A Unified Data‐Driven Method to Derive Hydrologic Dynamics From Global SMAP Surface Soil Moisture and GPM Precipitation Data. Issue 2 (1st March 2020)

• Record Type: Journal Article
• Authors: Mao, Yixin; Crow, Wade T.; Nijssen, Bart
• Is Part Of: Water resources research. Volume 56:Issue 2(2020)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-03-01
• Subjects: SMAP satellite -- soil moisture dynamics -- regression -- hydrologic model -- data‐driven; Hydrology -- Periodicals; 333.91
• DOI: 10.1029/2019WR024949
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: Data sets provided by the Soil Moisture Active Passive (SMAP) and the Global Precipitation Measurement (GPM) satellite missions contain rich information about land surface hydrologic processes.In this study, a unified regression method is proposed and applied to these global data sets to investigate surface soil moisture (SSM) dynamics. Two forms of regressors are implemented: 1) the linear regressors of SSM and precipitation flux and 2) the linear regressors of SSM and precipitation flux with an additional interaction term. Regression results based on 3 years of global SMAP and GPM data show that the unified regression method can identify the SSM characteristics found by several recent studies, including the SSM exponential decay rate, the fraction of precipitation retained in the surface soil layer, and the effective depth of hydrologic storage. Additionally, including the interaction regressor provides a novel way to derive the sensitivity of infiltration/runoff partitioning to antecedent SSM without the need for streamflow observations. These SMAP/GPM regression results are compared with those derived from a global SSM data set simulated by the variable infiltration capacity model. Relative to the satellite data, variable infiltration capacity retains moisture longer in the top layer, retains too much precipitation input in that layer, and exhibits levels of sensitivity of runoff/infiltration partitioning to top‐layer soil moisture that generally match SMAP especially in humid regions. This study demonstrates that the regression‐based method can recover useful process‐level insight from SMAP SSM retrievals and is a viable tool for evaluating the representation of surface processes in hydrologic models. Plain Language Summary: In the past few years, new satellite missions have been launched to make frequent observations of near‐surface soil moisture and rainfall over most global land areas. In this study, we analyze these satellite observations to learn how near‐surface soil moisture changes over time. Based on this analysis, we can determine for each location: 1) how fast near‐surface soil moisture declines after rainfall, 2) how much rainfall enters the soil versus how much runs off, and 3) how rainfall partitioning is affected by near‐surface soil moisture. The first two questions have been studied before, but our method is simpler and reaches the same conclusions. Our method is also able to answer the third question, which has not been looked at before. Finally, we analyze a model‐based near‐surface soil moisture data set in the same way and compare the results with our observation‐based finding. In this way, we pinpoint aspects of the mathematical model that do not correctly mimic the real‐world evolution of near‐surface soil moisture. This insight will guide model upgrades that better capture reality. Key Points: The regression derives multiple metrics of surface soil moisture dynamics from SMAP and GPM data that agree with recent studies The regression resolves the effect of antecedent surface soil moisture on infiltration/runoff partitioning without streamflow data The regression is capable of recovering process‐level insight from the satellite data which informs hydrologic model representation