A consortium of GAPP-sponsored groups is undertaking the development of an uncoupled North American Land Data Assimilation System (LDAS). NCEP, NASA/GSFC, NWS Hydrology Lab, NESDIS, Princeton University, University of Washington, University of Maryland, and Rutgers University have undertaken the development and prototype realtime demonstration of national, realtime, hourly, 15-km, uncoupled land-surface models forced by observed precipitation and observed GOES-derived solar insolation. This system is referred to as the LDAS, as future enhancements will include the assimilation of satellite-derived land-surface fields such as skin temperature, soil moisture, and snowpack.

The hallmarks of this LDAS project are 4 separate realtime land-surface models (LSMs) running in parallel in the NCEP computer environment on a common 0.125 degree grid using common a) NCEP-derived hourly atmospheric forcing, b) common land-surface characteristics (to the extent possible), and c) a common streamflow routing model. The land-surface state variables (e.g. soil moisture and temperature, snowpack) of the 4 LSMs will all be continuously cycled forward in realtime. The participating LSMs so far include the NOAH LSM, the MOSAIC LSM, the VIC-3L LSM, and the Sacramento Model.

In this talk we intercompare the realtime LDAS LSM results for the simulation of the snowpack over the CONUS and of streamflow from about 25 small to medium sized catchments. With respect to snow, comparisons are presented between the LDAS LSMs and three sources of snow-extent validation data: the Interactive Multisensor Snow and Ice Mapping System (IMS) produced by the Satellite Analysis Branch (SAB) in NOAA/SSD, MODIS-based snow extent from NASA's Earth Observing System, and snow maps from the National Operational Hydrologic Remote Sensing Center (NOHRSC). The snow validation is carried out over the NWS river forecast centers (RFC) regions using data from February 1997 through October 2001, both for the region average and by elevation bands within the RCF regions. The validation of the snow water equivalent estimations from the models is done using a selection of SNOTEL stations. For the streamflow validation, a common routing model was used to calculate the river discharge at approximately 25 small to medium basins starting in October 1997. These predicted discharges are compared to USGS reported values.