Application of Kalman Filtering for Soil Moisture Data Assimilation in GSFC's Land Data Assimilation System
Xiwu Zhan, Jared Entin, Paul R. Houser, Rolf H. Reichle and Jeffrey P. Walker
It has been documented that land surface data fields retrieved from satellite or aircraft remote sensing instruments contain uncertainties due to imperfect instrument calibration and inversion algorithms, geophysical noise, representativeness error, communication breakdowns, and other sources. Data assimilation systems have been used extensively in meteorology to expose significant defects in satellite data processing schemes, technology limits, bias, and noise. As a validation effort for the planned global soil moisture product from the Advanced Micriwave Scanning Radiometer (AMSR-E) on NASA_s Earth Observing System AQUA (formerly PM) plateform, this study attempts to assimilate a prototype surface soil moisture data set derived from the brightness temperature observations by the Tropical Rainfall Measurement Mission (TRMM) Microwave Imager (TMI) for the area of Southern Great Plains Hydrology Experiment 1999 in Okalahoma. The Extended Kalman Filter and/or the Ensemble Kalman Filter data assimilation techniques will be implemented in the Land Data Assimilation System (LDAS) of NASA-Goddard Space Flight Center. The Mosaic land surface model in the LDAS will be run for the year of 1999. Direct estimates of surface soil moisture from the Mosaic model, the prototype soil moisture data set from TMI and the Kalman Filter assimilation from LDAS will be compared with each other and with the observations by the Electronically Scanned Thinned Array Radiometer (ESTAR) and the ground gravimetric sampling. By means of this cross validation, the nature and uncertainties of the satellite retrieved surface soil moisture data sets will be examined. Results of these validation activities will be presented.