Assessing the influence of model inputs on performance of the EMT+VS soil moisture downscaling model for a large foothills region in northern Colorado
Date
2024
Journal Title
Journal ISSN
Volume Title
Abstract
Soil moisture is an important driving variable of the hydrologic cycle and a key consideration for decision-making in off-road vehicle mobility, crop modeling, drought forecasting, flood prediction, and a variety of other applications. Soil moisture can be estimated at coarse resolutions (>1 km) using satellite remote sensing or land surface models; however, coarse resolution estimates are unsuitable for many applications. Downscaling these products to finer resolutions (~10 m) creates soil moisture maps that are more useful. This study applies the Equilibrium Moisture from Topography, Vegetation, and Soil (EMT+VS) model to Maxwell Ranch, a 4,000-ha cattle ranch in Northern Colorado that represents a diverse range of topographic, vegetation, and soil characteristics and a wide range of soil moisture conditions. The EMT+VS model is a physically based geo-information method that downscales coarse resolution soil moisture estimates using ancillary fine resolution datasets of topography and vegetation. Input data to the EMT+VS model contain inherent sources of error that can impact the uncertainty of downscaled estimates. The objective of this study is to identify sources of uncertainty in inputs and assess their influence on the error of the EMT+VS model output. The study finds changes in vegetation input or digital elevation model (DEM) resolution introduce substantial errors in the EMT+VS model output; however, these errors can be mostly overcome when recalibration with local in-situ data is possible. The highest errors (RMSE = 0.20 cm3/cm3) tend to occur in locations with thick vegetation and high contributing area, which are difficult to accurately estimate with available remote sensing data sources.
Description
Rights Access
Subject
EMT+VS
downscaling
soil moisture