A Synthetic Study of Near-Surface Soil Moisture Assimilation for North America

Jeffrey P Walker and Paul R Houser

Power Point Presentation

Accurate initialization and forecasting of land 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 circulations, through its control on partitioning of the available energy into latent and sensible heat exchange. To overcome such limitations in the NASA Seasonal-to-Interannual Prediction Project (NSIPP), a one-dimensional Kalman filter assimilation of near-surface soil moisture observations has been added to the catchment-based Land Surface Model (LSM) used by NSIPP.

A set of numerical experiments have been undertaken using the NSIPP LSM, off-line from the GCM, to evaluate the assimilation procedure. In this study, "true" soil moisture data were generated by spinning-up the LSM and then running the LSM for 1987 using the ISLSCP forcing data sets. By setting the initial soil moisture prognostic variables to arbitrarily wet values uniformly throughout North America, and the LSM forced with the ISLSCP data sets, a degraded simulation run was made. The final simulation run assimilated near-surface soil moisture "observations" from the "true" soil moisture data into the degraded simulation once every 3 days.

This study has illustrated that by assimilating near-surface soil moisture observations, as would be made with a remote sensing satellite, errors in forecast soil moisture profiles as a result of poor initialization may be removed. After only 1 month of assimilation the rms error in the profile storage of soil moisture was reduced to 3%v/v, while after 12 months of assimilation the rms error in the profile storage was as low as 1%v/v. This study has also shown how the assimilation of near-surface soil moisture can improve the forecasts of runoff and evapotranspiration.