Assimilating satellite soil moisture observations in a hyper-resolution land surface model

Abstract:

The goal of this work is to explore how satellite observations could advance hyper-resolution land surface modeling. A hyper-resolution atmospheric forcing dataset (temperature, pressure, humidity, wind speed, precipitation, incident longwave and shortwave radiation) is developed from coarse resolution products using a set of (i) physically-based approaches that rely on correlations with landscape variables (such as topography, temperature lapse rate corrections, surface roughness, and vegetation index) and (ii) statistical approaches, including a random forest classification and a regression algorithm. A proof-of-concept has been implemented over Oklahoma, where high-resolution observations are available for validation purposes. Hourly NLDAS-2 (North America Land Data Assimilation System) atmospheric variables at 0.125° have been downscaled to a hyper resolution (500 m) over the study area for the calendar year 2015. These downscaled products have first been used to force a land surface model for soil moisture estimation. Then, a land data assimilation system is adopted to merge the SMAP (Soil Moisture Active Passive) Level 3 products into Noah-MP. 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 (Radiometer, Radar and Radiometer/Radar). Validation is performed against ground observations of soil moisture across Oklahoma. Preliminary results from the land surface model at 500 m forcing dataset has a good agreement with ground-based information. 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. Therefore, the assimilation of SMAP Radiometer soil moisture retrievals have the potential to improve the estimation of surface and root zone soil moisture, especially at higher elevations, increasing the correlation and reducing the random error between the model and the ground observations. This work will result in a radical improvement over the current state-of-the-art forcing data and will move into the era of hyper-resolution land modeling.