Validation Procedure for Remotely Sensed Soil Moisture in Hydrological Simulation at Data-sparse Non-complex Terrains

Abstract

Soil moisture is a state variable that quantifies the amount of water held in the unsaturated layers of the soil profile. A good understanding of its quantity and distribution patterns is vital for research and development. Because it is highly dynamic both in space and time, the most practical way to measure it on a large scale is by satellite remote sensing. High quality groundbased reference soil moisture measurements are required to validate remotely sensed data, but the high cost of implementing a standard ground sampling infrastructure is a major impediment. This study proposes an alternative validation approach for data-sparse regions. It begins with the introduction of a temperature correction term to the gravimetric algorithm, which improves reference measurements by up to 0.55% of their values in the temperature range 10–35℃. Next, a sampling design is developed for localized validation of remotely sensed soil moisture by clustering a large heterogeneous surface to smaller units of noncomplex terrains where landscape-defining characteristics are largely homogeneous, thus permitting the computation of areal soil moisture as a simple arithmetic mean of near-linear point measurements. This method yields results that only marginally differ from those obtained with a spatially distributed sampling method, indicating the potential of the proposed sampling design for localized validations at non-complex terrains in data-sparse regions. Additionally, an analysis of the temporal stability of Soil Moisture Active Passive (SMAP) surface soil moisture in a watershed proved the feasibility of using this concept as a basis for clustering the watershed to wetness classes, permitting time- and cost-efficient monitoring of soil moisture in a long-term basis by focusing on a few representative areas. Finally, a qualitative evaluation of SMAP revealed that while the repetitiveness and the retrieval accuracy of the data are suitable for applications in hydrology, its 9 km spatial resolution is too coarse to capture the land surface heterogeneity. This study recommends a downscaling procedure to improve its resolution before application.