Characterization of the Spatial and Temporal Controls on Soil Moisture and Streamflow Generation in a Semi-Arid Headwater Catchment

Christopher Jason Williams, Boise State University

Abstract

In recent years, increased demands have been placed on hydrologists modeling catchment scale hydrology. Historically, it has been acceptable to predict streamflow at the catchment outlet without partitioning physical processes (lumped modeling) that control catchment scale responses. Hydrologists are now challenged with partitioning processes controlling hillslope hydrology and stream responses and distributing those processes (distributed modeling) over various scales to improve hydrologic modeling. Increased demand for distributed versus lumped hydrologic models has led to the derivation of multiple topographic and wetness indices to parameterize soil moisture patterns. Many of the commonly derived indices assume steady state flows, common in humid environments, and are poor predictors of hydrologic processes occurring in semi-arid environments. The purposes of this study are to facilitate understanding of hillslope flowpaths and streamflow initiation and cessation in the semi-arid, snow-dominated Upper Dry Creek Experimental Watershed (UDCEW) and to improve the predictive capability of terrain and wetness based indices for modeling soil moisture patterns and hillslope processes occurring in the UDCEW.

Soil moisture patterns for the UDCEW were characterized from measured near-surface (upper 30 cm of soil profile) and modeled (SHAW model) deep (up to 1.2 m) soil moisture contents. Mapping soil moisture patterns identified a saturated subsurface source area that developed near the channel head early in the fall wet-up period. Streamflow initiated as the size of the subsurface source area increased during the late fall wet-up or early in the hydrologically wet season and ceased as the subsurface source area decreased during the late spring drydown. Streamflow generation at the site is then partially dependent on the distribution of fall (wet-up) season soil water to a saturated subsurface source area adjacent to the channel head.

Seasonal soil moisture patterns were statistically compared with site characteristics (surface and bedrock topography, soil depth, soil structure, and vegetation patterns) to determine the temporal and spatial variability of the controls on soil moisture. Results suggest that the controls on soil moisture patterns and hillslope processes at the site vary with hydrologic regime and are different from hydrologic controls commonly observed under humid, steady state conditions. Modified topographic wetness indices were created to characterize the observed hydrologic controls. Observed soil moisture patterns were more positively correlated with modified indices than indices commonly reported in literature. Performance of modified and commonly used indices exhibited seasonality. In the near surface, indices inclusive of soil depth best explained soil moisture variability during the wet-up, hydrologically wet, and drydown periods and indices that parameterized water input and evapotranspiration best explained soil moisture variability during the dry period. For deep soil moisture parameterization, indices inclusive of soil depth best explained soil moisture during the wet-up and hydrologically wet periods, and soil moisture during the dry and drydown periods was best explained by a process based index of water input and evapotranspiration. The findings suggest that, for semi-arid climates like the UDCEW, a multi-index approach to distributed modeling may be necessary to accurately simulate seasonal variation in soil moisture patterns.