Field-scale land surface modeling over continental extents: Applications in satellite remote sensing of soil moisture

Abstract:

Existing land surface models (LSM) struggle to accurately represent the observed field-scale (~100 meters) spatial heterogeneity of soil moisture due to the over-simplistic parameterizations of sub-grid heterogeneity and the coarseness of the model input data. This is especially relevant in the context of satellite remote sensing of soil moisture since land surface models are seen as important tools with which to validate high-resolution soil moisture retrievals. To address this challenge, we have developed HydroBloks, a semi-distributed land surface model that uses hydrologic response units (HRUs) to represent the observed field-scale spatial heterogeneity of soil moisture while maintaining the computational efficiency of existing LSMs. To accomplish this goal, HydroBloks couples the Noah-MP land surface model to the Dynamic TOPMODEL hydrologic model. The HRUs are defined by clustering proxies of the drivers of spatial heterogeneity using field-scale land data (e.g., NLCD). This allows for each HRU's results to be readily mapped out in space, enabling model application and validation at sub-100 meter scales. In this study, HydroBloks is implemented at three USDA watersheds over the contiguous United States (Little Washita, Little River, and Walnut Gulch). HydroBloks is run at each watershed between 2004 and 2014 using a 100 Latin Hypercube Sample to account for model parameter uncertainty. Each catchment's model ensemble is constrained and validated using available in-situ top-layer soil moisture observations. The results from this study provide insight into the strengths and weaknesses of existing soil moisture networks and the model’s potential applications for improved design of in-situ soil moisture networks.