Satellite-Derived Fields of Land Surface Variables Used by the National Centers for Environmental Prediction Numerical Weather Prediction Models

To improve the characterization of the spatial and temporal variability of water and energy cycles at land-atmosphere interface, the National Centers for Environmental Prediction (NCEP) numerical weather prediction models use satellite-derived land surface characteristics. Their current usage for weather forecasting in the coupled regional (Eta) and global medium range forecast (MRF) models can be categorized by two modes: 1) as an input for boundary conditions during the period of forecast, and 2) for validation of the model produced variables. The current paper focuses on first mode, i.e. the surface fields that are being currently used at NCEP for input, including surface type, green vegetation fraction, snow cover and snow-free albedo. The first three fields have been incorporated in the land surface scheme of both Eta and MRF models, whereas the forth (surface albedo) is being tested for incorporation in the Eta model and will be used for validation (the second mode) of the model-generated albedos in MRF. This paper reviews the methodologies used for producing the four surface fields and describes the way these fields are used in the NCEP models, focusing on the regional Eta model with a 32-km spatial resolution. The surface type is "fixed" for the period of forecast in both Eta and MRF. The green vegetation fraction and snow-free albedo, originally produced as climatological monthly fields, are interpolated on each day of the year and are "fixed" for the 2-day period of forecast by Eta but vary during the 7-day period in MRF. The snow cover is daily product produced operationally and delivered to the NCEP models on a regular basis. The surface fields, based on different datasets with different resolution, are averaged into the model-resolution grid. The surface type classification is prescribed from the 1-km EROS dataset based on the full resolution NOAA AVHRR observations. The vegetation fraction and snow-free albedo fields are based on a 5-year average monthly NOAA/AVHRR 0.15- deg dataset. The 25-km snow maps are produced by visual analysis of satellite data from polar and geostationary satellites and some available ground observations. This study will include description of planned enhancements in the input surface fields, e.g. derivation of the three predominant sub-grid model surface types and the corresponding vegetation fraction and albedo, based on 1-km data. Improvement and automation of the snow product will be discussed.