Submitted to: International Geoscience and Remote Sensing Symposium Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: 6/30/2003
Publication Date: 7/22/2003
Citation: Crow. W.T., Wood, E.F., Schaake, J. 2002. Integration of remote sensing soil moisture products into hydrologic prediction systems using ensemble-based assimilation strategies [abstract]. 2003 International Geoscience and Remote Sensing Symposium Proceedings. Interpretive Summary:
Technical Abstract: Efficient use of spaceborne soil moisture products for hydrologic monitoring and forecasting applications requires optimal strategies for integrating soil moisture retrievals with current operational land surface modeling and prediction capabilities. Hydrologic forecasts (of e.g. streamflow) are increasingly being presented within a probabilistic framework based on the Monte Carlo generation of an ensemble of hydrologic model predictions. Much of the uncertainty expressed in these ensembles originates from the probablistic representation of error in the quantitative precipitation forecast (QPF) used to force the hydrologic model. Likewise, ensemble-based data assimilation techniques such as the Ensemble Kalman filter (EnKF) have recently been applied to the problem of assimilating remote soil moisture observations into hydrologic models. Here, as in the hydrologic forecasting case, individual ensemble members are taken to represent a random sample of equally likely surface state forecasts given past states and observations. Such convergence in the probabilistic methodology applied to hydrologic forecasting and data assimilation implies that the Ensemble Kalman filter presents an effective framework to integrate: short-term quantitative precipitation forecasts, dynamic land surface model predictions, and remote observations of the near-surface for hydrologic modeling and forecasting purposes. This paper will detail the development of an ensemble-based forecasting strategy for streamflow and surface soil moisture which utilizes a probabilistic description of uncertainty in quantitative precipitation forecasts for model forecasting and incorporates remote soil moisture observations for model updating (and thus re-initialization for the next forecast period) via the EnKF. Preliminary results will focus on artificially generated soil moisture retrievals and testing components of the data assimilation system in isolation (e.g. assessing QPF skill and the assimilation of surface soil moisture via the EnKF). The impact of forecast lead time and error in remotely-derived soil moisture will also be discussed.