Experiments with EKF-LDAS :
With EKF-LDAS, numerical experiments are conducted for answering the following questions:
§  How is EKF data assimilation superior to direct insertion of observation data?
§  How do the model and observation error covariances affect EKF data assimilation?
§  How does EKF correct model errors caused by model initialization or forcing?
§  How can EKF data assimilation help data validation?
Experiment Settings:
§  Initial soil moisture:           0.3 fc or  0.7fc
§  SWDN:                100% or 120%
§  DA method:           none or EKF or DI
§  Mdl error variance:           0.01 or 0.001fc
§  Obs error variance:          0.05 or 0.005 or 0.5fc
§  Land surface model:        Mosaic model
§  Forcing data:           NASA forcing
§  Model time:                      from 00z07July1999  
                                                 to 23z20July1999
Soil Moisture Observation Data:
§MOS: Synthetic top layer soil moisture from Mosaic model with correct initiali-zation and forcing for the SGP99 area (12x21 1/8° grids) for noon time daily.
§TMI: Surface soil moisture derived by Jackson et al from TMI 10.7GHz brightness temperature for the SGP99 area and downloaded from Goddard DAAC.