This project addresses the question "How can weather forecasting be improved by better data assimilation?" Our particular emphasis will be on new ensemble-based data assimilation methods that can derive soil moisture estimates from a combination of passive and active microwave observations. Synthetic data-denial experiments clearly show that weather and climate predictions improve substantially when initial and boundary conditions incorporate accurate soil moisture information. Development of soil moisture maps and related land-atmosphere flux estimates will require optimal use of all available sources of information, including remotely sensed data, ground-based data, and process-oriented models of the coupled soil-vegetation-atmosphere system. This can be accomplished with appropriately designed data assimilation algorithms. Ensemble data assimilation methods provide a particularly promising option for the soil moisture problem. They are efficient, flexible, and robust and can be readily adapted to particular operational needs.
In this project we are 1) seeking to provide better understanding of ensemble data assimilation, 2) developing a practical soil moisture data assimilation algorithm based on ensemble concepts, and 3) testing this algorithm using passive and active L-band microwave observations and other data from the 1997 and 1999 Southern Great Plains field experiments (SGP97 and SGP99). We believe that our investigation of the fundamental properties of the ensemble approach will enable us to develop improved ensemble algorithms that meet the demanding requirements of operational soil moisture applications.