Location: Water Management and Systems ResearchTitle: Incorporating probabilistic variations in soil moisture downscaling
|KIM, BORAN - Colorado State University|
|NIEMANN, JEFFREY - Colorado State University|
Submitted to: Annual American Geophysical Union Hydrology Days
Publication Type: Proceedings
Publication Acceptance Date: 4/1/2022
Publication Date: 9/21/2022
Citation: Kim, B., Niemann, J.D., Green, T.R. 2022. Incorporating probabilistic variations in soil moisture downscaling [abstract]. 42nd Annual AGU Hydrology Days. 56-57.
Interpretive Summary: n/a
Technical Abstract: Soil moisture is a key variable for many applications including agricultural production and vehicle mobility. These applications require not only accurate estimates of soil moisture over large regions but also soil moisture patterns that exhibit realistic statistical properties, such as the range of values and the spatial correlation structure from fine spatial resolutions (~10 m grid cells) up to large spatial extents (~10 km regions). Satellites such as NASA’s Soil Moisture Active Passive (SMAP) provide soil moisture data nearly globally but at resolutions that are too coarse for such applications (~9 km), so downscaling is used to estimate fine resolution soil moisture patterns from the coarse data. Downscaling methods are often based on the dependence of soil moisture on regional topographic, vegetation, and soil characteristics. However, soil moisture patterns can also include significant random variations, which most downscaling methods neglect. The Equilibrium Moisture from Topography, Vegetation, and Soil (EMT+VS) downscaling model considers both the dependence on regional characteristics and some random variability, but the random variability was developed by considering only small spatial extents (<0.5 km). The objective of this research is to generalize the random components of the EMT+VS model to allow consideration from fine resolutions to large extents. Soil moisture measurements are considered for the 285 m by 540 m Cache la Poudre experimental watershed in Colorado and a 50 km by 75 km region in Arizona. The spatial structures of the random variations are analyzed using geostatistical methods. The EMT+VS model is then generalized and shown to produce soil moisture patterns with statistical properties similar to the observations. The improved downscaling method is expected to produce more realistic soil moisture patterns, which will improve predictions of agricultural productivity and vehicle mobility.