Extracting hydrologic information from the Soil Moisture Active Passive (SMAP) satellite data for improved hydrologic modeling

Abstract

Soil moisture is a key component of the water cycle. The launch of the new-generation satellite mission, the NASA Soil Moisture Active/Passive (SMAP) mission, in 2015 opens up unprecedented opportunities for researchers to learn more about surface soil moisture behavior and to then improve hydrologic modeling. In this dissertation, various ways of extracting hydrologic information from the SMAP surface soil moisture data are explored and their ability to improve hydrologic modeling (both model results and process representation) is assessed. First, data assimilation techniques are applied to incorporate the SMAP data to update modeled soil moisture states. Then, rainfall correction techniques are applied to use SMAP to back-correct rainfall estimates from the Global Precipitation Measurement (GPM) mission. These updated soil moisture states and precipitation estimates are then combined to improve simulated streamflow. This work shows that SMAP soil moisture assimilation only slightly nudges rainfall and streamflow estimates in the correct direction. One main reason for the small hydrologic improvement is that the Kalman-filter-based soil moisture data assimilation techniques are only able to correct zero-mean random error in a hydrologic simulation system, but not the often more substantial systematic error. These findings motivate the last part of this dissertation, in which surface soil moisture dynamics are directly derived from SMAP via a data-driven, multivariate regression approach. The SMAP-derived dynamics include surface moisture decay rate, fraction of precipitation retained in the surface layer, and the dependency of the infiltration/runoff partition process on antecedent moisture level. These governing dynamics derived from SMAP are compared with those derived from a model-based global dataset, and inaccuracies in the model setup are pointed out, including slow surface moisture decay, small sensitivity of the infiltration/runoff partition process to the top-layer moisture, and lack of spatial variation in surface soil moisture dynamics. This work demonstrates the potential of using the extracted information from SMAP to evaluate and improve process representation in hydrologic models.