Soil Moisture Variability and Hydrological Impact Assessment of Land Cover Change

Abstract

Water availability in a region depends on how precipitation over the region is transformed into various forms after reaching ground such as evaporation losses, runoff, infiltration, soil moisture, and ground water storage etc. Land Use / Land Cover (LU/LC) changes adversely affect the aforementioned components. Particularly, the effects of LU/LC changes on catchment hydrological responses, especially vegetative cover (forest, scrubs and cropland), affect the evapotranspiration. Further, rapid urbanization due to LU/LC changes leads to extent of impervious surface and thereby, impacts the infiltration rates as well as recharge. The LU/LC change impact on the hydrologic system is region specific, and each region is characterized by its own hydrology, terrain, climate and also anthropogenic factors. Therefore, a detailed assessment of LU/LC change impacts on hydrology is required, specifically at the region with seasonally limited water availability. It is emphasized by many researchers that the physically-based, distributed hydrological models along with remote sensing capabilities are more suitable for assessing the LU/LC change impacts on the hydrologic system. Further, Soil moisture, being a critical state variable, its knowledge is of paramount importance in several hydrological applications (e.g., runoff modeling and flood forecasting, agricultural monitoring and drought monitoring). The magnitude of soil moisture variability often under estimated and the spatial pattern of soil moisture is not consistent, and it is largely varying across the site and climate with the influence of heterogeneity in LU/LC, topography, soil properties, precipitation and evapotranspiration. Hence, the characterization of soil moisture variability is essential. The work reported in this thesis aims at understanding the soil moisture variability and land cover change impacts in an agricultural dominant semi-arid basin. The Variable Infiltration Capacity (VIC) model, a physically based, semi distributed hydrological model was used to simulate the hydrologic responses of the basin for different LU/LC scenarios (the year 2000 and 2010) with multiple soil layersiv parameterization (three soil layers: 0 – 10 cm, 10 – 45 cm and 45 – 100 cm). The total drainage area of the basin was discretized into number of model grids (5.5 km resolution: totally 1694 grids), and the input parameterization of the model was made at each grid level. The major input parameters to the model are meteorological forcing (Precipitation, Tmax, Tmin), soil characteristics, land surface vegetation classes (vegetation parameter & library) and topography. This study demonstrated a methodological frame work for improved vegetation parameterization to the model simulation. Moderate Resolution Imaging Spectroradiometer (MODIS)-derived 8-day Lear Area Inded (LAI) time-series data was used to sub-group agricultural dominant areas into major crop groups and corresponding monthly vegetation phenology in terms of LAI, albedo, height, root distribution were arrived. This exercise enabled improved definition of vegetation parameterization for the study area, incorporating the region specific conditions. Firstly, the model was calibrated and validated using the observed stream flow data collected at two different locations for the period 1994 – 2001. The model parameter values were adopted for each model grid (about 5.5 km) based on the saturated hydraulic conductivity at that grid by trial and error method. To assess the hydrological impacts of LU/LC change on the flow regime of the basin, the model was run using the two LU/LC conditions separately with the same observed meteorological forcing and soil data. The changes attributed to LU/LC at basin level indicate that the surface runoff and baseflow decreased by 18.86 and 5.83% respectively. The evapotranspiration increased by 7.8%, mainly because of the agricultural crops. The variability in hydrological components and the spatial variation of each component attributed to LU/LC was further assessed at the basin grid level. The majority of the basin grids showed an increase in evapotranspiration (80 % of basin grids) and subsequent decrease in runoff and baseflow (79 and 85% of basin grids, respectively) with resepect to LU/LC change. Further, the spatio-temporal variation of soil moisture was assessed using the model simulated soil moisture along with three different satellite derived surface soil moisture products (SM-CCI, SM-TRMM and SM-AMSRE). It was found from the analysis that the impacts of LU/LC changes on soil moisture were more evident in the deeper layers (45 cm and 100 cm). The soil moisture decreased byv an average of 14.43 and 18.21% (percentage change), particularly in dry periods at second and third layers, respectively. Further, the modeled soil moisture along with three different satellite surface soil moisture products were investigated for its spatio-temporal variability in the basin. The soil moisture in the top layer (up to 10 cm) showed high temporal variations. However, the mean soil moisture was found almost constant during the summer and winter seasons. The basin showed high variability in soil moisture during the intermediate wetness condition. Further, the spatial variability of the soil moisture during the wetting period (June-August) was high compared to drying period (December – February). Based on the analysis performed in this study, 29 (out of total model grids - 1694) representative grid locations were identified in the basin. These locations could be effectively considered for installing observational network mainly for monitoring soil moisture in near real-time.