How Does Assimilating SMAP Soil Moisture Improve Characterization of the Terrestrial Water Cycle in an Integrated Land Surface‐Subsurface Model?

Abstract Land surface modeling combined with data assimilation can yield highly accurate soil moisture estimates on regional and global scales. However, most land surface models often neglect lateral surface and subsurface flows, which are crucial for water redistribution and soil moisture. This study applies the Community Land Model (CLM) and the coupled CLM‐ParFlow model over a 22,500 area in western Germany. Soil moisture retrievals from the Soil Moisture Active Passive mission are assimilated with the Localized Ensemble Kalman Filter (with and without parameter estimation). The simulated soil moisture, evapotranspiration (ET) and groundwater level are evaluated using in situ observations from a Cosmic‐Ray Neutron Sensor network, Eddy Covariance (EC) stations and groundwater measurement wells. The assimilation improves the median correlation between simulated and measured soil moisture from 0.72 0.79 to 0.79 0.83 and decreases the median unbiased Root Mean Square Error (ubRMSE) from 0.063 0.060 / to 0.050 0.045 /. ET characterization shows a limited improvement with a highest ubRMSE reduction of 15% at the Rollesbroich1 site with the CLM‐ParFlow model. The assimilation does not improve the groundwater level characterization. Furthermore, the joint state‐parameter update does not outperform state‐only update. Overall, the simulation of full 3D subsurface hydrology with the ParFlow model component results in additional model outputs like groundwater levels and river stages, and a better soil moisture characterization (compared to CLM stand‐alone), but it does not make soil moisture assimilation more efficient to correct model states.