Objective: Develop an optimal methodology for initializing the land
surface for long-term climate prediction.
•Avoid GCM bias: develop off-line Mosaic Catchment Model forced with observations.
•Develop a Kalman filter based data assimilation capacity for NSIPP land.
•Identify and use
relevant observations of land forcing and states.
•Understand the
spatial-temporal sensitivity of coupled predictions to the initial land state.
•
Specific Goals:
•Spatial/Temporal
Domain: 1o global
(catchment based); long-term
retrospective; start with North America in collaboration with LDAS.
•Model: Catchment Model
forced by GCM (ECMWF) and observed precipitation/radiation (GOES, AVHRR,
SSM/I, TRMM). Runoff routing
model.
•Potential
Assimilation Data: Surface Temperature(AVHRR, MODIS, GOES), Snow Cover/Water (AVHRR,
GOES, SSM/I,AMSR), Soil Moisture(SSM/I(M. Owe),AMSR,GRACE,TRMM,Runoff).
•Assimilation: 1-D Kalman
filter, with a land surface Observation Operator
observation-model translation scheme.
•Spin-Up: Off-line LSM will
"run
up" or "spin up" to a NSIPP initialization.
•Evaluation/Validation: Fully-coupled
retrospective NSIPP simulations; comparison with independent
in-situ (Field Experiments, Snow, Runoff) and
remote-sensed observations.