Spatial and temporal modeling of soil moisture using remote sensing

Most hydrologic models which have been developed for application to mountain watersheds with high relief and heterogeneous vegetation types have input data requirements, including those relating to soil moisture, that are difficult to satisfy for all but a few well-instrumented, experimental watersheds. Modeling of the soil moisture content is a vital component of a viable rainfall-runoff model. For this study, the soil moisture component of the Ekalaka Rangeland Hydrology and Yield Model, developed by the Agricultural Research Service in Boise, Idaho, was modified to accomplish the initial step of modeling soil moisture in a mountain watershed with various types of vegetative cover.; The Tony Grove Creek watershed in the Wasatch Mountains in Northern Utah was selected as a test site. The soil moisture model was implemented on the watershed at specific locations identified by a mathematical classification algorithm, which was applied to Landsat TM satellite data. The classification routine yields information on the vegetation types existing on the watershed and assigns a set of coefficients to each pixel in the watershed. These coefficients are called membership values and indicate similarity to a user-determined number of classes.; The number of classes was determined by field surveys of the number of distinct vegetation groups observed, and the actual vegetation type represented by each class was also determined by field surveys. Membership values were applied to distribute over the entire watershed the point soil moisture results obtained from the model at the prototypical sites. This technique for spatial extrapolation was first used by Leu (1985) in a snowpack distribution application.; The results obtained from applying MSMAC as a point model were reasonably accurate for all classes in the watershed. The spatial distribution of extrapolated soil moisture corresponded accurately with the soil moisture measurements made at "blind" sites throughout the watershed. In summary, the use of a point model to simulate soil moisture at a specified number of points can be combined with satellite information to provide a reasonable estimation of the spatial distribution of soil moisture for a watershed. These results can greatly improve the accuracy of a rainfall/runoff model for the watershed.