A Synthetic Study on the Influence of Error in Surface Soil Moisture Observations on Assimilation

Accurate initialization and forecasting of surface soil moisture in fully-coupled climate system models is critical for seasonal-to-interannual climatological and hydrological prediction, because of its feedback to precipitation and atmospheric circulation. To provide accurate initial soil moisture, the NASA Seasonal-to-Interannual Prediction Project (NSIPP) has added the capability to assimilate near-surface soil moisture data to its land surface model.

Vegetation cover impedes remotely sensed observations of the near-surface soil moisture. To examine the amount of error there may be in these observations, while still being useful for assimilation, a set of numerical experiments has been undertaken using the NSIPP land surface model off-line from the GCM. In this study, "true" soil moisture data were generated for North America by spinning-up the land surface model and then running for 1987 using the ISLSCP forcing data sets. By adding perturbations to the initial soil moisture and the forcing data, a degraded simulation was made to imitate the likely error in soil moisture forecasts as a result of erroneous initial conditions and forcing. The final simulations using the perturbed initial condition and forcing data assimilated degraded near-surface soil moisture "observations" from the "true" simulation. The “observations” were degraded by various amounts to imitate erroneous observations.