Abstract:
Satellite L-band (1.4 GHz) microwave observations from the Soil Moisture and Ocean Salinity (SMOS; launched in 2009) mission and the Soil Moisture Active Passive (SMAP; to be launched in November 2014) mission provide information on surface soil moisture and soil temperature. By merging these satellite observations into a land data assimilation system, estimates of root zone soil moisture can be obtained that combine the information in the satellite observations with that provided by the land modeling system and its surface meteorological forcing inputs.

This presentation will provide an overview of the soil moisture assimilation system developed for the SMAP Level 4 Surface and Root Zone Soil Moisture (L4_SM) data product. The product will provide global estimates of soil moisture and related land surface states and fluxes on a 9 km grid every three hours with a latency of a few days. The L4_SM system consists of a variety of components, including the NASA Catchment land surface model, an L-band microwave radiative transfer model (RTM), and an ensemble Kalman filter. Global soil and RTM parameter datasets were developed specifically for the L4_SM system using recent soil texture databases and advanced optimization techniques that ensure unbiased land model estimates to the extent possible. Residual systematic differences between the assimilated microwave brightness temperature observations and the corresponding model estimates are addressed through climatological rescaling.

The algorithm calibration and validation effort is based on a variety of approaches. In situ observations from dense networks play a key role as they permit validation at the spatial scale of the assimilation estimates. Furthermore, the assimilation system is verified by analyzing its internal diagnostics, including the observation-minus-forecast residuals and the analysis increments. We demonstrate the calibration and validation effort by evaluating a prototype L4_SM data product that is based on the assimilation of SMOS observations. This evaluation illustrates the challenges that result from inconsistencies in the complex set of ancillary system inputs (land model and RTM parameters, rescaling coefficients, and model and observation error parameters). Such inconsistencies are commonplace because of unavoidable changes in operationally produced, global surface meteorological forcing data in connection with the complex and costly process of generating a complete set of ancillary inputs for the soil moisture assimilation system.