Downscaling soil moisture over regions that include multiple coarse-resolution grid cells

Many applications require soil moisture estimates over large spatial extents (30-300 km) and at fine-resolutions (10-30 m). Remote-sensing methods can provide soil moisture estimates over very large spatial extents (continental to global) at coarse resolutions (10-40 km), but their output must be downscaled to reach fine resolutions. When large spatial extents are considered, the downscaling procedure must consider multiple coarse-resolution grid cells, yet little attention has been given to the treatment of multiple grid cells. The objective of this paper is to compare the performance of different methods for addressing multiple coarse grid cells. To accomplish this goal, the Equilibrium Moisture from Topography, Vegetation, and Soil (EMT + VS) downscaling model is generalized to accept multiple coarse grid cells, and two methods for their treatment are implemented and compared. The first method (fixed window) is a direct extension of the original EMT + VS model and downscales each coarse grid cell independently. The second method (shifting window) replaces the coarse grid cell values with values that are calculated from windows that are centered on each fine grid cell. The window values are weighted averages of the coarse grid values within the window extent, and three weighting methods are considered (box, disk, and Gaussian). The methods are applied to three small catchments with detailed soil moisture observations and one large region. The fixed window typically provides more accurate estimates of soil moisture than the shifting window, but it produces abrupt changes in soil moisture at the coarse grid boundaries, which may be problematic for some applications. The three weighting methods produce similar results.