Scale-dependent soil and climate variability effects on watershed water balance of the SWAT model

The water balance of large watersheds in Texas was studied using a process level watershed model called the soil and water assessment tool (SWAT). The major components of the SWAT watershed model are evapo-transpiration (ET), soil water storage (SW), and water yield (WYLD), which is the sum of surface runoff and subsurface flow, Important input variables controlling the water balance in watersheds are the soils and climate. In this paper, changes in mean and variance of water balance components due to variability in soils and climate were assessed for six different watersheds (Hydrologic Cataloging Units, HCUs) in Texas covering wet to semi-arid regions using 1:250,000 scale data. Bulk density, soil available water capacity, and moist soil albedo were selected as soil parameters. Soil heterogeneity in watersheds was defined in terms of textural classes in the soil textural triangle. Spatial variability of precipitation between neighboring weather stations was quantified using power spectra. The impact of geographic scales on changes to the mean of the water balance components was determined by studying the Seco Creek sub-watershed within the Hondo HCU using 1:24,000 scale data. Results from scales of observation show that changes to mean SW was high as a function of increasing scale from 1:250,000 to 1:24,000, while mean ET sensitivity remained about the same. The direct influence of soil properties such as bulk density, available water capacity, on the variance of ET, SW, and WYLD was about equal when using 1:24,000 scale data. For watersheds in wet climate composed of heterogeneous soils (loam fine sands and fine sandy loams), the means of the water components were relatively sensitive to climate and soils variability, and soil heterogeneity. Watersheds composed of shallow soils in semi-arid climate showed sensitivity of mean water balance components due to moist albedo and available water capacity. Changes to mean and variance of water balance components as a function of geographic scale suggest the presence of scale-dependent water balance 'uncertainty' laws. (C) 2002 Elsevier Science B.V. All rights reserved.