Over the past several years, under the sponsorship of the NOAAGEWEX Continental Scale International Project (GCIP), the Environmental Modeling Center (EMC) of the National Centers for Environmental Prediction (NCEP) has joined with hydrologists in the NWS Office of Hydrology (OH) and satellite land-surface remote sensing experts in NESDIS to develop and implement a series of advancements to the land-surface and hydrology physics of the NCEP mesoscale Eta Model over the U.S., and its associated Eta-based 4-D Data Assimilation System (EDAS). These advancements include a new land-surface model (LSM) with four soil layers, explicit vegetation effects, snowpack physics, sub-grid runoff treatments, satellite-derived daily snow cover updates, and a satellite-derived seasonal cycle of green vegetation fraction.
We will first briefly describe a) the model configuration and b) off-line uncoupled development and testing that has spanned various retrospective experiments including ISLSCP/FIFE, ISLSCP/GSWP, and PILPS 2a, 2b, 2c, and 2d. Presented results will include our latest off-line tests at the Valdai, Russia site in PILPS 2d, focusing on our proposed upgrades to frozen soil and patchy snow cover physics.
Next we will present experiences from the operational coupled model, implementation milestones, and validation examples of surface fluxes, skin temperature (from satellite), shelter temperature and dewpoint, PBL profiles, and precipitation skill scores. Substantial validation and assessment efforts have been carried out by collaborating with extramural GCIP-sponsored investigators. Early first-year problems with a low-level warm/dry bias in the summer of 1996 have been largely solved. Our present attention is focused on 1) a low-level cool/moist bias in the early spring over moist soils with little vegetation and 2) overly large ground-heat flux over moist soils. One hallmark of our land-surface modeling system is the significant use of four satellite-derived land-surface databases, from which we will present examples. We recently implemented a satellite-based daily update at 23-km of the N. Hemisphere snow cover and sea ice. Additionally, we specify our seasonal vegetation cycle using a global, 0.144 degree, 5-year climatology of monthly green vegetation fraction, derived from a carefully cloud-screened 1985-1990 climatology of AVHRR-based NDVI. In the validation area, we apply GOES-derived, hourly, 0.50 degree retrievals of skin temperature and downward surface solar insolation.
In the data assimilation arena, the large Eta domain (all of North and Central America) has prompted us to test and operationally implement (on 03 June 1998) a fully continuous Eta-based cycling of all model prognostic state variables, including the land states of soil moisture/temperature. For the time being, we are cycling the land-surface states without any nudging of the soil moisture. This omission of nudging may have to be revisited after one full year of continuous cycling, depending on the extent of land-surface drift that is encountered.
To address the possibility of drift in the continuously cycled, coupled, land-surface state (primarily arising from precipitation and surface insolation biases in the coupled systems), we are testing the EDAS assimilation of a) a newly developed hourly, national, 4-km, radar/gage rainfall analysis and b) the hourly, 0.50 degree, national, GOES-based retrieval of skin temperature cited earlier. The skin temperature assimilation tests are using a variational technique that utilizes the adjoint of the LSM.
Finally, we will summarize the configuration of our more aggressive attempt to avoid land-surface state drift, known as "LDAS". We have just embarked on the development of an uncoupled, national, Land-Data Assimilation System (LDAS), in which the Eta LSM will be uncoupled from the Eta mesoscale model and executed in a real-time stand-alone mode forced by the cited, model-independent, hourly analyses of observed precipitation and solar insolation. This LDAS will provide an alternative setting for the assimilation of satellite-derived skin temperature and satellite-derived soil moisture.