Assimilating satellite-based soil moisture observations in a land surface model: The effect of spatial resolution

Abstract

This article focuses on developing a data assimilation system that combines the modeled surface moisture estimates and satellite observations. Specifically, model states simulated by the Noah-MP land surface model are updated using an Ensemble Kalman Filter with products from the NASA SMAP (Soil Moisture Active Passive) satellite mission. The land surface model is run on two different regular grids, one at 12.5 km and the other at 500 m to produce surface and root zone soil moisture estimates across Oklahoma during April-July 2015. In the first case, the model was forced with the NLDAS-2 (North America Land Data Assimilation System) dataset and in the second with a downscaled version of the same dataset. Ground observations from the Oklahoma Mesonet network are compared to surface and root zone soil moisture output simulated by three different Noah-MP model runs i) an open loop simulation (in which no satellite data are assimilated); ii) assimilation of the 36 km SMAP radiometer-only product, and iii) assimilation of the 9 km SMAP radiometer-radar combined product. Results show that SMAP soil moisture retrievals improve the model performance (i.e., with respect to the open loop run) and that forcing the land surface model with higher resolution atmospheric forcings yields higher correlations and smaller errors in soil moisture simulations with respect to the original NLDAS-2 dataset. Although root zone soil moisture is not directly assimilated (since satellite observations are limited to the top 5 cm of the soil column), the assimilation of SMAP products at the surface is transferred to lower layers by the modeled physical processes and is shown to improve root zone soil moisture estimates as well.

Keywords

• Soil moisture
• Data assimilation
• Land surface modeling