth including a lack of regulatory certainty that prevents this emerging market from developing into an established market, significant up-front costs, and a lack, until recently, of established methodologies for measuring forest carbon.  A forest carbon project is any project whose outcomes create an intentional change in land management practices within a specified area that result in an increase in forest carbon storage levels more than would otherwise occur in the absence of the project. Forest carbon projects generate carbon offsets through either the application of specialized forest management practices or through the protection or restoration of threatened or degraded land. Forest carbon projects can be grouped into the following broad categories of project activities: improved forest management, afforestation and reforestation, and Reduced Emissions from Deforestation and Degradation (REDD) (Willey and Chameides, 2007; Ravindranath and Ostwald, 2008).  Improved forest management projects involve enacting forest management strategies on existing forests that result in an increase in carbon stocks, or a reduction in greenhouse gas emissions. This can be accomplished through any of the following: reduced harvest volumes through less cutting; reduced impact logging; converting logged or to-be-logged forests into protected forests; the conversion of low-productive forests to high-productive forests through fertilization; or the planting of superior trees, control of competing vegetation, or thinning (Chénost, 2010). Afforestation and reforestation projects involve the conversion of non-forest land into forest land via the planting or replanting of trees. These projects typically involve the conversion of former forested land, such as cleared farmland, back into forest. Although less common, afforestation 4  and reforestation projects can also involve the conversion of historically un-forested land into forested land. REDD projects involve encouraging governments, and other parties who have been responsible for deforestation and forest degradation, to stop these activities, through the use of some form of compensation. These projects are often useful for protecting high-value forests under threat.  In order for forest-derived carbon offsets to be bought and sold within a carbon economy, forest carbon projects must conform to particular protocols and principles as described by a given regulatory or voluntary standard. These standards describe a range of criteria that a project must meet in order for it to qualify for that standard. An example of a criterion that all standards share is the definition of baseline and project scenarios, and the amount of carbon sequestered and stored within each. Being able to accurately measure the amount of carbon within each is critical to a project’s success. 1.3 Measuring Forest Carbon  For the emerging carbon market to be successful, accurate and efficient measurement techniques are required. This means that cost-effective, objective, and transparent approaches must be developed to measure forest carbon accurately as well as to provide the ability to monitor and quantify any changes that happen to the forest over time.  The process of measuring carbon stocks is relatively straightforward and techniques for doing so are well established, particularly for the main carbon pools of the forest which include aboveground live biomass, belowground biomass and dead woody biomass (Ravindranath and Ostwald, 2008). These three carbon pools are typically most affected by the actions of a forest carbon project, and therefore are the main focus of a measurement regime (Willey and Chameides, 2007; Subedi et al., 2010; Schelhaas et al., 2004; Ravindranath and Ostwald, 2008; Olander and Ebeling, 2010). These pools also contain the majority of carbon within forest ecosystems, with the exception of the soil carbon pool, which although significant, is usually least affected by a forest carbon project and therefore not a focus of most forest carbon standards and measurement techniques (Willey and Chameides, 2007; Ravindranath and Ostwald, 2008). Carbon estimation techniques for measuring these pools can be grouped into two general categories: ground-based and modeling. 5   Ground-based techniques are generally considered to be the most accurate and precise method of measuring forest carbon stocks (Willey and Chameides, 2007; Ravindranath and Ostwald, 2008). The optimal way to measure forest carbon stocks requires removing all aboveground and belowground vegetation and debris, sifting through soils and oven-drying and weighing all material which contains carbon (Ravindranath and Ostwald, 2008). This provides a dry biomass weight which can be converted into carbon. Techniques based on forest inventory also exist. These techniques make use of inventory data collected from permanent sample plots laid out in a statistically well-designed manner. The measurements collected from the inventory can be converted to aboveground and belowground carbon estimates using either biomass expansion factors or allometric biomass regression equations (Brown, 2002). However while generally considered to be both accurate and precise, these techniques can be time-consuming and cost prohibitive, particularly if the project area is large and the terrain difficult to access. Therefore these techniques are best used to verify and confirm carbon estimates derived from modeling techniques, which are both less invasive and more practical.  Currently, no single, definitive modeling technique for measuring forest carbon exists (Oregon State University, 2008; Ravindranath and Ostwald, 2008; Goetz et al., 2009; Cao and Valsta, 2010; Tupek et al., 2010). Rather a variety of techniques are available, ranging from site-specific data collection to broad regional and national estimates. These modeling techniques have been developed by governments, industry and non-profit organizations for specific locations and scales. Modeling techniques utilize computer applications and allow for the input of surrogate data from which forest volume can be extracted and the calculation of carbon stocks resulting from project activities. Examples of surrogate data include remotely sensed satellite and air photo imagery, and Light Detection and Ranging (LiDAR). Biomass (and hence carbon) estimates can also be extracted from existing forest inventory data such as the BC provincial and national forest inventory, which in many cases have been derived from remotely sensed data. Examples of modeling techniques which utilize this type of data as input include: CBM-CFS3 (Natural Resources Canada, 2011), FORECAST (University of British Columbia, 2011), CO2FIX (European Forest Institute, 2011), CARBINE (UK Forestry Commission, 2011), BIOMITRE (International Energy Agency Bioenergy, 2011), Private Woodland Planner (Enfor Consultants Ltd., 2007), COLE (United States Department of Agriculture, 2011), FORCARB2 6  (United States Department of Agriculture, 2011) and TimberCAM (New South Wales Government, 2011). 1.3.1 Forest Carbon Pools  The amount of carbon that moves between, and is stored, within the various carbon pools that make up a forest’s biomass depends greatly on forest life cycle stages. The rate of carbon uptake into forests from the atmosphere is highest when forests are young, productive and growing (Odum, 1969; Tan et al., 2011). At this life cycle stage, most carbon is stored within the living biomass of growing trees, understory vegetation and soil. However, as forests grow older increasing amounts of carbon within living biomass pools are transferred to either soil, or dead and decaying biomass pools, which release their carbon gradually back into the atmosphere. Although the carbon sequestration rate within younger growing forests is greatest, the carbon pools stored within older forests is much larger. Many experts suspect that old growth forests within BC contain some of the largest forest carbon pools in the world (BC Ministry of Forests, Mine and Lands, 2010).  Forest carbon is stored within five carbon pools: aboveground living biomass, belowground biomass, litter, deadwood and soil. Harvested wood products are also considered a forest carbon pool in certain cases where the harvested wood is used for long-lasting construction materials. Total forest carbon is equal to the sum of all of these pools. Measuring forest carbon requires the estimation of carbon stored within each of these pools, with the proportions of carbon stored within each varying depending on geographic region, forest age and type. Within BC the size and flux of forest carbon pools varies. A comprehensive study by the Canadian Forest Service found that of the four terrestrial ecozones of BC (Figure 1) the size of forest carbon pools was consistently highest within the Pacific Cordilleran ecozone and lowest within the Taiga Cordillera ecozone (Kurz, 1996). 7   Figure 1. Terrestrial Ecozones of British Columbia (Ecological Stratification Working Group, 1995).   A number of factors determine the pools that must be quantified for a forest carbon project, in addition to the methods and frequency for doing so. These factors include relevance to project type (i.e. how project actions will impact these pools over the length of the project), chosen standard(s), monitoring period, monitoring cost and impact of additional jurisdictional regulations. The following account provides a description of each pool along with an overview of how each can be measured.  Aboveground Biomass  Aboveground biomass (AGB) includes all living vegetation biomass above the soil including stems, stumps, branches, bark, seeds and foliage (Ravindranath and Ostwald, 2008). For purposes of carbon inventory, this can be categorized into two groups: live trees and understory vegetation.  8  Live trees are the largest and most important pool for carbon inventory within forested areas. Pool size varies greatly depending on the number, size, and species of trees, lifecycle stage and yearly growth cycle. The size of this pool can be large, particularly if tree size is significant. For example a 450 year old Douglas fir stand located in the Oregon Cascade Mountains contains an estimated 584 metric tons of carbon per hectare (Mg C ha-1), and a 150 year old hemlock-spruce stand situated along the Oregon coast contains an estimate 626 Mg C ha-1 (Willey and Chameides, 2007). During field work for this research an old growth western red cedar stand located within the study area near the city of Vancouver was found to contain an estimated 1,055 Mg C ha-1 using techniques described within the methodology section. Methods used for measuring the live tree pool are straightforward and well-developed (Ravindranath and Ostwald, 2008; Willey and Chameides, 2007) and generally involve the measurement of tree stem diameter at breast height (DBH) of all trees. The DBH, along with tree species type and its associated wood density estimate, is used to predict tree volume via allometric equations.  Understory vegetation also contains carbon, although within BC forests the size of the understory vegetation carbon pool is usually insignificant in terms of quantifying carbon for forest carbon projects (Mark Harmon, personal communication, December 2011). The measurement of understory vegetation may only be necessary for very young, fast-growing forests, where carbon storage in significant enough to contribute towards overall carbon store and flux, or in the absence of a significant overstory (Mark Harmon, personal communication, December 2011). In this case, understory vegetation may reach a height of several meters and contain up to 30 Mg C ha-1 (Willey and Chameides, 2007).  Belowground Biomass  The belowground biomass (BGB) pool includes all carbon stored within living biomass below the ground, the most significant of which is contained within living roots (Ravindranath and Ostwald, 2008). The most accurate method of measuring BGB requires the extraction of roots from a sample volume of soil, weighing the roots and extrapolating this data over the study area. This technique is invasive and time consuming and usually not required for forest carbon projects except in special cases. Less invasive and resource-intensive estimation methods are therefore often favored. These include root to shoot ratios (proportion of AGB) and biomass 9  equations (Ravindranath and Ostwald, 2008; Cairns et al., 1997). Both of these methods provide estimates that meet the requirements of most forest carbon project standards.   Deadwood Biomass  The deadwood biomass (DW) pool includes all non-living woody biomass not contained within the soil. This includes standing dead trees, stumps and coarse woody debris. The size of the DW pool varies considerably depending on forest conditions and can be significant, particularly within older stands in temperate rainforests. A number of conditions are typically associated with large DW pools. These include; forests that are un-managed and where dead wood is not likely to be removed, forests where decomposition rates are lower, forests where site productivities are high and trees grow large, and forests with cooler temperatures and drier conditions (Mark Harmon, personal communication, October 2008; Ravindranath and Ostwald, 2008).  Methods for measuring DW vary depending on the components being measured. Required information for measuring standing dead trees includes DBH, species, height and decay stage. The combination of these four attributes is enough to estimate a volume and hence carbon content estimates for the tree. For stumps a median diameter is required which allows for the carbon content to be calculated via biomass equations. Coarse woody debris can be the most difficult to measure as often it can be hard to discern exactly where each piece begins and ends due to advance decay rates which can obscure the exact dimensions of each piece.  Litter  The litter biomass pool (LT) includes all non-living organic debris and unattached plant material located on the forest floor. This includes leaves, needles, bark and other pieces of unattached vegetation. Litter also includes what is classified as fine woody debris – twigs and small pieces of wood with diameters less than 3cm (Ravindranath and Ostwald, 2008; Willey and Chameides, 2007).  Litter volume varies considerably depending on forest type and time of year. However litter volume is normally not significant in terms of carbon storage (less than 10% of total biomass, Ravindranath and Ostwald, 2008). As such, its measurement is often not required for forest carbon projects with the exception of fine woody debris, which if left undisturbed can be 10  measured every 3-5 years. Fine woody debris is best measured using a number of transects within the study area and measuring the frequency and size class of fine woody debris that intersects with each transect. These numbers can then be used to extrapolate fine woody debris volume, biomass and carbon within the entire study.  Soil  In BC soil carbon levels can be significant and are highest within the Pacific Cordilleran and Interior ecoclimatic provinces (Kurz, 1996). Soil stores carbon within both its organic and mineral components but as forest carbon project activities usually only impact the top organic component of soil, where soil carbon is highest, most measurement methods focus only on this upper layer (Ravindranath and Ostwald, 2008).  Measurement techniques are complex and can include diffuse reflectance spectroscopy (interaction of infrared light with matter), soil carbon dynamic models (e.g. CENTURY and RothC), wet digestion or titrimetric determination and simple regression models (Ravindranath and Ostwald, 2008). Simple regression models are the most practical for forest carbon models and involve the measurement of the depth of the top organic layer and subsequent regression analysis to estimate soil carbon.  1.3.2 Carbon Estimation Models  Three forest carbon estimation techniques have been chosen for evaluation here: the Canadian Forest Service Carbon Budget Model (CBM-CFS3), Private Woodland Planner (PWP), and VRI Biomass Equations. All three techniques are deterministic, and estimate forest carbon through the use of built-in biomass conversion calculations. Some of these techniques also account for carbon flows within different management regimes, and have separate modules for estimating bioenergy and carbon within post-harvesting forest products. Background information and a detailed methodology of use for each model are provided for each model in section 2.3.2 Calculating Carbon from Forest Carbon Estimation Techniques.  Each model has its similarities and differences relating to cost, availability, usability, technical capacity, and efficiency. For example, CBM-CFS3 and PWP are both available at no cost, and require only an accompanying basic spreadsheet program and computer of moderate ability. 11  Conversely the VRI Biomass Equations model requires additional Geographic Information System (GIS) software, which can be expensive and difficult to use. Technical capacity associated with each modeling technique also varies. The CBM-CFS3 and VRI biomass equation models both require a high level of user skill to setup and operate, as well as a sophisticated understanding of GIS and forestry principles. Both also require careful and extensive data input preparation, which can add significantly to the time requirements. In contrast, the PWP modeling technique is the easiest of the three modeling techniques to use, requires only a basic understanding of general forestry and harvesting knowledge, and requires minimal time to generate reports.  1.3.3 Available Data Sources  The data sources utilized for each evaluated technique can be separated into two categories: data collected from sample plots directly, and data collected remotely by satellite or air-borne sensors. Data collection from sample plots involves directly measuring the frequency, volume and decay class of a variety of forest features including trees, vegetation, dead wood and soil. Data collected remotely are (in most cases) derived originally from satellite and air photo imagery then later converted into a GIS dataset from which biomass estimates can be extrapolated (i.e. surrogate data). 1.3.4 Challenges  There are various challenges associated with measuring forest carbon. Forest ecosystems are inherently complex and the carbon stored within is difficult to conceptualize, let alone measure. Additionally, carbon is in constant flux and the resources needed for accessing and measuring forest elements that contain carbon can be high. There are also challenges associated with the data and techniques used for estimating forest carbon. High quality data may not always be readily available, their quality is often poor and their accuracy difficult to verify. Even more significant, there are varying methods and techniques available and there remains a general lack of consensus over the “best” methods.  Total error in the measurement of a given carbon pool is based on three types of error: sampling error, measurement error and regression error (Brown, 2002). Sampling error can occur if the 12  sample size is too small (it should be determined based on variance of the population, not by limitations of resources), if the location of measured elements is incorrect, or if the measured sample is not representative. Measurement error can occur if forest attributes are measured incorrectly, instruments are defective, techniques are inappropriate, the sample is not complete (missing, incomplete, not recorded observations) or data are missing (missing data, e.g., height of standing dead trees). Regression error can occur if the algorithms and calculations used to convert volume measurements to biomass and carbon estimates are incorrect.  The quantification of uncertainty ranges between 9-11% to as high as 35% for carbon inventories and research has found that soil and tree carbon had the greatest influence on carbon stock estimates and that growth and removals had the most influence on estimated change in carbon stocks (Heath and Smith, 2000; Kellwe et al., 2001; Brown, 2002; Chave et al., 2004). Brown (2002) examined model uncertainty for the five most numerous tree species in Oregon and found that model selection error may introduce 20 to 40% uncertainty into a live-tree carbon estimate, possibly making this form of error the largest source of uncertainty in the estimation of live-tree carbon stores. The effect of model selection could be even greater if models are applied beyond the height and DBH ranges for which they were developed (Brown, 2002).  Limitations to forest carbon quantification techniques exist when compared to ground inventory methods (Kivari et al., 2010 and Brown, 2002, amongst others). Stinson et al. (2001) discuss how modeled estimates of forest productivity was typically driven by empirical wood volume yield data that integrate the long-term impacts of climate and site on growth. As a result, a weakness in this approach is that it provides no sensitivity to current climate and does not account for departures from productivity as observed in the forestry data (such as may be caused by CO2 fertilization, N deposition, or climate change). Kivari et al. (2010) concluded that research is needed to determine whether or not these factors are causing significant changes in forest productivity that are not adequately described by empirical yield data; this same study also found that human errors are unavoidable in complex systems. Further, Kivari et al. (2010) identified the potential error in measured tree attributes relating to the estimation of tree height from tree diameter, for example some trees have large diameters but short heights.  13  Model selection uncertainty is potentially large enough that it could limit the ability to track forest carbon with the precision and accuracy required by carbon accounting protocols. The ideal technique exhibits the following attributes: accuracy, little to no cost, ease of use (existing tools are not particularly user friendly and usually require advanced knowledge to use), flexibility and efficiency. In reality however while some come close, no existing technique exhibits all of these attributes. The science is still developing, especially in new methods of measuring forest carbon stores on a broad scale. 1.4 Study Area  The study area for this project is the 21,882 hectare Indian River Watershed, located approximately 30 kilometers northeast of the city of Vancouver (49°31’00” N, 122°54’01” W) within the Canadian province of British Columbia (Figure 2). The Indian River Watershed is located within the core territory of the Tsleil-Waututh, a Coast Salish People who have inhabited the surrounding lands and waters since “Time out of Mind” (Tsleil-Waututh Nation, 2000). The area is important in the history, culture and economy of the Tsleil-Waututh people (Tsleil- Waututh Nation, 2007). Inlailawatash is the name of the historic Tsleil-Waututh village site located near the mouth of the Indian River, the principal stream within the watershed and the site of first contact with Spanish explorers in 1792 (Tsleil-Waututh Nation, 2007). Two small Indian reserves and several fee simple land parcels owned by the Tsleil-Waututh now occupy this former village site (Tsleil-Waututh Nation, 2007). The area is home to the salmon-bearing Indian River and contains numerous wildlife resources including elk, bear, deer, cougar, eagles and beaver.  14    Figure 2. Map showing the Indian River Watershed (Study Area) in relation to the Tsleil-Waututh Nation Traditional Territory and the Metro Vancouver region (Tsleil-Waututh Nation, 2011; GeoBC 2011).   The Indian River Watershed is typical for coastal BC – steep and rugged, and contains three separate Biogeoclimactic Ecosystem Classification zones: Coastal Western Hemlock (CWH), Mountain Hemlock (MH) and Coastal Mountain Heather Alpine (CMA) (Tsleil-Waututh Nation, 2007). The watershed is heavily vegetated with forests dominating the landscape. Dominant coniferous tree species include western hemlock (Tsuga heterophylla), western red cedar (Thuja plicata) and Douglas-fir (Pseudotsuga menziesii). Lesser amounts of Sitka spruce (Picea sitchensis), western white pine (Pinus monticola) and Pacific or Western yew (Taxus brevifolia) also exist in select areas. Deciduous tree species include red alder (Alnus rubra), bigleaf maple (Acer macrophyllum), vine maple (Acer circinatum), and black cottonwood (Populus 15  balsamifera ssp. trichocarpa), much of which is located at lower elevations of the watershed along the Indian River.  The study area has been heavily impacted by past industrial activities, including approximately 60 years of forest harvesting, the establishment of a 500 kilowatt hydroelectric line through the valley bottom, and the construction of over 300 kilometers of logging and maintenance roads, many of which have not been properly decommissioned (Tsleil-Waututh Nation, 2007). These activities have resulted in two predominant forest types within the study area: second growth managed forest, predominantly younger age classes (20-60 years) located within the government-defined timber harvesting land base  (lower elevation more accessible areas), and un-managed (never cut) old growth forest located at inaccessible high elevations (Tsleil-Waututh Nation, 2007). 1.5 Literature Review  The following section provides a review and synthesis of the theory behind carbon management in forests and forest carbon quantification techniques. The literature reviewed originates from a variety of sources including research papers, books, reports, news and media.  To assist with the focus and organization of this review the literature has been classified into two themes: (1) carbon management in forests, and (2) forest carbon quantification techniques. Literature within the first theme focuses on the impacts of traditional forest management practices on carbon stores in forests, and on the principles and policies for effective management of forest carbon. Literature within the second theme focuses on the variety of techniques and methods available for measuring and accounting for forest carbon. 1.5.1 Carbon Management in Forests  Forests are the world’s largest terrestrial carbon pools, and store and sequester large amounts of carbon. As such their role within the global carbon cycle has always been important. In the mid- 1990s Dixon (1994) and Brown (1996) quantified global forest carbon stocks and explained how the slowing of deforestation, combined with an increase in forestation, could conserve or sequester significant levels of carbon. Brown (1996) outlined three categories of forestry practices that can conserve and sequester carbon: (1) management for conservation of carbon storage by slowing deforestation, changing harvesting regimes, and protecting forests from other 16  anthropogenic disturbances; (2) management for expanding carbon storage by increasing the area and/or carbon density in native forests, plantations, and agroforestry and/or in wood products; and (3) management for substitution by increasing the transfer of forest biomass carbon into products such as biofuels and long-lived wood products.  There has been a variety of research that has examined the impacts of forest management practices on forest carbon sequestration rates and stores. Humans have historically impacted forest carbon in North America since colonization. Birdsey, Pregitzer and Lucier (2006) examined the impacts of past forest management practices on forest carbon within the United States between the years 1600 and 2100. They found that overall, forest carbon storage fluctuated greatly between periods of land clearing and re-growth and that traditional harvesting and managing of forests for wood has had both positive and negative impacts of carbon stores. These impacts are examined in more detail by Dixon et al. (1991), who assessed forest management activities which specifically enhanced forest carbon sequestration. Dixon et al. (1991) identified a number of promising forest practice activities and assessed associated costs and benefits. Studies have also compared the forest carbon benefits associated with particular management regimes. In a recent study, modeled forest growth, timber yield and carbon storage from three proposed forest management plans were compared for Oregon's Elliott State Forest. The results of this analysis indicate that reduced timber harvests are associated with increased carbon storage, and that carbon sequestration rates from a no-harvest scenario would be significant (Ecotrust, 2011).  There are several resources available that explain the concept of managing forests for carbon, as opposed to timber harvesting for wood. Although this is a relatively new concept, the general principles behind the management techniques are not. Forest managers already have a good foundation of tools and techniques to build upon, and a number of papers and guidebooks have been published explaining these carbon management practices to forests managers. For example, Deusen (2010) presented an analytical approach for owners to compare traditional forest management with regular harvests with allowing the trees grow to accumulate more carbon. A derivative of land expectation value, called rotation equivalent value, has been shown to be a useful decision tool for comparing carbon storage with other, more traditional management 17  options. Ryan (2010) provided an overview of forest carbon and explained various techniques for increasing the amount of carbon stored within a forest. 1.5.2 Forest Carbon Quantification Techniques  Measuring carbon in forests with accuracy and precision is gaining global attention as countries seek to quantify the amount of carbon stored within their forests (Brown, 2002). Many established methods for measuring carbon exist, and vary depending on scale. The best techniques are based on a combination of permanent sample plots and remotely sensed estimates of aboveground biomass (Brown, 2002). Volume measurements collected from the permanent sample plots can be converted into aboveground and belowground biomass estimates using either biomass expansion factors or allometric regression equations. The remotely sensed estimates can take the form of existing forest inventories, air photo or satellite imagery, or some derivative of remotely-sensed imagery such as the Normalized Difference Vegetation Index (NDVI). The trend towards estimating carbon stocks using remote sensing data is growing and new remotely sensed data technologies are continuously being developed (Brown, 2002). This trend can be observed within Canada by looking at the development and progression of forest inventories since 1981.  Canada’s first national forest inventory, CanFI81, was a computer-based system that converted data from provincial and territorial inventories into a national classification scheme that estimated non-merchantable biomass from merchantable biomass estimates, while focusing on generating aboveground biomass estimates for inventoried forest land (Penner et al., 1997; Canada’s National Forest Inventory, 2011). CanFI81 demonstrated how the conversion factors are relatively insensitive to changes in stand age, density, site quality and size distribution, likely due to published biomass and volume prediction equations that only utilized DBH and height as the independent variables. Consequently, the resulting estimates provide a regional summary of aboveground biomass at the time of inventory. Penner et al. (1997) also concluded that biomass equations were best at predicting stemwood biomass, so the volume/biomass conversion factor is the most reliable. The equations for the non-stemwood biomass components were not as precise.  Subsequent Canadian forest inventories began to utilize regression analysis to examine the relationship between inventory volume estimates and remotely sensed derivatives of biomass. 18  The latest Canadian forest inventory was compiled in 2001 (CanFI2001), and demonstrated a progression in the techniques used for inventorying forest carbon, most notable being the utilization of remotely sensed satellite imagery (Power and Gillis, 2006; Canada’s National Forest Inventory, 2011). Several notable studies utilizing a similar approach soon followed. In 2001, Myneni et al. (2001) analyzed data from remote-sensing and forest inventories to identify the size and location of forest carbon sinks within northern hemisphere forests. Several other studies conducted by Donga (2003), González-Alonso (2006) and Myeong et al. (2006) found strong statistical relationships between satellite measurements of NDVI and inventory estimates of forest woody biomass. These studies further demonstrate that a regression model can be used to represent the relationship between forest biomass and remotely-sensed data. Additionally, Zhenga (2007) found that the simple ratio was a significant predictor of aboveground biomass in select circumstances.  Within BC, the statistical relationships between ground-measured volume and calculated biomass values stored within provincial inventory datasets have been examined. Kivari et al. (2010) summarize a methodology developed by CFS and BCMoFR staff to attach aboveground live tree biomass estimates to Vegetative Resources Inventory polygons in BC’s Vegetative Resources Inventory Management System Database. The methodology utilizes biomass equations developed by Lambert et al. (2005), Ung et al. (2008), Standish et al. (1985) and Boudewyn et al. (2007). The results from this analysis show a strong relationship between calculated biomass values and ground-measured volume at the plot level.  Several recently published papers show that laser-based systems are at least as accurate at measuring forest carbon as ground-based techniques. Mascaro et al. (2011) compared forest carbon estimates derived from plot measurements with estimates derived from a LiDAR-based system for Barro Colorado Island, Panama. They found that “lidar-based uncertainties of aboveground carbon stocks are indistinguishable from errors obtained when doing the most detailed plot-based estimates.” Asner (2011) describes a new technique for determining forest carbon stocks from an advanced system dubbed the Carnegie Airborne Observatory (CAO), which is made up of a combination of LiDAR and other optical sensors. The CAO not only identifies forest structure from its LiDAR sensor, but can also identify individual tree species and deduce their unique carbon content based on their chemical and spectral properties 19  such as photosynthetic pigment concentrations, water content of leaves, defense compounds like phenols, structural compounds such as lignin and cellulose, as well as phosphorous and other micronutrients. More studies indicate that accurate forest carbon estimates can be estimated in a practical manner which combines airborne laser technology, satellite mapping and field measurements (Asner, 2012; Baccini et al., 2012; Baccini et al., 2008; Saatchi et al., 2011). The advantages of these types of techniques are that they can produce broad-scale and high- resolution aboveground carbon estimates, in rugged geographic locations, relatively affordably (Asner, 2012). In addition, these techniques mitigate the challenges of measuring aboveground carbon density of areas with high spatial variation, which field measurements usually cannot capture (Asner, 2012). 1.6 Research Objective  The objective of this thesis is to compare three forest carbon measurement techniques, Canadian Forest Service Carbon Budget Model (CBM-CFS3), Vegetative Resource Inventory (VRI) Biomass Equations, and Private Woodland Planner (PWP), with ground-collected forest carbon estimates and evaluate them through statistical analysis. This process is separated into the following four steps:  1. Establishment of a control dataset consisting of forest carbon estimates derived from established permanent forest sample plots. Forest carbon estimates were calculated from inventory and volume data collected in the field. This control dataset consisted of 19 permanent forest sample plots located within 8 stratifications based on age class and ecological zone. Within each sample plot, carbon was estimated for individual forest carbon pools as well as for the forest ecosystem as a whole. 2. Calculation of the amount of forest carbon within each permanent forest sample plot using forest carbon modeling techniques. The forest carbon measurement techniques were chosen based on suitability to measuring forest carbon within BC, and the types of input data that are available. 3. Comparison of forest carbon estimates from each forest carbon measurement technique with forest carbon estimates derived from the control. 4. Determination, using statistical analysis, of which of the evaluated techniques most accurately measured forest carbon within the study area. 20  2. METHODS  2.1 Introduction  The following section outlines the methodology used to establish an evaluation process for the identification of a forest carbon estimation technique that best estimates forest carbon within BC forests. The methodology for this project is summarized as follows:  1. Creation of a detailed carbon inventory from forest carbon plots established within the study area. This carbon inventory is referred to as the control. 2. Creation of a detailed carbon inventory of areas represented by the control from three forest carbon estimation techniques. The carbon inventory collected via these techniques is hereafter referred to as the treatments. 3. Comparison of each treatment with the control. 4. Quantitative evaluation of each treatment via statistical analysis, specifically: a. Are the results derived from each treatment statistically different from the control? b. Are the results from each treatment statistically different from each other? 5. Qualitative evaluation of each treatment. 2.2 Establishing Forest Carbon Plots 2.2.1 Sampling Technique  A modified stratified random sampling technique was chosen for identifying the stratifications within which forest carbon plots were to be established. This sampling technique was chosen to ensure that the sample would be as close to representative of the study area conditions as possible, given project constraints. The study area was stratified into 8 homogeneous strata based on biogeoclimactic zone and age class (Figure 3).  21    Figure 3. Spatial distribution of stratifications (GeoBC, 2011).   Although there are four biogeoclimactic zones within the study area (Coastal Western Hemlock very wet maritime 1 and 2, Mountain Hemlock and Alpine Tundra), only the two Coastal Western Hemlock zones were included within the sampling analysis. The AT zone was excluded because it contains no forest and the MH zone was excluded due to its inaccessibility (see section 2.2.2 for more information on the definition of inaccessibility). Age classes were separated into four categories: Age Class 1 (0 to 40), Age Class 2 (40 to 80), Age Class 3 (80 to 250) and Age Class 4 (greater than 250). The four largest stratifications within the study area in order of decreasing area are CWH vm 2 / Age Class 1 (3,212 hectares), CWH vm 2 / Age Class 4 (2,951 22  hectares), CWH vm 1 / Age Class 1 (2,392 hectares), and CWH vm 1 / Age Class 2 (1,533 hectares). Together these four stratifications make up almost 80% of the study area, with the other four stratifications making up the remaining 20%. Table 1 lists each stratification in order of decreasing area for the study area.   Table 1. Stratifications in order of decreasing area for the study area.  Stratifications Area (ha) 1 CWH vm 2 / Age Class 1 3,211.9 2 CWH vm 2 / Age Class 4 2,951.4 3 CWH vm 1 / Age Class 1 2,392.0 4 CWH vm 1 / Age Class 2 1,533.2 5 CWH vm 2 / Age Class 3 955.4 6 CWH vm 1 / Age Class 4 937.1 7 CWH vm 1 / Age Class 3 609.4 8 CWH vm 2 / Age Class 2 203.3  2.2.2 Spatial Location of Forest Carbon Plots  Forest carbon plots were located via the following steps:  1. Identification of Candidate Forest Stands The first step in spatially locating forest carbon plots involves the identification of candidate forest stands within which forest carbon plots can be placed. In order to keep within the constraints of this project the location of candidate forest stands was restricted to areas which could be readily accessed by boat, all-terrain vehicle (ATV) and foot within a 3 hour time restriction (one-way). This constraint was defined quantitatively as follows: within 15 kilometers of the marine access point in North Vancouver and having at least one portion of its area located within 200 meters of a road that an ATV could be driven on.  Candidate forest stands that did not meet these criteria were removed from the analysis. A Geographic Information System (GIS) was used to identify the candidate forest stands based on these constraints (Figure 4).  23    Figure 4. Identified candidate forest stands intersecting or adjacent to 200m road buffer (GeoBC, 2011).   2. Random Selection of Candidate Forest Stands for Forest Carbon Plot Placement  Once identified, candidate forest stands were randomly ranked within each stratum (Figure 5). The three highest ranked candidate forest stands were selected as candidates for forest carbon plot placement. In the event that a candidate forest stand could not be accessed it was removed from the selection and the next highest ranked candidate forest stand was selected in its place.  24    Figure 5. Randomly ranked candidate forest stands (GeoBC, 2011).   3. Forest Carbon Plot Placement  Within the study area, terrain might be impassible or hazardous, or a candidate forest stand might be too small or irregularly shaped and the edge of the stand and adjacent stands might adversely affect interior forest condition. When this occurred, forest condition recorded within a forest carbon plot might not be representative of a candidate forest stand’s true condition. Thus the methodology for locating plot centers within a forest candidate stand was modified to mitigate these concerns.  25  The actual location of forest carbon plot centers was established subjectively within a section of forest representative of wider stand conditions and not influenced by edge effects such as roads. While this method was not random, it saved much time and effort given the rough terrain within the study area and also minimized the risk of having a forest carbon plot situated within an area that is not representative of wider stand conditions. Once a forest carbon plot center was identified the path to and from the plot center to the road was marked at regular intervals with flagging tape, and the plot center was permanently marked using a wooden stake and UTM coordinates recorded using a handheld GPS unit (Figure 6).    Figure 6. Forest carbon plot locations (GeoBC, 2011). 26  2.2.3 Forest Carbon Plots  A total of 19 permanent forest carbon plots were placed amongst the eight strata (Table 2). Stratum 1-2 is the largest stratum in area and also contains the greatest number of forest carbon plots (5), while stratum 2-1 and stratum 2-3, which is the smallest stratum in area, contain the fewest, with one each. The largest proportion of forest carbon plots, 12 in total, were placed within the Coastal Western Hemlock (CWH) very wet maritime 1 stratum, principally due to its lower elevation and easier accessibility. The remaining seven plots were placed within the Coastal Western Hemlock (CWH) very wet maritime 2 stratum. Table 2 shows the distribution of forest carbon plots by strata.   Table 2. Forest carbon plot frequency by strata.  Strata Plot Number Frequency Strata Area (ha) 1-1 3,5 2 669.0 1-2 1,7,10,16,18 5 989.8 1-3 9,15 2 136.4 1-4 2,6,11 3 228.1 2-1 13 1 337.3 2-2 4,8,17 3 120.6 2-3 19 1 7.4 2-4 12,14 2 95.9   2.2.4 Field Data  The methodology used for the collection of biomass data from each plot was developed by Dr. Mark Harmon of Oregon State University. In 2008 Dr. Harmon hosted a two and a half day workshop at the H. J. Andrews Experimental Forest, located east of Eugene, Oregon. The H.J. Andrews Forest is situated in the western Cascade Range of Oregon and is broadly representative of Pacific Northwest coniferous forests and associated wildlife and stream ecosystems (Oregon State University, 2008). The workshop was attended by forestry and carbon professionals from across North America including a large proportion from the Pacific Northwest. Topics covered include biological and economic principles for managing forests for carbon sequestration, as well as field and laboratory exercises on how to measure the amount of carbon stored within forests. The forest carbon methodology taught at this workshop will hereby be referred to as the Oregon State University (OSU) methodology. For this project the OSU methodology was used during the 27  collection and calculation of forest carbon for each plot. The methodology is based upon accepted principles of forest inventory, soil sampling, and ecological survey (Oregon State University, 2008). The OSU methodology for the establishment of plot design and setup is based upon USDA Forest Inventory and Analysis (FIA) National Program standards (Oregon State University, 2008; USDA Forest Service, 2011).  The methodology for establishing each plot’s center and boundaries prior to measurement is straightforward. Once a location for the plot center is identified (see section 2.2.2 Spatial Location of Forest Carbon Plots for more information on how plot centers were identified), a 3 foot long wooden stake. To aid in the future identification of the plot center, UTM coordinates were recorded using a handheld GPS and the stake was spray-painted in a bright florescent color. Plot type was identified as a fixed 8.46 meter circular radius with a total area of 225m2. The circumference of the plot was marked with bright flagging tape using a measuring tape at a distance of precisely 8.46 meters from the wooden stake at the plot center. The circumference was marked at both regular intervals and at locations where a tree, stump, or dead fallen tree crossed the plot boundary. In these situations flagging tape is used to delineate these exact locations for subsequent measurement purposes. Finally photographs of the plot were taken from various locations and at various angles. UTM coordinates for all 19 plots are listed in Table 3.  28  Table 3. Forest carbon plot UTM Easting and Northing coordinates.  Plot Number Location (UTM Easting) Location (UTM Northing) 1 507863 5482596 2 504911 5489388 3 508024 5484282 4 511107 5485796 5 507889 5485578 6 506360 5486949 7 510826 5485681 8 509921 5487802 9 505691 5488461 10 506956 5484504 11 506056 5488779 12 509545 5487381 13 509323 5488437 14 512965 5484782 15 506575 5486394 16 508575 5480285 17 513231 5485297 18 510220 5485587 19 509340 5486455  All 19 forest carbon plots were sampled between September 25, 2009 and May 20, 2010. The author was present and supervised the identification and data collection from all plots with the assistance of the following field assistants; Raelene Esteban, Josh George, Charles George, Sophie Middleton, Lianzhen Xu, Ed Thomas, Dave Thomas, Lindsay Skrivanos, and Andy Spence. Sample field notes are shown in Figure 7.  29    Figure 7. Sample field notes.  Within each plot a total of seven carbon pools were measured: live trees with stem diameter less than 10cm, live trees with diameter at breast height (DBH) greater than 10cm, standing dead trees, stumps, coarse woody debris, fine woody debris, and soil. Measured attributes for each pool are described as follows and summarized in Table 4:  • Live Trees (Diameter < 10cm) – this pool consists of all live trees with stem diameters less than 10 cm in diameter. The number and species of trees located within the plot is recorded.  • Live Trees (DBH > 10cm) – this pool consists of all live trees with DBH greater than 10 cm. The species and DBH for each tree within the plot is recorded. DBH is measured using a diameter tape and is defined as being 1.3 meters above the ground on the upper slope side of 30  the tree (Watts and Tolland, 2005). For trees intersecting the plot circumference, the approximate proportion of the tree that is located within the plot is recorded.  • Standing Dead Trees – this pool consists of all dead trees with DBH greater than 10 cm. The species (when possible), height, DBH, and decay class of all standing dead trees within the plot is recorded. Since the top of a standing dead tree may be broken and missing, its height is measured using a measuring tape and a Suunto clinometer. The distance from the observer to the base of the tree, in combination with the angle from the observer to the top of the tree, are used to trigonometrically calculate the tree's height (Watts and Tolland, 2005). The height of ground between the tree and observer must also be noted and factored into the calculations. The decay class of the tree is based on BC's Vegetative Resource Inventory Ground Sampling Procedures (Ministry of Forests and Range, 2007). Tree species is recorded as long as the tree is not in an advanced state of decay and its bark still identifiable.  • Tree Stumps – this pool consists of all tree stumps located within the plot. The species (when possible), decay class (Figure 8), height, and average diameter of tree stumps is recorded. The average diameter of a stump is defined as being half way between the ground and the top of the stump. The height of the stump is defined as being the average height of the stump on all sides (Ministry of Forests and Range, 2007; Oregon State University, 2008). For stumps that are broken or uneven across the top, broken sections are visually folded down to compensate for the missing parts so that the stem height can be estimated (Figure 8) (Ministry of Forests and Range, 2010). 31   Figure 8. Measuring the height of broken stumps (Ministry of Forests and Range, 2010).  • Coarse Woody Debris (CWD) – this pool consists of all dead and downed wood with a diameter at either end greater than 10 cm at the point where the piece crosses the plot circumference. Pieces of coarse woody debris include downed or suspended dead tree boles with or without roots attached, and freshly fallen trees with green foliage but no connection between roots and the ground. Dead branches still connected to trees or pieces buried beneath soil layers are not considered course woody debris. The length, diameter at both ends, and decay class for all coarse woody debris located within the plot is recorded using a measuring tape and diameter tape. The decay class for each piece is also recorded and is adapted from the Ministry of Forests and Range’s Vegetation Resource Inventory Ground Sampling Procedures (Figure 9) (Ministry of Forests and Range, 2010). 32    Figure 9. Decay class classification used for dead wood identification (Ministry of Forests and Range, 2010).   • Fine Woody Debris – this pool consists of all small and fine dead woody debris with diameters less than 10cm. Fine woody debris pieces were separated into two categories: pieces with diameters less than 3cm, and pieces with diameters between 3cm and 10cm. Pieces with diameters less than 2mm were not measured, this includes very small twigs and needles. Fine woody debris was measured using four 5m transects, radiating out from the plot center at 90 degrees right angles from each other. The initial direction of the first transect was chosen randomly. A measuring tape was used to delineate each transect and the number and class of fine woody debris pieces that intersected with each transect were recorded. If a piece was found adjacent but not intersecting a transect then it was not recorded. The number of pieces within each size category for each transect was then averaged between all four transects.  • Soil - this pool consists of the organic soil layer. The organic soil layer overlays the inorganic mineral soil layer and is measured by estimating its depth at four dispersed locations within the plot. Soil pits were dug using a shovel until the organic soil layer could be distinguished from the inorganic soil layer and its depth measured using a ruler or measuring tape. The depth for all four soil pits is then recorded and averaged by plot. 33   Table 4 provides a summary of each measured forest carbon pool and its associated measured parameters.   Table 4. Summary of carbon pools and associated measured parameters.   2.3 Techniques and Processes  2.3.1 Calculating Carbon from Field Data  Total carbon storage for each plot was estimated using a simple simulation model programmed in Excel and developed by Harmon (2001). Field data for each carbon pool (see section 2.2.4) is inputted into the spreadsheet on the appropriate worksheet and biomass and carbon stores are calculated automatically. Calculations for each component are listed in Appendix 1.  Carbon Pool Measured Parameters Comments Live Trees (<10cm) Species, frequency Only trees with stem diameter <10cm were measured Live Trees (>10cm) Species, frequency, volume (using diameter at breast height) and decay class Only trees with DBH >10cm were measured Standing Dead Trees Species (if possible), frequency, volume (using diameter at breast height), height and decay class Only standing dead trees with DBH >10cm were measured  Stumps Species (if possible), frequency, volume (using diameter at breast height and average height), and decay class  Coarse Woody Debris Frequency, decay class and volume (estimated via measured length and diameters at both ends)  Fine Woody Debris Frequency of two size classes (size class I = <3cm, size class II = >3cm and <10cm) 4 transects, 5 meters in length each, were established radiating outwards from the plot center. The radiating direction of each transect was determined randomly Soil Depth of organic top layer  Generally determined to be 10-15cm in depth 34  2.3.2 Calculating Carbon from Forest Carbon Estimation Techniques  2.3.2.1 Canadian Forest Service Carbon Budget Model  The operational-scale carbon budget model of the Canadian forest sector (CBM-CFS3) was first developed in 2002 by Natural Resource Canada’s Canadian Forest Service in partnership with the Canadian Model Forest Network (Natural Resources Canada, 2011). CBM-CFS3 is used by the Government of Canada for reporting to the United Nations Framework Convention on Climate Change on greenhouse gas contributions from Canada’s forests (Natural Resources Canada, 2011). The model was developed in response to the forest industry’s need for an operational-scale carbon accounting tool and was developed specifically for measuring carbon within forests in Canada (Natural Resources Canada, 2011)  CBM-CFS3 consists of a database containing the carbon contents for unique spatial stratifications within Canada. This information is stored within 1,000's of records, each characterizing a specific combination of forest ecosystem type, age, and area (Kurz and Apps, 1999). The model simulates, on a yearly basis, carbon dynamics of aboveground and belowground forest biomass and dead organic matter (Kull et al., 2006). Required input information for measuring existing carbon stocks include volume-over-age curves, detailed forest inventory (including merchantable stem volume, leading species, area, and age), disturbance information, harvest schedule, and land-use change information (Natural Resources Canada, 2011).  Additional information sources are required to model and compare various impacts of forest management scenarios on future carbon stocks (Natural Resources Canada, 2011).  The model bases its calculations of forest biomass using a series of volume-to-biomass estimation conversion factors developed as part of Canada’s national forest inventory (Natural Resources Canada, 2011; Power and Gillis, 2006). Belowground biomass conversion factors are estimated from published regression equations and methods developed specifically for CBM- CFS3 (Natural Resources Canada, 2011). Dead organic matter carbon pools such as litter, woody debris and soil carbon are simulated in CBM-CFS3 and described in Kurz and Apps (1999). The 35  model produces detailed reports for each unique spatial unit. Table 5 contains a summary of ground-measured carbon pools and associated measured CBM-CFS3 carbon pools.  Version 1.2, released August 2010 and downloaded from the National Forest Information System, was used for the analysis. All recorded carbon pool parameters were taken from year 0. Parameters for use of the model were as follows:  Basic specification:  Administrative Boundary: British Columbia  Ecological Boundary: Pacific Maritime  Species: hemlock, cedar and amabilis fir  Non-Forest Soil Types: none selected Optional specification:  Stands: 1  Time steps: 100  Maximum age for growth curves: 300  Growth curve interval: 10  Optional classifiers: none Stand attributes:  Leading Species: stand specific  Land cover: Forest only  Stand age: stand specific  Stand area: stand specific Disturbance types:  Leave all values as default Disturbance events:  No disturbance events scheduled Growth and yield information:  Values unique by plot Modified project parameters:  None made 36   2.3.2.2 Vegetative Resource Inventory Biomass Equations  The Vegetative Resource Inventory (VRI) Biomass Equations modeling technique estimates forest carbon using VRI-derived merchantable volume estimates, and volume to biomass to carbon conversion equations based on the Tree2co2e methodology (Northway, 2008). The technique uses existing forest inventory data from the VRI dataset that are freely available, has simple software requirements (Microsoft Excel) and requires straightforward biomass calculations that are easy to understand and can be altered depending on project area and specifics. Recent developments have seen the addition of biomass estimates to the VRI dataset, however analysis for the present project was completed prior to this information becoming available.  Merchantable tree volume data found within the BC Ministry of Forests, Lands and Natural Resource Operations VRI dataset was used as the primary input source for the VRI Biomass Equations modeling technique. VRI data collection is carried out in two phases: the photo interpretation phase and the ground sampling phase (Ministry of Forests and Range, 2010). During the photo interpretation phase, vegetative polygon characteristics are estimated from aerial photography or other data sources. During the ground sampling phase, the vegetative polygon characteristics are checked to ensure that a given characteristic is present within a given vegetation polygon. Subsequent ground measurements are used to provide a level of certainty regarding the accuracy of the dataset (Ministry of Forests and Range, 2010). Table 5 contains a summary of ground-measured carbon pools and associated measured VRI Biomass Equation carbon pools.  A VRI dataset was downloaded for the study area as a VRI Forest Vegetation Composite Polygons and Rank 1 GIS shapefile from the BC Government GeoBC Data Distribution Service website http://www.for.gov.bc.ca/hts/vri/1. The VRI dataset was compiled from aerial photography acquired in 2005. Using ArcMap software the dataset was clipped to the candidate  1 An updated Vegetative Resources Inventory polygon dataset was released June, 2011 to reflect changes due to harvest and growth for 2010. In addition to these changes the new dataset includes the inclusion of four new attributes pertaining to aboveground biomass (Kivari and Otukol, 2010.). 37  forest stands within the study area. The attribute table of the resulting shapefile was then exported into DBF format and imported into a Microsoft Excel spreadsheet. All subsequent calculations were then performed within this spreadsheet. Dataset fields used included area, species, species percent and volume.  Volume to biomass to carbon calculations were based on the Tree2co2e methodology (Northway, 2008), and include the following steps:  1. Multiply VRI merchantable volume estimate (based on primary utilization figures and measured in cubic meters) by 5002 to determine merchantable tree biomass. This figure represents an approximate average density of wood (in kg/m3) for tree species within British Columbia. 2. Multiply by 2. This figure represents the ratio between merchantable biomass and total tree biomass, the rationale being that approximately half of a tree’s biomass is located within the bole, the other half is located within the branches, leafs and roots. 3. Multiply by 0.46. 1 kilogram of tree contains an average of 0.46kg of carbon. This figure is independent of species, actual biomass to carbon ratios vary by species. 4. Convert to tonnes (1 kilogram = 0.001 tonne) and calculate total carbon dioxide equivalent (CO2E) (multiply kilograms carbon by 3.66)  Total CO2E is therefore equal to: VRI Volume * 500 * 2 * 0.46 * 3.66 2.3.2.3 Private Woodland Planner  Private Woodland Planner was developed by Enfor Consultants Ltd. and is a software model developed specifically to assist small woodland managers in assessing the timber and non-timber forest products values on their property (Enfor, 2007). Stand inventory and yield curves are used as input into the model with carbon values a by-product of this analysis (Enfor, 2007). PWP is a simple, straightforward model that can be used by anyone to obtain a quick estimate of forest  2 Nelson et al. (1985) lists basic density of wood figures for tree species within Western Canada. Actual average wood density for tree species within the study area varies and is in fact closer to 400 kg/m3. The implications of basing this calculation on a 500 kg/m3 wood density figure, as opposed to a 400 kg/m3 wood density figure, are discussed further in Section 4. 38  carbon. Version 2.0, released May 2007, is available from http://www.enfor.com/software/pwp, was used for the analysis.  PWP estimates both the total carbon amount sequestered within the stand to date, and the amount of carbon being sequestered annually. The amount of carbon sequestered to date includes total aboveground gross biomass, including the bole, branches, stump and top (Enfor, 2007; Enfor, 2012) PWP does not include belowground biomass within its reported forest carbon amount, the reasoning behind this omission being that belowground biomass is generally not recoverable and would be left on site regardless of harvest activity  (Enfor, 2012)  Carbon estimates were derived from calculated timber yields drawn from BC Ministry of Forests yield models, VDYP (Variable Density Yield Projection, version 6.6d4) for natural stands and TIPSY (The Table Interpolation Program for Stand Yields, version 3.2m) for unmanaged stands (Enfor, 2007). Natural stand calculations were developed for use on stands older than 30 years. Private Woodland Planner’s carbon dioxide approximation is based on broad averages (Enfor, 2007; Enfor, 2012). The model does not use Biomass Expansion Factors in its calculation of carbon, rather net merchantable volume is increased by 20% to account for non-merchantable above ground biomass (Enfor, 2012). PWP uses a rough calculation of 1m3 = 1 tonne CO2 sequestered, and reports carbon in CO2e. Table 5 contains a summary of ground-measured carbon pools and associated measured PWP carbon pools.  Parameters used in the model were as follows:  Location: Squamish, Coast Forest Region, Squamish Forest District FIZ: C Utilization: 17.5 PSYU: 193 Forest Management Type: Unmanaged Timber Land (TIPSY) Harvest Percent: 100% Area, Species, Species %, Age Class, Site Class: Values unique by plot   39  Table 5. Summary of ground-measured carbon pools and corresponding measured model carbon pools.  Ground- Measured Carbon Pool Canadian Forest Service Carbon Budget Model VRI Biomass Equations Private Woodland Planner Live Trees (<10cm) Measured.  Model uses a non-linear expansion factor to account for stem wood biomass of very small live sapling trees. Not measured. Not measured. Live Trees (>10cm) Measured.  Model uses logistic-type non-linear expansion factors to predict the proportions of total tree biomass  from merchantable stem wood. This includes branches, top, foliage, fine roots, and coarse roots. Live non-merchantable trees and submerchantable trees are also included. Measured.  Model uses a simple factor of 2 to represent the ratio between merchantable biomass and total tree biomass. Measured.  Model calculates total aboveground gross biomass, using a biomass expansion factor of 20% to account for non- merchantable above ground biomass. Standing Dead Trees Not measured.  Factored within total tree biomass estimate? Not measured. Not measured. Stumps Not measured.  Factored within total tree biomass estimate? Not measured. Not measured. Coarse Woody Debris Measured.  Dynamics simulated as part of DOM pools. Not measured. Not measured. Fine Woody Debris Measured.  Dynamics simulated as part of DOM pools. Not measured. Not measured. Soil Measured.  Dynamics simulated as part of DOM pools. Not measured. Not measured. Other - Shrubs and Mosses Not measured (unless a volume-over-age curve is provided for a shrub or moss component. At this time, the model disregards this carbon pool and operates under the assumption that its carbon content is neutral. Not measured. Not measured. 40  2.4 Analysis  The methodology for the comparison of the control and treatment carbon estimates is detailed in sections 2.4.1 and 2.4.2. Section 2.4.1 details procedures for determining whether the control and treatment datasets are normally distributed. Section 2.4.2 details procedures for comparing forest carbon estimate results derived from the control with those derived from the three techniques. All analysis was conducted using the SAS® Analytics software package. 2.4.1 Data Distribution  It is assumed that the distribution of carbon estimates for each plot follows a normal distribution, which is found to be the basis of variation in a large number of biometrical characters in forestry (Kuehl, 1999; Food and Agriculture Organization, 2000). This is supported by the central limit theorem, which states that the mean of any set of variates with any distribution having a finite mean and variance tends to the normal distribution. In addition all commonly used statistics (e.g. standard deviation) assume that data are normally distributed (Food and Agricultural Organization, 2000). Therefore before subsequent comparison analysis can be conducted it must first be established that the data collected from the 19 forest sample plots accurately reflect real-world conditions within the study area and are therefore normally distributed. If the data are found to be not normally distributed then the validity of the subsequent comparison analysis cannot be established.  Normality tests were conducted on two separate datasets, both of which consisted of the residuals of the control and treatment estimates. The first dataset includes all observations from the control and three treatments (19 from the control, 19 from CBM-CFS3, 17 from VRI Biomass Equations, and 19 from PWP for a total of 74 observations), and the second dataset includes all observations minus two outliers (72 observations in total). Outliers are observations that are removed from other values in a random sample from a population. These were identified within SAS as extreme observations.  Four test statistics were used in SAS to detect the presence of non-normality: the Shapiro-Wilk test, the Kolmogorov-Smirnov test, the Cramer-von Mises test, and the Anderson-Darling test. The null hypothesis of each normality test is that the dataset is normally distributed. The 41  normality assumption is based on the residuals of the control and treatment estimates. The results from each test were compared with the chosen significance level (α) or p-value, which was set to 0.01. If the result from each test is greater than the p-value, then the test fails to reject the null hypothesis and thus the dataset is considered normally distributed. Comparison of variance (ANOVA) and f and Pr>F values were also examined for each analysis.  2.4.2 Comparison Analysis  The comparison analysis is separated into two components, differential analysis and statistical analysis. There are two main assumptions of this analysis. The first assumption is that tests for normality were met. The second assumption is that carbon estimates derived from the control were closest to reality.  Differential analysis involves the comparison of carbon estimates in a non-statistical manner. Total forest carbon estimates from the control and treatments were first compared with each other for each plot and displayed graphically. Scatter plots were then created in order to examine the correlation between each treatment with the control. For each scatter plot the variance, slope, and offset were examined as indicators of comparison. A separate Least Squares Mean analysis, using a Bonferroni statistical value of 0.0166 for three comparisons, was also conducted in order to examine any similarities or differences between the means of the control and treatments. The means for each treatment were then plotted against the means for the control for a visual comparison.  42  3. RESULTS 3.1 Control Data  The following section and sub-sections (3.1.1 and 3.1.2) contains a summary of results from the inventory volume data measurements and forest carbon estimates collected from the forest carbon sample plots. 3.1.1 Inventory Results  A large variety of species including red alder, black cottonwood, bigleaf maple, vine maple, western hemlock, western redcedar, Douglas-fir, amabilis fir, and Sitka spruce were observed within the forest carbon plots. The leading/dominant species within 14 of the plots was western hemlock, followed by Douglas-fir and western redcedar. Higher elevation plots located primarily within the Coastal Western Hemlock (CWH) very wet maritime 2 stratum were dominated by western hemlock, western redcedar and Douglas-fir, whereas lower elevation plots located primarily within the Coastal Western Hemlock (CWH) very wet maritime 1 stratum contained a mixture of coniferous and deciduous species. No yellow cedar was observed within the 19 plots. Table 6 lists species composition for all 19 plots as compared with the species composition from the VRI dataset. Species composition differences are explored further within the Discussion section.  43  Table 6. Forest carbon plot stand characteristics. Plot Number Species Composition (Observed)1 Species Composition (VRI)1 1  HW 37%, MB 17%, CW 12%, DR 12%, ACT12% DR70%, ACT20%, MB5%, HW5% 2 HW 40%, CW 40%, MB 20% HW50%, CW30%, FD15%, BA5% 3 CW 64%, HW 32%, SS 4% FD 80%, HW 20% 4 FD 42%, HW 33%, DR 17%, CW 8% HW 70%, DR 20%, BA 10% 5 HW 89%, SS 11% HW 70%, CW 25%, FD 5% 6 HW 92%, CW 8% HW 50%, FD 40%, CW 10% 7 HW 64%, CW 27%, DR 9% HW 70%, DR 20%, BA 10% 8 FD 50%, HW 50% HW 40%, FD 40%, CW 20% 9 HW 45%, MB 45%, ACT 10% HW 50%, CW 30%, FD 15%, BA 5% 10 HW 80%, SS 20% HW 100% 11 HW 100% HW 60%, CW 40% 12 HW 70%, CW 30% HW 70%, CW 20%, BA 10% 13 CW 57%, HW 43% CW 40%, HW 40%, FD 20% 14 HW 86%, CW 14% HW 80%, BA 15%, CW 5% 15 HW 100% HW 95%, CW 5% 16 HW 100% DR 70%, HW 20%, MB 5%, ACT 5% 17  HW 83%, CW 17% HW 70%, YC 30% 18 HW 93%, CW 7% FD 80%, HW 15%, CW 5% 19 FD 36%, HW 36%, CW 28%  HW 60%, CW 30%, YC 10% 1 Species codes are as follows: HW = western hemlock, CW = western redcedar, FD = Douglas-fir, BA = amabilis fir, SS = Sitka spruce, YC = yellow-cedar, DR = red alder, ACT = black cottonwood, MB = bigleaf maple, and MV = vine maple.  Average tree size also differed markedly between species. The largest observed trees were black cottonwood, with an average DBH of 52.18 cm. However black cottonwood trees were only observed within two plots, both of which were located adjacent to the Indian River where growing conditions are good. The next largest observed trees were western redcedar, with an average size of 48.32 cm DBH. Unlike black cottonwood, the number of western redcedar trees were more numerous and observed within 12/19 plots. The largest observed tree within all plots was in fact a western redcedar which had a DBH of 224.5 cm, more than twice as large as any other tree within the 19 plots. Following western redcedar the next largest trees (average DBH, maximum diameter) were Douglas-fir (37.08 cm, ), western hemlock (35.19 cm), red alder (29.01 cm), Sitka spruce (27.60 cm), bigleaf maple (14.62cm ), and vine maple (12 cm). 3.1.2 Carbon Results  Field volume data measurements were converted into forest carbon estimates for each individual forest carbon pool using the OSU methodology (see section 2.2.4 Field Data). The Live Trees (>10cm diameter) carbon pool was found to contain an estimated 66.47% of total carbon within all 19 plots (Figure 10). 44    Figure 10. Total forest carbon percentage by forest carbon pool for all 19 plots.  The Live Trees (DBH > 10cm) carbon pool contains the majority of carbon within the study area (66% of all carbon). Within this pool 55% of all carbon is stored within western hemlock trees, followed by western redcedar (28%), Douglas-fir (8%), and black cottonwood (5%) (Figure 11). This is not surprising given that western hemlock trees are found within all 19 plots. The remaining 4% of carbon within the Live Tree (DBH > 10cm) carbon pool is split between red alder, Sitka spruce, bigleaf maple, and vine maple.  45   Figure 11. Carbon pool size percentage by tree species within the Total live tree (>10cm) carbon pool.  The next largest carbon pool was Forest Floor, containing 16.51% of total carbon, followed by Dead and Down Trees (11.71%), Stumps (2.66%), Fine Woody Debris (2.23%), Standing Dead Trees (0.27%), and Live Trees (<10cm diameter) (0.15%).  Table 7 contains a summary of calculated carbon for each forest carbon pool by forest carbon sample plot. Average forest carbon for each forest carbon pool for all 19 plots is shown in Figure 12. For the complete set of results see Appendix 2.   46  Table 7. Summary of estimated forest carbon pool size (Mg C ha-1) by forest carbon sample plot, estimated using the OSU methodology. n/a = not applicable and is used where measurements were not recorded. All figures rounded to 1 decimal place.     Live Trees (>10cm diameter) Live Trees (<10cm diameter) Dead and Down Trees Standing Dead Trees Stumps Fine Woody Debris Forest Floor Total Plot 1 332.7 0.0 25.9 13.7 0.0 5.6 257.6 635.3 Plot 2 985.3 6.1 15.4 0.0 31.8 8.3 14.0 1054.8 Plot 3 129.9 0.0 93.5 2.5 15.9 8.9 65.2 318.5 Plot 4 152.7 0.0 89.7 1.3 6.5 7.0 25.0 282.2 Plot 5 155.4 0.0 68.3 0.0 70.8 9.6 45.5 349.6 Plot 6 335.5 0.0 119.2 0.0 0.9 10.8 38.6 505.0 Plot 7 196.4 0.0 30.0 0.0 4.9 2.5 11.7 245.4 Plot 8 130.9 0.0 13.4 0.0 26.1 20.8 70.0 260.7 Plot 9 188.2 0.0 45.7 0.0 2.4 9.5 76.5 322.2 Plot 10 318.1 0.0 16.7 0.0 0.9 4.4 77.3 417.3 Plot 11 349.3 0.0 107.1 0.0 0.5 16.5 71.9 545.3 Plot 12 168.9 4.9 47.6 0.7 3.6 15.9 83.7 325.3 Plot 13 93.2 0.0 55.5 0.0 15.6 13.8 35.0 213.0 Plot 14 312.6 n/a 8.9 1.1 1.6 4.8 108.4 437.4 Plot 15 181.1 n/a 33.0 0.2 3.5 4.8 48. 270.9 Plot 16 265.1 0.0 32.8 0.0 0.8 6.3 41.1 346.9 Plot 17 211.8 0.0 36.0 0.0 0.0 8.1 69.2 325.1 Plot 18 98.6 0.0 6.4 0.0 7.3 3.9 21.7 138.0 Plot 19 261.6 0.0 13.1 0.0 1.8 2.6 48.3 327.4 Total 4868.1 10.9 857.9 19.5 194.9 163.5 1208.9   The size and distribution of carbon estimates varied across forest carbon pools. The mean carbon quantity for all plots was 385.28 Mg C ha-1 with estimates from several carbon pools relatively consistent across all plots (Figure 13). For trees with diameters greater than 10 cm carbon pool estimates were fairly consistent with a mean value of 256.21 Mg C ha-1 across all plots. Individual estimates varied between 93.1 Mg C ha-1 and 349.3 Mg C ha-1 except for one plot where the live tree (diameter > 10cm) carbon pool was calculated at 985.3 Mg C ha-1. The forest floor carbon pool featured a single large observation, 256.7 Mg C ha-1, likely due to the plot being located on a deep sediment deposit adjacent to the Indian River. Remaining forest floor carbon pool estimates were also fairly consistent with a range of between 11.7 Mg C ha-1 to 108 Mg C ha-1 and a mean value of 63.63 Mg C ha-1.  47   Figure 12. Total average forest carbon (Mg C ha-1) by forest carbon pool for all 19 plots.  Estimates of coarse woody debris, stumps and fine woody debris carbon pools were not as consistent. Coarse woody debris estimates varied the most with values ranging between 6.4 Mg C ha-1 to 199.2 Mg C ha-1 with a mean value of 45.15 Mg C ha-1. The mean value of the stump carbon pool was calculated at 10.26 Mg C ha-1 although one plot contained 70.8 Mg C ha-1. Finally, fine woody debris values ranged between 2 Mg C ha-1 and 20 Mg C ha-1 with a mean value of 8.6 Mg C ha-1. Most plots did not contain small live trees (with diameters less than 10 cm) or standing dead tree carbon pools. As such their absence nullifies the meaningfulness of any analysis of patterns within these two carbon pools. 48  0 200 400 600 800 1000 1200 1400 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Plot C ar bo n (M g C  h a- 1)  Figure 13. Total forest carbon (Mg C ha-1) by plot, estimated using the OSU methodology.  3.2 Treatment Data  Treatment data carbon estimates have been calculated using all three carbon estimate techniques/models: CBM-CFS3, VRI Biomass Equations and PWP. 3.2.1 Canadian Forest Service Carbon Budget Model (CBM-CFS3)  CBM-CFS3 was used to calculate carbon for softwood and hardwood stemwood, foliage, sapling stemwood, merchantable stem bark, branches, tops and stumps, other aboveground biomass, soil carbon and dead organic matter. Complete forest carbon pool results from the CBM-CFS3 analysis are available in Appendix 3. Figure 14 contains a summary of calculated Mg C ha-1 for each plot. Overall total carbon results appear fairly consistent except for one high value and one low value.  The mean value for all plots is 574.85 Mg C ha-1.  49  0 200 400 600 800 1000 1200 1400 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Plot C ar bo n (M g C  h a- 1)  Figure 14. Total forest carbon (Mg C ha-1) by plot, estimated using CBM-CFS3.  3.2.2 Vegetative Resource Inventory (VRI) Biomass Equations  Total forest carbon per hectare was calculated within the VRI Biomass Equations modeling technique. Unlike the CBM-CFS3, there was marked variation in the results, with values ranging between 150 Mg C ha-1 and 1200 Mg C ha-1. The mean value for all plots was 549.24 Mg C ha-1. Figure 15 contains a summary of calculated Mg C ha-1 for each plot.    50  0 200 400 600 800 1000 1200 1400 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Plot C ar bo n (M g C  h a- 1)  Figure 15. Total forest carbon (Mg C ha-1) by plot, estimated using VRI Biomass Equations.  3.2.3 Private Woodland Planner (PWP)  Total carbon per hectare was calculated for all plots using the PWP software application. Results from this analysis vary considerably, ranging between 75 Mg C ha-1 and 775 Mg C ha-1. The mean value for all plots was 359.86 Mg C ha-1, the lowest mean value of the three techniques. Figure 16 contains a summary of calculated Mg C ha-1 for each plot.    51  0 200 400 600 800 1000 1200 1400 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Plot C ar bo n (M g C  h a- 1)  Figure 16. Total forest carbon (Mg C ha-1) by plot, estimated using Private Woodland Planner.  3.3 Control and Treatment Data Analysis  The following section and sub-sections (3.3.1 and 3.3.2) contains a summary of results from the analysis and comparison of the control and treatment forest carbon estimates. 3.3.1 Data Distribution  Results from the data distribution analysis indicate that the control dataset is not normally distributed. Of the four normality tests conducted (Table 8), only one, Cramer-von Mises, has a p-value greater than 0.01.   Table 8. Normality tests based on all observations.  Normality Test p-value Shapiro-Wilk 0.0041 Kolmogorov-Smirnov <0.01 Cramer-von Mises 0.0134 Anderson-Darling 0.0082  Normality assumptions were not met in the three remaining tests, Anderson-Darling, Kolmogorov-Smirnov and Shapiro-Wilk, due to the presence of several outliers, two of which 52  were significant. The first outlier, with an estimated value of 1054.78 Mg C ha-1, is a part of the control dataset and is located within Plot 2, which is located within a 253 year projected age stand in the Coastal Western Hemlock very wet maritime 1 biogeoclimatic zone. A large proportion of biomass within this plot is stored within a single large cedar tree. The mass of this tree is estimated at 26,834.08 kg, with the resulting carbon per hectare estimate calculated at 596.31 Mg C ha-1. Observations of surrounding forest outside the plot did not reveal other trees of this size, suggesting that the size of this tree is not characteristic of the forest stand.  This outlier exists not because of measurement or model error, but rather because of the selection of plot locations within the candidate stands.  The second outlier, with an estimated value of 1226.84 Mg C ha-1, is a part of the VRI Biomass Equations treatment dataset and is located within Plot 15, located within a 183 year projected age stand in the Coastal Western Hemlock very wet maritime 2 biogeoclimatic zone. This value is considered an outlier because its value is much higher than the estimate derived from the control for the same plot (270.92 Mg C ha-1). As it was assumed that the control is the most accurate estimate it is clear that this data point is also an outlier. Based on the assumption that the values were outliers and not representative of the sample, both were removed from the dataset and the analysis was conducted a second time.   53   Figure 17. Normal Probability Plot. The Y axis is in the units of the "residuals of the ANOVA or Regression analysis". The X axis is the expected normal value (Kuehl, 1999) or the number of standard deviations are the residuals from the centre (zero).   The resulting re-analysis of the control and treatment estimates (analysis based on 72 observations and on the residuals) indicated that the data were normally distributed (Table 9). All four tests had a p-value greater than 0.01. Analysis of variance (ANOVA), an assumption of which is that the data are normally distributed, also improved, from 0.0392 for all observations to 0.2359 for all observations without the two outliers.   Table 9. P-Value tests based on observations without outliers.  Normality Test p-value Shapiro-Wilk 0.0154 Kolmogorov-Smirnov 0.0287 Cramer-von Mises 0.0257 Anderson-Darling 0.0221  54  The removal of outliers also improved the f and Pr>F values (Table 10). Table 10. f and Pr>F values.  Observations F Value Pr > F All Observations 4.45 0.0064 All Observations – Outliers 4.17 0.0090  3.3.2 Comparison Analysis  The following section summarizes the results of the comparison analysis between the control and treatments. Figure 18 provides a comparison of total forest carbon by plot for the control and all treatments.  0 200 400 600 800 1000 1200 1400 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Pl ot Carbon (Mg C ha-1) PWP VRI CBM Control  Figure 18. A comparison of total forest carbon (Mg C ha-1) by plot for the control and all treatments. 55   3.3.2.1 Differential Analysis  Appendix 7 (Comparisons between control and treatment estimates by plot) compares carbon estimates for each plot and shows which treatment estimate was closest to the control estimate. Resulting values were negative for treatment estimates that estimate less carbon per hectare than control estimates and positive for treatment estimates that estimate higher carbon per hectare than control estimates.  This analysis revealed a number of trends. CBM-CFS3 overestimated carbon for 17 out of 19 plots by an average of 246 tonnes per plot and underestimated carbon for the remaining 2 plots by an average of 287 tonnes per plot. VRI Biomass Equations overestimated carbon for 12 out of 17 plots by an average of 341 tonnes per plot and underestimated carbon for 5 out of 17 plots by an average of 139 tonnes per plot. Finally PWP underestimated carbon for 10 out of 19 plots by an average of 196 tonnes per plot and overestimated carbon for the remaining 9 plots by an average of 164 tonnes per plot.  56  -800 -600 -400 -200 0 200 400 600 800 1000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Pl ot Carbon (Mg C ha-1) PWP VRI CBM  Figure 19. A comparison of mean values (Mg C ha-1) by plot. The net difference between the means of each treatment are shown relative to the means of the control.  Furthermore, analyzing the proximity of treatments estimates to the control revealed the following additional observations. Out of all of the forest carbon estimates derived from the treatments only two estimated by CBM-CFS3 (11.8% of the total) were found to be closer to the control than any other treatment estimate. The VRI Biomass Equations fared slightly better with four estimates (or 23.5% of the total) being close to the control. Surprisingly, the remaining 11 plots (64.7% of the total) were best predicted by PWP. Overall by strata, PWP provided the results that were closest to the control (-70.7955 Mg C ha-1) and CBM-CFS3 the farthest from the control (1570.324 Mg C ha-1). Overall results from the VRI Biomass Equations were slightly 57  lower, at 1134.771 Mg C ha-1. A summary of the differences between the control and treatment estimates is shown in Table 11.  Table 11. Summary of differences between control and treatment estimates (G = Control, M1 = CBM-CFS3, M2 = VRI Biomass Equations, M3 = PWP). All units in Mg C ha-1.  PLOT G M1 M2 M3 M1-G M2-G M3-G  Sum 7320.238 10922.19 9337.077 6837.363 3601.955 3393.797 - 482.875 AVG 385.2757 574.8522 549.2398 359.8612 189.5766 199.6351 - 25.4145   In addition, three scatterplots were created to depict the relationship between the control (independent variable) and each treatment (dependent variable) (Figures 19, 20, and 21). 19 plots were included within each analysis except for VRI Biomass Equations, which included 17 plots (see Appendix 4: VRI Biomass Equations Summary of Results). For each scatterplot a regression line was added, along with a corresponding correlation coefficient R2 (R-squared), to determine the percentage of variance explained by the model. If x and y were perfectly related, than R2 would be one. Therefore the lower the R2 value, the less related the control is with the treatment.   R2 = 0.0505 0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400 Control C B M -C FS 3     Figure 20. Scatterplot showing the relationship between forest carbon estimates (Mg C ha-1) of the control and the CBM-CFS3 treatment for 19 plots. 58  R2 = 0.0371 0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400 Control VR I B io m as s Eq ua tio ns                                 Figure 22. Scatterplot showing the relationship between forest carbon estimates (Mg C ha-1) of the control and the PWP treatment for 19 plots. Figure 21. Scatterplot showing the relationship between forest carbon estimates (Mg C ha-1) of the control and the VRI Biomass Equations treatment for 17 plots. R2 = 0.0633 0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400 Control PW P 59  R2 values for each regression line are as follows: for the control and CBM-CFS3 R2 equals 0.051, for the control and VRI Biomass Equations R2 equals 0.037, and for the control and PWP R2 equals 0.063. The R2 value for the control and PWP is highest, and tells us that 6.3% of the variation between the control and PWP is explained by the model. Likewise R2 for the control and CBM-CFS3 is the lowest, with only 3.7% of the variation explained by the model.  Examination of the slope and offset of the regression line can also be a useful indicator for evaluating the comparison of two datasets with each other. The more that two datasets are similar to each other, the more the slope of the regression line will align in a 1:1 or x = y orientation. Likewise, the offset of the regression line for two similar datasets will pass closer through the origin (0,0) of a scatterplot. In this case no regression line has a slope close to 1:1, nor does any pass through the origin. This indicates that each pair of control and treatment datasets are dissimilar to each other, with a weak correlation between the control and treatment datasets.  Several factors explain the low R2 values: the presence of outliers, which tend to decrease the correlation between two datasets, and differences in values between the control and treatment datasets. This second factor likely has the biggest impact on the low R2 values, and further supports the results of the statistical analysis (see section 3.3.2.2 Statistical Analysis).  3.3.2.2 Statistical Analysis  Least Square Mean values (average carbon estimate per plot) for the control and each treatment are listed in Table 12. The analysis for all observations, including outliers, identified one comparison, between the control and CBM-CFS3, where CBM-CFS3 values were statistically different (P < 0.01) from the control. Comparisons between the control and VRI Biomass Equations and PWP were not found to be statistically different. Results from a second analysis with outliers removed found only that the difference between the control and CBM-CFS3 was more significant, and likewise the differences between the control and VRI Biomass Equations and PWP were slightly less significant.   60  Table 12. Least Square Mean values (Mg C ha-1) for the control and each treatment.  Technique Least Square Mean (Mg C ha-1) Control 385.28 CBM-CFS3 574.85 VRI Biomass Equations 549.24 PWP 359.86  Overall, the CBM-CFS3 and VRI Biomass Equations were more similar to each other than to the control. Both overestimate carbon, VRI Biomass Equations by an average of 199.64 Mg C ha-1 and CBM-CFS3 by an average of 189.57 Mg C ha-1. This resulted in the VRI Biomass Equations overestimating carbon by 32.98% on average and CBM-CFS3 overestimating carbon by 29.85% on average. On the other hand, PWP underestimated carbon by 7.06% on average, but the values were the most similar to the control values. The analysis indicated that carbon estimates from the control and PWP were not statistically different.  61  4. DISCUSSION  This thesis presents an approach for evaluating forest carbon quantification modeling techniques within a coastal British Columbia context. Forest carbon estimates were gathered from 19 permanent forest plots (the control) within a southern coastal BC watershed (the Indian River Watershed) and compared with estimates derived from three modeling techniques (the treatments). The three modeling techniques that were evaluated were the Canadian Forest Service Carbon Budget Model (CBM-CFS3), Vegetative Resource Inventory (VRI) Biomass Equations, and Private Woodland Planner (PWP). Results of the comparative analysis found that estimates derived from the control were most similar to those derived from PWP, and least similar to estimates derived from CBM-CFS3.  In this section I will discuss the following two topics: an interpretation of the results and how I believe they contribute to knowledge in the field of forest carbon quantification, and guidelines and recommendations for further research and considerations for forest carbon project developers.  4.1 Interpretation of the Results  The hypothesis prior to the analysis was that CBM-CFS3 would be the most accurate modeling technique. This was based on the synthesis of information from various sources, including training courses, conversations with academics and forestry professionals, literature and media sources, and background information about the model developers. Results from differential and least squares mean analysis however indicated that the forest carbon estimates calculated by PWP were most similar to those derived from the control, followed by the VRI Biomass Equations modeling technique, and lastly by CBM-CFS3.  These results are surprising. Although the results of this analysis show PWP forest carbon estimates are closest to those derived from the control, there are limitations to how each model calculates forest carbon. Both PWP and VRI Biomass Equations only estimate live tree biomass, and do not include any other forest biomass component including stumps, standing dead trees, coarse woody debris, fine woody debris, and soil. As a result both may be underestimating the 62  amount of forest carbon contained within the forest stand represented by each forest carbon plot. In addition, the average woody density figure used in the VRI Biomass Equations modeling technique is likely too high (see section 2.3.2.2 for further explanation), with the effect of overestimating carbon estimates.  There are additional limitations with PWP. PWP is a simple model. It is designed to assist small forest land managers with assessing timber and non-timber product values using very broad forest and inventory parameters. Its primary purpose is to assess timber growth and yield, not measure carbon. It is not designed specifically to generate carbon estimates; rather forest carbon is an additional reporting parameter within the application. The spatial resolution of required input data is low, and the model is not designed to estimate forest carbon at the stand scale. In addition, PWP forest carbon estimates do not include any forest biomass component that is not a part of live trees, and it’s conversion parameters from tree volume to CO2e carbon is overly simplistic. These limitations in PWP’s ability to estimate forest carbon likely results in estimates that are underestimated In reality, the amount of carbon stored within wood varies by species, age and decay stage. Conversely, it is possible that forest carbon estimates from CBM-CFS3 would have been closer to those derived from the ground-based technique had the ground estimates been more representative of true forest conditions. It is unknown how and even if these changes would have impacted the control. However, other research evaluating the accuracy of CFS-CBM3 applied elsewhere has demonstrated that the model is quite effective at measuring forest carbon, and it is unreasonable to think the model would be any less effective at estimating forest carbon within the study area, had a number of circumstances been different. These include having access to better quality input data, having a more representative sample for the comparison, and the difficulties associated with comparing a tool designed to measure carbon at the stand scale with carbon estimates derived at the plot scale. The inherent difficulties in comparing forest carbon estimates derived from different techniques is an especially important factor to consider.  While the overall intended purpose of each technique is the same, that is, to estimate the amount of carbon stored within an area of forest, the manner in which they do so differs. The technology that CBM-CFS3 is built upon differs from the technology that was used to develop the VRI Biomass Equations technique. CBM-CFS3 was designed for foresters and modelers to estimate 63  forest carbon accurately using site-specific forest inventory and ecological parameters. CBM- CFS3 does not make use of volume estimates that are incorporated within the input VRI dataset; rather, the model estimates biomass through the ‘growing’ of forest inventory from ecological characteristics derived from the dataset. As such, the comparison of aboveground biomass between CFS-CBM and other methods that also utilize VRI volume estimates is difficult. CBM- CFS3 uses an aggregated average, while the inventory estimates should be more site-specific (Steven Northway, personal communication, January 2009).  There are also differences in how the each modeling technique calculates live tree carbon. All three models use merchantable tree volume as their primary input upon which total tree and aboveground biomass is extrapolated from. CBM-CFS3 uses researched and published logistic- type non-linear expansion factors to predict the proportions of total tree biomass, whereas VRI Biomass Equations uses a simple factor of 2 to represent the ratio between merchantable biomass and total tree biomass. PWP uses an entirely different method to calculate live tree carbon, it uses a biomass expansion factor of 20% to account for non-merchantable aboveground biomass. In addition, PWP makes the basic assumption that 1 cubic meter of wood, regardless of species or age, is equivalent to a metric tonne of carbon.  Likewise, there is a notable difference between VRI stand attributes, which were used as input for two of the modeling techniques, and measured inventory from the ground survey. This is to be expected. VRI is obtained remotely from air-based sensors, with attributes extrapolated by technicians in a lab. During this process some ground sampling is performed to determine how much of a given characteristic is actually present within a given polygon, and a Net Volume Adjustment Factor is applied to account for errors in the estimates of net tree volume. This technique is necessary for the capture of data over large areas in an efficient manner. VRI volume estimates only represent merchantable volume, that is, the volume of the merchantable tree stem at the 125 utilization level. Ground based estimates are much more accurate, with individual trees and most other biomass-containing features measured by hand. This of course can only be done for very small areas at high cost. Given these differences it is difficult to compare forest carbon estimates derived from each technique.  64  Elsewhere the comparison of techniques for estimating forest carbon stocks has had mixed results. Forest carbon estimates have in the past been characterized by uncertainty (DeFries et al., 2002; Houghton, 1998; Pellitier et al., 2011; Grassi et al., 2008; Harmon and Marks, 2002). For example, Kolchugina and Vinson (1993) compared two methods for measuring forest carbon within the former Soviet Union. One method was based on forest statistical data, the other on ecosystem data. Although estimates compared well within live plant carbon pools, large differences were observed within coarse woody debris carbon pools, where the forest statistical method exceeded the ecosystem method by a factor of 2.4. Likewise, Zhaodi et al. (2010) evaluated three forest volume to biomass conversion methods for measuring the forest carbon inventory of China’s forests. Results from all three methods varied considerably, with large variances and differences in forest carbon estimates of up to 300% between methods.  Elsewhere, Cao et al. (2010) compared methods for estimating forest carbon within six Scots Pine stands in Finland. They found that the various methods (four in total) resulted in different estimates of average carbon stocks, ranging between 51-63% variation for each stand.  However despite these examples, a larger proportion of studies have had greater success. Fazakas et al. (1999) used data from the Swedish national forest inventory, Landsat TM and a classification approach to estimate regional forest biomass and volumes. The accuracy of these estimates at the plot level was poor. However when aggregated at the stand level, results reached relatively low root mean square errors (RMSE) of 8.7% for biomass and 4.6% for volumes. Schmid et al. (2006) compared three models (two empirical, one process) to assess aboveground and belowground carbon pools within several forested regions in Switzerland for four management scenarios. Results found differences in forest carbon estimates but similar projected patterns of net carbon fluxes between each scenario. Differences were attributed to specific model formulations and approaches. Most recently, Asner (2012) predicted aboveground carbon density across regions of Madagascar with high precision (r2 = 0.88) and accuracy (RMSE = 21.1 Mg C ha-1) using a technique combining airborne laser technology, satellite mapping and field measurements. Airborne LiDAR-based mapping of aboveground tropical forest carbon has proven highly precise and accurate, with errors recently becoming indistinguishable from those derived from field measurements (Mascaro et al., 2011).  65  Regardless of the outcomes, this research has resulted in new knowledge gained within the field of forest carbon quantification. While there were shortcomings and issues with the comparison analysis, this was the first time a comparison exercise of this type had been done using these particular modeling techniques in the study area. The results and conclusions from this research contribute to the knowledge in the field of forest carbon quantification in several ways. They verify the large size of forest carbon stores contained within southern British Columbia coastal forests, several plots were found to contain in excess of 1,000 Mg C ha-1. Additionally, measurements gathered from the 19 plots provided useful information on the proportion of carbon stored within each forest carbon pool, and in particular validates research that confirm that trees and soil are by far among the largest forest carbon pools within forests of this type, and typically dominate all other carbon stored in other carbon pools.  New knowledge was also gained from the analysis on the performance of the three modeling techniques. PWP performed surprisingly well in comparison to the other modeling techniques, and the results of the analysis indicate that this relatively simple model, despite initial misgivings, may be used to measure forest carbon accurately and in a quick and efficient manner. Also, the relationship between VRI volume and biomass is relatively strong, with biomass, and hence carbon, readily extrapolated from provincially available VRI to quantify forest carbon over larger areas. Paired with a well designed sampling strategy, this technique is a useful tool for the measurement of forest carbon stocks. Finally, although CBM-CFS3 did not meet expectations in the analysis, better input data and sample size representativeness, combined with a shift in focus from the plot to the stand level, would results in better results. Much research and development has gone into this Canadian-built model and I believe its merits exist, given proper parameters. In addition, it was found that quality of input data and sample design significantly affected results for all three modeling techniques. Utilization of more accurate input data would have resulted in potentially different results.  4.2 Guidelines, Recommendations and Considerations  Given these conclusions, several guidelines, recommendations and considerations for future research and analysis can be made. 66   Guidelines for good sampling design are critical for obtaining a representative sample and ensures accurate representation of the population (i.e. the collected sample reflects true conditions) and refers to the selection of a subset of a population in order to determine characteristics of the whole population (Food and Agriculture Organization, 2000; Kuehl, 1999). If it is unknown whether a sample reflects true conditions, then the reliability of any results based upon a representative sample is difficult to determine. The sample obtained for the study area consisted of 19 plots, and is not considered statistically valid according to sample size calculations (add reference to Tony's book). Based on a confidence level of 95%, a precision level of ±10% (38.5 Mg C ha-1) would require approximately 145 plots in order for the sample to be  statistically representative of study area conditions. Sample size was also calculated at two other levels of precision. A ±5% (19.26 Mg C ha-1) precision level would require approximately 569 plots, whereas a ±20% (77.054 Mg C ha-1) precision level would require approximately 36 plots. However, due to project constraints (geography, weather, daylight, terrain, transportation, distance, etc.), obtaining a larger sample was not possible. As such, several strategies were employed to minimize sampling error and maximize sample representation. These included stratification of the study area from a single, heterogeneous area, into smaller, more homogenous areas, randomly ranking and selecting candidate forest stands, placing plots in areas which best mimic wider stand conditions, and avoiding areas where abnormal stand conditions exist. If given the opportunity to conduct further research within the study area several additional strategies could be incorporated into the methodology.  Forest carbon project proponents are required to use the best available inventory data to measure baseline and project carbon stocks, and to ensure that these estimates are conservative. Future research in the study area should incorporate the use of VRI and statistically valid ground sample data as the base inventory for project development, with estimates and projections updated to account for any disturbance that might have occurred during each reporting period.  Project proponents are also required to achieve an accurate and unbiased representation of the population whenever possible. Sample size should be maximized, with plots spatially located in as many areas as possible. One recommended strategy for achieving this is to establish plots in two phases. The first phase, or testing phase, would include the initial establishment of 2-4 plots 67  in 2-4 diverse locations within the study area, classified by forest type or ecological zone. The purpose of this initial phase is to gain a better understanding of the logistics involved in sampling forest carbon from plots, and to get a measure of variability which will assist in the calculation of plots needed within the sampling design. The second phase would then include the establishment of all remaining plots. The incorporation of these phases within a randomized stratified sampling strategy, without concessions for access or resource constraints, should increase the probability of obtaining a more accurate and unbiased representation of the population.  Forest carbon project proponents are also required to manage for any uncertainty associated with a project. A certain level of uncertainly will always be associated with modeled forest carbon estimates, however future research should incorporate a thorough assessment of this uncertainty to determine key potential sources of error. Error can occur during all phases of a project, and in particular during the process of placing sampling plots, and during the measurement process. Even slight differences in methodology can result in inconsistent results between measurers and plots. A few examples include the measurement of the organic soil depth, which can be difficult to ascertain, particularly when soil type, moisture and light levels between plots differ and make locating the transition point between the organic and inorganic layer difficult to find. Tree height can be also be challenging to estimate in areas of dense canopy closure, as can be tree diameter, which can vary depending on the height off the ground at which the field technician chooses to take the measurement at.  Several strategies can be utilized to mitigate for these possible effects of error. Ensuring that both the measurer and methodology remain consistent throughout the inventory collection process is one such strategy, as was carefully documenting forest conditions within each plot. However other strategies for mitigating error were not used. This includes increasing the sample size and locating plots randomly. This would have increased the probability of obtaining a more accurate and unbiased representation of the population. Additionally, further resources would have allowed for the re-measurement of plots in order to minimize measurement error and ensure accurate estimates. Other strategies include using sampling results to re-calibrate model estimates, comparing results with independent data and to the results of other methods, re- sampling plots, conducting a thorough analysis of uncertainty related to each step in the 68  conversion, adding additional models to the analysis, and improving the quality of input data if available.  The impact of a forest carbon project on carbon stocks requires the measurement of both baseline and project scenario carbon stocks, the difference between these two being defined as the net carbon benefit, or additionality. Accurately measuring both the baseline and the additionality is critical to a successful forest carbon project, with exact requirements for measuring forest carbon pools varying depending on chosen standard, methodology and project type. There are many methods for measuring baseline and additionality carbon stocks. These methods range from site- specific data collection, to broad national and global estimates that use remotely-collected forestry data (such as Vegetation Resources Inventory, air photos, LiDAR, and satellite imagery). However for most forest carbon projects, utilizing a ground-based quantification method can be inefficient and costly. As such, a model should be used to create a coarse-scale approximation of forest carbon stock that can be combined with field sampling to verify the model results. Given these recommendations and considerations, the best strategy for quantifying forest carbon stocks is to combine a suitable model, which can be used to approximate forest carbon over the study area, with field sampling to validate model results. The combination of these two methods results in the most efficient method for estimating forest carbon stocks.  4.3 Conclusion   Results from this research indicate that forest carbon estimates from PWP are closest to estimates derived from the ground-based technique, while estimates from CFS-CBM3 were most different. These results were the opposite from the original hypothesis, which predicted that estimates from CFS-CBM3 would be closest to estimates derived from the ground-based technique within the study area. Estimates from the VRI Biomass Equations model were also found to be farther from the control than originally hypothesized. Given these results, several factors, both anticipated and unanticipated, are likely responsible.  These factors include resource constraints, which limited the number of plots that could be established and restricted the location of where plots could be placed to areas that were readily 69  accessible. In addition, constraints on available resources also limited the number of modeling techniques that could be evaluated. Another factor which likely influenced these results was the quality of data that was used as input into the models. The data used in this research was the best available.  The results from this research could nonetheless contribute to the field of forest carbon quantification. Although several components of the analysis were not statistically valid, sampling methodology was framed by “real-world” constraints, and reflects conditions that possibly exist under normal project scenarios. During the development of a forest carbon project, there are constraints to resources that will exist and must be considered. In contrast, much of the research that had been conducted in the field of forest carbon quantification does not share these constraints. Such research is conducted under ideal circumstances where a sampling strategy is not constrained by limitations to budget, access or time constraints.  The results of this research demonstrate a continued need to further develop efficient and accurate methods for measuring forest carbon stocks under realistic conditions, and show how despite some issues, models for doing so are necessary. Because of this, model development will continue and techniques will continue to be refined and become more accurate. This refinement will provide project developers with the ability to better measure carbon within forests, and as such lead to greater economic opportunities, improved forest conservation, and increased mitigation of the effects of climate change.       70  References Asner, G.P., Clark, J.K., Mascaro, J., Vaudry, R., Chadwick, K.D., Vieilledent, G., Rasamoeline, M., Balaji, A., Kennedy-Bowdoin, T., Maatong, L., Colgan, M.S., and Knapp, D.E. 2012. Human and environmental controls over aboveground carbon storage in Madagascar. Carbon Balance and Management. 7:2. Asner, G.P., Mascaro, J., Muller-Landau, H.C., Vieilledent, G., Vaudry, R., Rasamoelina, M., Hall, J.S., and van Breugel, M. 2011. A universal airborne LiDAR approach for tropical forest carbon mapping. Oecologia. DOI 10.1007/s00442-011-2165-z. Asner, G.P., Powell, G.V.N., Mascaro, J., Knapp, D.E., Clack, J.K., Jacobson, J., Kennedy- Bowdoin, T., Balaji, A., Paez-Acosta, G., Victoria, E., Secada, L., Valqui, M., and Hughes, R.F. 2010. High-resolution forest carbon stocks and emissions in the Amazon. Proceedings of the National Academy of Sciences. 10716738-16742. Baccini, A., Goetz, S.J., Walker, W.S., Laporte, N.T., Sun, M., Sulla-Menashe, D., Hackler, J., Beck, P.S.A., Dubayah, R., Friedl, M.A., Samanta, S., and Houghton, R.A. 2012. Estimated carbon dioxide emissions from tropical deforestation improved by carbon- density maps. Nature Climate Change. doi:10.1038/nclimate1354. Baccini, A., Laporte,. N., Goetz,. S.J., Sun, M., and Dong, H. 2008. A first map of tropical Africa's above-ground biomass derived from satellite imagery. Environmental Research Letters. 3:045011. BC First Nations Forestry Council. 2008. Carbon credit opportunities for First Nations in BC [online]. Available from http://www.fnforestrycouncil.ca/documents/CarbonCredit Opportunities.pdf [accessed December 18, 2008]. BC Ministry of Forests, Mine and Lands. 2010. The state of British Columbia’s forests, 3rd ed. Forests Practices and Investment Branch [online]. Available from http://www.for.gov.bc.ca/ hfp/sof/index.htm#2010_report [accessed 8 June 2011]. BC Ministry of Forests and Range. 2007. Vegetation Resources Inventory Ground Sampling Procedures. Prepared by Ministry of Forests and Range Forest Analysis and Inventory Branch for the Terrestrial Ecosystems Task Force Resources Information Standards Committee [online]. Available from http://www.for.gov.bc.ca/hfd/library/documents/bib4661 2_2007.pdf [accessed 20 December 2011].  Bhishma, P.S., Pandey, S.S., Pandey, A., Rana, E.B., Bhattarai, S., Banskota, T.R., Charmakar, S., and Ta,rakar, R. 2011. Forest Carbon Stock Measurement: Guidelines for measuring carbon stocks in community-managed forests. Asia Network for Sustainable Agriculture and Bioresources (ANSAB), Federation of Community Forest Users, Nepal (FECOFUN), International Centre for Integrated Mountain Development (ICIMOD), Norwegian Agency for Development Cooperation (NORAD). 978-9937-2-2612-7. 71  Birdsey, R., Pregitzer, K., and Lucier, A. 2006. Forest Carbon Management in the United States - - 1600-2100. Journal of environmental quality. v.35, no.4, p. 1461-1469. Boudewyn, P., Song, X., Magnussen, S., and Gillis, M.D. 2007. Model-based, volume-to- biomass conversion for forests and vegetated land in Canada. Info report – BC-X-411 Brown, S. 1999. Guidelines for Inventorying and Monitoring Carbon Offsets in Forest based Projects. Winrock International [online]. Available from   http://www.winrock. org/REEP/Guidelines.html [accessed 18 December 2008]. Brown, S., Sathaye, J., Cannell, M., and Kauppi, P. 1996. Management of forests for mitigation of greenhouse gas emissions. Climate Change Ch. 24:773-797. Brown, S. 2002. Measuring carbon in forests: current status and future challenges. Environmental Pollution. 116 (3): 363-372. Brown, S. 2002. Measuring, monitoring and verification of carbon benefits for forest based projects. Philosophical Transactions: Mathematical, Physical & Engineering Sciences 360. 1797:1669-1683. Cairns, M.A., Brown, S., Helmer, E.H., and Baumgardner, G.A. 1997. Root biomass allocation in the world's upland forests. Oecologia. 111:1-11. California Climate Action Registry. 2007. Forest Sector Protocol [online]. Available from http://www.climateregistry.org/resources/docs/protocols/industry/forest/forest_project_pro tocol_version_2.1_sept2007.pdf [accessed 20 February 2011]. Canada's National Forest Inventory. 2011 [online]. Available from https://nfi.nfis.org/history.php ?lang=en [accessed on 15 October 2011].  Cao, T., Valsta, L., and Mäkela, A., 2010. A comparison of carbon assessment methods for optimizing timber production and carbon sequestration in Scots pine stands. Forest Ecology and Management. 260 1726-1734. CBM-CFS3 Tutorial 1: How to use standard input data to perform carbon analyses with the Operational-Scale Carbon Budget Model (CBM-CFS3). Available from: Carbon Accounting Liaison Officer, Canadian Forest Service, Northern Forestry Centre, 5320– 122 Street, Edmonton, AB T6H 3S5. Chave, J., Churst, G., Condit, R., Aguilar, S., Hernandez, A., Lao, S., and Perez, R. 2004. Error propagation and scaling for tropical forest biomass estimates. In Tropical forests and global atmospheric change. Edited by Malhi, Y., and Phillips, O. Oxford University Press. 155-166. Chénost, C., Gardette, Y.M., Demenois. J., Grondard, N., Perrier. M., and Wemaëre, M. 2010. Bringing forest carbon projects to the market [online].United Nations Environment Programme. Available from http://www.unep.fr/energy/activities/forest_carbon/pdf/ 72  Guidebook%20English%20Final%2019-5-2010%20high%20res.pdf [accessed 22 May 2010]. Cooper, C. F. 1983. Carbon storage in managed forests. Canadian Journal of Forest Research. 13:155-166. Davies, B., Dettman, S., Goslin, M., Mertins, M., and Silverman, H. 2011. Carbon Analysis of Proposed Forest Management Regimes on the Elliott State Forest [online]. Ecotrust. Available from http://www.ecotrust.org/forestry/ [accessed 13 November 2011]. DeFries, R. S., Houghton, R.A., Hansen, M.C., Field, C.B., Skole, D., and Townshend, J. 2002. Carbon emissions from tropical deforestation and regrowth based on satellite observations from the 1980s and 1990s. Proc. Natl Acad. Sci. USA 99, 14256-14261. Desai, A.R., Paw, U.K.T., Cook, B.D., Davis, K.J., and Carey, E.V. 2005. Comparing net ecosystem exchange of carbon dioxide between an old-growth and mature forest in the upper Midwest, USA. Agricultural and Forest Meteorology. 128: 33-55. Dixon, R.K., Brown, S., Houghton, R.A., Solomon, A.M., Trexler, M.C., and Wisniewski, J. 1994. Carbon pools and flux of global forest ecosystems. Science. 263:185-190. Dixon, R. K., Schroeder, P. E., and Winjum, J.K. 1991. Assessment of promising forest- management practices and technologies for enhancing the conservation and sequestration of atmospheric carbon and their costs at the site level. US Environmental Protection Agency, Environmental Research Laboratory. Pages: 600-603. Dong, J., Kaufmann, R.K., Myneni, R.B., Tucker, C.J., Kauppi, P.E., Liski, J., Buermanna, W., Alexeyev, V., and Hughes, M.K. 2003. Remote sensing estimates of boreal and temperate forest woody biomass: carbon pools, sources, and sinks. Remote Sensing of Environment. 84 393–410.  Ecological Stratification Working Group. 1995. A National Ecological Framework for Canada. Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological Resources Research and Environment Canada, State of the Environment Directorate, Ecozone Analysis Branch, Ottawa/Hull. Report and national map at 1:7500 000 scale.  EcoRegions Working Group. 1989. Ecoclimatic Regions of Canada, First Approximation. Ecological Land Classification Series, No. 23, Environment Canada, Ottawa 119p.  Ecosecurities. 2010. The forest carbon offsetting report [online]. Available from http://www.ecosecurities.com/Registered/ForestCarbonOffsettingReport2010.pdf [accessed 15 November 2011]. Enfor Consultants Ltd. 2007. Private Woodland Planner User Guide Version 2.0 [online]. Available from http://forestcarbon.ca/?Page=\software\pwp\ [accessed 4 May 2010]. 73  European Forest Institute, CASFOR-II: Modelling Carbon Sequestration in Forested Landscapes, CO2FIX v3.2 [online]. Available from http://www.efi.int/projects/casfor/models.htm [accessed 11 November 2011]. Fahey, J.T., Woodbury, P.B., Battles, J.J., Goodale, C.L., Hamburg, S.P., Ollinger, S.V., and Woodall, S.W. 2010. Forest carbon storage: ecology, management, and policy. Frontiers in Ecology and the Environment. 8: 245–252. First Nations Summit. 2011. BC Forest Carbon Offset Protocol. The First Nations Summit, West Vancouver, BC. Food and Agriculture Organization, 2000. Forestry Statistics and Data Collection [online]. Available from ftp://ftp.fao.org/docrep/fao/003/X6831E/X6831E00.pdf [accessed 9 January 2012]. Food and Agriculture Organization, 2011. State of the World’s Forests 2011 [online]. Available from http://www.fao.org/docrep/013/i2000e/i2000e.pdf [accessed 12 November 2011. Fazakas, Z., Nilsson, M., and Olsson, H., 1999. Regional forest biomass and wood volume estimation using satellite data and ancillary data. Agricult. Forest Meteorol. 98-99, 417- 425. Gayton, D.V., 2008. Impacts of Climate Change on British Columbia’s Biodiversity: A Literature Review. Forest Research Extension Society (FORREX) Series 23.  Goetz, S.J., Baccini, A., Laporte, N.T., Johns, T., Walker, W., Kellndorfer, J., Houghton, R.A., and Sun, M. 2009. Mapping and monitoring carbon stocks with satellite observations: a comparison of methods. Carbon Balance and Management. 4:2.  Golinkoff, J, Hanus, M and Carah, J. 2011. The use of airborne laser scanning to develop a pixel- based stratification for a verified carbon offset project. Carbon Balance and Management. 6:9.  González-Alonso, F., Merino-De-Miguel, S., Roldán-Zamarrón, A., García-Gigorro, S., and Cuevas, J. M. 2006. Forest biomass estimation through NDVI composites. The role of remotely sensed data to assess Spanish forests as carbon sinks. International Journal of Remote Sensing. 27:24, 5409 – 5415.  Goodale, C.L., Apps, M.J., Birdsey, R.A., Field, C.B., Heath, L.S., Houghton, R.A., Jenkins, J.C., Kohlmaier, G.H., Kurz, W., Liu, S., Nabuurs, G-J., Nilsson, S., and Shvidenko, A.Z. 2002. Forest Carbon Sinks in the Northern Hemisphere. Ecological Applications. 12:891– 899.  Grassi, G., Monni, S., Frederici, S., Achard, F. and Mollicone, D. 2008. Applying the conservativeness principle to REDD to deal with the uncertainties of the estimates. Environ. Res. Lett. 3, 035005.  74  Greig, M., and Bull, G. 2009. Carbon management in British Columbia’s forests: opportunities and challenges. FORREX Forum for Research and Extension in Natural Resources. Forest Research Extension Society (FORREX) Series 24.  Guan, D., Wu, J.B., Zhao, X.S., Han, S.J., Yu, G.R., Sun, X.M., and Jin, C.J. 2006. CO2 fluxes over an old temperate mixed forest in northeastern China. Agricultural and Forest Meteorology. 137: 138-149.  Hamann, A. and T. Wang. 2006. Potential Effects of Climate Change on Ecosystems and Tree Species Distribution in British Columbia. Ecology. 87(11): 2773-2786.  Hamilton, K., Chockalingam, U., and Bendana, M. 2010. State of the forest carbon markets 2009 [online]. Available from http://www.forest-trends.org/documents/files/doc_2384.pdf [accessed 15 October 2010]. Harmon, M.E. 2001. Carbon cycling in forests: simple simulation models [online]. H. J. Andrews Research Report. Available from http://www.fsl.orst.edu/lter/pubs/ [accessed 8 October 2008]. Harmon, M.E., Marks, B. 2002. Effects of silvicultural practices on carbon stores in Douglas-fir western hemlock forests in the Pacific Northwest, U.S.A.: results from a simulation model. Can. J. For. Res. 32, 863-877. Hollinger, D.Y., Kelliher, F.M., Byers, J.N., Hunt, J.E., McSeveny, T.M., and Weir, P.L. 1994. Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. Ecology. 75: 134-150.  Houghton, R. A. 2005. Aboveground forest biomass and the global carbon balance. Glob. Change Biol. 11, 945-958.  International Energy Agency Bioenergy, BIOmass based Climate Change MITigation through Renewable Energy Systems, Software Tools, The BIOMITRE Software Tool [online]. Available from http://www.ieabioenergy-task38.org/softwaretools/ [accessed 11 November 2011].  IPCC. 2007a. Summary for Policymakers. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.  IPCC. 2007b. Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, 2007. B. Metz, O.R. Davidson, P.R. Bosch, R. Dave, L.A. Meyer (eds). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.  Iverson, C. 2009. In search of standards for forest carbon offset projects in BC - a review of Georgian and Californian state standards [online]. Undergraduate Report. University of 75  British Columbia. Available from https://circle.ubc.ca/bitstream/handle/2429/16099/ IversonChad_FRST_497_Graduating_Essay_2008.pdf?sequence=1 [accessed 5 September 2011].  Kellwe, M., Palace, M., and Hurtt, G. 2001. Biomass estimation in the Tapajos National Forest, Brazil: examination of sampling and allometric uncertainties. Forest Ecology and Management. 154:371-382. Kivari, A., Xu, W., and Otukol, S. 2010. Volume to biomass conversion for British Columbia forests [online]. Forest Analysis and Inventory Branch, Ministry of Forests and Range. Available from http://www.for.gov.bc.ca/hts/vridata/standards/news/ Volume_to_Biomass_Conversion_Report.pdf [accessed 20 May 2011] Knohl, A., Schulze, E.D., Kolle, O., and Buchmann, N. 2003. Large carbon uptake by an unmanaged 250-year-old deciduous forest in central Germany. Agricultural and Forest Meteorology. 118: 151-167. Kolchugina, T.P., and Vinson, T.S. 1993. Comparison of two methods to assess the carbon budget of forest biomes in the Former Soviet Union. Oregon State University. Water, Air, and Soil Pollution. 70: 207-221. Kuehl, O., R. 1999. Statistical Principles of Research Design and Analysis. Belmont, CA: Duxbury Press. Kull, S., Kurz, W.A., Rampley, G., Banfield, G.E., Schivatcheva, R.K., and Apps, M.J. 2006. Operational scale carbon budget model of the Canadian forest sector (CBM-CFS3) version 1.0: user’s guide [online]. Available from http://nofc.cfs.nrcan.gc.ca/ bookstore_pdfs/26683.pdf [accessed 11 January 2008] Kurz, W. A. 2010. Experiences in implementing a Tier 3 approach. In IPCC Experts Meeting, Sidney, Australia, 9-11 August 2010. Edited by H.S.Eggleston, N. Srivastave, K. Tanabe, J. Baasansuren and M. Fukuda. Carbon Budget Model of the Canadian Forest Sector, Victoria, BC. pp. 39-47. Kurz, W.A.; Apps, M.J. 1999. A 70-year retrospective analysis of carbon fluxes in the Canadian forest sector. Ecol. Appl. 9(2):526–547. Kurz, W. A., Apps, M. J., Comeau, P. G., and Trofymow, J.A. 1996. The carbon budget of British Columbia’s forests, 1920-1989: preliminary analysis and recommendations for refinements. Canadian Forest Service, Victoria, BC. Lambert M.-C., Ung, C.H. , and Raulier, F. 2005. Canadian national tree aboveground biomass equations. Can. J. For Res. 35: 1996-2018. Law, B.E., Goldstein, A.H., Anthoni, P.M., Unsworth, M.H., Panek, J.A., Bauer, M.R., Fracheboud, J.M., and Hultman, N. 2001. Carbon dioxide and water vapor exchange by young and old ponderosa pine ecosystems during a dry summer. Tree Physiology. 21: 299-308. 76  LiveSmart BC. 2011. Effects of Climate Change. British Columbia Provincial Government [online]. Available from http://www.livesmartbc.ca/learn/effects.html [accessed 14 October, 2011]. Mascaro, J., Detto, M., Asner, G.P., and Muller-Landau, H.C. 2011. Evaluating uncertainty in mapping forest carbon with airborne LiDAR. Remote Sensing of Environment. doi:10.1016/j.rse.2011.07.019  Ministry of Forests and Range. 2010. Vegetation Resources Inventory Ground Sampling Procedures - VRI Ground Sampling Procedures. Version 4.9. Forest Analysis and Inventory Branch [online]. Available from http://www.ilmb.gov.bc.ca/risc/index.html [assessed 22 December 2012].  Myeong, S., Nowak, D.J., and Duggin, M.J. 2006. A temporal analysis of urban forest carbon storage using remote sensing. Remote Sensing of Environment. 101 277–282. Myneni, R.B., Dong, J., Tucker, C.J., Kaufmann, R.K., Kauppi, P.E., Liski, J., Zhou, L., Alexeyev, V., and Hughes, M.K. 2001. A large carbon sink in the woody biomass of northern forests. PNAS. u vol. 98 u no. 26. Natural Resources Canada, Canadian Forest Service, Climate Change, Carbon Budget Model [online]. Available from http://cfs.nrcan.gc.ca/pages/94 [accessed 11 November 2011]. Nelson, R.W., Dobie, J., and Wright, D.M. 1985. Conversion factors for the forest products industry in western Canada. Forintek Canada Corp., Western Forest Products Laboratory. Information Report VP-X-97. New South Wales Government, Department of Primary Industries, TimberCAM, What is TimberCAM [online]? Available from http://www.dpi.nsw.gov.au/forests/info/timbercam [accessed 11 November 2011]. Northway, S. 2008. Tree2co2e methodology handout. University of British Columbia, Faculty of Forestry. Odum, E.P. 1969. The strategy of ecosystem development. Science. 164: 262-270. Olander, J. and Ebeling (Editors), J. 2011. Building Forest Carbon Projects: Step-by-Step Overview and Guide. In Building Forest Carbon Projects, Washington, DC: Forest Trends. Oregon State University. 2008. Carbon Management Workshop. Mark Harmon, College of Forestry, Forest Ecosystems and Society, 210 Richardson Hall, Oregon State University, Corvallis, Oregon. Patenaude, G., Milne, R., Dawson, T.P. 2005. Synthesis of remote sensing approaches for forest carbon estimation: reporting to the Kyoto Protocol. Environmental Science and Policy. 8 161-178. 77  Paw, U.K.T., Falk, M., Suchanek, T.H., Ustin, S.L., Chen, J.Q., Park, Y.S., Winner, W.E., Thomas, S.C., Hsiao, T.C., Shaw, R.H., King, T.S., Pyles, R.D., Schroeder, M., and Matista, A.A. 2004. Carbon dioxide exchange between an old-growth forest and the atmosphere. Ecosystems. 7: 513-524. Pellitier, J., Ramankutty, N. and Potvin, C. 2011. Diagnosting the uncertainty and detectability of emission reductions for REDD+ under current capabilities: An example for Panama. Environ. Res. Lett. 6, 024005.  Penner, M., Power, K.,  Muhairwe, C., Tellier, R., and Wang, Y. 1997. Canada’s forest biomass resources: deriving estimates from Canada’s forest inventory. Pacific Forestry Centre Canadian Forest Service. Information Report BC-X-370.  Phillips, O. L., Malhi, Y., Higuchi, N., Laurance, W.F., Núñez, R.M., Vásquez, R.M., Laurance, S.G., Ferreira, L.V., Stern, M., Brown, S., and Grace, J. 1998. Changes in the carbon balance of tropical forests; Evidence from long-term plots. Science. 282, 439-442.  Piao, S., Fang, J., Zhu, B., and Tan, K. 2005. Forest biomass carbon stocks in China over the past 2 decades: estimation based on integrated inventory and satellite data. Journal of Geographical Research. vol. 110, G01006.  Power, K., and Gillis, M. 2006. Canada’s Forest Inventory 2001. Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria, BC. Information Report BC- X-408.  Province of British Columbia. GeoBC [online]. Available from http://geobc.gov.bc.ca/ [accessed 22 January 2012].  Ravindranath, N.H., and Ostwald, M. 2008. Carbon inventory methods: handbook for greenhouse gas inventory, carbon mitigation and roundwood production projects. Series: Advances in Global Change Research. Vol. 29 XX, 308 p.  Robinson, A.P., Duursma, R.A., and Marshall, J.D. 2005. A regression-based equivalence test for model validation: shifting the burden of proof. Tree Physiology 25. Heron Publishing, Victoria, Canada. pp. 903–913. Roser, C., Montagnani, L., Schulze, E.D., Mollicone, D., Kolle, O., Meroni, M., Papale, D., Marchesini, L.B., Federici, S., and Valetini, R. 2002. Net CO2 exchange rates in three different successional stages of the "Dark Taiga" of central Siberia. Tellus 54: 642-654. Ryan, M.G., Harmon, M.E., Birdsey, R.A., Giardina, C.P., Heath, L.S., Houghton, R.A., Jackson, R.B., McKinley, D.C., Morrison, J.F., Murray, B.C., Pataki, D.E., and Skog, K.E. 2010. A synthesis of the science on forests and carbon for U.S. forests. Issues in Ecology. Report Number 13. Saatchi, S.S., Harris, N.L., Brown, S., Lefsky, M., Mitchard, E.T.A., Salas, W., Zutta, B.R., Buermann, W., Lewis, S.L., Hagen, S., Petrova, S., White, L., Silman, M., and Morel, A. 78  2011. Benchmark map of forest carbon stocks in tropical regions across three continents. Proceedings of the National Academy of Sciences. 108:9899-9904. Schelhaas, M.J., van Esch, P.W., Groen, T.A., de Jong, B.H.J, Kanninen, M., Liski, J., Masera, O., Mohren, G.M.J., Nabuurs, G.J., Palosuo, T., Pedroni, L., Vallejo, A., and Vilén, T. 2004. CO2FIX V 3.1 – A modelling framework for quantifying carbon sequestration in forest ecosystems. Wageningen, Alterra, Alterra-rapport 1068. 120 blz.; 60 figs.; 4 tables.; 95 refs. Schmid, S., Thürig, E., Kaufmann, E., Lischke, H., and Bugmann, H. 2006. Effect of forest management on future carbon pools and fluxes: A model comparison. Forest Ecology and Management. 237 65-82. Spittlehouse, D.L. and Stewartz, R.B.,2003. Adaptation to climate change in forest management. BC Journal of Ecosystems and Management. Volume 4, Number 1. Standish, J.T., Manning, G.H., and Demearschalk, J.P. 1985. Development of biomass equations for British Columbia tree species. Info Report BC-X-264. Pacific forest research centre. Stinson, G., Kurz, W.A., Smyth, C.E., Neilson, E.T., Dymond, C.C., Metsaranta, J.M., Boisvenue, C., Rampley, G.J., Li, Q., White, T.M., and Blain, D. 2011. An inventory- based analysis of Canada’s managed forest carbon dynamics, 1990 to 2008. Global Change Biology. 17, 2227–2244. Tan, Z. H., Zhang, Y.P., Schaefer, D., Yu, G.R., Liang, N., and Song, Q.-H. 2011. An old- growth subtropical Asian evergreen forest as a large carbon sink. Atmospheric Environment. 45: 1548-1554. Tans, P., and Keeling, R. NOAA Research, Trends in Atmospheric Carbon Dioxide [online]. Available from http://www.esrl.noaa.gov/gmd/ccgg/trends/ [accessed 11 November 2011]. Tsleil-Waututh Nation. 2007. Indian river watershed bioregional atlas. 1st ed. Tsleil-Waututh Nation. 2000. Our Land to Share. Tupek, B., Zanchi, G., Verkerk, P.J., Churkina, G., Viovy, N., Hughes, J.K., Lindner, M. 2010. A comparison of alternative modeling approaches to evaluate the European forest carbon fluxes. Forest Ecology and Management. 260 241-251. Tvinnereim, E. 2010. The Global Carbon Market in 2020 [online]. Available from http://www.worldcommercereview.com/publications/article_pdf/109 [accessed 12 November 2011]. Tvinnereim, E., Zelljadt, E., Yakymenko, N., and Mazzacurati, E. 2011. Carbon 2011. In Point Carbon’s 7th annual conference, Carbon Market Insights 2011 in Amsterdam, 1-3 March 2011. 79  UK Forestry Commission, The CARBINE carbon accounting model [online]. Available from http://www.forestry.gov.uk/fr/INFD-633DXB [accessed 11 November 2011]. Ung C.H., Bernier, P., and Guo, X.J. 2008. Canadian national biomass equations: new parameter estimates that include British Columbia data. Can. J. For. Res. 38:1123-1132 United Nations Framework Convention on Climate Change (UNFCCC). 2006. Procedures to demonstrate the eligibility of lands for afforestation and reforestation project activities (Version 02) [online]. Executive Board Report (EB 26, Report, Annex 18). Available at: http://cdm.unfccc.int/EB/026/eb26_repan18.pdf [accessed 10 September 2010]. United Nations. 2006. Procedures to demonstrate the eligibility of lands for afforestation and reforestation project activities [online]. United Nations Framework Convention on Climate Change (UNFCCC). Available from http://cdm.unfccc.int/EB/026/ eb26_repan18.pdf [accessed 10 September 2010]. United States Department of Agriculture, Forest Service, Carbon, Tools for carbon inventory, management, and reporting [online]. Available from http://nrs.fs.fed.us/carbon/tools/ [accessed 11 November 2011]. United States Department of Agriculture, Forest Service, Publications and Data, FORCARB2: An updated version of the U.S. Forest Carbon Budget Model [online]. Available from http://nrs.fs.fed.us/pubs/35613. [Accessed 11 November 2011]. University of British Columbia, Faculty of Forestry, UBC Forest Ecosystem Simulation Research Group, FORECAST Ecosystem Model [online]. Available from http://www.forestry.ubc.ca/ecomodels/moddev/forecast/forecast.htm [accessed 11 November 2011]. USDA Forest Service. 2011. Forest Inventory and Analysis National Program - Forest Inventory and Analysis (FIA) National Program standards. Phase 2 Measurement Field Guide, Version 5.1 [online]. Available from http://www.fia.fs.fed.us/library/field-guides- methods-proc/docs/Complete%20FG%20Document/core_ver_5-1_10_2011.pdf [accessed 19 December 2011]. van der Werf, G.R., Morton, D.C., DeFries, R.S., Olivier, J.G.J, Kasibhatla, P.S., Jackson, R.B., Collatz, G.J., and Randerson, J.T. 2009. CO2 emissions from forest loss. Nature Geoscience. 2:737-738 (http://biology.duke.edu/jackson/ng09.pdf). Van Deusen, P.C. 2010. A method to evaluate the option of storing carbon in your forest. Can. J. For. Res. 40: 2243–2247. Waage, S., and Hamilton, K. 2011. Investing in forest carbon: lessons from the first 20 years. Available from The Katoomba Group, Ecosystem Marketplace and Forest Trends [online]. Available from http://www.forest-trends.org/documents/files/doc_2677.pdf [accessed 15 November 2011].  80  Walter, M. and Kahlert, G. 2010. Forest Carbon Standards – A WWF Assessment Guide. WWF Germany [online]. Available from http://wwf.panda.org/what_we_do/footprint/climate_carbon _energy/forest_climate/publications/?193463/WWFs-Review-of-Forest-Carbon- Standards [accessed 15 November 2011].  Watson, R.T., Meira Filho, L.G., Sanhueza, and E., Janetos, A. 1992. Greenhouse Gases: Sources and Sinks. WMO/UNEP Science Assessment of Ozone Depletion: 1991 [online]. Available from http://www.ipcc.ch/ipccreports/1992%20IPCC%20Supplement/IPCC_Suppl_Report_199 2_wg_I/ipcc_wg_I_1992_suppl_report_section_a1.pdf [accessed 14 October 2011].  Watts, S. B. and L. Tolland (Editors). 2005. Forestry Handbook for British Columbia Fifth Edition 773 pages Forestry Undergraduate Society University of British Columbia  Willey, Z., and Chameides, B. 2007. Harnessing farms and forests in the low-carbon economy – how to create, measure, and verify greenhouse gas offsets. The Nicholas Institute for Environmental Policy Solutions. Duke University Press, Durham & London.  Wilson, S.J. and Hebda, R.J. 2008. Mitigating and adapting to climate change through the conservation of nature. The Land Trust Alliance of British Columbia [online]. Available from http://www.treesintrust.com/external/LTA_ClimateChange.pdf [accessed 15 November 2011].  Zhaodi, G., Fang, J., Pan, Y., and Birdsey, R. 2010. Inventory-based estimates of forest biomass carbon stocks in China: A comparison of three methods. Forest Ecology and Management. 259 1225-1231. Zheng, G., Chen, J.M., Tian, Q.J., Ju, W.M., and Xia, X.Q. 2007. Combining remote sensing imagery and forest age inventory for biomass mapping. Journal of Environmental Management. 85 616–623. 81  Appendices  Appendix 1: Volume to Biomass Calculations (Harmon, 2001) Living vegetation carbon stores are modeled using a Chapman-Richards function:  Lt= Lmax (1-exp [B1*t] )B2  where Lt is the live vegetation carbon store as time t, Lmax is the maximum carbon store of live vegetation possible (a function of site productivity), B1 is a parameter that determines the time required to approach the maximum store, and B2 is a parameter that determines the lag in vegetation growth after disturbance (Cooper 1983). Whenever a disturbance occurs time is reset to 0 so that live biomass will decrease to 0 as well.  Detritus is simulated as one pool with an average decomposition rate, a loss from fire, and inputs from normal mortality as well as that associated with major disturbances:  Dt= D t-1 + Mt + DMt- Kt - Ft -SFt  Where Dt is the store of carbon in detritus at time t, D t-1 is the same but for the previous year, Mt is the input from mortality associated with competition and minor disturbances, DMt is the mortality associated with major disturbances (timber harvest, fire, wind), Kt is the loss from decomposition in year t, Ft is the loss from fire in year t, and SFt is the loss to soil formation in year t. The first type of mortality inputs are calculated as:  Mt= m* Lt  where m is the mortality rate-constant. The mortality inputs associated with major disturbances are calculated as:  DMt= (1-h) * Lt 82   where h is the fraction of live vegetation that is removed by harvest. Losses from decomposition are calculated as:  Kt=k* D t-1  where k is the decomposition rate-constant (the average for all types of detritus). Losses from fire are calculated as:  Ft=f* D t-1  where f is the fraction of detritus that is removed by fire. The loss of detritus to soil formation is:  SFt = sf * D t-1  where sf is the soil formation rate. Soil carbon stores (St) are controlled by inputs from detritus and losses via decomposition: St = St-1 + SFt – KSt  where St-1 is the soil store the year before t and KSt is the decomposition loss from soil as determined by:  KSt = ks * St-1  where ks is the rate-constant describing soil carbon decomposition. Forest products are input from living vegetation periodically by harvest. The amount of harvest that ends up in forest products after manufacturing is variable, as is the longevity of the products themselves. Only one aggregated pool of forest products is considered:  Pt=Pt-1 + MFt - PKt  83  where Pt is the forest products store as year t, Pt-1 is the same for the previous year MFt is the input from manufacturing and PKt is the loss from decomposition, incineration, and other mechanisms that release forest products to the atmosphere. The input from manufacturing is computed as a fraction of the harvested carbon:  MFt = mf * h * Lt  where mf is the manufacturing efficiency expressed as a fraction of the harvest turned into long-term forest products, h is the fraction of live carbon harvested, and Lt is the live carbon store the year of the harvest. The loss of products from decomposition, incineration, etc. is calculated as:  PKt = pk * Pt  where pk is the rate-constant for loss of forest products. The total carbon stores at time t are calculated as:  Tt = Lt + Dt + St + Pt  where the stores are defined as above. The flux in carbon stores (or net change) is calculated as:  ΔTt=Tt – Tt-1  where Tt is the total store at time t and Tt-1 is the total carbon store the year before t. 84  Appendix 2: Forest Carbon Plot Summary of Results  Plot    Date Collected Live Trees (>10cm diameter) Live Trees (<10cm diameter) Dead and Down Trees Standing Dead Trees Stumps Fine Woody Debris Forest Floor Flux Total Plot 1 25-Sep-09 332.649 0 25.81853 13.68792 0 5.560927 257.6 -46.6202 635.32 Plot 2 2-Oct-09 985.3321 6.081291 15.3457 0 31.82343 8.328382 13.95333 4.111167 1054.78 Plot 3 15-Oct-09 129.9294 0 93.46582 2.513293 15.84595 8.872246 65.205 -15.2284 318.46 Plot 4 29-Oct-09 152.7368 0 89.69201 1.250614 6.501662 7.022141 24.955 -3.36965 282.16 Plot 5 26-Nov-09 155.405 0 68.28386 0 70.82311 9.617062 45.4825 -10.493 349.61 Plot 6 2-Dec-09 335.5046 0 119.1637 0 0.902827 10.81695 38.64 -2.9573 505.03 Plot 7 3-Dec-09 196.3669 0 29.9487 0 4.935687 2.512191 11.6725 1.664424 245.44 Plot 8 8-Dec-09 130.8552 0 13.43212 0 26.09317 20.27486 70.035 -9.38362 260.69 Plot 9 9-Dec-09 188.2094 0 45.69086 0 2.335814 9.46335 76.475 -12.9767 322.17 Plot 10 15-Dec-09 318.0564 0 16.70559 0 0.910243 4.362297 77.28 -12.1217 417.31 Plot 11 17-Dec-09 349.3146 0 107.0906 0 0.531441 16.48537 71.91333 -17.3873 545.34 Plot 12 29-Jan-10 168.9371 4.85 47.61997 0.70664 3.632685 15.84703 83.72 -17.4028 325.32 Plot 13 5-Feb-10 93.14711 0 55.53893 0 15.61415 13.77247 34.96 -7.00678 213.03 Plot 14 19-Feb-10 312.5644 n/a 8.864595 1.090222 1.630521 4.841115 108.4067 -19.4935 437.40 Plot 15 5-Mar-10 181.0834 n/a 33.01812 0.200444 3.515264 4.802718 48.3 -8.20126 270.92 Plot 16 29-Apr-10 265.9977 0 32.78976 0 0.75221 6.256735 41.055 -6.45033 346.85 Plot 17 6-May-10 211.7439 0 35.9551 0 0 8.127554 69.23 -13.3643 325.06 Plot 18 13-May-10 98.5905 0 6.375456 0 7.33824 3.916062 21.735 0.756484 137.96 Plot 19 20-May-10 261.6348 0 13.09169 0 1.78152 2.592942 48.3 -7.44945 327.40 85  Appendix 3: Canadian Forest Service Carbon Budget Model Summary of Results  Plot Foliage(SW) + Foliage(HW) Other(SW) + Other(HW) Merch(SW )+ Merch(HW) Biomass Aboveground Soil Carbon Biomass   DOM Total Ecosystem Total Ecosystem (per/ha) 1 38 186 346 570 2395 685 2826 3512 642.06 2 271 1334 2439 4044 6915 4941 8578 13518 605.37 3 12 87 75 174 753 212 1036 1249 538.83 4 50 274 285 609 1768 744 2377 3121 157.75 5 285 1560 1179 3023 10181 3693 13843 17536 687.04 6 514 2811 2520 5844 18739 7139 25236 32375 600.95 7 171 929 969 2069 6005 2527 8070 10597 535.63 8 682 3734 3083 7498 25424 9160 34632 43792 344.10 9 272 1416 2009 3697 6995 4517 8571 13087 379.50 10 85 370 993 1448 3251 1769 4328 6097 949.07 11 246 1208 2220 3674 6430 4488 7979 12467 616.03 12 94 448 932 1475 2393 1802 2989 4791 632.55 13 369 1880 698 2947 11504 3600 15589 19189 449.69 14 66 295 746 1107 1739 1352 2185 3537 696.04 15 360 1451 4575 6386 9897 7801 12338 20139 784.46 16 80 377 818 1275 3315 1521 5788 7309 670.59 17 58 286 552 896 888 1095 1581 2675 530.40 18 113 693 1377 2183 4165 2668 8131 10798 626.25 19 50 262 394 707 746 863 1322 2185 475.89               86   Appendix 4: VRI Biomass Equations Summary of Results   Plot Carbon (stem only CO2E kg per/ha)  Carbon (tree CO2E kg per/ha) Carbon (tree CO2E tonnes per/ha)  1 193445.64 386891.28 386.89128 2 n/a n/a n/a 3 93018.9 186037.8 186.0378 4 203294.7 406589.4 406.5894 5 137886.84 275773.68 275.77368 6 178040.7 356081.4 356.0814 7 203294.7 406589.4 406.5894 8 152029.08 304058.16 304.05816 9 n/a n/a n/a 10 526125 1052250 1052.25 11 360711.3 721422.6 721.4226 12 413407.98 826815.96 826.81596 13 59599.44 119198.88 119.19888 14 506763.6 1013527.2 1013.5272 15 613419.66 1226839.3 1226.8393 16 243953.64 487907.28 487.90728 17 338571.96 677143.92 677.14392 18 185111.82 370223.64 370.22364 19 259863.66 519727.32 519.72732  87   Appendix 5: Private Woodland Planner Summary of Results   Plot   Total abovegroundaboveground Biomass (unmanaged)  AbovegroundAboveground biomass (per hectare) Total flux (unmanaged)  Total flux (per hectare)  1 1507 275.5078 5 0.914093 2 10513 470.7974 11 0.492606 3 256 110.44 22 9.49094 4 4415 223.1601 214 10.81682 5 4191 164.199 271 10.6175 6 11367 210.9966 609 11.30438 7 4415 223.1601 214 10.81682 8 23010 180.8053 1439 11.3072 9 15905 461.219 14 0.405977 10 3979 619.3767 76 11.83027 11 11625 574.4201 8 0.3953 12 3797 501.3137 4 0.528116 13 3016 70.67949 406 9.514547 14 2993 588.9877 2 0.393577 15 19705 767.5498 46 1.791794 16 3523 323.2288 61 5.596638 17 2191 434.4292 7 1.387953 18 4691 272.0619 174 10.0914 19 1676 365.0303 6 1.306791          88  Appendix 6: Summary of Control and Treatment Results  (G = Control, M1 = CBM-CFS3, M2 = VRI BIOMASS EQUATIONS, M3 = PWP) Plot  Strata  G  M1  M2  M3  3 1-1 318.458 642.0593 386.8913 275.5078 5 1-1 349.6116 605.3685  470.7974 7 1-2 245.4359 538.8266 186.0378 110.44 10 1-2 417.3145 157.7537 406.5894 223.1601 16 1-2 346.8514 687.0423 275.7737 164.199 18 1-2 137.9553 600.9515 356.0814 210.9966 1 1-2  635.3163 535.6349 406.5894 223.1601 9 1-3 322.1744 344.1036 304.0582 180.8053 15 1-3 270.92 379.5016  461.219 2 1-4 1054.783 949.0676 1052.25 619.3767 6 1-4 505.0281 616.0255 721.4226 574.4201 11 1-4 545.3354 632.5504 826.816 501.3137 13 2-1 213.0327 449.6912 119.1989 70.67949 4 2-2 282.1582 696.0406 1013.527 588.9877 8 2-2 260.6904 784.455 1226.839 767.5498 17 2-2 325.0566 670.5874 487.9073 323.2288 19 2-3 327.401 530.3962 677.1439 434.4292 12 2-4 325.3177 626.2469 370.2236 272.0619 14 2-4 437.3975 475.8897 519.7273 365.0303  89   Appendix 7: Differences Between Control and Treatment Estimates by Plot   (G = Control, M1 = CBM-CFS3, M2 = VRI BIOMASS EQUATIONS, M3 = PWP)  Plot Strata C T1 C - T1 T2 C - T2 T3 C - T3 3 1-1 318.458 538.8265746 220.3685746 186.0378 -132.4202 110.4400345 -208.0179655 5 1-1 349.6116 687.0423407 337.4307407 275.77368 -73.83792 164.1990448 -185.4125552 7 1-2 245.4359 535.6348564 290.1989564 406.5894 161.1535 223.1601294 -22.2757706 10 1-2 417.3145 949.0675882 531.7530882 1052.25 634.9355 619.3767317 202.0622317 16 1-2 346.8514 670.5873718 323.7359718 487.90728 141.05588 323.2288016 -23.62259841 18 1-2 137.9553 626.2469262 488.2916262 370.22364 232.26834 272.0618939 134.1065939 1 1-2  635.3163 642.0592698 6.742969822 386.89128 -248.42502 275.5077789 -359.8085211 9 1-3 322.1744 379.5016341 57.32723406   461.2190334 139.0446334 15 1-3 270.92 784.4550221 513.5350221 1226.83932 955.91932 767.5498391 496.6298391 2 1-4 1054.783 605.3685144 -449.4144856   470.7973955 -583.9856045 6 1-4 505.0281 600.9514988 95.92339881 356.0814 -148.9467 210.9966235 -294.0314765 11 1-4 545.3354 616.0254573 70.69005731 721.4226 176.0872 574.4201445 29.08474448 13 2-1 213.0327 449.6912459 236.6585459 119.19888 -93.83382 70.67949334 -142.3532067 4 2-2 282.1582 157.7537404 -124.4044596 406.5894 124.4312 223.1601294 -58.9980706 8 2-2 260.6904 344.1035957 83.41319568 304.05816 43.36776 180.8052552 -79.88514478 17 2-2 325.0566 530.3961613 205.3395613 677.14392 352.08732 434.4291549 109.3725549 19 2-3 327.401 475.8897068 148.4887068 519.72732 192.32632 365.030274 37.62927399 12 2-4 325.3177 632.550402 307.232702 826.81596 501.49826 501.3136874 175.9959874 14 2-4 437.3975 696.0406171 258.6431171 1013.5272 576.1297 588.9877204 151.5902204  90   Appendix 8: Closest Estimates to the Control by Plot  (G = Control, M1 = CBM-CFS3, M2 = VRI Biomass Equations, M3 = PWP)  Plot Strata C T1 T2 T3 3 1-1 318.458 538.8266 186.0378 110.44 5 1-1 349.6116 687.0423 275.7737 164.199 7 1-2 245.4359 535.6349 406.5894 223.1601 10 1-2 417.3145 949.0676 1052.25 619.3767 16 1-2 346.8514 670.5874 487.9073 323.2288 18 1-2 137.9553 626.2469 370.2236 272.0619 1 1-2  635.3163 642.0593 386.8913 275.5078 9 1-3 322.1744 379.5016  461.219 15 1-3 270.92 784.455 1226.839 767.5498 2 1-4 1054.783 605.3685  470.7974 6 1-4 505.0281 600.9515 356.0814 210.9966 11 1-4 545.3354 616.0255 721.4226 574.4201 13 2-1 213.0327 449.6912 119.1989 70.67949 4 2-2 282.1582 157.7537 406.5894 223.1601 8 2-2 260.6904 344.1036 304.0582 180.8053 17 2-2 325.0566 530.3962 677.1439 434.4292 19 2-3 327.401 475.8897 519.7273 365.0303 12 2-4 325.3177 632.5504 826.816 501.3137 14 2-4 437.3975 696.0406 1013.527 588.9877  91   Appendix 9: Differences Between Control and Treatments by Strata   (G = Control, M1 = CBM-CFS3, M2 = VRI BIOMASS EQUATIONS, M3 = PWP)  Strata  G-AVG  M1-AVG  M1-G  M2-AVG  M2-G  M3-AVG  M3-G  1-1 334.0348 623.7139 289.6791 386.8913 52.8565 373.1526 39.11779 1-2 356.5747 504.0418 147.4671 326.2143 -30.3603 186.3912 -170.183 1-3 296.5472 361.8026 65.25541 304.0582 7.511 321.0121 24.46494 1-4 701.7155 732.5478 30.83232 866.8295 165.114 565.0369 -136.679 2-1 213.0327 449.6912 236.6585 119.1989 -93.8338 70.67949 -142.353 2-2 289.3017 717.0277 427.7259 909.4246 620.1229 559.9221 270.6204 2-3 327.401 530.3962 202.9952 677.1439 349.7429 434.4292 107.0282 2-4 381.3576 551.0683 169.7107 444.9755 63.61788 318.5461 -62.8115 Total   1570.324  1134.771  -70.7955 

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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072837/manifest

Comment

Related Items