Use of Global Datasets for Downscaling Soil Moisture with the EMT+VS Model

摘要

Satellite remote sensing and land-surface models provide coarse-resolution (9--40 km) soil moisture estimates, but various applications require fine-resolution (10--30 m) soil moisture patterns. The Equilibrium Moisture from Topography, Vegetation, and Soil (EMT+VS) model downscales soil moisture using fine-resolution topography, vegetation, and soil data. It has been shown to reproduce temporally unstable soil moisture patterns (i.e. patterns where the spatial structure varies in time). It can also reproduce hillslope dependent patterns (wetter locations occur on hillslopes oriented away from the sun) and valley dependent patterns (wetter locations occur in valley bottoms). However, the EMT+VS model requires several parameters to characterize the local climate, soil, and vegetation characteristics. In previous applications, the parameters were calibrated using point soil moisture data, but many regions of interest may not have such data. The purpose of this study is to evaluate EMT+VS model performance when the parameters are estimated from global datasets without site-specific calibration. Reliable and accessible global datasets were identified and methods were developed to estimate the parameters from the datasets. The global model (without site-specific calibration) was applied to six study sites, and its results were compared to local soil moisture observations and the results from the locally calibrated model. The use of global datasets decreased downscaling performance and the spatial variability of soil moisture was underestimated. Overall, only 5 of the 16 parameters can be estimated from global datasets. However, the global model still provides more reliable soil moisture estimates than the coarse-resolution input for most sampling dates at all six study sites.