Assessing the influence of model inputs on performance of the EMT + VS soil moisture downscaling model for a large foothills region in Northern Colorado

Abstract

Soil moisture is an important variable in the hydrologic cycle and a key consideration for off-road vehicle mobility and many other applications. Soil moisture can be estimated at coarse resolutions (∼10 km grid cells) using microwave satellites or land surface models, but these estimates must be downscaled to be useful for such applications. The Equilibrium Moisture from Topography Plus Vegetation and Soil (EMT + VS) model has been proposed as a tool to downscale soil moisture to 10–40 m resolutions. However, it was developed by considering small catchments and determining the model inputs from field data. Although it produced accurate results, it has not been tested using the spatial extents and data from microwave satellites or land surface models. The objective of this study is to test the EMT + VS soil moisture downscaling method for such conditions, examining how the choice of model inputs affects the accuracy of the resulting soil moisture estimates. The model is applied to a 4,000-ha ranch in Northern Colorado that includes diverse topography, vegetation, and soils. The results are compared to in situ soil moisture observations collected at 86 locations on 10 dates that span a wide range of conditions. The study finds that the EMT + VS model can reproduce much of the space–time variation in the observations, and the model outperforms its coarse resolution input and two other soil moisture estimation methods. The model can produce similar results using a variety of coarse resolution soil moisture inputs and fine resolution (30 m, 10 m, or 3 m) topographic inputs, but the choice of satellite, date, and index used to characterize vegetation impacts the accuracy of the soil moisture results.