"Forestry, Faculty of"@en . "DSpace"@en . "UBCV"@en . "Asadian, Yeganeh"@en . "2010-04-14T21:36:15Z"@en . "2010"@en . "Master of Science - MSc"@en . "University of British Columbia"@en . "Vegetation interception loss plays an important role in controlling the water balance of a watershed, especially where urban development has taken place. The aim of this study is to document the importance of urban trees as a form of \u00E2\u0080\u0098green infrastructure\u00E2\u0080\u0099 to reduce stormwater runoff and rainwater intensity, and cause a delay in precipitation reaching the ground. A 21 months study was carried out in the North/West Vancouver in British Columbia to determine how effective urban trees are to intercept and detain rainwater. We applied a unique methodology for measuring rain/throughfall under 54 different urban trees using a system of PVC pipes hung beneath the canopy to capture the throughfall where it drained into a rain gauge attached to a data logger. To ensure that the study adequately captured the range of throughfall variability, trees were selected to sample different landscape sites (streets, parks, and natural forested areas), elevations, tree type, health condition and species, including Douglas-fir (Pseudotsuga menziesii), Western red cedar (Thuja plicata), Bigleaf maple (Acer macrophyllum), Oak (Quercus sp.), Copper beech (Fagus sylvatica), Horse chestnut (Aesculus hippocastanum), Cherry (Prunus sp.), and Poplar (Populus sp.). Interception loss and throughfall were monitored from February 2007 until November 2008. Rainfall interception varied seasonally for all species. Interception losses accounted for on average 76.5% and 56.4% of gross precipitation for coniferous and deciduous trees, respectively. The interception loss varied depending on canopy structure, climatic conditions, and rainfall characteristics. The results showed that urban trees intercept and evapotranspire more rain than trees in forested environments. Together with the delay in runoff trees can act as an effective stormwater management tool on individual properties."@en . "https://circle.library.ubc.ca/rest/handle/2429/23517?expand=metadata"@en . "RAINFALL INTERCEPTION IN AN URBAN ENVIRONMENT by YEGANEH ASADIAN B.Sc., The University of British Columbia, 2007 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) April 2010 \u00C2\u00A9 Yeganeh Asadian, 2010 ii Abstract Vegetation interception loss plays an important role in controlling the water balance of a watershed, especially where urban development has taken place. The aim of this study is to document the importance of urban trees as a form of \u00E2\u0080\u0098green infrastructure\u00E2\u0080\u0099 to reduce stormwater runoff and rainwater intensity, and cause a delay in precipitation reaching the ground. A 21 months study was carried out in the North/West Vancouver in British Columbia to determine how effective urban trees are to intercept and detain rainwater. We applied a unique methodology for measuring rain/throughfall under 54 different urban trees using a system of PVC pipes hung beneath the canopy to capture the throughfall where it drained into a rain gauge attached to a data logger. To ensure that the study adequately captured the range of throughfall variability, trees were selected to sample different landscape sites (streets, parks, and natural forested areas), elevations, tree type, health condition and species, including Douglas-fir (Pseudotsuga menziesii), Western red cedar (Thuja plicata), Bigleaf maple (Acer macrophyllum), Oak (Quercus sp.), Copper beech (Fagus sylvatica), Horse chestnut (Aesculus hippocastanum), Cherry (Prunus sp.), and Poplar (Populus sp.). Interception loss and throughfall were monitored from February 2007 until November 2008. Rainfall interception varied seasonally for all species. Interception losses accounted for on average 76.5% and 56.4% of gross precipitation for coniferous and deciduous trees, respectively. The interception loss varied depending on canopy structure, climatic conditions, and rainfall characteristics. The results showed that urban trees intercept and evapotranspire more rain than trees in forested environments. Together with the delay in runoff trees can act as an effective stormwater management tool on individual properties. iii Table of Contents Abstract ......................................................................................................................................... ii\u00E2\u0080\u00A9 Table of Contents......................................................................................................................... iii\u00E2\u0080\u00A9 List of Figures ................................................................................................................................v\u00E2\u0080\u00A9 List of Tables .............................................................................................................................. viii\u00E2\u0080\u00A9 Acknowledgements ...................................................................................................................... ix\u00E2\u0080\u00A9 Chapter 1 Introduction .................................................................................................................1\u00E2\u0080\u00A9 1.1 Interception Process ...............................................................................................................4\u00E2\u0080\u00A9 1.1.1 The Terminology of Canopy Interception ..................................................................................... 4\u00E2\u0080\u00A9 1.1.2 Controls on Interception Loss........................................................................................................ 5\u00E2\u0080\u00A9 1.2 Methods for Measuring Rainfall Interception........................................................................7\u00E2\u0080\u00A9 1.2.1 Rainfall Interception Models ......................................................................................................... 9\u00E2\u0080\u00A9 1.3 Objectives.............................................................................................................................12\u00E2\u0080\u00A9 1.4 Thesis Outline ......................................................................................................................13\u00E2\u0080\u00A9 Chapter 2 Study Area and Methods ..........................................................................................14\u00E2\u0080\u00A9 2.1 Site Description....................................................................................................................14\u00E2\u0080\u00A9 2.2 Throughfall Measurements ..................................................................................................16\u00E2\u0080\u00A9 2.2.1 Tree Selection .............................................................................................................................. 16\u00E2\u0080\u00A9 2.2.2 Tree Classification ....................................................................................................................... 21\u00E2\u0080\u00A9 2.2.3 Field Instrumentation ................................................................................................................... 25\u00E2\u0080\u00A9 2.2.4 Meteorological Station................................................................................................................. 26\u00E2\u0080\u00A9 2.2.5 Data Collection and Calibration................................................................................................... 27\u00E2\u0080\u00A9 2.2.6 Testing the Efficiency of the Throughfall Gauges....................................................................... 28\u00E2\u0080\u00A9 2.3 Methods for Calculation of Interception Loss .....................................................................30\u00E2\u0080\u00A9 iv 2.3.1 Canopy Parameters ...................................................................................................................... 31\u00E2\u0080\u00A9 2.4 Statistical Analysis ...............................................................................................................33\u00E2\u0080\u00A9 Chapter 3 Results.........................................................................................................................34\u00E2\u0080\u00A9 3.1 Overview of Study Period....................................................................................................34\u00E2\u0080\u00A9 3.1.1 Rainfall Event Summary.............................................................................................................. 35\u00E2\u0080\u00A9 3.2 Relation between Event Throughfall and Gross Precipitation.............................................39\u00E2\u0080\u00A9 3.2.1 Event Throughfall Patterns .......................................................................................................... 40\u00E2\u0080\u00A9 3.2.2 Interception Loss.......................................................................................................................... 42\u00E2\u0080\u00A9 3.2.3 Urban Trees vs. Control in Forest................................................................................................ 43\u00E2\u0080\u00A9 3.2.4 Climate and Precipitation Variability During an Event ............................................................... 44\u00E2\u0080\u00A9 3.3 Variables Influencing Interception Loss ..............................................................................53\u00E2\u0080\u00A9 3.3.1 Throughfall \u00E2\u0080\u0093 Assessed Attributes Relationships ........................................................................ 59\u00E2\u0080\u00A9 3.4 Efficiency of Throughfall Gauges........................................................................................63\u00E2\u0080\u00A9 Chapter 4 Discussion ...................................................................................................................66\u00E2\u0080\u00A9 4.1 Canopy Interception Loss ....................................................................................................66\u00E2\u0080\u00A9 4.1.1 Event Based Analysis .................................................................................................................. 67\u00E2\u0080\u00A9 4.1.2 Controls on Interception Loss...................................................................................................... 71\u00E2\u0080\u00A9 4.1.3 Outcome of Throughfall Gauges ................................................................................................. 72\u00E2\u0080\u00A9 Chapter 5 Conclusion..................................................................................................................74\u00E2\u0080\u00A9 5.1 Review of Key Findings ......................................................................................................74\u00E2\u0080\u00A9 5.2 Future Research....................................................................................................................76\u00E2\u0080\u00A9 References.....................................................................................................................................77\u00E2\u0080\u00A9 Appendix A...................................................................................................................................83\u00E2\u0080\u00A9 v List of Figures Figure 2.1: Study Sites in Northlands Golf course and Roche Point Park in City of North Vancouver with latitude and longitude equal to 49\u00C2\u00BA 19' 02'' N and 122\u00C2\u00BA 58' 01'' W respectively (Ortho-photos provided by District of North Vancouver\u00E2\u0080\u0099s GIS Department).............................. 17\u00E2\u0080\u00A9 Figure 2.2: Study Sites in Mahon Park, Mosquito Park, and the District Hall Area, in District of North Vancouver with latitude and longitude equal to 49\u00C2\u00BA 20' 00'' N and 123\u00C2\u00BA 04' 00'' W respectively (Ortho-photos provided by District of North Vancouver\u00E2\u0080\u0099s GIS Department). ........ 18\u00E2\u0080\u00A9 Figure 2.3: Study Sites at Caulfield Area in District of West Vancouver with latitude and longitude equal to 49\u00C2\u00BA 21' 00'' N and 123\u00C2\u00BA 15' 02'' W respectively (Ortho-photos provided by District of West Vancouver\u00E2\u0080\u0099s GIS Department)........................................................................... 19\u00E2\u0080\u00A9 Figure 2.4: Study Sites in Lighthouse Park in District of West Vancouver with latitude and longitude equal to 49\u00C2\u00BA 20' 02'' N and 123\u00C2\u00BA 16' 00'' W respectively (Ortho-photos provided by District of North Vancouver\u00E2\u0080\u0099s GIS Department).......................................................................... 20\u00E2\u0080\u00A9 Figure 2.5: Locations along the longest diameter where hemispherical photographs were taken 22\u00E2\u0080\u00A9 Figure 2.6: The throughfall gauges measuring system................................................................. 26\u00E2\u0080\u00A9 Figure 2.7: Example of reference rainfall gauge installed on the rooftop of North Vancouver\u00E2\u0080\u0099s City Hall to measure gross precipitation (above canopy rainfall). ............................................... 27\u00E2\u0080\u00A9 Figure 2.8: Random distribution of point throughfall collectors under a Douglas-fir (Tag# 6) in Lighthouse Park ............................................................................................................................ 29\u00E2\u0080\u00A9 Figure 3.1: Vancouver\u00E2\u0080\u0099s Climate Data for 2007 and 2008 (Location of rain gauge: Latitude 49.33618\u00C2\u00BA N and Longitude 123.07814 \u00C2\u00BA W)................................................................................ 34\u00E2\u0080\u00A9 Figure 3.2: Rainfall event histograms of (a) gross precipitation; (b) average rainfall intensity; (c) event duration ............................................................................................................................... 37\u00E2\u0080\u00A9 Figure 3.3: Summer season histograms of events used in the study:(a) gross precipitation; (b) average rainfall intensity; (c) event duration ................................................................................ 38\u00E2\u0080\u00A9 Figure 3.4: Winter season histograms of events used in the study: (a) gross precipitation; (b) average rainfall intensity; (c) event duration ................................................................................ 38\u00E2\u0080\u00A9 Figure 3.5: Throughfall as a function of gross precipitation (season: summer) for (a) coniferous and (b) deciduous trees. ................................................................................................................ 39\u00E2\u0080\u00A9 vi Figure 3.6: Throughfall as a function of gross precipitation (season: winter) for (a) coniferous and (b) deciduous trees. ................................................................................................................ 39\u00E2\u0080\u00A9 Figure 3.7: Summer event throughfall and average event temperature (a) coniferous; and (b) deciduous trees ............................................................................................................................. 41\u00E2\u0080\u00A9 Figure 3.8: Winter event throughfall and average event temperature (a) coniferous; and (b) deciduous trees ............................................................................................................................. 41\u00E2\u0080\u00A9 Figure 3.9: Box plot of percentage interception loss for different species in summer. The outliers are presented by black circles. ...................................................................................................... 42\u00E2\u0080\u00A9 Figure 3.10: Box plot of percentage interception loss for different species in winter.................. 43\u00E2\u0080\u00A9 Figure 3.11: Meteorological and throughfall data for rainfall event two (Douglas-fir) ............... 48\u00E2\u0080\u00A9 Figure 3.12: Meteorological and throughfall data for rainfall event two (Western red cedar) .... 49\u00E2\u0080\u00A9 Figure 3.13: Meteorological and throughfall data for rainfall event two (Deciduous trees)........ 50\u00E2\u0080\u00A9 Figure 3.14: Relationship between throughfall and gross precipitation; and residual plot for a single standing Western red cedar in a good condition (tag # 591) during summer .................... 54\u00E2\u0080\u00A9 Figure 3.15: Relationship between throughfall and gross precipitation; and residual plot for a single standing Western red cedar in a good condition (tag # 591) during winter....................... 54\u00E2\u0080\u00A9 Figure 3.16: Relationship between throughfall and gross precipitation; and residual plot for a single standing Bigleaf maple in a good condition (tag # 597) during summer........................... 55\u00E2\u0080\u00A9 Figure 3.17: Relationship between throughfall and gross precipitation; and residual plot for a single standing Bigleaf maple in a good condition (tag # 597) during winter ............................. 55\u00E2\u0080\u00A9 Figure 3.18: Box plot of slope coefficients for different species (summer). The outliers are presented by black circles............................................................................................................. 57\u00E2\u0080\u00A9 Figure 3.19: Box plot of slope coefficients for different species (winter).................................... 57\u00E2\u0080\u00A9 Figure 3.20: Box plot of R2 coefficients for different species (summer). The outliers are presented by black circles............................................................................................................. 58\u00E2\u0080\u00A9 Figure 3.21: Box plot of R2 coefficients for different species (winter). The outliers are presented by black circles. ............................................................................................................................ 58\u00E2\u0080\u00A9 vii Figure 3.22: The relationship between throughfall measured by innovative throughfall gauges and bottles for 5 selected trees (6 events)..................................................................................... 65\u00E2\u0080\u00A9 viii List of Tables Table 1.1: The original and revised Rutter & Gash model........................................................... 11\u00E2\u0080\u00A9 Table 2.1: Characteristics of the selected trees including their location and tag numbers........... 23\u00E2\u0080\u00A9 Table 3.1: Rainfall events classification summary ....................................................................... 35\u00E2\u0080\u00A9 Table 3.2: Percentage of average interception loss during summer ............................................. 43\u00E2\u0080\u00A9 Table 3.3: Percentage of average interception loss during winter................................................ 43\u00E2\u0080\u00A9 Table 3.4: Rainfall depth from two other nearby stations ............................................................ 44\u00E2\u0080\u00A9 Table 3.5: Event characteristics.................................................................................................... 45\u00E2\u0080\u00A9 Table 3.6: Lag time between gross precipitation and throughfall ................................................ 47\u00E2\u0080\u00A9 Table 3.7: Percentage of interception for the selected rainfall events.......................................... 52\u00E2\u0080\u00A9 Table 3.8: Seasonal regression (TF vs. Pg) coefficients for different tree species, types and health conditions...................................................................................................................................... 60\u00E2\u0080\u00A9 Table 3.9: Summary of results for ANOVA analysis................................................................... 62\u00E2\u0080\u00A9 Table 3.10: Seasonal storage capacity (mm) and canopy cover (c) estimations .......................... 63\u00E2\u0080\u00A9 Table 3.11: Regression coefficients from throughfall gauges and bottles for the selected trees (n = 6 events) .................................................................................................................................... 65\u00E2\u0080\u00A9 ix Acknowledgements I would like to express my gratitude to my supervisors Dr. Dan Moore and Dr. Markus Weiler for all their support throughout this project, for keeping me focused, for listening to my concerns, and for providing me with advice that enabled the production of this thesis. Sincere appreciation is extended to Dr. Hans Schreier for his encouragement and useful suggestions throughout this research work. I am heartily thankful to Martin Tromp, Gerald Lohmann, Sophie Bachmair, Nils Ilchmann, Selina Agbayani, Maheen Javed and Eoin O\u00E2\u0080\u0099Neill for their assistant in the field and invaluable advice on fieldwork and data analysis. I also offer my enduring gratitude to Richard Boase and Kim Stephens for their unconditional support, enthusiasm and inspiration whenever I needed the most. I would also like to thank all those responsible for providing financial support. Thank you to District of North Vancouver, City of North Vancouver, District of West Vancouver, Greater Vancouver Regional District, Province of British Columbia, Real State Foundation of British Columbia, Inter-Governmental Water Balance Model Partnership, and Canadian Water Network. I also like to express my appreciation to the North Shore Mentally Handicapped Association for their assistance in assembling the structures for the throughfall gauges. I am as ever, especially indebted to my mom (Soheila), grand parents (Jafar and Pooran), Yalda and Mohammad for their love and support throughout my life. This milestone could not have been possible without the support and sense of humor of my friends. 1 Chapter 1 Introduction Water quality concerns have intensified and stormwater management practices have come under scrutiny as development occurs on an increasing percentage of the available land area. Urbanization has resulted in profound changes to natural watershed conditions by altering terrain, vegetation, soil characteristics, and surface conditions. Urban development impacts climatic conditions and alters the hydrological processes in urban watersheds (McPherson et al. 1997; Sanders 1986). The losses in vegetation cover and increase in impervious surfaces, such as paved roads, sidewalks, and concrete buildings, increases the total amount of runoff, the flashiness of runoff event, flooding, erosion, and the cost of stormwater management. Villarreal and Bengtsson (2004) noted that stormwater runoff prior to development was regulated by trees, vegetation and natural soils where they function together to reduce stormwater runoff. Trees reduce stormwater runoff by canopy interception loss, which is the proportion of incident precipitation that is intercepted, stored and subsequently evaporated from the leaves, branches and stems of vegetation. Interception provides two major roles in a watershed. First, interception is an important part of the water balance, serving as either a loss or gain of water to the watershed. Second, interception plays an important role in protecting the mineral soil surface from the energy of rainfall. Reduction of raindrop energy by interception minimizes soil detachment and reduces subsequent erosion as well as protection of soil structure and infiltration capacity. Interception can be impacted by management which affects the amount, type, and 2 distribution of vegetation in a watershed. Another portion of the same rainfall event infiltrates into the ground, decreasing peak flows and the total amount of runoff. Trees also slow storm flow event by reducing the volume of water that must be managed at any one time and also reduce the rainfall intensity. Trees are generally overlooked in urban planning, but they are an essential component of the urban infrastructure, capable of controlling the hydrological processes, regulating air and water quality, reducing Urban Heat Islands (UHI) and absorbing CO2 (Sanders 1986; Taha 1997). Stormwater managers have started to use trees as a tool to help reduce stormwater generation and in this way reduce the cost of constructing traditional stormwater control infrastructure. The value of trees for stormwater management has been calculated based on the avoided costs of handling stormwater runoff (McPherson et al. 1997; Villarreal et al. 2004; Zipperer et al. 1997). McPherson et al. (2005) reported that in some cities in the USA the urban tree investment can be between $13 \u00E2\u0080\u0093 $65 per tree annually in planting and maintenance cost. In return, gains in stormwater services are between $1.37 \u00E2\u0080\u0093 $3.09 per dollar invested that would have been spent otherwise toward traditional stormwater management. Another study has estimated the worth of the USA\u00E2\u0080\u0099s urban forests as $400 billion in terms of stormwater management mitigation alone (American Forests 1996). These studies demonstrate the importance of trees as source controls capable of treating stormwater at the site level by reducing the runoff component within the hydrological cycle. 3 Relatively few investigations have focused on the effects of urban forests on rainfall interception and runoff reduction. Some studies have focused on conceptualizing the influence of urban forests on hydrology by using satellite remote sensing to estimate vegetation coverage. However, the resolutions were too low to distinguish between conifers and deciduous trees (American Forests 1996). Previous studies primarily were conducted in naturally forested areas (Carlyle- Moses 2004; Crockford 1990; Domingo et al. 1998; Link et al. 2004; Llorens et al. 2007; Llorens et al. 1997; Nadkarni et al. 2004). Intuitively, the next step is to conduct a study that provides a fundamental understanding of spatial and temporal throughfall at the single tree level in an effort to improve understanding of rainfall interception in urban settings. This thesis is an attempt to achieve this link by presenting results of an investigation of temporal and spatial changes in throughfall and rainfall interception using an innovative study design for urban trees on the North Shore (North and West Vancouver) in British Columbia. The target users who will benefit from the study of these processes are academics, city/municipal engineers, planners, decision makers, and developers involved in urban planning and management of stormwater. Sections 1.1 and 1.2 provide a focused literature review and identify knowledge gaps that this thesis will aim to fill through the objectives stated in Section 1.3. Section 1.1 focuses on the processes of canopy interception. Section 1.2 concentrates on the methods to measure rain interception evaporation process. It also deliberates on how interception loss by forest canopies can be estimated by analytical models. 4 1.1 Interception Process 1.1.1 The Terminology of Canopy Interception Tree canopy interception accounts for storing precipitation temporary in the canopy and releasing it slowly to the ground and back into the atmosphere. Interception by the forest canopy is defined as: Equation 1.1 where Ic is the canopy interception loss (mm); Pg is the gross precipitation, measured above canopy or in an open area (mm); TF is the throughfall, precipitation that passes through the canopy or as drip from vegetation (mm); and SF is the stemflow, water that flows down the stems to the ground surface. Net precipitation (Pn) is also commonly referred to as the quantity of rain water that actually reaches the ground. It is the sum of throughfall and stemflow. The division of a given quantity of rainfall into the above pathways is highly dependent on forest structure, canopy structure and meteorological factors (Crockford et al. 2000; Nadkarni et al. 2004; Xiao et al. 2002). A number of interception studies have been conducted in tropical (Jetten 1996), temperate broadleaf, and temperate conifer forest (Link et al. 2004; Rutter et al. 1975; Toba et al. 2005) to evaluate the net interception loss over a period of time, estimate the components of interception, and assess/improve interception models. Interception loss is commonly 29-40% of gross 5 precipitation in coniferous forests and between 10-20% in deciduous forests (Carlyle-Moses 2004; Crockford 1990; Link et al. 2004; Llorens et al. 2007; Llorens et al. 1997). 1.1.2 Controls on Interception Loss The rainfall interception capacity of the natural/urban forest is strongly influenced by forest structure, which includes species, dimensions, and understory (Keim et al. 2005; Xiao et al. 2002; Xiao et al. 2000b). The size and shape of the canopy, which can be described as tree architecture (e.g., foliation period, leaf and stem surface areas, gap fractions, and surface detention storage capacity), affect the amount, intensity, and spatial distribution of throughfall. Variation in these characteristics creates variation in interception. The structure of the overstory, mainly the branch and leaf angles, concentrates throughfall in drip points, which can result in greater amounts and intensities of rainfall at these particular places (Brooks et al. 2003; Bryant et al. 2005; Keim et al. 2005; Masukata et al. 1990; Xiao et al. 2002; Xiao et al. 2000b). Many other locations in a dense forest canopy may receive no throughfall at all. Relationships have been verified for the amount of throughfall captured and forest types in many parts of the world. Moreover, species variation has been distinguished to play an important role in influencing the result of interception loss. It has been suggested that conifers tend to have greater interception capacity than broadleaf species. It is noted that during a rainfall event, raindrops can run together on deciduous leaves, forming large raindrops which can fall as throughfall; however, the needles of conifers do not allow for this to take place (Ford et al. 1978; Iroum\u00C3\u00A9 et al. 2002; Keim et al. 2006; Lee et al. 2005; Lundberg et al. 1997; Toba et al. 2005). 6 Observations of interception loss indicate that the precipitation and associated meteorological characteristics as well as vegetation characteristics influence the interception process. In general, the precipitation and storm factors which affect interception loss are precipitation amount, intensity, duration, frequency of events, and antecedent rainfall conditions (Asdak et al. 1998; Brooks et al. 2003; Horton 1919). For instance, low intensity long duration frontal storms may generate different interception losses than high intensity, short duration convectional storms. Furthermore, the intercepted water is lost to evaporation during periods of rain, as well as during breaks within it, and the amount lost depends on meteorological conditions including temperature, humidity, and wind speed (Dunkerley 2000). In urban settings, field observations and experimental measurements of rainfall interception processes are sorely needed in order to better understand these processes. Urban tree interception processes are somewhat different from those reported for natural forests as a result of various factors such as edge effect, growing spaces (isolation: greater distances between individuals), open canopies, age, canopy structure, diversity, and microclimate (higher temperatures and wind penetration and associated rainfall) (Xiao et al. 2002). These characteristics define the storage capacity for each stand or individual tree, and control the evapotranspiration rate (Guevara- Escobar et al. 2007; McJannet et al. 2007a; McJannet et al. 2007b; Sanders 1986; Wang et al. 2008; Xiao et al. 2002; Zipperer et al. 1997). 7 1.2 Methods for Measuring Rainfall Interception The net precipitation method has been used in most interception studies to quantify interception loss, based on gross precipitation, throughfall, and stem flow (Calder 1996; Gash 1979; Horton 1919; Rutter et al. 1975). There is a wide variety of measurement techniques with associated accuracies conducted in both the field and laboratories. The two commonly used methods in the field are: point and area measurements (Aston A. R. 1979; Li et al. 1997; Lloyd et al. 1988). Point measurements involve collecting throughfall directly beneath the canopy at randomly or systematically selected points, while area measurements involve determination of the amount of throughfall for a defined area underneath the canopy. Throughfall has been measured with stationary precipitation funnels (Carlyle-Moses 2004; Crockford 1990; Horton 1919), with plastic sheet (Calder et al. 1976), with roving precipitation funnels (Lloyd et al. 1988), and with troughs (Kelliher et al. 1992). It has been indicated that most of these common sampling techniques entail large errors in the estimated interception. Time resolution of gross or net precipitation is highly dependent on the type of data collection. For instance, if gauges are manually emptied, the time resolution is per day or per storm event. There is a higher temporal resolution if tipping bucket rain gauges are used. Marsalek (1981) suggested that tipping bucket rain gauges are not suited for long term monitoring, due to both catching and counting inaccuracies, related to the positioning and mechanics/electronics of the instrument employed. Mechanical errors as a result of inherent characteristics of the counting device have a strong influence on the measurement of rainfall intensity, with increasing impact as the rate increases (Brooks et al. 2003; Lundberg et al. 1997; Marsalek 1981). Consequently, tipping bucket rain 8 gauges under-catch during high rainfall rates. Weighing devices are not widely used. A few studies that have applied this technique have reported good time resolution. Lundberg (1997) noted that a major cause of error with collecting gross precipitation is associated with under- catch, which is due to the wind around the precipitation gauges. Gross precipitation is measured above the canopy or in an open area away from any obstructions. Wind loss problems tend to be higher above the canopy due to higher wind speeds. Plastic sheets can obtain spatially correct averages; however, losses through adhesion, blockage, holes in the sheets, and evaporation can cause errors. This method is not favored for prolonged periods, as the areas underneath the sheets are deprived from receiving water (Calder 1996). Comparing funnels to troughs, it can be concluded that troughs provide a better representation of average areas (Crockford 1990). A disadvantage of using these methods is wetting losses through adhesion of water to the collectors/containers. Moreover, other losses such as splashing, combined with blockage of collection gutter during large rainfall events, may produce large measurement errors (Horton 1919; Link et al. 2004; Lundberg et al. 1997; Xiao et al. 2000). In this study, we developed and applied a rainfall interception measuring system that evaluates the net precipitation directly underneath individual trees (representative canopy area averages). This method is applicable in both urban settings and mixed forest stands. It has the ability to function during the quickly changing climatic conditions. It provides a high temporal resolution, with indication of the time and date for each tipping. 9 1.2.1 Rainfall Interception Models Models developed to estimate the interception loss from forests vary from simple empirical relationships to physically based conceptual models. They provide estimation of interception loss from climate data. The most commonly applied models were found to be the original and sparse Rutter and Gash models. Link et al. (2004) suggests that empirical models are developed for a given set of conditions in specific vegetation covers, consequently there is limited utility outside the conditions for which they were developed. These models have been successfully used in a wide range of canopy conditions, from closed forests to isolated trees. In closed and sparse forests the models are generally applied at the stand level whereas in isolated or widely separated (savanna-type) trees a tree-based approach has been preferred (Aboal et al. 1999; Carlyle-Moses et al. 1999; Deguchi et al. 2006; Dykes 1997; Lloyd et al. 1988; Muzylo et al. 2009; N\u00C3\u00A1var et al. 1999). Rutter et al. (1971) were the first to present a conceptual, physically based model. The original Rutter model characterizes a running water balance of rainfall input storage and output in the form of drainage and evaporation. Drainage and evaporation rates/amount are dependent on the amount of water stored in the canopy during each event. This model was revised by adding a stemflow component, where a portion of the rainfall input is directly diverted to stems and trunks (Rutter et al. 1975). Gash\u00E2\u0080\u0099s analytical model (1979) is a simpler storm-based model that incorporates some of the features of linear regression models in the physical background of the Rutter model. This model 10 segregates rainfall input as a series of discrete storms that are divided into intervals sufficiently long enough for canopy to dry out completely. This assumption is valid as it can occur in forest canopies and urban trees. Subsequently, each event is separated into three phases: canopy wetting up, saturation and drying. The separation stresses the significance of the meteorological factors in combination with vegetation characteristics. The Gash model (1995) was revised to incorporate the canopy cover fraction parameter in order to take into account the evaporation per unit area of the canopy rather than unit area of the ground. Canopy cover or closure is a measure of the fraction of the landscape covered by vegetation. Canopy cover, such as LAI (leaf area index), measures the amount of leaf material in an ecosystem, which imposes important controls on photosynthesis, respiration, rain interception, and other processes that link vegetation to climate. LAI estimation is calculated based on the light penetration through vegetation, where the difference between the sky and the canopy is recorded. Using this method of calculation provides permanent temporal and spatial data of the canopy (Deguchi et al. 2006; Martens et al. 1993). The separate components of Rutter and Gash models are shown in Table 1.1. 11 Table 1.1: The original and revised Rutter & Gash model Source Interception Model Components of Interception Property Output Variable Rutter (1971) Dr \u00E2\u0080\u0093 Canopy drainage C \u00E2\u0080\u0093 Canopy storage K & b \u00E2\u0080\u0093 Constants characteristics of the canopy Empirical Ic \u00E2\u0080\u0093 Interception loss Revised Rutter (1975) C \u00E2\u0080\u0093 Canopy storage p \u00E2\u0080\u0093 Free throughfall coefficient R \u00E2\u0080\u0093 Precipitation rate E \u00E2\u0080\u0093 Evaporation rate K & b \u00E2\u0080\u0093 Constants characteristics of the canopy Empirical Ic \u00E2\u0080\u0093 Interception loss Gash (1979) \u00E2\u0080\u0093 Mean rainfall rate w \u00E2\u0080\u0093 Mean evaporation rate S \u00E2\u0080\u0093 Canopy storage capacity p \u00E2\u0080\u0093 Free throughfall coefficient pt \u00E2\u0080\u0093 Stemflow portioning coefficient Empirical Ic \u00E2\u0080\u0093 Interception loss TF \u00E2\u0080\u0093 Throughfall SF \u00E2\u0080\u0093 Stemflow Revised Gash (1995) \u00E2\u0080\u0093 Mean rainfall rate c \u00E2\u0080\u0093 canopy cover c \u00E2\u0080\u0093 Mean evaporation rate/unit cover Sc \u00E2\u0080\u0093 canopy capacity/unit are of cover Empirical Ic \u00E2\u0080\u0093 Interception loss TF \u00E2\u0080\u0093 Throughfall SF \u00E2\u0080\u0093 Stemflow 12 1.3 Objectives Following from the knowledge gaps identified in the preceding literature review, the objectives of this thesis are: (1) to quantify rainfall interception by urban trees, using the innovative throughfall gauges; (2) to determine the dominant variables (vegetation or meteorological characteristics) influencing the interception process; (3) to derive seasonal storage capacity (S) for different tree species; and (4) to test the effectiveness of spatial representation of results produced by throughfall gauges. 13 1.4 Thesis Outline Chapter 2 provides information about the study area and methods used. In Chapter 3 the results of the investigation are presented. Firstly, an overview of the climate conditions for the period of 2007 and 2008 are shown. The seasonal linear regression of the divided incident rainfall events versus the throughfall depth is illustrated. Seasonal throughfall patterns and temperature for both coniferous and deciduous trees are evaluated. Event based temperature, throughfall amount and intensities are shown for four selected events. The linear regressions of rainfall versus throughfall depth are examined for different species, tree type and condition. Furthermore, these regressions are used to determine the key canopy parameters, and evaluate how these parameters change seasonally. The effectiveness of the throughfall gauges is compared to the bottles installed underneath five selected trees. Chapter 4 discusses the pattern of throughfall in urban trees, and the most effective variables influencing the results. Chapter 5 summarizes the key findings of the thesis and makes suggestions for future research. 14 Chapter 2 Study Area and Methods 2.1 Site Description This study was conducted in the District of West Vancouver, and the District and City of North Vancouver in British Columbia. Together these three municipalities are commonly referred to as the North Shore. North and West Vancouver communities are highly urbanized cities with increasing urban development that has resulted in the creation of larger proportions of impervious surfaces. The dominant land use in these municipalities is residential, followed by industrial and commercial. The North Shore sprawls in an east-west direction across the Coast Mountain slopes, and can be characterized by rugged and steep terrain. The three municipalities are described individually for a better understanding of their location/characteristics. The city of North Vancouver with an elevation range of 0\u00E2\u0080\u009380 m for urban areas falls in the Regional District of Greater Vancouver. It is surrounded on three sides by the District of North Vancouver and bounded by Burrard Inlet to the south. The District of North Vancouver with urban areas\u00E2\u0080\u0099 elevation range of 0\u00E2\u0080\u0093200 m is in the Regional District of Metro Vancouver. This District is surrounded by the Coast Mountains to the North, Burrard Inlet to the south, Capilano River to the west, and Indian Arm to the east. The District of West Vancouver is also in the Regional District of Metro Vancouver. This district municipality with an elevation range of 0\u00E2\u0080\u0093 15 300 m for urban areas is located northwest of the city of Vancouver on Burrard Inlet and Howe Sound, and is adjoined by the District of North Vancouver to its east (Environment Canada 2007). Major rivers and creeks flow through these areas, including Capilano River, MacKay Creek, Mosquito Creek, Lynn Creek, and Seymour River. The major concern regarding these waterways is the direct drainage of stormwater runoff into the rivers leading to flooding and non- point sources of pollution (Environment Canada 2007). The regional climate is characterized by cool, wet winters and warm, moderate summers. In Vancouver 166 days per year have measurable precipitation, on average. These coastal rainfall events are described as having long durations and low intensities. The average annual precipitation near sea level is about 1200 mm but reaches up to 3000 mm at higher elevations, with most of the rainfall occurring in November to February period. The amount of precipitation varies with elevation, increasing by about 100 mm for every 100 m rise in altitude. Consequently, the North Shore receives more rain and snowfall at higher elevations during the winter. The average annual temperature is 10\u00C2\u00BA C at sea level. 16 2.2 Throughfall Measurements 2.2.1 Tree Selection The study design focuses on the direct measurement of throughfall for 54 urban trees located on private and public properties along streets, parks, and forested areas. Different species including coniferous, deciduous, and some ornamental trees were randomly selected. The main coniferous trees assessed were Douglas-fir (Pseudotsuga menziesii) and Western red cedar (Thuja plicata). Bigleaf maple (Acer macrophyllum) was the deciduous tree examined. The evaluated ornamental trees included: Cherry (Prunus sp.), Copper beech (Fagus sylvatica), Horse chestnut (Aesculus hippocastanum), Oak (Quercus sp.), and Poplar (Populus sp.). The locations of the study sites on the North Shore are shown in Figure 2.1 to 2.4. 17 Figure 2.1: Study Sites in Northlands Golf course and Roche Point Park in City of North Vancouver with latitude and longitude equal to 49\u00C2\u00BA 19' 02'' N and 122\u00C2\u00BA 58' 01'' W respectively (Ortho-photos provided by District of North Vancouver\u00E2\u0080\u0099s GIS Department). 18 Figure 2.2: Study Sites in Mahon Park, Mosquito Park, and the District Hall Area, in District of North Vancouver with latitude and longitude equal to 49\u00C2\u00BA 20' 00'' N and 123\u00C2\u00BA 04' 00'' W respectively (Ortho-photos provided by District of North Vancouver\u00E2\u0080\u0099s GIS Department). 19 Figure 2.3: Study Sites at Caulfield Area in District of West Vancouver with latitude and longitude equal to 49\u00C2\u00BA 21' 00'' N and 123\u00C2\u00BA 15' 02'' W respectively (Ortho-photos provided by District of West Vancouver\u00E2\u0080\u0099s GIS Department). 20 Figure 2.4: Study Sites in Lighthouse Park in District of West Vancouver with latitude and longitude equal to 49\u00C2\u00BA 20' 02'' N and 123\u00C2\u00BA 16' 00'' W respectively (Ortho-photos provided by District of North Vancouver\u00E2\u0080\u0099s GIS Department). 21 2.2.2 Tree Classification The trees were classified into different types: dominant, co-dominant, single, and within-forest (control). Dominant trees were described as the trees with crowns receiving full light from above and partly from the side. They were typically larger than the average trees in the stand with crowns that extended above the general level of the canopy. Crowns were also well developed. Trees with crowns receiving full light from above, but comparatively little from the sides, were defined as co-dominant trees. Co-dominant crowns formed the general level of the main canopy in even-aged groups of trees. Single standing trees were exposed to light and wind from all sides. Forested areas were used as control sites, where trees were embedded within large groups of trees independent from any edge effect. These areas have little or no development (Brooks et al. 2003; Oke et al. 1989; Zipperer et al. 1997). Tree health condition was also assessed as it reflected the structural integrity. This assessment helped indicate patterns of throughfall for individual trees. The rating of tree condition involved analysis of the tree crown and the density of foliage. Four different classes were assessed based on density of the canopies: good and poor were used to describe coniferous trees, while good/broad and poor/less broad were used to describe deciduous trees. Tree health conditions in control sites were not evaluated as they were considered to be away from urban areas. These controls were assumed to be representative of health conditions in forested settings, which are naturally variable. Table 2.1 shows the location and assessed attributes for the selected trees. 22 Hemispherical photography was used to estimate the Leaf Area Index (LAI) for all selected tree canopies. A Nikon digital camera with a Nikon fisheye converter (FC-E8) was used to take a series of upward photographs from the ground to produce circular images that record the size, shape, and location of gaps in tree canopies. For evaluation we divided the longest diameter of the tree into three sections: the edges and the center, where hemispherical photographs were taken 1 m above the ground (Figure 2.5). Hemiview software was applied to obtain the LAI and Gap Fractions. Moreover, an average of the results obtained from the three images was taken to assess the horizontal/vertical heterogeneity in canopies. This assessment provided us with an estimate of canopy coverage (c \u00E2\u0080\u0093 canopy cover). Figure 2.5: Locations along the longest diameter where hemispherical photographs were taken 23 Table 2.1: Characteristics of the selected trees including their location and tag numbers Tag # Tree species Longitude Latitude Tree type Condition Height (m) DBHa (cm) Crown Spread (m) 1 Douglas-fir 49.33842249 -123.263099 Single Poor 30 70 10 2 Bigleaf maple 49.33817063 -123.2630977 Single Poor 23 48 25 3 Bigleaf maple 49.33799775 -123.263964 Single Good 30 120 25 4 Douglas-fir 49.33820413 -123.2641853 Co-dominant Good 26 45 12 5 Western red cedar 49.33784443 -123.2641421 Control (Forest) 24 45 18 6 Douglas-fir 49.3383561 -123.2645991 Co-dominant Poor 25 40 9 7 Bigleaf maple 49.33797061 -123.2640327 Control (Forest) 27 65 15 8 Bigleaf maple 49.33845242 -123.2657421 Dominant Good 24 65 14 9 Bigleaf maple 49.33840761 -123.2656731 Dominant Poor 28 55 23 10 Bigleaf maple 49.33811137 -123.2654099 Co-dominant Poor 29 60 12 11 Douglas-fir 49.33810585 -123.2638957 Dominant Good 33 120 11 12 Douglas-fir 49.33793529 -123.2637434 Control (Forest) 32 45 11 13 Douglas-fir 49.33806203 -123.2633862 Single Good 27 50 18 14 Western red cedar 49.33886334 -123.2630601 Single Poor 30 125 10 15 Western red cedar 49.33827854 -123.263112 Single Good 19 40 12 16 Douglas-fir 49.33710994 -123.2627754 Dominant Poor 35 150 13 579 Western red cedar 49.346794 -123.245806 Co-dominant Good 35 24 10 580 Western red cedar 49.343614 -123.251533 Dominant Poor 30 29 12 581 Bigleaf maple 49.346631 -123.242867 Co-dominant Good 32 18 24 582 Western red cedar 49.345947 -123.242853 Dominant Good 30 24 10 583 Western red cedar 49.346544 -123.242794 Co-dominant Poor 30 18 8 17 Copper beech 49.323514 -123.083294 Single Good/Broad 25 120 25 18 Copper beech 49.323481 -123.083311 Single Good/Broad 25 100 24 19 Oak 49.323789 -123.083492 Single Good/Broad 15 60 14 20 Copper beech 49.323475 -123.081867 Single Poor/Less Broad 23 100 12 21 Copper beech 49.323506 -123.081839 Single Poor/Less Broad 24 100 16 22 Horse chestnut 49.32399445 -123.094307 Single Good/Broad 13 75 15 23 Horse chestnut 49.3237155 -123.0944303 Single Poor/Less Broad 13 60 12 24 Cherry 49.32412075 -123.0938531 Single Good/Broad 14 55 12 25 Cherry 49.324328 -123.093592 Single Poor/Less Broad 18 40 13 26 Oak 49.32508438 -123.0924237 Single Poor/Less Broad 19 40 11 27 Poplar 49.32359091 -123.0927787 Single Good/Broad 22 55 12 28 Poplar 49.32374374 -123.0928891 Single Poor/Less Broad 23 45 10 585 Douglas-fir 49.335931 -123.078469 Single Good 35 78 14 586 Douglas-fir 49.336042 -123.078936 Dominant Poor 32 66 10 587 Western red cedar 49.3362 -123.078917 Co-dominant Good 29 70 13 588 Western red cedar 49.336311 -123.078894 Co-dominant Poor 29 43 9 590 Douglas-fir 49.336425 -123.079685 Single Poor 39 68 10 591 Western red cedar 49.3386 -123.078558 Single Good 32 95 15 200 Western red cedar 49.3186613 -122.9811076 Dominant Poor 25 90 10 598 Douglas-fir 49.31679946 -122.9820302 Dominant Good 35 43 9 599 Western red cedar 49.31692549 -122.9826631 Single Poor 38 68 10 1408 Bigleaf maple 49.32380575 -122.976936 Co-dominant Good/Broad 27 69 10 1409 Douglas-fir 49.3224305 -122.9826474 Co-dominant Good 25 37 8 1411 Western red cedar 49.32359887 -122.9769773 Dominant Good 31 83 8 4607 Bigleaf maple 49.31963279 -122.981176 Dominant Poor/Less Broad 33 61 14 100 Western red cedar 49.30903392 -122.9690737 Control (Forest) 24 35 10 592 Bigleaf maple 49.3106711 -122.9693066 Dominant Good/Broad 38 110 26 593 Bigleaf maple 49.3087099 -122.968331 Co-dominant Poor/Less Broad 36 78 10 594 Bigleaf maple 49.30905167 -122.9681795 Control (Forest) 34 45 14 595 Douglas-fir 49.30899795 -122.9690875 Control (Forest) 37 35 11 596 Douglas-fir 49.30926791 -122.9694863 Co-dominant Poor 39 45 12 24 Tag # Tree species Longitude Latitude Tree type Condition Height (m) DBHa (cm) Crown Spread (m) 597 Bigleaf maple 49.30906995 -122.9692525 Single Good/Broad 29 50 19 1033 Bigleaf maple 49.30918659 -122.9681519 Single Poor/Less Broad 27 68 22 a DBH = diameter at breast height 25 2.2.3 Field Instrumentation The throughfall gauge under each tree consisted of four components: the wooden frame, the PVC (Polyvinyl chloride) pipes, the tipping bucket rain gauge and the data logger. The wooden structure included a platform where the rain gauge was placed. Four metal rods supported a wooden roof and held the platform together. This frame was mounted directly to the trunk of the tree. Two PVC pipes were used per throughfall unit. These pipes were hung from branches using ropes and bolts, at an angle. The two pipes were positioned underneath the canopy based on the shape and structure of individual tree in a way that the entire diameter of the canopy was covered. Each pipe was approximately 3 m long, with three 0.85 m by 0.028 m slits cut on top along the length of each pipe providing a total orifice area of 0.1428 m2. The throughfall was captured by these openings and drained into a tipping bucket rain gauge with a total surface area of 0.034 m2 (RAINEW, RainWise Inc., Bar Harbor, ME). Data loggers (HOBO, Onset Computer Corporation, Pocasset, MA) attached to the rain gauges recorded both canopy\u00E2\u0080\u0099s air temperature and rainfall events. The temperatures recorded by the data loggers accounted for within-canopy temperature variation. Trees modify canopy microclimate along a vertical gradient, which is suggested to have a minor impact on canopy interception responses (Brooks et al. 2003; Jetten 1996). Overall, this flexible system allowed independent movement of the different components of each throughfall gauge without causing any serious damage to the entire structure (Figure 2.6). 26 Figure 2.6: The throughfall gauges measuring system 2.2.4 Meteorological Station Gross precipitation was measured using control units of the same design (Figure 2.7). These units were positioned on the rooftops of buildings away from any structures that may block rainfall. Additional climate stations were set up in each municipality to capture the meteorological variability along the elevation gradient. These climate stations were within a 5 km radius of the study sites. Each station was equipped to measure barometric pressure, temperature, humidity, rainfall, wind speed and direction. These supplemental records were 27 utilized to validate the tipping rain gauge data, thus ensuring correct identification of rainfall events. Figure 2.7: Example of reference rainfall gauge installed on the rooftop of North Vancouver\u00E2\u0080\u0099s City Hall to measure gross precipitation (above canopy rainfall). 2.2.5 Data Collection and Calibration The throughfall gauges were operated from February 15, 2007, to November 28, 2008. The data loggers were programmed to record the number of tips and the air temperature every 15 minutes. The number of tips was related to the depth of rainfall/throughfall. The climate stations and rain gauges were twice calibrated after the installation in the field during the study period. The throughfall gauges required regular maintenance to clear obstructions including: dirt, debris leaves and insects. The tipping bucket rain gauges were frequently wiped clean to remove mud and dirt. 28 2.2.6 Testing the Efficiency of the Throughfall Gauges Additional point collectors were installed from September 16 to November 30, 2008, to ensure that the throughfall gauges were capturing the variability in throughfall. Five trees (coniferous and deciduous) situated in the District of West Vancouver (Lighthouse Park and privately owned lands) were used for this assessment. The chosen trees with tag numbers 580, 579, 11, 6, and 2 were located far from trails and human sight to minimize vandalism. Throughfall collectors were built using 0.15 m length of PVC pipes fitted within the mouth of Nalgene wide mouth bottles (capacity 2.5 L) using couplers. Application of the silicon sealant both inside and outside of the bottles prevented leakage. Wooden stakes were driven into the ground and Dutch taped to individual bottles following each rotation. Staking helped in upholding the vertical position and minimizing the movement of the bottles. This type of set up had three main advantages: (1) greater receiving surface area (81.1 cm2); (2) less throughfall loss due to splashing; and (3) less evaporation (due to the light color of the bottles). The throughfall collectors were manually emptied every 10 to 14 days. Ten throughfall collectors were placed randomly under the selected trees as shown in Figure 2.8. The point throughfall collectors were moved after each rainfall event. Lloyd and Marques (1988) suggested that for a certain number of gauges, roving provides more accurate estimates of cumulative throughfall rather than using fixed collectors. This roving method is not accurate if used for season long or annual time periods. Also, it does not increase the accuracy for a single event estimates when assessing temporal or spatial variability (Loustau et al. 1992). 29 Figure 2.8: Random distribution of point throughfall collectors under a Douglas-fir (Tag# 6) in Lighthouse Park 30 2.3 Methods for Calculation of Interception Loss Rainfall events were defined as storms with cumulative gross precipitation exceeding 1 mm, with a minimum of 6 hours without precipitation between events (Harr 1977; Pierson 1980). Cumulative precipitation records for each rain gauge and event were manually evaluated to identify gauges that had clogged or failed during individual events. Events were categorized into two seasons. Seasons were classified as summer (April \u00E2\u0080\u0093 October) and winter (November \u00E2\u0080\u0093 March). This division was particularly applied due to deciduous trees not retaining their leaves all year around. In this study, stemflow was not measured since it was considered to be a minor component of the water balance for mature canopies, especially conifers, where the branches slope downward from the stem. This structural characteristic minimizes the probability for intercepted water to be routed to the stem, even if a small amount of precipitation intercepted in the upper canopy still contributes to stemflow. In addition, the bark is ridged and ruffled, which enhances absorption of water. The absorption of water by epiphytes and various moss species on branches and tree trunks also plays a role in controlling the stemflow. Consequently, we assumed stemflow to be insignificant, based on the results of previous research studies (Brooks et al. 2003; Crockford 1990; Link et al. 2004; Llorens et al. 2007). We computed the total volume of throughfall captured underneath each tree for individual events by using the total number of tips and the obtained average volume from the calibrations. The 31 difference between the surface areas of the PVC pipes and the rain gauge (ratio: 4.2) was taken into account. For each event, canopy interception was computed as the difference between Pg and TF (TF = Pn). The data for Pg were obtained from the reference gauges on the rooftops and standard climate stations in the municipalities. The rainfall and throughfall intensities for each event were determined as the ratio of the total depth of storm precipitation to its duration. 2.3.1 Canopy Parameters Canopy interception parameters are commonly derived from the relationship between cumulative Pg and Pn volumes on a weekly or event basis (Gash 1979; Leyton et al. 1967; Link et al. 2004). We used the method (mean method) applied by Klassen et al. (1998), where it takes a linear regression between interception (I) and Pg for rain events sufficiently large to saturate the canopy has the form Equation 2.1 where E is the mean wet canopy evaporation rate and R is the mean rainfall rate. Consequently we plotted the sum of throughfall (TF) against the Pg with the expression as follows: 32 Equation 2.2 A regression of TF against Pg therefore had a slope of (1- E/R) and the negative intercept was S. 33 2.4 Statistical Analysis Statistical analyses were completed using R Version 2.8.1 and S-PLUS Version 8.0.4. Linear regressions were performed on untransformed data of rainfall versus throughfall, since exploratory analyses showed residuals to be normally distributed with no heteroscedasticity in the data. All the results for this study were chosen based on goodness of fit and residual analysis. 34 Chapter 3 Results 3.1 Overview of Study Period Vancouver\u00E2\u0080\u0099s climate for the year 2007 and 2008 is shown in Figure 3.1. These data were obtained from the District of North Vancouver\u00E2\u0080\u0099s climate station, British Columbia. It is observed that during the month of November until March the highest amount of precipitation is received particularly during the year 2007. The highest mean temperatures were observed in the month of July and August. Figure 3.1: Vancouver\u00E2\u0080\u0099s Climate Data for 2007 and 2008 (Location of rain gauge: Latitude 49.33618\u00C2\u00BA N and Longitude 123.07814 \u00C2\u00BA W) 35 3.1.1 Rainfall Event Summary The total number of events assessed from the climate stations on the North Shore equaled 318. Rainfall events for the study period were classified based on the total amount of precipitation and average rainfall intensity as shown in Table 3.1. Table 3.1: Rainfall events classification summary Average Rainfall Intensity (mmh-1) <1 1-2 2-3 >3 <10 Medium 10-30 Low Medium 30-50 - 50-70 - Gross Precipitation (mm) >70 - - Low Figure 3.2 illustrates the amount, average intensity, and duration frequency of classified rainfall events. It is evident that events with gross precipitation less than 10 mm were the most frequent followed by the 10 \u00E2\u0080\u0093 30 mm category. The highest frequency was seen for average rainfall intensities of 2 \u00E2\u0080\u0093 3 mmh-1 followed by 1 \u00E2\u0080\u0093 2 mmh-1. Events with duration between 12 \u00E2\u0080\u0093 24 h were the most frequent, followed by 6 \u00E2\u0080\u009312 h, and less than 6 h. Overall, it can be noted that the frequencies decreased as the amount and duration of events increased. The total number of 172 events was utilized. The selected events represented the overall variability of all rainfall events during the study period. It is important to note that many events were eliminated due to the 36 problems associated with throughfall gauges. Figures 3.3 and 3.4 illustrate the seasonal histograms for the selected events. The most common event duration for summer was approximately 8 h, while for winter it was 12 h. The highest frequency of average rainfall intensity was observed to be 1.18 m mh-1 and 1.20 mmh-1 for summer and winter respectively. The highest frequencies recorded for the selected events during summer and winter were 4.31 mmh-1 and 5.14 mmh-1. 37 Figure 3.2: Rainfall event histograms of (a) gross precipitation; (b) average rainfall intensity; (c) event duration 38 Figure 3.3: Summer season histograms of events used in the study:(a) gross precipitation; (b) average rainfall intensity; (c) event duration Figure 3.4: Winter season histograms of events used in the study: (a) gross precipitation; (b) average rainfall intensity; (c) event duration 39 3.2 Relation between Event Throughfall and Gross Precipitation For the 54 selected trees during 2007 and 2008, 7042 events were recorded in total. Numerous events were eliminated due to clogging of rain gauges by leaves in late autumn and ice during the winter season when temperature fell below 0\u00C2\u00BA C. Also, many rain gauges/data loggers were vandalized or stolen; thus, there are missing data for some of the selected trees. Seasonal throughfall depth as a function of gross precipitation for both coniferous and deciduous trees is shown in Figures 3.5 and 3.6. The highest R2 for throughfall and gross precipitation can be seen in deciduous trees during winter (Figure 3.6b), while the lowest R2 is seen for coniferous trees during the summer (Figure 3.5a). Figure 3.5: Throughfall as a function of gross precipitation (season: summer) for (a) coniferous and (b) deciduous trees. Figure 3.6: Throughfall as a function of gross precipitation (season: winter) for (a) coniferous and (b) deciduous trees. 40 3.2.1 Event Throughfall Patterns Figures 3.7 and 3.8 provide an overview of the average temperature and throughfall depth for individual events in each season for coniferous and deciduous trees. The seasonal throughfall pattern is similar for both coniferous and deciduous trees; however, the only difference can be noted to be the magnitude. The highest amount of throughfall is seen during the summer season, particularly in the month of October (Figure 3.7a). Event throughfall as a percentage of gross precipitation for coniferous trees averaged 35.5% and 46.6% for summer and winter respectively. While the event throughfall as a percentage of gross precipitation for deciduous trees averaged 57.6% and 74.9% for summer and winter respectively. It is evident that there is a higher throughfall percentage in deciduous trees particularly when they lose their leaves between mid October through April, while the highest throughfall amount was recorded to be 237.9 mm. Within-canopy seasonal mean temperatures were similar for both coniferous and deciduous trees. Figures 3.7 a and b show that the highest mean temperature was seen during the month of April and May (18.8 \u00C2\u00BAC). However the mean temperatures remained relatively high throughout June to September. The lowest mean temperature was -0.5 \u00C2\u00BAC during the month of January (Figures 3.8 a and b). 41 Figure 3.7: Summer event throughfall and average event temperature (a) coniferous; and (b) deciduous trees Figure 3.8: Winter event throughfall and average event temperature (a) coniferous; and (b) deciduous trees 42 3.2.2 Interception Loss Figures 3.9 and 3.10 illustrate the variation in interception losses for different species. The results combined the variability among trees and storm events. For the summer season, the highest interception loss was demonstrated by Douglas-fir, and the lowest by Cherry trees. Copper beech had the highest variability in interception losses during summer; however, it is important to note that variability in the results was high for all species. Figure 3.9 emphasizes the outliers for Douglas-fir and Western red cedar. During winter the highest interception loss was yet again shown by Douglas-fir, followed by Western red cedar. Poplar and Oak had the lowest interception losses. Western red cedar had the highest variability for winter. Figure 3.9: Box plot of percentage interception loss for different species in summer. The outliers are presented by black circles. 43 Figure 3.10: Box plot of percentage interception loss for different species in winter 3.2.3 Urban Trees vs. Control in Forest Tables 3.2 and 3.3 illustrate the average interception loss (%) for all tree species within urban and forested areas for summer and winter season. In general, seasonal average interception losses are higher by both coniferous and deciduous trees located in the urban environments when compared to the ones in forested areas. It is important to indicate that the coniferous trees situated within forested areas showed 3.8 % higher average interception losses during winter season. Table 3.2: Percentage of average interception loss during summer Coniferous Deciduous Control in Forest 78.8 55.6 Trees in Urban Environment 81.7 67.1 Table 3.3: Percentage of average interception loss during winter Coniferous Deciduous Control in Forest 74.1 36.5 Trees in Urban Environment 71.4 45.8 44 3.2.4 Climate and Precipitation Variability During an Event Four discrete events were chosen between February 2007 and November 2008 for 9 trees (Tag numbers 586, 588, 590, 591, 18, 22, and 28) in the District of North Vancouver. Overall, the selected events fell into the high frequency categories, in terms of gross precipitation, intensity, and duration where they emphasized the seasonal variability (Figures 3.3 and 3.4). Canopy interception was derived from the difference between the Pg and TF for individual events. The data for Pg were obtained from the reference climate station on the rooftop of the District of North Vancouver; however for comparison Table 3.4 includes Pg from a standard climate station in the District of North Vancouver and a non-standard rain gauge on the rooftop of North Vancouver\u00E2\u0080\u0099s City Hall. These supplemental records were utilized to validate the tipping rain gauge data, thus ensuring correct identification of rainfall events. Table 3.4: Rainfall depth from two other nearby stations Rain Gauge on the Rooftop of District of North Vancouver Elevation = 130 m Standard Climate Station in the District of North Vancouver Elevation = 130 m Rain Gauge on the Rooftop of City of North Vancouver Elevation = 110 m Event Duration of Measurement PG (mm) PG (mm) PG (mm) 1 March 10 \u00E2\u0080\u009311, 2007 113.5 112.7 110.4 2 October 17 \u00E2\u0080\u0093 23, 2007 187.5 190.1 185.2 3 January 1 \u00E2\u0080\u00932, 2008 27.4 29.2 25.8 4 June 9, 2008 24.3 25.2 24.8 Table 3.5 highlights the event characteristics. Selected events generated 352.8 mm of gross precipitation with a maximum hourly rainfall intensity of 4.47 mmh-1. This intensity corresponds to a 2 year event in this area (Denault et al. 2006). These obtained results reflect on the rainfall 45 characteristics in the North Shore, where frontal system produces long durations and relative low rainfall intensities. Table 3.5: Event characteristics Even t Duration of Measurement Duration (h) PG (mm) Average rainfall intensity (mmh-1) Maximum rainfall intensity (mmh-1) Average Temperature (\u00C2\u00BAC) Wind Speed ms-1 1 March 10 \u00E2\u0080\u009311, 2007 36.5 113.5 3.09 4.62 9.3 0.44 2 October 17 \u00E2\u0080\u0093 23, 2007 141 187.5 1.33 4.52 6.9 0.03 3 January 1 \u00E2\u0080\u00932, 2008 10.5 27.4 2.55 4.51 3.5 0.03 4 June 9, 2008 14.75 24.3 1.62 4.22 8.2 0.03 Precipitation and above and within canopy climate data for coniferous (Douglas-fir, and Western red cedar) and deciduous (Copper beech and Horse chestnut) for event 2 are shown in Figures 3.11, 3.12, and 3.13. It is important to note that there are no data available for Poplar tree (#28) due to rain gauge failure (clogging) for event 2. These figures show the effect of the urban tree canopies on throughfall intensity, and demonstrate the range of conditions controlling interception loss during the rainfall event. Event 2 began at 0400 hours on October 17 and lasted 141 hours. During this period, 187.5 mm of precipitation was recorded by the reference rain gauge on the rooftop of District of North Vancouver. The precipitation intensity, humidity, wind speed, and temperature were typified as moderately low. Figures 3.11a, 3.12a, and 3.13a illustrate that there was not much variation between the measured temperatures above and below the canopy for all selected trees. Wind speed was below 0.1 ms-1. Average humidity was above 95%. The amount of throughfall captured underneath each canopy averaged 45.6% and 90.9% (85.6 mm and 170.4 mm) for 46 coniferous and deciduous trees, respectively. Figures 3.11b, 3.12b, and 3.13b show that throughfall levels for both species are not constant, but are dynamic. The difference in gross precipitation and net precipitation magnitude is shown in Figures 3.11c, 3.12c, and 3.13c. Table 3.6 presents the delay in throughfall reaching the ground for all study sites. The delay ranged from 4.25 \u00E2\u0080\u0093 6.25 hours for event 2. This delay did not affect the peak in net precipitation; however, as shown by Xiao et al. (2002) this would delay the peak runoff for a storm. Throughfall ceased roughly between 0.25 \u00E2\u0080\u0093 1.50 hours after the rainfall stopped for tag numbers: 586, 590, 591, and 22. However, for the trees with tag numbers 588 and 18, the throughfall ceased approximately 12.50 hours before the end of event 2. The average rainfall intensity for event 2 was determined by dividing the gross precipitation by the rainfall duration. Figures 3.11c, 3.12c, and 3.13c illustrate the impact of canopy on throughfall intensity, and exemplify how the climatic conditions control evaporation during the rainfall event. Both temperature and wind are suggested to play an important role in driving the evaporation rate; however, wind was omitted due to low velocities (Brooks et al. 2003; Link et al. 2004) 47 Table 3.6: Lag time between gross precipitation and throughfall Event 1 Event 2 Event 3 Event 4 Tag # Start Time (h) Start Time (h) Start Time (h) Start Time (h) 586 2.0 6.25 1.0 5.0 588 2.25 6.25 4.25 2.75 590 1.5 4.5 3.5 1.75 591 2.25 6.5 4.0 6.0 18 1.0 6.0 2.25 5.0 22 0.75 4.25 3.00 3.5 28 1.25 - 3.00 1.5 48 Figure 3.11: Meteorological and throughfall data for rainfall event two (Douglas-fir) 49 Figure 3.12: Meteorological and throughfall data for rainfall event two (Western red cedar) 50 Figure 3.13: Meteorological and throughfall data for rainfall event two (Deciduous trees) 51 Table 3.7 summarizes the interception losses for the selected trees and events. When evaluating the average interception losses, it is evident that two of the cedar trees (#588, 591) overall showed the highest interception losses. Both Western red cedar trees were co-dominant; however, one was of good health condition and the other poor. Events 2 and 4 occurred during summer, when both Douglas-fir and Western red cedar had high interception losses compared to the deciduous trees. The highest interception loss was in event 4 by a dominant Douglas-fir of a poor condition. Based on the results, compared to Western red cedar the Douglas-fir trees showed a wider range of interception losses during the seasons. It is important to note that deciduous trees demonstrated lower interception losses than coniferous trees. The lowest and highest interception loss for deciduous trees was by a single standing Horse chestnut tree in good condition during event 2 and a single standing Copper beech in good condition during event 4 correspondingly. This high variability in the results was due to the species defoliation with the onset of the cooler weather in autumn and winter. Event 4 was the smallest precipitation event and had the highest interception losses with 24.3 mm of gross precipitation over 14.75 hours duration. The average temperature was recorded as 8.2 \u00C2\u00BAC with maximum rainfall intensity of 4.22 mmh-1. Event 3 had the shortest duration in comparison to the other selected events. In general, rainfall type plays a role in determining interception loss. For instance, a low intensity, long-duration frontal rainfall generates different interception loss than a high intensity short duration convectional storm (Deguchi et al. 2006; Pypker et al. 2005; Xiao et al. 2000b). 52 Table 3.7: Percentage of interception for the selected rainfall events Event 1 Event 2 Event 3 Event 4 Tag # Inet (%) Inet (%) Inet (%) Inet (%) 586 15.92 47.32 38.09 82.21 588 36.21 67.10 50.83 74.89 590 13.52 30.64 20.72 11.42 591 18.77 72.42 53.60 82.14 18 19.63 14.88 10.65 59.03 22 10.10 3.40 40.01 51.20 28 5.62 - 16.07 27.66 53 3.3 Variables Influencing Interception Loss The potential influence of tree species, condition and type was tested with linear regressions for all the selected trees. The analysis included a factor variable (season) in interception loss between summer and winter. Histograms and quantile\u00E2\u0080\u0093quantile (QQ) plots of regression indicated that residuals were approximately normally distributed with no apparent heteroscedasticity. It is important to note that due to structural damage, the tree with tag number 16 did not produce any reliable data; thus, no substantial analysis was conducted on this site. Figures 3.14 to 3.17 illustrate the seasonal variation in the throughfall for a coniferous and a deciduous tree. As seen in Figures 3.14 and 3.15 for Western red cedar the intercepts were -1.43 mm and -4.82 mm while the slopes were equivalent to 0.30 and 0.61 for summer and winter, respectively. Calculated R2 values were 0.90 (summer) and 0.89 (winter). Bigleaf maple intercepts were -5.2 mm for summer and -2.88 mm for winter as seen in Figures 3.16 and 3.17. The slopes were recorded to be 0.90 and 0.89 with R2 equivalent to 0.98 and 0.97 for summer and winter correspondingly. The data appeared to have a clear seasonal trend in the residual plots, where more throughfall was seen during winter. 54 Figure 3.14: Relationship between throughfall and gross precipitation; and residual plot for a single standing Western red cedar in a good condition (tag # 591) during summer Figure 3.15: Relationship between throughfall and gross precipitation; and residual plot for a single standing Western red cedar in a good condition (tag # 591) during winter y = 0.30x \u00E2\u0080\u0093 1.43 R2 = 0.90 y = 0.61x \u00E2\u0080\u0093 4.82 R2 = 0.89 55 Figure 3.16: Relationship between throughfall and gross precipitation; and residual plot for a single standing Bigleaf maple in a good condition (tag # 597) during summer Figure 3.17: Relationship between throughfall and gross precipitation; and residual plot for a single standing Bigleaf maple in a good condition (tag # 597) during winter y = 0.90x \u00E2\u0080\u0093 5.20 R2 = 0.98 y = 0.89x \u00E2\u0080\u0093 2.87 R2 = 0.98 56 Appendix A provides a summary of seasonal regression coefficients for individual trees. Equation 3.1 is used to determine whether the regressions between TF and Pg significant. The significance level, \u00CE\u00B1, was set to 0.05. The hypotheses tested were: H0: \u00CE\u00B2 = 0 H1: \u00CE\u00B2 \u00E2\u0089\u00A0 0 The test statistic is: Equation 3.1 In all cases, the null hypothesis can be rejected, since the p-values for the F-statistic are less than 0.05. Seasonal regressions for all selected trees were significant. Seasonal box plots of slopes and R2 coefficients for different species are illustrated in Figures 3.18 to 3.21. Deciduous trees had higher slopes for both summer and winter with not much variation in range. Coniferous trees showed lower slopes, which can be explained by the lower amounts of throughfall passing through the canopies. Variation in range for coniferous trees was higher for both seasons. There are strong relations between species and throughfall for winter and summer: all R2 for both seasons were above 0.9. There are outliers present for R2 coefficients for Bigleaf maple and Douglas-fir trees during both seasons (Figures 20 and 3.21); however, for the slope coefficients it is only evident during the summer season. 57 Figure 3.18: Box plot of slope coefficients for different species (summer). The outliers are presented by black circles. Figure 3.19: Box plot of slope coefficients for different species (winter) 58 Figure 3.20: Box plot of R2 coefficients for different species (summer). The outliers are presented by black circles. Figure 3.21: Box plot of R2 coefficients for different species (winter). The outliers are presented by black circles. 59 3.3.1 Throughfall \u00E2\u0080\u0093 Assessed Attributes Relationships Table 3.8 shows the seasonal coefficients for different tree species, condition and type. These attributes are important in predicting the amount of throughfall, which depends on rainfall depth and seasonal variability. It is crucial to mention that the regressions obtained are from averaging the values obtained from individual trees. When evaluating regressions, it is evident that single standing Horse chestnut trees in poor condition had the highest slope for both seasons. Poplar and Copper beech trees, both single standing and in poor conditions demonstrated high slope values during winter season. During the summer season, Bigleaf maple trees (control in forest) and Poplar trees in good condition (single standing) had high slopes. R2 values were highest in deciduous trees. When contrasting control trees, Bigleaf maple showed the highest slope and R2 than Western red cedar and Douglas-fir for both seasons. Douglas-fir control trees demonstrated the lowest slope and R2 during the summer. 60 Table 3.8: Seasonal regression (TF vs. Pg) coefficients for different tree species, types and health conditions Species Condition Type Winter b0b Winter b1 c Winter R2 Winter SEa Summer b0 b Summer b1 c Summer R2 Summer SEa Douglas-fir Good Co-dominant -3.96 0.74 0.94 4.27 -2.36 0.59 0.95 3.03 Douglas-fir Good Dominant -6.26 0.65 0.87 6.27 -2.17 0.31 0.64 4.5 Douglas-fir Good Single -4.68 0.62 0.81 5.01 -3.43 0.52 0.67 5.23 Douglas-fir Poor Co-dominant -5.54 0.78 0.96 3.76 -3.58 0.66 0.93 3.18 Douglas-fir Poor Dominant -4.92 0.68 0.93 6.89 -1.93 0.51 0.46 3.77 Douglas-fir Poor Single -4.21 0.86 0.98 3.32 -2.92 0.74 0.92 4.00 Douglas-fir Control -4.71 0.72 0.95 3.52 -2.05 0.51 0.77 4.09 Western red cedar Good Co-dominant -3.87 0.65 0.91 3.87 -2.25 0.46 0.9 3.26 Western red cedar Good Dominant -4.06 0.74 0.95 4.08 -4.44 0.75 0.96 3.81 Western red cedar Good Single -4.11 0.62 0.93 4.26 -2.3 0.48 0.92 2.65 Western red cedar Poor Co-dominant -5.06 0.64 0.89 5.9 -1.44 0.34 0.85 2.95 Western red cedar Poor Dominant -3.34 0.6 0.93 3.75 -2.17 0.43 0.89 3.47 Western red cedar Poor Single -5.97 0.71 0.94 4.46 -2.98 0.49 0.94 3.34 Western red cedar Control -4.22 0.7 0.91 4.07 -2.78 0.6 0.9 3.14 Bigleaf maple Good/Broad Co-dominant -3.81 0.94 0.98 3 -4 0.86 0.97 3.68 Bigleaf maple Good/Broad Dominant -3.17 0.89 0.99 2.72 -3.88 0.65 0.91 4.9 Bigleaf maple Good/Broad Single -3.92 0.91 0.96 3.65 -5.18 0.88 0.97 4.02 Bigleaf maple Poor/Less broad Co-dominant -3.31 0.92 0.98 2.73 -4.6 0.86 0.95 4.41 Bigleaf maple Poor/Less broad Dominant -3.5 0.94 0.98 3.08 -4.7 0.81 0.93 5.44 Bigleaf maple Poor/Less broad Single -2.89 0.93 0.99 2.79 -3.22 0.82 0.95 3.28 Bigleaf maple Control -2.74 0.94 0.99 2.33 -3.79 0.89 0.99 2.88 Cherry Good/Broad Single -4.28 0.95 0.98 3.22 -4.5 0.92 0.97 4.25 Cherry Poor/Less broad Single -2.1 0.94 0.99 1.53 -2.31 0.8 0.94 2.52 Copper beech Good/Broad Single -3.65 0.83 0.96 3.61 -5.34 0.81 0.9 6.06 Copper beech Poor/Less broad Single -3.91 0.96 0.98 2.63 -5.81 0.93 0.96 4.95 Horse chestnut Good/Broad Single -3.86 0.92 0.99 3.35 -4.72 0.89 0.96 4.36 Horse chestnut Poor/Less broad Single -3.4 0.99 0.99 2.36 -4.58 0.96 0.98 3.35 Oak Good/Broad Single -2.33 0.91 0.99 2.96 -5.48 0.89 0.93 6.26 Oak Poor/Less broad Single -1.94 0.95 0.99 1.54 -2.02 0.85 0.98 1.72 Poplar Good/Broad Single -1.64 0.91 0.99 1.09 -3.54 0.92 0.99 2.97 Poplar Poor/Less broad Single -2.53 0.96 0.99 2.06 -2.51 0.83 0.96 2.55 aSE = Standard Error b b0 = Intercept c b1 = Slope 61 A two-way analysis of variance is used to test the equality of seasonal means of slopes and intercepts obtained from regressions for Douglas-fir, Western red cedar, and Bigleaf maple. For the rest of species, the analysis of variance could not have been completed, since they did not have at least two observations for each factor level. The significance level, \u00CE\u00B1, was set to 0.05. Two sets of hypotheses were developed for health condition and tree type. (1) Health condition: H0: Seasonal mean of slopes and intercepts for species\u00E2\u0080\u0099 health conditions are equal. H1: Seasonal mean of slopes and intercepts for species\u00E2\u0080\u0099 health conditions are not equal. (2) Tree type: H0: Seasonal mean of slopes and intercepts for different tree types are equal. H1: Seasonal mean of slopes and intercepts for different tree types are not equal. The probability values for ANOVA tests were all greater than 0.05 except for Douglas-fir trees\u00E2\u0080\u0099 summer slope, with trees in poor condition having significantly higher slopes than trees in good condition (Table 3.9). With that exception, all the other null hypotheses cannot be rejected. 62 Table 3.9: Summary of results for ANOVA analysis Condition: Poor \u00E2\u0080\u0093 Good Winter Intercept Winter Slope Summer Intercept Summer Slope Species F P-Value F P-Value F P-Value F P-Value Douglas-fir 0.003 0.96 2.14 0.18 0.04 0.85 7.28 0.03 Western red cedar 0.54 0.48 0.06 0.81 1.70 0.23 3.07 0.12 Bigleaf maple 1.01 0.34 0.48 0.50 0.08 0.78 0.32 0.59 Canopy storage capacity (S) for all the selected species was estimated from the regressions of TF against Pg where the negative intercept was S. Table 3.10 illustrates average storage capacities for different species for each season. Average winter S for deciduous species were not calculated as they lose their leaves during this period. It is evident that all deciduous trees demonstrated high storage capacity during the summer season, when they retain their leaves. On the other hand, coniferous trees showed higher storage capacity during winter throughout the dormant season. Table 3.10 shows the estimated summer canopy cover (c) for different tree species. In general, all tree species showed high canopy coverage during the summer season. Bigleaf maple trees showed the highest canopy cover, as a result of having larger leaves with greater surface areas. No hemispherical photographs were taken during the winter as we assumed that the coniferous Type: Dominant \u00E2\u0080\u0093 Co-dominant \u00E2\u0080\u0093 Single Winter Intercept Winter Slope Summer Intercept Summer Slope Species F P-Value F P-Value F P-Value F P-Value Douglas-fir 0.22 0.81 0.58 0.58 0.75 0.51 3.96 0.07 Western red cedar 0.54 0.6 0.04 0.96 1.88 0.21 1.83 0.22 Bigleaf maple 0.11 0.89 0.22 0.80 0.01 0.98 1.77 0.23 63 trees maintained the same canopy cover throughout the year, while deciduous trees canopy cover approximated to 0 due to defoliation. Table 3.10: Seasonal storage capacity (mm) and canopy cover (c) estimations Species Average Winter S (mm) Average Summer S (mm) Summer Canopy Gap Fraction Summer Canopy Cover (c) Douglas-fir 4.9 2.8 0.10 0.90 Western red cedar 4.4 2.6 0.10 0.90 Bigleaf maple - 4.2 0.08 0.92 Cherry - 3.4 0.10 0.90 Copper beech - 5.5 0.13 0.87 Horse chestnut - 4.6 0.11 0.89 Oak - 3.7 0.09 0.91 Poplar - 3.0 0.10 0.90 \u00E2\u0080\u00A9\u00E2\u0080\u00A9 3.4 Efficiency of Throughfall Gauges This section examines the effectiveness of the innovative throughfall gauges. Sources of error were the precision of data logger time accuracy (\u00C2\u00B11 min), the tipping buckets (\u00C2\u00B10.23 mm) and volume measurement (\u00C2\u00B11 mL). The measurement errors were small. Figure 3.22 represents the relationship between the results obtained from the roving bottles and the pipes (throughfall gauges). It appears that there is a strong relationship between the two measurement techniques. Regression analysis for individual trees was conducted (Table 3.11). All trees, except one (Douglas-fir, # 11) had R2 values above 0.9. F-statistic test is used to determine whether the regressions between the results obtained from the bottles and the results obtained from the pipes are significant. The significance level, \u00CE\u00B1, was set to 0.05. The hypotheses tested were: 64 H0: The regression is not significant. H1: The regression is significant. The probability values for the F statistics tests were less than 0.05 (Table 3.11); therefore, the null hypothesis can be rejected. Regressions for all selected trees were significant. Equation 3.2 shows the chi-square (\u00CF\u00872) test used to examine the significance of the differences between the throughfall obtained from the bottles and pipes. The significance level, \u00CE\u00B1, was set to 0.05. The hypotheses tested were: H0: There is no difference between the throughfall captured by pipes and bottles H1: There is a difference between the throughfall captured by pipes and bottles Equation 3.2 The probability values for the \u00CF\u00872 tests were less than 0.05. Therefore, there is a difference between the throughfall captured by pipes and bottles as seen in Table 3.11. Consequently the null hypothesis can be rejected. The difference between the amount of throughfall captured by pipes and bottles for the duration of mid September to the end of August was 157.8 mm. It is evident that pipes are capturing 1.17 times more throughfall than bottles. 65 Figure 3.22: The relationship between throughfall measured by innovative throughfall gauges and bottles for 5 selected trees (6 events) Table 3.11: Regression coefficients from throughfall gauges and bottles for the selected trees (n = 6 events) Tag # Species Intercept Slope R2 F Significance F Chi-Squared \u00CF\u00872 df \u00CF\u00872 Calculated \u00CF\u00872 Critical 2 Bigleaf maple -3.98 1.12 0.94 61.62 0.0015 120.32 6 120.32 12.59 6 Douglas-fir 0.96 1.26 0.90 37.48 0.0040 70.98 6 70.98 12.59 11 Douglas-fir -4.43 0.92 0.76 12.91 0.0230 32.02 6 32.02 12.59 579 Western red cedar -13.59 1.79 0.95 73.55 0.0010 185.06 6 185.06 12.59 580 Western red cedar -23.64 2.15 0.92 48.83 0.0022 44.18 6 44.18 12.59 66 Chapter 4 Discussion 4.1 Canopy Interception Loss High temporal resolution and measurements of canopy rainfall interception at the individual tree level can be obtained using the innovative throughfall measurement system. The throughfall results indicate that interception is influenced by seasonal differences in foliation periods and rainfall characteristics. Figure 3.5 suggests that the relationship between throughfall and incident precipitation is higher during winter for deciduous trees. This may be as a result of having no foliage, where throughfall amounts were close to the amount of gross precipitation for all deciduous trees. In addition, for some deciduous trees there were far more winter than summer events. Our data show that, on average, deciduous trees\u00E2\u0080\u0099 interception loss was about 45% of gross precipitation during the winter. Additionally, during the summer season, deciduous trees demonstrated the highest relationship as seen in Figure 3.6. This can be explained by the high surface areas of the leaves. With reference to Figures 3.7 and 3.8 it can be said that the throughfall pattern is similar for both coniferous and deciduous trees; however, the magnitude differs. Moreover, temperature seemingly plays a role in controlling the amount of throughfall reaching the ground. During the months of May \u00E2\u0080\u0093 June (Figures 3.7 and 3.8) throughfall amount was noticed to be lower. This can both be due to high temperatures and the gross precipitation characteristics (duration, 67 amount, and intensity). The highest amount of throughfall was observed from October \u00E2\u0080\u0093 March. Figure 3.8a shows a higher throughfall amount during March for coniferous trees. This can be explained by having higher rain gauge failures for deciduous tree species where numerous events were eliminated. The interception loss for deciduous trees averaged to 67.1% and 45.8% of the gross precipitation for the summer and winter respectively. Based on our results, coniferous trees intercepted more rainfall than deciduous trees all year around. For coniferous trees our results demonstrate that on average the interception loss was approximately 81.7% and 71.4% of gross precipitation for summer and winter respectively. When comparing interception losses amongst species, both Douglas-fir and Western red cedar demonstrated outliers during the summer season (Figure 3.9). These outliers can be as a result of canopy structures\u00E2\u0080\u0099 characteristics, climatic conditions or sampling error during events. In general, urban trees showed higher interception losses than control trees within forested areas. 4.1.1 Event Based Analysis Interception loss is determined as the difference between gross precipitation and the sum of throughfall and stemflow. Based on the obtained throughfall data, the average Inet% for the four events ranged between 44.7% and 23.5% of gross precipitation for coniferous and deciduous trees respectively. Based on the variability in rainfall amount, intensity and duration, the interception losses for coniferous trees in the selected events ranged from 11.4% and 82.2%, which were 2.8 mm and 19.9 mm of gross precipitation correspondingly. While for deciduous 68 trees the interception loss varied between 3.4% and 59.0%, which were 6.4 mm and 14.3 mm of gross precipitation in that order. The lowest interception losses occurred during events 1 and 2 for a Horse chestnut and a Poplar. The results suggest that interception losses for both coniferous and deciduous trees are higher within urban environments compared to trees within forested areas. Link et al. (2004) suggested that annual interception losses in temperate forests were observed to range from 11% to 36% of gross precipitation in deciduous, and from 9% to 48% of gross precipitation in coniferous canopies. Bryant et al. (2005) reported 22.3% interception loss in a pine forest. Possible factors contributing to these differences are UHIs, greater distances between trees (edge effect), and open grown canopies. UHIs cause local scale variation in temperature differences between urban and natural forested areas. This is due to replacement of natural vegetation by man made structures where resulting in less evapotranspiration (Taha 1997). In addition, urban trees are isolated with greater distances between them, making them more exposed during severe weather events, unlike trees within forested areas where they are surrounded by other trees (Aboal et al. 1999; Nadkarni et al. 2004). Urban tree canopies are classified as open grown trees due to no inter-tree competition; consequently, they have larger structural dimensions (e.g., larger storage capacity) than trees in forests (Brooks et al. 2003; Horton 1919; Xiao et al. 2002; Xiao et al. 2000b; Zipperer et al. 1997). Tree health condition and type were assumed to affect interception rate. Single standing trees in good health were expected to have a higher interception rate. This was demonstrated in some, 69 but not in all the events. A co-dominant (poor health condition) and a single (good health condition) Western red cedar intercepted at higher rates compared to other trees of better health conditions and types. A dominant Douglas-fir with poor canopy condition showed the highest interception loss for event 4. Western red cedar trees generally had higher interception losses compared to Douglas-firs. This is due to the differences in canopy structure between the two tree species. For the event in June the interception losses were relatively high for all species. The high rates can be explained by small rainfall events, where most of the water from the event is used to wet the crown surfaces. It is evident that interception loss for the deciduous trees is very low during event 1, 2, and 3, where they do not possess any foliation. The time delay in throughfall penetrating through the canopy was greatest for events 2 and 4, which ranged from 1.5 \u00E2\u0080\u0093 6.5 h. Event 1 did not show significant delay in throughfall (0.75 \u00E2\u0080\u0093 2.25 h). Event 3 had moderately higher time delays that were lower than event 1. It can be suggested that tree type and health condition played an important role in controlling the time delay. For the four events it was noticed that a Western red cedar tree that was single standing and in a good health condition showed a longer time delay in throughfall. In general coniferous trees had the highest time delay in comparison to deciduous trees. Additionally when evaluating the time delays amongst all species the single standing Poplar in a poor condition showed the earliest time delay in throughfall, followed by Horse chestnut. Trees generally dampen rainfall intensity; however, there were instances where the throughfall intensity was equivalent to or higher than the actual rainfall intensity. The highest throughfall 70 intensities in event 2 were seen at 63.5 h with rainfall intensity recorded to be 2.41 mmh-1. The throughfall intensity exceeded the actual rainfall intensity except for a Western red cedar tree (tag # 591). The throughfall intensities for the two Douglas-fir trees (tag # 586 and 590) and a Western red cedar (tag # 588) were 6.54 mmh-1, 9.30 mmh-1 and 8.0 mmh-1, respectively. Throughfall intensities for Copper beech and Horse chestnut were recorded to be 23.96 mmh-1 and 17.37 mmh-1. This variation can be explained by rainfall characteristics, meteorological factors and structure of the canopy. It is evident in Figure 3.11c - Figure 3.13c that high throughfall intensities are delayed in time for lower rainfall intensities. It is suggested that crown density wetness is a factor, where as the crown dampens the drip becomes larger, consequently resulting in higher throughfall intensities (Brooks et al. 2003; Crockford et al. 2000). The observed reduction in throughfall intensity by tree canopies serves two purposes. First, it delays the water reaching the ground by temporary storage of the water on the tree. This storage both reduces and delays the peak in the stormwater runoff. Second, it protects the mineral soil surface from the energy of raindrops reaching the ground at maximum velocity. Reduction of raindrop energy by interception minimizes soil detachment and subsequent erosion, which in turn protects soil structure and infiltration capacity leading to less stormwater runoff (Pypker et al. 2005; Xiao et al. 2002). All the selected events demonstrated a reduction in raindrop energy by having lower intensities captured underneath the canopy. The differences in the magnitudes of rainfall intensity for the events were dependent on climatic conditions. Tree\u00E2\u0080\u0099s health condition, type and specie can be suggested to contribute to the differences in throughfall intensities. 71 4.1.2 Controls on Interception Loss When we compared the differences in interception losses between summer and winter for individual trees, we found a relationship between the amount of troughfall captured and the season. The canopy structure (leaves) is primarily responsible for two main effects on throughfall. First, it affects the ratio of throughfall to gross precipitation. Deguchi et al.(2006) suggested a decrease in the number of leaves appeared to cause an increase in throughfall. The second effect is the throughfall spatial variability caused by seasonal changes in the canopy structure. This variability can be the result of sampling error or the spatial heterogeneity associated with the canopy structure (Deguchi et al. 2006; Price et al. 2003). It can be noted that the changes in throughfall amount is attributed to the fact that LAI is one of the main factors determining interception loss, in terms of the relationship between LAI and S (Ford et al. 1978; Jetten 1996; Pypker et al. 2005). Interception in tree canopies is affected by the amount, intensity, and duration of precipitation, as well as the air temperature and wind speed (Crockford et al. 2000; Deguchi et al. 2006; Loustau et al. 1992). In addition to the meteorological factors, there is a strong relationship between throughfall, tree species and health condition. Tree type did not seem to play a significant role in determining the amount of throughfall reaching the ground. Xiao and McPherson (2002) suggested that annual rainfall interception in Santa Monica\u00E2\u0080\u0099s urban trees as follows: deciduous trees were responsible for 3.9%, coniferous trees accounted for 23%, and the remaining 73.1% was contributed by palm 72 trees, and broadleaf evergreen trees (Xiao et al. 2000a). Interception can be suggested to be a function of growth form of trees, tree canopy density and structure. 4.1.3 Outcome of Throughfall Gauges Compared to the traditional methods of measuring gross precipitation and throughfall (measured with funnels, troughs and plastic sheet net rainfall gauges), the innovative design used in this study had high accuracy and time resolution. In addition, these throughfal gauges were designed to overcome the weakness of traditional system, in order to capture the throughfall amount for single standing trees in urban environments. An expected advantage with the hanging pipes draining into shielded rain gauges and data loggers was minimized evaporation. An unanswered issue with the pipe system was the separation of drip from the canopy, as the method was designed to represent an average canopy area sampling. Moreover, random sampling of the length and width of the slits for the PVC pipes for 10 throughfall gauges illustrated that the surface areas remained unchanged over the two-year period. It can be noted that there was a lot of tension on the pipes as they hung from the trees, thus resulting in the slits appearing much smaller. One practical problem was associated with litter from the trees, which gradually accumulated in the pipes and rain gauges where they were removed manually every month. We assumed that the water absorbed by this debris was negligible. Table 3.11 shows that there is a significant difference between the throughfall captured by the pipes and the bottles underneath the canopies. Our results showed that the pipes captured 1.17 times more than the bottles. It is suggested that this is due to the effect of plant community structure, where secondary 73 interception occurs in stratified forest communities where water drips from the canopy and is intercepted by lower plants (Asdak et al. 1998; Brooks et al. 2003; Dunkerley 2000). In many sites, especially Lighthouse Park, Roche Point Park, and Northlands Golf course the ground was covered by seedlings, shrubs and numerous types of ferns (0.5 \u00E2\u0080\u0093 1.5 m in height). Also the branches of many dominant and co-dominant trees seemingly overlaid one another, making it a denser environment for rainwater to penetrate. The estimated understory shrub coverage for the selected trees with tag number 2, 6, 11, 579, and 580 were as following: 55%, 10%, 20%, 70%, and 5% respectively. It is important to note that the selected Douglas-fir tree with tag number 11 had dense coniferous branches overlapping more than 20% of its canopy area particularly where the right pipe was located. Also, there were many smaller/shorter trees and seedling located within the vicinity of the canopy. As a result lower throughfall amount can be seen in both the pipes and the bottles. 74 Chapter 5 Conclusion 5.1 Review of Key Findings The goal of this project was to shed some light on rainfall interception by single and stands of trees in urban environments of the North Shore of British Columbia. Rainfall interception was influenced by three factors: characterization and magnitude of the rainfall event, tree species and health condition, and meteorological factors. The trees\u00E2\u0080\u0099 ability to dampen rainfall intensity and delay the stormwater to reach the ground was emphasized. Location and seasons contributed to the discrepancy of the interception losses. The inter-species variation on interception was evident as Western red cedar showed higher interception losses, longer time delays and lower throughfall intensities when compared to Douglas-fir and deciduous trees during event analysis. It is important to note that Douglas-fir followed by Western red cedar showed the highest interception losses during both summer and winter. Interception losses for coniferous and deciduous trees averaged to be 76.5% and 56.4% respectively. On average control trees located in forested areas showed 1.12 times less interception loss than urban trees. The innovative throughfall gauges confirmed that the design worked effectively in capturing a good representation of throughfall from individual tree canopies. Our results suggested that 1.17 times more throughfall was captured by the pipes when compared to the bottles installed underneath the canopies. Based on the previous methods/research (roving bottles and rain 75 gauges), the obtained interception losses from the throughfall data attained from the pipes should be higher than what we estimated. The data obtained from this research will be used in the Water Balance Model powered by QUALHYMO (QUALity HYdrology MOdels). As urban development progresses, extensive areas of our natural landscape are converted into impervious surfaces such as roads, parking lots, driveways and buildings. Human-made drainage systems such as sewers and storm drains are built to enhance runoff movement through cities and into drainage and natural waterways, where the water quality is negatively impacted when runoff carries contaminants. Therefore, this tool will help determine the water balances at individual properties in order to provide data and additional information for different stakeholders to utilize in the planning of future urban development. It is also important to note that we are in need of better stormwater management. Using natural vegetation as a low impact development and best management practice can be an effective technique as it controls stormwater runoff on site, mitigating the impacts of urbanization on urban hydrology at a local scale. 76 5.2 Future Research There are many avenues down which this study can be expanded upon. Future research projects could focus on (a) measurement of stem and epiphyte moisture; (b) measurement of spatial variability of throughfall; (c) deeper look at the affects of trees\u00E2\u0080\u0099 health condition on interception loss; (d) identifying soil moisture pattern; (e) assessment of seasonal variability of LAI; (f) application of different model in combination with detailed micrometeorological measurements to evaluate the effects of interception loss for urban trees; and (g) determining the amount of stormwater runoff which can be reduced and delayed from individual properties by urban trees. 77 References Aboal, J.R., Jim\u00C3\u00A9nez, M.S., Morales, D. and Hern\u00C3\u00A1ndez, J.M., 1999. Rainfall interception in laurel forest in the Canary Islands. Agricultural and Forest Meteorology, 97(2): 73-86. American Forests, 1996. Urban ecological analysis report, Phase 1: Economic benefits and costs of the urban forest in low income and non-low income communities. American Forests, Final report NA-94-0297. American Forests. Washington, D.C. Asdak, C., Jarvis, P.G., van Gardingen, P. and Fraser, A., 1998. Rainfall interception loss in unlogged and logged forest areas of Central Kalimantan, Indonesia. Journal of Hydrology, 206(3-4): 237-244. Aston A. R., 1979. Rainfall interception by eight small trees. Journal of Hydrology, 42: 383-396. Brooks, K.N., Ffolliott, P.F., Gregersen, H.M. and DeBano, L.F., 2003. Hydrology and the management of watersheds. Iowa State Press, Iowa. Bryant, M.L., Bhat, S. and Jacobs, J.M., 2005. Measurements and modeling of throughfall variability for five forest communities in the southeastern US. Journal of Hydrology, 312(1-4): 95-108. Calder, I.R., 1996. Dependence of rainfall interception on drop size .1. Development of the two- layer stochastic model. Journal of Hydrology, 185(1-4): 363-378. Calder, I.R. and Rosier, T.P.W., 1976. The design of large plastic-sheet net rainfall gauges. Journal of Hydrology, 30: 403-405. Carlyle-Moses, D.E., 2004. Throughfall, stemflow, and canopy interception loss fluxes in a semi-arid Sierra Madre Oriental matorral community. Journal of Arid Environments, 58(2): 181- 202. Carlyle-Moses, D.E. and Price, A.G., 1999. An evaluation of the Gash interception model in a northern hardwood stand. Journal of Hydrology, 214(1-4): 103-110. Crockford, D.P.R., 1990. Partitioning of rainfall in a eucalypt forest and pine plantation in southeastern australia: II stemflow and factors affecting stemflow in a dry sclerophyll eucalypt forest and a pinus radiata plantation. Hydrological Processes, 4(2): 145-155. 78 Crockford, R.H. and Richardson, D.P., 2000. Partitioning of rainfall into throughfall, stemflow and interception: effect of forest type, ground cover and climate. Hydrological Processes, 14(16- 17): 2903-2920. Deguchi, A., Hattori, S. and Park, H.-T., 2006. The influence of seasonal changes in canopy structure on interception loss: Application of the revised Gash model. Journal of Hydrology, 318(1-4): 80-102. Denault, C., Millar, R.G. and Lence, B.J., 2006. Assessment of possible impacts of climate change in an urban catchment. Journal of the American Water Resources Association, 42(3): 685-697. Domingo, F., S\u00C3\u00A1nchez, G., Moro, M.J., Brenner, A.J. and Puigdef\u00C3\u00A1bregas, J., 1998. Measurement and modelling of rainfall interception by three semi-arid canopies. Agricultural and Forest Meteorology, 91(3-4): 275-292. Dunkerley, D., 2000. Measuring interception loss and canopy storage in dryland vegetation: a brief review and evaluation of available research strategies. Hydrological Processes, 14: 669- 678. Dykes, A.P., 1997. Rainfall interception from a lowland tropical rainforest in Brunei. Journal of Hydrology, 200(1-4): 260-279. Environment Canada, 2007. Available on-line at: http://www.climate.weatheroffice.ec.gc.ca/climateData/canada_e.html (Accessed on April 27, 2008). Ford, E.D. and Deans, J.D., 1978. The effects of canopy structure on stemflow, throughfall and interception loss in a young Sitka Spruce plantation. Journal of Applied Ecology, 15(3): 905- 917. Gash, J.H.C., 1979. An analytical model of rainfall interception by forests. Quarterly Journal of the Royal Meteorological Society, 105(443): 43-55. Gash, J.H.C., Lloyd, C.R. and Lachaud, G., 1995. Estimating sparse forest rainfall interception with an analytical model. Journal of Hydrology, 170(1-4): 79-86. 79 Guevara-Escobar, A., Gonz\u00C3\u00A1lez-Sosa, E., V\u00C3\u00A9liz-Ch\u00C3\u00A1vez, C., Ventura-Ramos, E. and Ramos- Salinas, M., 2007. Rainfall interception and distribution patterns of gross precipitation around an isolated Ficus benjamina tree in an urban area. Journal of Hydrology, 333(2-4): 532-541. Harr, R.D., 1977. Water flux in soil and subsoil on a steep forested slope. Journal of Hydrology, 33: 37-58. Horton, R.E., 1919. Rainfall interception. Monthly Weather Review, 47(9): 603-623. Iroum\u00C3\u00A9, A. and Huber, A., 2002. Comparison of interception losses in a broadleaved native forest and a Pseudotsuga menziesii (Douglas fir) plantation in the Andes Mountains of southern Chile. Hydrological Processes, 16(12): 2347-2361. Jetten, V.G., 1996. Interception of tropical rain forest: performance of a canopy water balance model. Hydrological Processes, 10(5): 671-685. Keim, R.F., Skaugset, A.E. and Weiler, M., 2005. Temporal persistence of spatial patterns in throughfall. Journal of Hydrology, 314(1-4): 263-274. Keim, R.F., Skaugset, A.E. and Weiler, M., 2006. Storage of water on vegetation under simulated rainfall of varying intensity. Advances in Water Resources, 29(7): 974-986. Kelliher, F.M., Whitehead, D. and Pollock, D.S., 1992. Rainfall interception by trees and slash in a young Prinus Radiata D. Don Stand. Journal of Hydrology, 131: 187-204. Klaassen, W., Bosveld, F. and de Water, E., 1998. Water storage and evaporation as constituents of rainfall interception. Journal of Hydrology: 212\u00E2\u0080\u0093213, 36\u00E2\u0080\u009350. Lee, Y.-H. and Mahrt, L., 2005. Effect of stability on mixing in open canopies. Agricultural and Forest Meteorology, 135(1-4): 169-179. Leyton, L., Reynolds, E.R.C. and Thompson, F.B., 1967. Rainfall interception in forests and moorland. In: Sopper, W.E. and Lull, H.W. Editors, 1967. International Symposium on Forest Hydrology Pergamon Press, Oxford. Li, Y.C., Alva, A.K., Calvert, D.V. and Zhang, M., 1997. Stem flow, throughfall, and canopy interception of rainfall by citrus tree canopies. Hortscience, 32: 1059-1060. 80 Link, T.E., Unsworth, M. and Marks, D., 2004. The dynamics of rainfall interception by a seasonal temperate rainforest. Agricultural and Forest Meteorology, 124(3-4): 171-191. Llorens, P. and Domingo, F., 2007. Rainfall partitioning by vegetation under Mediterranean conditions. A review of studies in Europe. Journal of Hydrology, 335(1-2): 37-54. Llorens, P., Poch, R., Latron, J. and Gallart, F., 1997. Rainfall interception by a Pinus sylvestris forest patch overgrown in a Mediterranean mountainous abandoned area .1. Monitoring design and results down to the event scale. Journal of Hydrology, 199(3-4): 331-345. Lloyd, C.R., Gash, J.H.C. and Shuttleworth, W.J., 1988. The measurement and modelling of rainfall interception by Amazonian rain forest. Agricultural and Forest Meteorology, 43: 277- 294. Loustau, D., Berbigier, P., Granier, A. and El Hadj Moussa, F., 1992. Interception loss, throughfall and stemflow in a maritime pine stand. I. Variability of throughfall and stemflow beneath the pine canopy. Journal of Hydrology, 138: 151-163. Lundberg, A., Eriksson, M., Halldin, S., Kellner, E. and Seibert, J., 1997. New approach to the measurement of interception evaporation. Journal of Atmospheric and Oceanic Technology, 14(5): 1023-1035. Marsalek, J., 1981. Calibration of the tipping bucket rain gauge. Journal of Hydrology(53): 343- 354. Martens, S.N., Ustin, S.L. and Rousseau, R.A., 1993. Estimation of tree canopy Leaf-Area Index by gap fraction analysis. Forest Ecology and Management, 61(1-2): 91-108. Masukata, H., Ando, M. and Ogawa, H., 1990. Throughfall, stemflow and interception of rainwater in an evergreen broadleaved forest. Ecological Research, 5(3): 303-316. McJannet, D., Wallace, J. and Reddell, P., 2007a. Precipitation interception in Australian tropical rainforests: I. Measurement of stemflow, throughfall and cloud interception. Hydrological Processes, 21(13): 1692-1702. McJannet, D., Wallace, J. and Reddell, P., 2007b. Precipitation interception in Australian tropical rainforests: II. Altitudinal gradients of cloud interception, stemflow, throughfall and interception. Hydrological Processes, 21(13): 1703-1718. 81 McPherson, E.G., Nowak, D., Heisler, G., Grimmond, S., Souch, C., Grant, R. and Rowntree, R., 1997. Quantifying urban forest structure, function, and value: the Chicago urban forest climate project. Urban Ecosystems, 1(1): 49-61. McPherson, G., Simpson, J.R., Peper, P.J., Maco, S.E. and Xiao, Q.F., 2005. Municipal forest benefits and costs in five US cities. Journal of Forestry, 103(8): 411-416. Muzylo, A., Llorens, P., Valente, F., Keizer, J.J., Domingo, F. and Gash, J.H.C., 2009. A review of rainfall interception modelling. Journal of Hydrology, 370(1-4): 191-206. Nadkarni, N.M. and Sumera, M.M., 2004. Old-growth forest canopy structure and its relationship to throughfall interception. Forest Science, 50(3): 290-298. N\u00C3\u00A1var, J., Carlyle-Moses, D.E. and Martinez M.\u00C2\u00A7, A., 1999. Interception loss from the Tamaulipan matorral thornscrub of north-eastern Mexico: an application of the Gash analytical interception loss model. Journal of Arid Environments, 41: 1-10. Oke, T.R., Crowther, J.M., McNaughton, K.G., Monteith, J.L. and Gardiner, B., 1989. The micrometeorology of the urban forest. Biological Sciences, 324(1223): 335-349. Pierson, T.C., 1980. Piezometric response to rainstorms in forested hillslope drainage depressions. Journal of Hydrology (New Zealand), 19(1-10). Price, A.G. and Carlyle-Moses, D.E., 2003. Measurement and modelling of growing-season canopy water fluxes in a mature mixed deciduous forest stand, southern Ontario, Canada. Agricultural and Forest Meteorology, 119(1-2): 69-85. Pypker, T.G., Bond, B.J., Link, T.E., Marks, D. and Unsworth, M.H., 2005. The importance of canopy structure in controlling the interception loss of rainfall: Examples from a young and an old-growth Douglas-fir forest. Agricultural and Forest Meteorology, 130(1-2): 113-129. Rutter, A.J., Kershaw, K.A., Robins, P.C. and Morton, A.J., 1971. A predictive model of rainfall interception in forests, I. Derivation of the model from observations in a plantation of Corsican pine. Agricultural Meteorology, 9: 367-384. Rutter, A.J., Morton, A.J. and Robins, P.C., 1975. A predictive model of rainfall interception in forests, II. Generalization of the model and comparison with observations in some coniferous and hardwood stands. The Journal of Applied Ecology, 12(1): 367-380. 82 Sanders, R.A., 1986. Urban vegetation impacts on the hydrology of Dayton, Ohio. Urban Ecology, 9(3-4): 361-376. Taha, H., 1997. Urban climates and heat islands: Albedo, evapotranspiration, and anthropogenic heat. Energy and Buildings, 25(2): 99-103. Toba, T. and Ohta, T., 2005. An observational study of the factors that influence interception loss in boreal and temperate forests. Journal of Hydrology, 313(3-4): 208-220. Villarreal, E.L. and Bengtsson, A., 2004. Inner city stormwater control using a combination of best management practices. Ecological Engineering, 22(4-5): 279-298. Wang, J., Endreny, T.A. and Nowak, D.J., 2008. Mechanistic simulation of tree effects in an urban water balance model. Journal of the American Water Resources Association, 44(1): 75-85. Xiao, Q. and McPherson, E.G., 2002. Rainfall interception by Santa Monica's municipal urban forest. Urban Ecosystems, 6(4): 291-302. Xiao, Q.F., McPherson, E.G., Ustin, S.L. and Grismer, M.E., 2000. A new approach to modeling tree rainfall interception. Journal of Geophysical Research-Atmospheres, 105(D23): 29173- 29188. Xiao, Q.F., McPherson, E.G., Ustin, S.L. and Grismer, M.E., 2000a. A new approach to modeling tree rainfall interception. Journal of Geophysical Research-Atmospheres, 105(D23): 29173-29188. Xiao, Q.F., McPherson, E.G., Ustin, S.L., Grismer, M.E. and Simpson, J.R., 2000b. Winter rainfall interception by two mature open-grown trees in Davis, California. Hydrological Processes, 14(4): 763-784. Zipperer, W.C., Sisinni, S.M., Pouyat, R.V. and Foresman, T.W., 1997. Urban tree cover: an ecological perspective. Urban Ecosystems, 1: 229-246. 83 Appendix A Seasonal regression coefficients for the 54 trees Tag # Winter b0 b Winter b1 c Winter R2 Winter SEa F statistic d.f. P-value Summer b0 b Summer b1 c Summer R2 Summer SEa F statistic d.f. P-value 1 -4.86 0.85 0.97 3.50 2141 66 < 2.2e-16 -2.17 0.63 0.88 4.13 518.7 73 <2.2e-16 2 -2.67 0.95 0.99 2.72 5188 70 < 2.2e-16 -3.58 0.90 0.97 3.23 1634 59 <2.2e-16 3 -4.97 0.93 0.93 4.53 1067 75 < 2.2e-16 -5.15 0.85 0.95 3.45 1331 73 < 2.2e-16 4 -4.59 0.85 0.98 2.73 4212 76 < 2.2e-16 -2.88 0.70 0.94 3.05 1152 73 < 2.2e-16 5 -6.04 0.75 0.94 4.38 1273 76 < 2.2e-16 -3.58 0.64 0.87 4.00 450.9 66 < 2.2e-16 6 -7.59 0.82 0.95 4.68 1286 67 < 2.2e-16 -5.07 0.75 0.92 3.93 785.8 68 < 2.2e-16 7 -2.10 0.95 0.99 2.03 7840 71 < 2.2e-16 -3.46 0.91 0.98 2.08 4215 73 < 2.2e-16 8 -2.90 0.86 0.98 2.93 3140 56 < 2.2e-16 -2.75 0.51 0.88 3.32 494.5 65 < 2.2e-16 9 -3.12 0.93 0.98 2.98 4009 71 < 2.2e-16 -3.81 0.85 0.95 3.30 1463 73 < 2.2e-16 10 -3.37 0.94 0.99 2.49 5855 71 < 2.2e-16 -4.86 0.84 0.92 4.27 850.8 75 <2e-16 11 -9.63 0.72 0.80 8.65 289.9 71 < 2.2e-16 -1.99 0.21 0.35 4.96 45.84 84 <1.63e-09 12 -7.37 0.80 0.94 5.06 1039 71 < 2.2e-16 -2.17 0.42 0.60 5.84 113.5 76 < 2.2e-16 13 -6.82 0.72 0.93 4.89 889.6 71 < 2.2e-16 -3.30 0.53 0.80 4.20 329.4 80 < 2.2e-16 14 -8.87 0.87 0.98 4.13 976.4 18 < 2.2e-16 -1.78 0.39 0.93 2.15 666.3 50 < 2.2e-16 15 -3.40 0.64 0.97 2.83 2137 71 < 2.2e-16 -3.18 0.66 0.94 2.92 1272 84 < 2.2e-16 16 -3.43 0.55 0.93 7.37 107.7 8 6.43e-06 0.00 NA 0.00 NA - - - 17 -4.48 0.73 0.94 5.00 1136 78 < 2.2e-16 -5.83 0.70 0.81 8.41 363.4 85 < 2.2e-16 18 -3.43 0.79 0.95 3.64 792.8 38 < 2.2e-16 -5.24 0.81 0.95 4.89 1067 58 < 2.2e-16 19 -2.33 0.91 0.99 2.96 5978 63 < 2.2e-16 -5.48 0.89 0.93 6.26 896.6 63 < 2.2e-16 20 -3.05 0.96 0.98 2.18 2111 39 < 2.2e-16 -4.96 0.91 0.95 4.88 1529 75 < 2.2e-16 21 -3.91 0.96 0.98 2.63 3193 58 < 2.2e-16 -5.81 0.93 0.96 4.95 1441 61 < 2.2e-16 22 -3.86 0.92 0.99 3.35 4844 73 < 2.2e-16 -4.72 0.89 0.96 4.36 1971 85 < 2.2e-16 23 -3.40 0.99 0.99 2.36 4186 62 < 2.2e-16 -4.58 0.96 0.98 3.35 3725 70 < 2.2e-16 24 -4.28 0.95 0.98 3.22 4741 78 < 2.2e-16 -4.50 0.92 0.97 4.25 2076 73 < 2.2e-16 25 -2.10 0.94 0.99 1.53 3813 41 < 2.2e-16 -2.31 0.80 0.94 2.52 816.4 49 < 2.2e-16 26 -1.94 0.95 1.00 1.54 4557 13 < 2.2e-16 -2.02 0.85 0.98 1.72 1317 32 < 2.2e-16 27 -1.64 0.91 1.00 1.09 26410 37 < 2.2e-16 -3.54 0.92 0.99 2.97 4168 61 < 2e-16 28 -2.53 0.96 1.00 2.06 9475 33 < 2.2e-16 -2.51 0.83 0.96 2.55 1643 69 < 2.2e-16 100 -2.40 0.65 0.87 3.76 277.6 42 < 2.2e-16 -1.99 0.56 0.93 2.27 765.3 55 < 2.2e-16 200 -2.15 0.50 0.91 3.44 635.8 66 < 2.2e-16 -2.42 0.45 0.94 3.85 1061 68 < 2.2e-16 579 -4.86 0.78 0.96 3.65 1954 76 < 2.2e-16 -2.40 0.54 0.86 3.62 505 79 < 2.2e-16 580 -4.53 0.70 0.94 4.05 1296 76 < 2.2e-16 -1.92 0.42 0.84 3.09 444.9 83 < 2.2e-16 581 -4.08 0.93 0.99 2.76 4698 65 < 2.2e-16 -3.53 0.81 0.96 2.72 1155 43 < 2.2e-16 582 -4.49 0.80 0.96 3.74 1915 72 < 2.2e-16 -4.37 0.78 0.95 3.05 1032 57 < 2.2e-16 583 -4.66 0.61 0.86 6.36 362.6 59 < 2.2e-16 -1.20 0.34 0.78 3.27 244.1 67 < 2.2e-16 585 -2.54 0.51 0.69 5.13 127.1 57 3.87e-16 -3.55 0.50 0.54 6.26 43.4 37 1.01e-07 586 -6.41 0.82 0.93 6.41 1088 84 < 2.2e-16 -3.86 0.51 0.92 3.77 710.4 62 < 2.2e-16 587 -2.89 0.51 0.87 4.08 375.8 58 < 2.2e-16 -2.11 0.38 0.93 2.91 734.2 55 < 2.2e-16 588 -5.45 0.68 0.91 5.43 840.6 82 < 2.2e-16 -1.69 0.35 0.91 2.64 845.1 79 < 2.2e-16 590 -3.57 0.88 0.98 3.14 5206 84 < 2.2e-16 -3.67 0.84 0.97 3.87 2105 67 < 2.2e-16 591 -4.82 0.61 0.89 5.70 612.5 79 < 2.2e-16 -1.43 0.30 0.90 2.37 809.1 89 < 2.2e-16 592 -3.45 0.92 0.99 2.51 7548 67 < 2.2e-16 -5.02 0.80 0.94 6.48 1131 66 < 2.2e-16 593 -3.24 0.89 0.98 2.96 2888 54 < 2.2e-16 -4.35 0.88 0.97 4.54 3024 82 < 2.2e-16 594 -3.39 0.93 0.98 2.63 2350 48 < 2.2e-16 -4.11 0.87 0.99 3.68 3898 49 < 2.2e-16 595 -2.06 0.64 0.96 1.97 855.5 37 < 2.2e-16 -1.94 0.60 0.94 2.34 837.3 52 < 2.2e-16 596 -3.49 0.73 0.98 2.84 2014 51 < 2.2e-16 -2.10 0.57 0.94 2.44 1237 73 < 2.2e-16 597 -2.87 0.89 0.98 2.76 1953 48 < 2.2e-16 -5.20 0.90 0.98 4.58 2600 40 < 2.2e-16 598 -2.90 0.57 0.95 3.89 1230 70 < 2.2e-16 -2.35 0.41 0.92 4.04 828.2 70 < 2.2e-16 599 -3.07 0.55 0.91 4.78 749.3 77 < 2.2e-16 -4.18 0.60 0.95 4.53 1389 70 < 2.2e-16 1033 -3.10 0.91 0.98 2.87 3029 51 < 2.2e-16 -2.86 0.75 0.94 3.33 1082 66 < 2.2e-16 1408 -3.54 0.95 0.98 3.25 2786 58 < 2.2e-16 -4.48 0.91 0.98 4.64 2966 70 < 2.2e-16 84 Tag # Winter b0 b Winter b1 c Winter R2 Winter SEa F statistic d.f. P-value Summer b0 b Summer b1 c Summer R2 Summer SEa F statistic d.f. P-value 1409 -3.33 0.62 0.90 5.82 658.5 76 < 2.2e-16 -1.84 0.48 0.96 3.02 1991 74 < 2.2e-16 1411 -3.62 0.68 0.94 4.42 980.9 61 < 2.2e-16 -4.50 0.73 0.97 4.56 1960 65 < 2.2e-16 4607 -3.89 0.94 0.98 3.18 3631 60 < 2.2e-16 -5.59 0.77 0.92 7.57 844.1 76 < 2.2e-16 aSE = Standard Error b b0 = Intercept c b1 = Slope "@en . "Thesis/Dissertation"@en . "2010-05"@en . "10.14288/1.0069783"@en . "eng"@en . "Forestry"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "Attribution-NonCommercial-NoDerivatives 4.0 International"@en . "http://creativecommons.org/licenses/by-nc-nd/4.0/"@en . "Graduate"@en . "Rainfall interception in an urban environment"@en . "Text"@en . "http://hdl.handle.net/2429/23517"@en .