An empirical approach towards improved spatial estimates of soil moisture for vegetation analysis

Landscape-level spatial estimates of soil water content are critical to understanding ecological processes and predicting watershed response to environmental change. Because soil moisture influences are highly variable at the landscape scale, most meteorological datasets are not detailed enough to depict spatial trends in the water balance at these extents. We propose a tactical approach to gather high-resolution field data for use in soil moisture models. Using these data, we (1) describe general soil moisture trends for a 6400 ha watershed in the Oregon Western Cascades, USA (2) use this description to identify environmental variables to stratify across in collecting data for a statistical explanatory model of soil moisture spatial pattern at the onset of seasonal drought, and (3) examine the spatial scale of variability in soil moisture measurements compared to the scale of variability in potential explanatory factors. The results indicate that soil moisture dynamics and controls are different for different soil depths across this mountainous watershed. Soil moisture variability exhibits complex spatial patterns that can be partially estimated (p to 50 percent of the variation accounted) with easily measurable climatic and terrain variables. The analysis incorporates both macroscale (climate) and mesoscale (topographic drainage and radiation) influences on the water balance. Without additional data on the distribution of edaphic and biotic factors, we are not able to model the variability of soil moisture at the microscale. The regression approach can be used to extrapolate field measurements across similar topographic areas to examine spatial patterns in forest vegetation and moisture-controlled ecological processes.