Rapid urban expansion and the associated landscape modifications have led to significant changes of surface processes in built environments. These changes further interact with the overlying atmospheric boundary layer and strongly modulate urban microclimate. To capture the impacts of urban land surface processes on urban boundary layer dynamics, a coupled urban land-atmospheric modeling framework has been developed. The urban land surface is parameterized by an advanced single-layer urban canopy model (SLUCM) with realistic representations of urban green infrastructures such as lawn, tree, and green roof, etc. The urban atmospheric boundary layer is simulated by a single column model (SCM) with both convective and stable schemes. This coupled SLUCM-SCM framework can simulate the time evolution and vertical profile of different meteorological variables such as virtual potential temperature, specific humidity and carbon dioxide concentration. The coupled framework has been calibrated and validated in the metropolitan Phoenix area, Arizona. To quantify the model sensitivity, an advanced stochastic approach based on Markov-Chain Monte Carlo procedure has been applied. It is found that the development of urban boundary layer is highly sensitive to surface characteristics of built terrains, including urban land use, geometry, roughness of momentum, and vegetation fraction. In particular, different types of urban vegetation (mesic/xeric) affect the boundary layer dynamics through different mechanisms. Furthermore, this framework can be implanted into large-scale models such as Weather Research and Forecasting model to assess the impact of urbanization on regional climate.

Land surface fluxes of energy and mass developed over heterogeneous mountain landscapes are fundamental to atmospheric processes. However, due to their high complexity and the lack of spatial observations, land surface processes and land-atmosphere interactions are not fully understood in mountain regions. This thesis investigates land surface processes and their impact on convective precipitation by conducting numerical modeling experiments at multiple scales over the North American Monsoon (NAM) region. Specifically, the following scientific questions are addressed: (1) how do land surface conditions evolve during the monsoon season, and what are their main controls?, (2) how do the diurnal cycles of surface energy fluxes vary during the monsoon season for the major ecosystems?, and (3) what are the impacts of surface soil moisture and vegetation condition on convective precipitation?

Hydrologic simulation using the TIN-based Real-time Integrated Basin Simulator (tRIBS) is firstly carried out to examine the seasonal evolution of land surface conditions. Results reveal that the spatial heterogeneity of land surface temperature and soil moisture increases dramatically with the onset of monsoon, which is related to seasonal changes in topographic and vegetation controls. Similar results are found at regional basin scale using the uncoupled WRF-Hydro model. Meanwhile, the diurnal cycles of surface energy fluxes show large variation between the major ecosystems. Differences in both the peak magnitude and peak timing of plant transpiration induce mesoscale heterogeneity in land surface conditions. Lastly, this dissertation examines the upscale effect of land surface heterogeneity on atmospheric condition through fully-coupled WRF-Hydro simulations. A series of process-based experiments were conducted to identify the pathways of soil moisture-rainfall feedback mechanism over the NAM region. While modeling experiments confirm the existence of positive soil moisture/vegetation-rainfall feedback, their exact pathways are slightly different. Interactions between soil moisture, vegetation cover, and rainfall through a series of land surface and atmospheric boundary layer processes highlight the strong land-atmosphere coupling in the NAM region, and have important implications on convective rainfall prediction. Overall, this dissertation advances the study of complex land surface processes over the NAM region, and made important contributions in linking complex hydrologic, ecologic and atmospheric processes through numerical modeling.

Biological soil crusts (BSCs) dominate the soil surface of drylands in the western United States and possess properties thought to influence local hydrology. Little agreement exists, however, on the effects of BSCs on runoff, infiltration, and evaporative rates. This study aims to improve the predictive capability of an ecohydrology model in order to understand how BSCs affect the storage, retention, and infiltration of water into soils characteristic of the Colorado Plateau. A set of soil moisture measurements obtained at a climate manipulation experiment near Moab, Utah, are used for model development and testing. Over five years, different rainfall treatments over experimental plots resulted in the development of BSC cover with different properties that influence soil moisture differently. This study used numerical simulations to isolate the relative roles of different BSC properties on the hydrologic response at the plot-scale. On-site meteorological, soil texture and vegetation property datasets are utilized as inputs into a ecohydrology model, modified to include local processes: (1) temperature-dependent precipitation partitioning, snow accumulation and melt, (2) seasonally-variable potential evapotranspiration, (3) plant species-specific transpiration factors, and (4) a new module to account for the water balance of the BSC. Soil, BSC and vegetation parameters were determined from field measurements or through model calibration to the soil moisture observations using the Shuffled Complex Evolution algorithm. Model performance is assessed against five years of soil moisture measurements at each experimental site, representing a wide range of crust cover properties. Simulation experiments were then carried out using the calibrated ecohydrology model in which BSC parameters were varied according to the level of development of the BSC, as represented by the BSC roughness. These results indicate that BSCs act to both buffer against evaporative soil moisture losses by enhancing BSC moisture evaporation and significantly alter the rates of soil water infiltration by reducing moisture storage and increasing conductivity in the BSC. The simulation results for soil water infiltration, storage and retention across a wide range of meteorological events help explain the conflicting hydrologic outcomes present in the literature on BSCs. In addition, identifying how BSCs mediate infiltration and evaporation processes has implications for dryland ecosystem function in the western United States.

Ecohydrological responses to rainfall in the North American monsoon (NAM) region lead to complex surface-atmosphere interactions. In early summer, it is expected that soil properties and topography act as primary controls in hydrologic processes. Under the presence of strongly dynamic ecosystems, catchment hydrology is expected to vary substantially in comparison to other semiarid areas, affecting our understanding of ecohydrological processes and the parameterization of predictive models. A large impediment toward making progress in this field is the lack of spatially extensive observational data. As a result, it is critical to integrate numerical models, remote sensing observations and ground data to understand and predict ecohydrological dynamics in space and time, including soil moisture, evapotranspiration and runoff generation dynamics. In this thesis, a set of novel ecohydrological simulations that integrate remote sensing and ground observations were conducted at three spatial scales in a semiarid river basin in northern Sonora, Mexico. First, single site simulations spanning several summers were carried out in two contrasting mountain ecosystems to predict evapotranspiration partitioning. Second, a catchment-scale simulation was conducted to evaluate the effects of spatially-variable soil thickness and textural properties on water fluxes and states during one monsoon season. Finally, a river basin modeling effort spanning seven years was applied to understand interannual variability in ecohydrological dynamics. Results indicated that ecohydrological simulations with a dynamic representation of vegetation greening tracked well the seasonal evolution of observed evapotranspiration and soil moisture at two measurement locations. A switch in the dominant component of evapotranspiration from soil evaporation to plant transpiration was observed for each ecosystem, depending on the timing and magnitude of vegetation greening. Furthermore, spatially variable soil thickness affects subsurface flow while soil texture controls patterns of surface soil moisture and evapotranspiration during the transition from dry to wet conditions. Finally, the ratio of transformation of precipitation into evapotranspiration (ET/P) and run off (Q/P) changed in space and time as summer monsoon progresses. The results of this research improve the understanding of the ecohydrology of NAM region, which can be useful for developing sustainable watershed management plans in the face of anticipated land cover and climate changes.

Climate and its influence on hydrology and weathering is a key driver of surface processes on Earth. Despite its clear importance to hazard generation, fluvial sediment transport and erosion, the drawdown of atmospheric CO₂ via the rock cycle, and feedbacks between climate and tectonics, quantifying climatic controls on long-term erosion rates has proven to be one of the grand problems in geomorphology. In fact, recent attempts addressing this problem using cosmogenic radionuclide (CRN) derived erosion rates suggest very weak climatic controls on millennial-scale erosion rates contrary to expectations. In this work, two challenges are addressed that may be impeding progress on this problem.

The first challenge is choosing appropriate climate metrics that are closely tied to erosional processes. For example, in fluvial landscapes, most runoff events do little to no geomorphic work due to erosion thresholds, and event-scale variability dictates how frequently these thresholds are exceeded. By analyzing dense hydroclimatic datasets in the contiguous U.S. and Puerto Rico, we show that event-scale runoff variability is only loosely related to event-scale rainfall variability. Instead, aridity and fractional evapotranspiration (ET) losses are much better predictors of runoff variability. Importantly, simple hillslope-scale soil water balance models capture major aspects of the observed relation between runoff variability and fractional ET losses. Together, these results point to the role of vegetation water use as a potential key to relating mean hydrologic partitioning with runoff variability.

The second challenge is that long-term erosion rates are expected to balance rock uplift rates as landscapes approach topographic steady state, regardless of hydroclimatic setting. This is illustrated with new data along the Main Gulf Escarpment, Baja, Mexico. Under this conceptual framework, climate is not expected to set the erosion rate, but rather the erosional efficiency of the system, or the steady-state relief required for erosion to keep up with tectonically driven uplift rates. To assess differences in erosional efficiency across landscapes experiencing different climatic regimes, we contrast new CRN data from tectonically active landscapes in Baja, Mexico and southern California (arid) with northern Honduras (very humid) alongside other published global data from similar hydroclimatic settings. This analysis shows how climate does, in fact, set functional relationships between topographic metrics like channel steepness and long-term erosion rates. However, we also show that relatively small differences in rock erodibility and incision thresholds can easily overprint hydroclimatic controls on erosional efficiency motivating the need for more field based constraints on these important variables.

Land-atmosphere interactions of semiarid shrublands have garnered significant scientific interest. One of the main tools used for this research is the eddy covariance (EC) method, which measures fluxes of energy, water vapor, and carbon dioxide. EC fluxes can be difficult to interpret due to complexities within the EC footprint (i.e. the surface conditions that contribute to the flux measurements). Most EC studies use a small number of soil probes to estimate the land surface states underlying the measured fluxes, which likely undersamples the footprint-scale conditions, especially in semiarid shrublands which are characterized by high spatial and temporal variability. In this study, I installed a dense network of soil moisture and temperature probe profiles in the footprint region of an EC tower at two semiarid sites: a woody savanna in southern Arizona and a mixed shrubland in southern New Mexico. For data from May to September 2013, I link land surface states to EC fluxes through daily footprints estimated using an analytical model. Novel approaches are utilized to partition evapotranspiration, estimate EC footprint soil states, connect differences in fluxes to footprint composition, and assess key drivers behind soil state variability. I verify the hypothesis that a small number of soil probes poorly estimates the footprint conditions for soil moisture, due to its high spatial variability. Soil temperature, however, behaves more consistently in time and space. As such, distributed surface measurements within the EC footprint allow for stronger ties between evapotranspiration and moisture, but demonstrate no significant improvement in connecting sensible heat flux and temperature. I also find that in these systems vegetation cover appears to have stronger controls on soil moisture and temperature than does soil texture. Further, I explore the influence of footprint vegetation composition on the measured fluxes, which reveals that during the monsoon season evaporative fraction tends to increase with footprint bare soil coverage for the New Mexico site and that the ratio of daily transpiration to evapotranspiration increases with grass coverage at the Arizona site. The thesis results are useful for understanding the land-atmosphere interactions of these ecosystems and for guiding future EC studies in heterogeneous landscapes.

The history of outdoor water use in the Phoenix, Arizona metropolitan area has given rise to a general landscape aesthetic and pattern of residential irrigation that seem in discord with the natural desert environment. While xeric landscaping that incorporates native desert ecology has potential for reducing urban irrigation demand, there are societal and environmental factors that make mesic landscaping, including shade trees and grass lawns, a common choice for residential yards. In either case, there is potential for water savings through irrigation schedules based on fluxes affecting soil moisture in the active plant rooting zone. In this thesis, a point-scale model of soil moisture dynamics was applied to two urban sites in the Phoenix area: one with xeric landscaping, and one with mesic. The model was calibrated to observed soil moisture data from irrigated and non-irrigated sensors, with local daily precipitation and potential evapotranspiration records as model forcing. Simulations were then conducted to investigate effects of irrigation scheduling, plant stress parameters, and precipitation variability on soil moisture dynamics, water balance partitioning, and plant water stress. Results indicated a substantial difference in soil water storage capacity at the two sites, which affected sensitivity to irrigation scenarios. Seasonal variation was critical in avoiding unproductive water losses at the xeric site, and allowed for small water savings at the mesic site by maintaining mild levels of plant stress. The model was also used to determine minimum annual irrigation required to achieve specified levels of plant stress at each site using long-term meteorological records. While the xeric site showed greater potential for water savings, a bimodal schedule consisting of low winter and summer irrigation was identified as a means to conserve water at both sites, with moderate levels of plant water stress. For lower stress levels, potential water savings were found by fixing irrigation depth and seasonally varying the irrigation interval, consistent with municipal recommendations in the Phoenix metropolitan area. These results provide a deeper understanding of the ecohydrologic differences between the two types of landscape treatments, and can assist water and landscape managers in identifying opportunities for water savings in desert urban areas.

Woody plant encroachment is a worldwide phenomenon linked to water availability in semiarid systems. Nevertheless, the implications of woody plant encroachment on the hydrologic cycle are poorly understood, especially at the catchment scale. This study takes place in a pair of small semiarid rangeland undergoing the encroachment of Prosopis velutina Woot., or velvet mesquite tree. The similarly-sized basins are in close proximity, leading to equivalent meteorological and soil conditions. One basin was treated for mesquite in 1974, while the other represents the encroachment process. A sensor network was installed to measure ecohydrological states and fluxes, including precipitation, runoff, soil moisture and evapotranspiration. Observations from June 1, 2011 through September 30, 2012 are presented to describe the seasonality and spatial variability of ecohydrological conditions during the North American Monsoon (NAM). Runoff observations are linked to historical changes in runoff production in each watershed. Observations indicate that the mesquite-treated basin generates more runoff pulses and greater runoff volume for small rainfall events, while the mesquite-encroached basin generates more runoff volume for large rainfall events. A distributed hydrologic model is applied to both basins to investigate the runoff threshold processes experienced during the NAM. Vegetation in the two basins is classified into grass, mesquite, or bare soil using high-resolution imagery. Model predictions are used to investigate the vegetation controls on soil moisture, evapotranspiration, and runoff generation. The distributed model shows that grass and mesquite sites retain the highest levels of soil moisture. The model also captures the runoff generation differences between the two watersheds that have been observed over the past decade. Generally, grass sites in the mesquite-treated basin have less plant interception and evapotranspiration, leading to higher soil moisture that supports greater runoff for small rainfall events. For large rainfall events, the mesquite-encroached basin produces greater runoff due to its higher fraction of bare soil. The results of this study show that a distributed hydrologic model can be used to explain runoff threshold processes linked to woody plant encroachment at the catchment-scale and provides useful interpretations for rangeland management in semiarid areas.

The North American Monsoon System (NAMS) contributes ~55% of the annual rainfall in the Chihuahuan Desert during the summer months. Relatively frequent, intense storms during the NAMS increase soil moisture, reduce surface temperature and lead to runoff in ephemeral channels. Quantifying these processes, however, is difficult due to the sparse nature of coordinated observations. In this study, I present results from a field network of rain gauges (n = 5), soil probes (n = 48), channel flumes (n = 4), and meteorological equipment in a small desert shrubland watershed (~0.05 km2) in the Jornada Experimental. Using this high-resolution network, I characterize the temporal and spatial variability of rainfall, soil conditions and channel runoff within the watershed from June 2010 to September 2011, covering two NAMS periods. In addition, CO2, water and energy measurements at an eddy covariance tower quantify seasonal, monthly and event-scale changes in land-atmosphere states and fluxes. Results from this study indicate a strong seasonality in water and energy fluxes, with a reduction in Bowen ratio (B, the ratio of sensible to latent heat fluxes) from winter (B = 14) to summer (B = 3.3). This reduction is tied to shallow soil moisture availability during the summer (s = 0.040 m3/m3) as compared to the winter (s = 0.004 m3/m3). During the NAMS, I analyzed four consecutive rainfall-runoff events to quantify the soil moisture and channel flow responses and how water availability impacted the land-atmosphere fluxes. Spatial hydrologic variations during events occur over distances as short as ~15 m. The field network also allowed comparisons of several approaches to estimate evapotranspiration (ET). I found a more accurate ET estimate (a reduction of mean absolute error by 38%) when using distributed soil moisture data, as compared to a standard water balance approach based on the tower site. In addition, use of spatially-varied soil moisture data yielded a more reasonable relationship between ET and soil moisture, an important parameterization in many hydrologic models. The analyses illustrates the value of high-resolution sampling for quantifying seasonal fluxes in desert shrublands and their improvements in closing the water balance in small watersheds.