A study of soil moisture variability and vegetation greenness dynamics in a mountainous rangeland watershed using direct measurements, remote sensing, and modeling.

摘要

Soil moisture content (&thetas;, m3 m−3) and vegetation greenness in a semi-arid mountainous rangeland watershed were evaluated using field measurements, remotely sensed data, a hydrologic model, and a Geographic Information System (GIS). Field research was carried out in the Reynolds Creek Experimental Watershed (RCEW, 234 km2) located on the northern flanks of the Owyhee mountains about 80 km (50 miles) southwest of Boise, Idaho, USA.; This research was divided into three objectives: (1) to characterize the spatial dependence and temporal stability of &thetas; in smaller sub-catchments, (2) to investigate the relationship between plant available soil water (PASW) and greenness of rangeland vegetation, and (3) to evaluate the effect of terrain, soil, and climate parameters on the spatial variability of &thetas; and other water balance parameters using a simulation model and GIS.; Geostatistical analysis of &thetas; measurements taken at small sub-catchments (0.13 and 0.26 km2) within the RCEW at a depth of 0–30 cm indicated that there was a spatial correlation in &thetas; for about 200 m. Spatial variability was associated with the season and orientation of the sub-catchments. A parametric approach of temporal stability analysis indicated that about five to six sampling locations are adequate to capture the catchment average &thetas; for these small sub-catchments.; Soil water content and the vegetative greenness dynamics at the scale of RCEW were evaluated using field &thetas; measurements, remote sensing data, and modeling. Vegetation indices (VI) derived from satellite images provided spatially distributed patterns on the growth and productivity of vegetation. The SAVI (Soil Adjusted Vegetation Index) was compared with the PASW derived from field &thetas; measurements and with ERHYM (Ekalaka Rangeland Hydrology and Yield Model) generated outputs of transpiration (T) and potential evapotranspiration (PET). The ERHYM-simulated &thetas; were compared with the field measured &thetas;. They were found to be in good agreement with an R2 of 0.71.; The deterministic source variables such as precipitation, aspect, slope, soil depth, soil type, and vegetation for the model-simulated &thetas;, T/PET, and actual PASW were evaluated using GIS overlay analysis. Spatial variability of model-predicted &thetas;, T/PET, and actual PASW increased with areal extent.; Overall, the characterization of spatial variability of &thetas; at small sub-catchments and watershed scale by evaluation of field measurements, remotely sensed data, and simulation models is a unique approach. Integration of remote sensing, modeling, and field measurements appears to be a viable means of studying &thetas; changes and vegetation greenness patterns in rangeland areas. (Abstract shortened by UMI.)