|Starks, Patrick - Pat|
Submitted to: American Society for Testing and Materials
Publication Type: Book / Chapter
Publication Acceptance Date: 3/1/2003
Publication Date: 3/1/2003
Citation: STARKS, P.J., ROSS, J.D., HEATHMAN, G.C. MODELING THE SPATIAL AND TEMPORAL DISTRIBUTION OF SOIL MOISTURE AT WATERSHED SCALES USING REMOTE SENSING AND GIS. AMERICAN SOCIETY FOR TESTING AND MATERIALS. 2003. p. 58-74. Interpretive Summary: In hydrology, the amount of water in the soil affects partitioning of rainfall into runoff and infiltration, thus impacting surface and groundwater recharge, flood forecasting, and flow routing modeling. Measurement of soil water content at a point is straightforward, but point measurements are inadequate for watershed hydrology due to variability of soil properties, land cover, and meteorological inputs over space. Passiv microwave remote sensing systems have been successfully used to provide regional estimates of surface (0-5cm layer) soil water content at the spatial resolution of the sensor. In this study, a simple one-dimensional soil water budget model was used to combine remotely sensed estimates of surface soil water content and spatial information on land cover, soil type and meteorology to predict the amount of water in the root zone over the 611 square kilometer Little Washita River Experimental Watershed (LWREW). Watershed averages generated from the model output compared well to measurements from those study sites that represented the dominant soil textures of the watershed. Preliminary results from this study indicate that it is possible to integrate remotely sensed estimates of surface soil water content with meteorological, soils, and vegetation data to extrapolate the amount of water in the root zone using a simple, one- dimensional model applied in a distributed fashion across a variable watershed. Watershed or regional soil water content maps generated from such a modeling approach would provide useful information to hydrologists, flood forecast centers, and agricultural concerns related to crop production and yield forecasting.
Technical Abstract: A simple one-dimensional soil water budget model was used to combine remotely sensed estimates of surface soil water content and spatial information on land cover, soil type and meteorological inputs to predict the amount of water in the root zone over the 611 square kilometer Little Washita River Experimental Watershed(LWREW). The model was first evaluated at the point scale by testing its performance at four test sites within th LWREW. Results indicated that modeled estimates of root zone soil water contentclosely matched measurements at two of the four sites, but underestimated measured values at the other two sites. The model was then revised to accommodate spatially distributed remotely sensed estimates of surface water content and geo-spatial data (land cover, soil texture, rainfall) on a 1 kilometer by 1 kilometer spatial resolution. Modeled surface water content from the four study sites was averaged for each day of the study period and used to represent the mean watershed value. These average watershed values were then compared to surface measurements from the remotely sensed data. A linear regression on the two data sets yielded correlation coefficient of 0.95. Daily modeled surface and root zone estimates from each 1 kilometer by 1 kilometer cell in the watershed were averaged and plotted as a time series along with the measured values at each of the four study sites. Graphical analysis of the time series plots revealed that the watershed averages, approximated the soil water content at those study sites that represented that dominant soil textures of the watershed.