Abstract:
The NASA Soil Moisture Active Passive (SMAP) mission generates, among other data sets, the Level-4 Soil Moisture (L4_SM) product, which provides global, 3-hourly, 9-km resolution estimates of surface and root-zone soil moisture and related land surface states and fluxes with a mean latency of ~2.5 days. The L4_SM algorithm is based on the assimilation of SMAP radiometer brightness temperature (Tb) observations into the NASA Catchment land surface model using a spatially distributed ensemble Kalman filter.

The L4_SM algorithm uses a “tau-omega” radiative transfer model (RTM) to convert the modeled soil moisture and temperature estimates into model forecasts of the observed Tbs. The RTM parameters are calibrated using Tb observations from SMAP and the Soil Moisture Ocean Salinity mission. Residual bias in the seasonal cycle of the modeled Tb is addressed by scaling – prior to assimilation – the assimilated observations to match the seasonally varying climatology of the calibrated modeling system. This approach is not perfect and results in a small bias in the observations-minus-forecast Tb residuals (typically less than 3 K in magnitude). This Tb bias then leads to a non-zero long-term mean of the soil moisture analysis increments, which implies a non-physical net addition or subtraction of water in the L4_SM system.

Another concern in the context of the long-term mean water balance are the perturbations that are applied to select model forcing and prognostic variables in the L4_SM system. These perturbations create the ensemble spread that describes the uncertainty in the modeled soil moisture and temperature, which is needed to assign the weights to the model and the observations in the analysis update. In very dry regions, however, only positive (wetting) perturbations are physically possible, which results in a net addition of water in the L4_SM system.

A preliminary analysis suggests that the global average fluxes are 2.4, 1.7, and 0.8 mm/d for precipitation, evapotranspiration and runoff, respectively. The imbalance of 0.16 mm/d is split between the analysis increments (0.06 mm/d) and the impact of the perturbations (0.1 mm/d). In the presentation, we quantify in more detail the natural water balance and the non-physical terms introduced by the assimilation component in the L4_SM system.