Downscaled NAM Data for Forecasting Applications

Local NWS offices have for years been running a SmartInit code to downscale gridded North American Mesoscale (NAM) model output for use in populating NDGD 5 km grids. The starting NAM data is a gridded data set with limited fields, and the algorithms have had to work with incomplete data for their computations.

NCEP is developing the capability to run the SmartInit code in house with the full native level model output as a starting point. Temperatures, dew points, and winds are corrected based on terrain differences between the 12 km native topography and the 5 km topography. Other fields such as cloud cover and probability of precipitation are computed based on algorithms, with the later using a combination of forecasted precipitation from the NAM and SREF.

These downscaled data sets will be available to help populate forecast grids as well as serve as a first guess for the Real-Time Mesoscale Analysis (RTMA) over Alaska and Puerto Rico. Bias correction will also be applied with the goal of developing a NAM MOS look-alike product.