Abstract:
The NASA Goddard Earth Observing System (GEOS) provides near-real time, global weather analysis and forecast products. In this work, a weakly-coupled land analysis was implemented in GEOS to assimilate L-band (1.4 GHz) passive microwave radiance observations from the NASA Soil Moisture Active Passive (SMAP) satellite mission. The SMAP observations are highly sensitive to surface (~0-5 cm) soil moisture.

In retrospective experiments for boreal summer 2017, the SMAP assimilation is shown to mitigate errors in the GEOS weather analysis, with improvements primarily in screen-level (2m) specific humidity (q2m) and maximum daily temperature (T2m_max). Regionally, the SMAP assimilation improves the RMSE of q2m and T2m_max by up to 0.4 g/kg and 0.3 K, respectively. Improvement in specific humidity extends into the lower troposphere (below ~700 mb), with relative improvements in bias of 15-25%.

In the weakly-coupled system, the impact of the SMAP analysis is communicated through the land-atmosphere processes encoded in the model; the impact is very similar when the GEOS atmospheric analysis is used in a simple three-dimensional variational (3D-VAR) configuration or in the hybrid four-dimensional ensemble-variational (4D-EnVAR) operational configuration. Finally, the beneficial impact of the SMAP analysis on near-surface atmospheric humidity and temperature is mostly maintained in medium-range weather forecasts with lead times up to 5 days.