A multi-scale analysis of vegetation and irrigation heterogeneity effects on ecohydrological function and ecosystem services in a semi-arid urban area

An ever-increasing proportion of humanity resides in cities, yet the factors driving resource use and shaping urban sustainability remain poorly understood. In arid and semi-arid climates, water is a limited and costly resource, and future supplies are threatened by increases in population and climate change. A majority of summertime household water use in many western cities goes to irrigating vegetation, and this can be viewed as a significant cost or, from an ecosystem services perspective, "disservice". However, urban vegetation also provides a wide variety of important provisioning and regulating ecosystem services. While the general value of green spaces to ecological services is widely recognized, the importance of specific structural and compositional characteristics on ecosystem services and disservices is unknown. This dissertation research focused on three broad objectives: (1) quantifying the compositional and structural variation in urban vegetation and characterizing consequences for water use patterns; (2) evaluating heterogeneity in irrigation practices and its relationship to water use by the two most common classes of urban vegetation, turfgrass and trees; and (3) quantifying a key ecosystem service, land surface temperature (LST) amelioration, in relation to vegetation composition, structure, and residential water use patterns. Because scale is a critical consideration in evaluating ecohydrological patterns and processes, I conducted my analyses at a range of spatial scales, from that of individual city parks to the city of Aurora, Colorado (> 200 km2). To develop high resolution land cover data essential for subsequent analyses, I compared the accuracy of different remote sensing classification approaches for mapping urban land cover (LC) and structure. Specifically, I compared classification accuracy of LC maps derived from lidar and 4-band multispectral data using three different approaches: (1) an object-oriented segmentation (OBIA) and Random Forest classification approach; (2) a pixel wise classification using Random Forests; and (3) a traditional pixel-wise maximum likelihood classification. I mapped six classes: trees, buildings, low-vegetation, low-impervious, bare soil, and water. To inform improved urban water conservation and planning, I evaluated spatial patterns and correlative relationships among physical land cover properties, socioeconomic and demographic characteristics, and single-family outdoor residential water use. Using the high resolution land cover maps and lidar-derived vertical structural data I developed, land cover composition and vertical structural characteristics for over 45,000 single-family detached residential parcels was analyzed. These data were combined with socioeconomic and demographic datasets from the 2010 US Census and local government agencies, and used in Random Forest regression analyses of outdoor water use from residential water meter records, with separate analyses conducted using parcels and census block groups as sampling units. Water use can be viewed as an ecological disservice—a cost incurred to maintain irrigated urban vegetation—but assessments of cost and benefit should consider a wider suite of ecosystem services. One such service provided by irrigated vegetation is amelioration of the urban heat island formation through moderation of land surface temperature (LST). Using land cover maps and lidar-derived vertical structural data (e.g., tree and building height), I evaluated the relative importance of land cover compositional and vertical structural variables in predicting LST derived from Landsat 5 TM thermal band data. After aggregating data using 2010 census blocks, I analyzed LST using the Random Forest machine learning algorithm. Finally, I evaluated water use by irrigated Poa pratensis turf and several common urban tree species in five city parks and recreational areas in Aurora. Two separate approaches were used to assess turf water use. Drainage lysimeters were installed in each study site and monitored to yield monthly and seasonal estimates of turf ET. Secondly, I used an infrared gas analyzer to quantify instantaneous ET from a small chamber sampled along a gradient of irrigation application and soil moisture availability. (Abstract shortened by UMI.