The current study was undertaken to improve the understanding of the long-term impacts of urbanization on hydrologic behavior and water supply in semi-arid regions. The study focuses on the Upper Santa Clara River basin in northern Los Angeles County which is undergoing rapid and extensive development. The Hydrologic Simulation Program- Fortran (HSPF) model is parameterized with land use, soil, and channel characteristics of the study watershed. Model parameters related to hydrologic processes are calibrated at the daily timestep using various spatial configurations of precipitation and parameters. Results indicate that the HSPF performs best with distributed precipitation forcing and parameters (distributed scenario), however the model performs fairly well under all scenarios. The model also shows slightly better performance during wetter seasons and years than during drier periods. Potential urbanization scenarios are generated on the basis of a regional development plan. The calibrated (and validated) model is run under the proposed development scenarios for a ten year period. Results reveal that increasing development increases total annual runoff and wet season flows, while decreases are observed in baseflow and groundwater recharge during both dry and wet seasons. As development increases, medium sized storms increase in both peak flow and overall volume, while low and high flow events (extremes) appear less affected. Urbanization is also shown to decrease recharge and, when considered at the regional-scale, could potentially result in a loss of critical water supply to southern California.

Urbanization is a demographic trend worldwide. It is found that half of all cities with populations greater than 100,000 are located in water stressed watersheds and are heavily dependent on imported water. Urban irrigation can exceed natural precipitation in these cities and is an important component of the water cycle. For instance, 14-30% of municipal water consumption in California is used for irrigation. Hence, understanding and quantifying the potential influence of urban anthropogenic soil moisture contribution on local and regional hydrological cycle is an imperative step toward sustainable and better managed water resources in water scarce regions.

The first part of current work examines the influence of irrigation on urban hydrological cycles through development of an irrigation scheme within the Noah Land Surface Model (LSM) - Urban Canopy Model (UCM) system. The model is run at a 30-m resolution for a two year period over a study domain in the Los Angeles metropolitan area. A sensitivity analysis indicates significant sensitivity relative to both the amount and timing of irrigation on diurnal and monthly energy budgets, hydrological fluxes and state variables. Monthly residential water use data and three estimates of outdoor water consumption are used to calibrate the developed irrigation scheme. Model performance is evaluated using a previously developed MODIS-(Moderate Resolution Imaging Spectroradiometer)-Landsat evapotranspiration (ET) and Landsat Land Surface Temperature (LST) products as well as hourly ET observations through the California Irrigation Management Information System (CIMIS). Results show that the Noah LSM-UCM realistically simulates the diurnal and seasonal variations of ET when the irrigation module is incorporated. However, without irrigation, the model produces large biases in ET simulations. The ET errors for the non-irrigation simulation are -56 and -90 mm/month for July 2003 and 2004, respectively, while these values reduce to -6 and -11 mm/month over the same two months when the proposed irrigation scheme is adopted. Results also show that the irrigation-induced increase in latent heat flux leads to a decrease in LST of about 2 °C in urban parks. The developed modeling framework can be utilized for a number of applications, ranging from outdoor water use estimation to climate change impact assessments.

In the second part of this work we investigate the utility of remote sensing based surface parameters in the Noah-UCM for a more accurate representation of developed surfaces in this modeling framework. Landsat and fused Landsat-MODIS data are utilized to generate high resolution (30 m) monthly spatial maps of green vegetation fraction (GVF), impervious surface area (ISA), albedo, leaf area index (LAI), and emissivity in the Los Angeles metropolitan area. The gridded remotely sensed parameter datasets are directly substituted for the land-use/lookup table based values in the Noah-UCM modeling framework. Model performance in reproducing ET and LST fields is evaluated utilizing Landsat based LST and ET estimates from CIMIS stations as well as in-situ measurements. Our assessment shows that the large deviations between the spatial distributions and seasonal fluctuations of the default and measured parameter sets lead to significant errors in the model predictions of monthly ET fields (RMSE= 22.06 mm/month). Results indicate that implemented satellite derived parameter maps, particularly GVF, enhance the Noah-UCM capability to reproduce observed ET patterns over vegetated areas in the urban domains (RMSE= 11.77 mm/month). GVF plays the most significant role in reproducing the observed ET fields, likely due to the interaction with other parameters in the model. Our analysis also shows that remotely sensed GVF and ISA improve the model capability to predict the LST differences between fully vegetated pixels and highly developed areas.

In the third part we explicitly address the impacts of urban irrigation by integrating the developed irrigation scheme within the coupled framework of the WRF-UCM over the semi-arid Los Angeles metropolitan area. We focus on the impacts of irrigation on the urban water cycle and atmospheric feedback in arid and semi-arid cities. Our objective is to build upon previous work, focusing on improving the representation of irrigated urban vegetated in the numerical weather prediction models which are now standard tools to study urban-atmosphere interactions. Our results demonstrate a significant sensitivity of WRF-UCM simulated surface turbulent fluxes to the incorporation of irrigation. Introducing anthropogenic moisture, the vegetated pixels show increased latent heat fluxes and decreased sensible and ground heat fluxes confirming irrigation induced shift in the energy partitioning toward elevated latent heat fluxes. The evaluation of the model performance via comparison against CIMIS based reference ET indicates that WRF-UCM, after adding irrigation, performs reasonably during the course of the month, tracking day to day variability of ET with notable fidelity. In the absence of irrigation, simulated ET fluctuations are similar to the CIMIS based ET and irrigated case, in the first few days of simulations. Toward the end of the month, however, the differences between simulated ET and CIMIS based ET0 become more significant in the not irrigated case. This is due to fact that, in the not irrigated simulation, the soil moisture is the only source of water in the absence of irrigation and significant precipitation. In the course of simulation, the soil moisture sorted in the soil layers is consumed, resulting in considerable decreases in the soil moisture levels in all layers. The soil moisture depletion leads to reduced latent heating and cooling effects of urban vegetation. Analysis of these results indicates the importance of accurate representation of urban irrigation in water scarce regions such as Los Angeles metropolitan area. Moreover, it is found that the initial soil moisture level plays a principal role in disguising the real impacts of irrigation in urban domains and to see the actual impacts of urban irrigation the simulations should be conducted for a longer period of time than one month which is rare in WRF-UCM studies.

This research contributes to a better understanding of the dynamics of single-family water consumption associated with vegetation in semi-arid cities. The innovative research approach couples long-term water consumption data with remote-sensing based products, socio-demographic, land cover, landscaping and climate data analyzed with multidisciplinary techniques. Accurate water demand forecasting and long-term conservation planning are required to meet future urban water needs relying on a set of integrated sustainable resources. In the face of climate change and urbanization growth, quantifying and predicting ecosystems costs related to irrigation and benefits are important and challenging. To address these needs, the first part of this study focuses on analyzing trends and determinants in single-family water use in Los Angeles indicating that the current water rate structure can be optimized to achieve higher water savings: Tier 2 water rates do not lead to more conservation behaviors, though they were implemented to do so. It also shows that residential landscape is primarily maintained by irrigation and is not correlated with the seasonal precipitation pattern. The analysis of the effectiveness of water restrictions during the last drought in Los Angeles confirms that there is still room for outdoor water conservation: there was no significant change in the landscape greenness during the implementation of the watering restrictions. Conservation programs focusing on efficient watering practices can lead to higher water savings while supporting healthy landscapes. Increased mandatory restrictions were more effective than voluntary restrictions during the drought period reaching a maximum of 23% water reduction in July-August 2009 in City-average single-family household water use. This suggests that outdoor water use savings may result from long-term conservation programs. This leads to the last part of this research work focusing on quantifying and predicting landscaping irrigation through the development of predictive models. The predictive regression model explores the relationship between single-family water use and vegetation greenness surplus estimated through remote-sensing vegetation indices that can be used as a predictive tool to target outdoor conservation measures. This project was integrated within a coupled socio-ecohydrological approach to address inter-disciplinary issues related to the urban landscape. It encourages the use of advanced models and remote-sensing products incorporated into policies that improve outdoor use modeling and predictions to guide future water demand management strategies under uncertain climate variability.

Wildfires alter land surfaces and land-atmosphere interactions, causing enhanced runoff and debris flows. The current study evaluates hydrologic behavior and recovery for three watersheds in the burned in the 2010 Bull Fire in the southern Sequoia National Forest. One unburned watershed was selected outside the fire perimeter for a control. The effects of wildfires have been extensively analyzed, but these studies typically focus on debris flows immediately following the fire and vegetation recovery on a plot-scale. This study attempts to evaluate hydrologic recovery through in-situ instruments in areas analyzed by a United States Forest Service Burned Area Emergency Response team. Many agencies attempt to predict post-fire runoff peaks and volumes to identify values-at-risk and determine if protective measures must be implemented. The four study sites are instrumented with tipping buckets and pressure transducers to measure precipitation and discharge throughout the approximate two year study period. Stream discharge is measured at five-minute intervals, the tipping buckets are event-based to track storm duration and intensity, and channel cross-sections are measured every two months to detail sediment deposition or scour that accompany rainfall-runoff events. Precipitation was found to be consistent between the study areas, but a link between precipitation and cross-section area changes was not determined. Geomorphic parameters were evaluated with an area-normalized discharge to determine correlations. Aspect, soil type, and watershed shape were found to be the controlling factors in elevated discharge. To judge watershed recovery, the response lag, lag-to-peak, and runoff ratio were calculated for 13 selected storms across all study sites. Variability of each parameter was determined to depend on the precipitation tipping bucket location and the time between storms. Recovery was unable to be determined based on the three rainfall-runoff hydrograph parameters. Future work involves relating a remote sensing vegetation index with the storm parameters to couple vegetation recovery with hydrologic recovery and gain insight into watershed recovery on a storm-by-storm basis.

This work has investigated the policy of wildfires, modeling techniques for post-fire assessment, and the influence of controlling variables on post-fire recovery. Post-fire mitigation and management require reliable predictions of immediate hydrologic consequences and long-term recovery to pre-fire conditions. This research shows that models used by agencies are not adaptable to all geographical and climatological conditions. Results show inconsistencies between model predictions for peak discharge events across the sites and less confidence associated with larger return periods (25- and 50-year peak flow events). Remote sensing techniques improve spatial and temporal resolution of data streams for model parameters and post-fire recovery predictions. This research shows that recovery is dependent on many variables, including burn severity, slope aspect, and vegetation biomass. The lack of vegetation recovery across watersheds results in significant changes in annual and seasonal discharge throughout the study period. Understanding these key controlling variables will improve post-fire hydrological predictions. Previously established remote sensing algorithms can be applied and adapted to burned areas to improve hydrologic and recovery predictions. This work encourages new tools that can be incorporated into policies that minimize development at the WUI, improve homeowner preparation in fire-prone areas, and improve post-fire recovery predictions. This work improves post-fire modeling and predictions primarily with remote sensing applications to guide accurate, efficient, and cost-effective management decisions.

Temperature rise, caused by urbanization, increases electricity demand, adversely affects human health, and alters hydrological and ecological patterns. Urban heat islands (UHI) are affected by both complex urban cover as well as surrounding environments. This study describes the variability of temperature and relative humidity as it is (1) influenced by local vegetation and land cover type over neighborhoods and (2) moderated by coastal climate across the urban gradient. Data were collected at four different times of the day during three different weather conditions using mobile transects over four urban neighborhoods in Southern California: Venice, Mid-Wilshire, Pomona, and Riverside. The four neighborhoods also form a climate gradient extending from the Venice (coast) to Riverside (99 miles inland). The influence of urban land cover on local climate proved dependent on the time of day and general atmospheric conditions (sunny, cloudy, etc). Temperature was statistically related to vegetation during cooler or early morning measurements, but less so during midday and extreme heat conditions. Inland neighborhood temperatures and relative humidity showed an increased dependence on vegetation from coastal areas because of the lack of other moisture sources. This study serves as an initial baseline for continued study of the urbanization of coastal and inland areas and relationship to local land cover and vegetation density.

Southern California's fire regime is characterized by periodic large-scale wildfires that occur in late summer and early fall, leaving little time for the landscape to recover before the winter rainy season. Recent trends suggest that fire frequency and intensity is increasing due to factors such as urban encroachment into wildlands and a changing climate. While there is an awareness of the hydrologic consequences of these fires, wildfire's potential effects on water quality in Southern California are less understood, especially at the urban-fringe. This study investigates the impacts of wildfire on regional soils with respect to mercury (Hg) storage, accumulation, and transport potential; on storm runoff chemistry in an urban fringe watershed highly impacted by regional atmospheric pollutants; and on procedures required to model post fire hydrologic response and predict contaminant loading. The regional soil work showed reduced influence of grain size on Hg storage and accelerated accumulation of Hg in burned soil surfaces over time, as well as sediment-facilitated Hg transport in post-fire storms. Analysis of post-fire runoff from an urban fringe watershed showed up to three orders of magnitude increases from pre-fire levels in the concentrations and loads of many trace metals, including cadmium (Cd) and lead (Pb). These results also highlighted a shift in the timing of chemical delivery in post-fire storms to coincide with, rather than precede, peak discharge, amplifying the fire's impacts on mass loading. The investigation into modeling these impacts allowed for initial estimates to be made of seasonal loading, the development of a model to simulate pre-fire hydrology, and an identification of steps required to develop a model application capable of simulating post-fire sediment and metal concentrations. The resultant body of work provides an increased understanding of wildfire's effects on Southern California's soils and water and on what is required to predict these impacts.

The current work focuses on assessing the impacts of future climate variability on water resources in southern California. Specifically, this dissertation work includes: (1) developing archetypal watersheds and climate scenarios to obtain regional changes to hydrology and sediment transport and (2) developing a statistical downscaling approach that considers regional climate heterogeneity (commonly neglected in downscaling methods) and using this data to drive hydrologic models. The archetypal or "representative" watersheds exemplify observed physiological features and allow us to model hydrologic trends of coastal watersheds in southern California. Future climate scenarios were developed using historical observations [1955 - 2006] and used as input to the Environmental Protection Agency's Hydrologic Simulation Program-Fortran (EPA HSPF). In the statistical downscaling approach, the CNRM-CM3 GCM model was used to develop daily precipitation and temperature. A k-means clustering analysis was utilized and.extensive testing of predictor-predictand relationships was performed to select optimal monthly predictors. Control, no-clustering method, and clustering approaches, based on mean/total annual temperature/precipitation, annual variance and elevation, were performed for daily temperature and precipitation. The developed downscaling approaches were applied to extreme future climate scenarios A2 (high carbon dioxide emission) and B1 (low emission) and change in hydrologic fluxes for the Ballona Creek Watershed were investigated.

Results from the archetypal framework indicate that precipitation variability is the primary variable in determining the magnitude of change in sediment and hydrologic fluxes. Highly vegetated systems, characterized by low annual flows typical of those found in Santa Barbara County, are expected to experience a significant loss of total annual flow and sediment flux due to rising temperatures and precipitation uncertainty. Highly urbanized (typical in Los Angeles) and moderately urbanized (typical in San Diego) watersheds, are expected to experience a significant change to storm dynamics (peak flow and storm sediments) due to climate change. The downscaling investigation shows that the optimal clustering methods to reconstruct temperature and precipitation were elevation and precipitation variance, respectively. Using the high (A2) and low (B1) emission scenarios, precipitation occurrence and quantity will increase throughout the year. As a result, hydrologic fluxes are expected to increase significantly in Ballona Creek, especially during dry periods. The development of the archetypal framework allows for a broad perspective of how future climate variability and regional land use patterns may influence hydrologic and sediment fluxes; changes in these fluxes may have significant implications for restoration and management of coastal wetlands, bays, and harbors. The statistical downscaling method captures daily temperature well, but further efforts may improve daily precipitation reconstructions. Results from this work have significant application in projects involving ecosystem impacts and regional sustainability studies.

Potential Evapotranspiration (PET) is used as a common input to calculate evaporative demand in hydrological, ecological and biological modeling. Dynamic and distributed measurement of PET is important for improved hydrologic predictions at the watershed scale since PET varies with time and space. In this work, an advanced dynamic PET estimation is proposed by integrating geostationary satellite products into a currently existing remote sensing-based PET algorithm and evaluated in the framework of operational hydrologic forecasting modeling. The development work is approached through a series of studies. At first, a previously developed Moderate Resolution Imaging Spectroradiometer (MODIS) based PET (MODIS-PET) product applied over several flux towers and basins in the Upper Colorado River Basin (UCRB) to determine its applicability and predictive ability in comparison to other ground based distributed PET methods. Results from this primary study indicate the MODIS-PET is an improved PET estimation method compared to the other two contemporary distributed PET products that were tested over this geographically complex study region. In addition to elevation and cloud cover, uncertainties are associated with the MODIS-PET algorithm pertaining from three model variables; land surface temperature, air temperature and surface emissivity. The crude hypothetical sinusoidal curve considered in the conversion of instantaneous MODIS-PET to the daily PET estimation can potentially be replaced with satellite data with improved temporal resolution. Hence, integration of Geostationary Operational Environmental Satellites (GOES), a series of geostationary satellites with frequent observations, data in the MODIS-PET algorithm is performed in the second part. The coupling of GOES within the MODIS-PET algorithm shows significant improvement over the previously developed stand-alone MODIS-PET product, especially for cloudy days and high temperature pixels. Finally, evaluation of these two remote sensing products (merged GOES-MODIS and stand-alone MODIS) is undertaken as lumped input in the National Weather Service (NWS) River Forecasting Centers (RFC) operational forecasting models in two mountainous watersheds from the UCRB. The preliminary results show that PETs estimated using the satellite data is a suitable replacement of the static PET in the snow-dominated basins of the UCRB region. The new remote-sensing PET available in near-real-time which we advocate will ultimately provide more reasonable representation of current climatological conditions for streamflow forecasting, drought monitoring and crop water demand.