Upgrades to the NCEP meteorological downscaling system for NAM

The NCEP NAM downscaling (named smartinit) system developed by NOAA/Global System Division and NCEP downscales meteorological model fields to an National Digital Forecast Database high resolution grid at either 5 or 2.5 km over the Continental U.S., Alaska, Hawaii and Puerto Rico. Downscaled NAM analysis fields are then input into to the NCEP Real-Time Mesoscale Analysis (RTMA) where fields are further adjusted to the nearby observations. NAM parent 12 km or nested grid hourly predictions are also downscaled to 84 forecast hours. 2 m temperature and dewpoint temperature, 10 m winds and surface pressure are adjusted to a high resolution topographical database. Simple terrain effects are also introduced for winds while temperatures are also downscaled to the high resolution topography using similarity theory. Smartinit diagnostic downscaled meteorological analyses and predictions that account for complex terrain and coastal effects are often used to drive air pollution models. Smartinit also sharpens gradients of all fields around coastal areas using the land mask dataset. Basically, temps, winds, and dew point from the downscaled NAM land-water point are replaced by downscaled winds from the nearest neighbor point that agrees with the Analysis (NDFD) land-water point. Snow depth is then adjusted depending on the interpolated temperature. Chance of wetting rain and mixed layer depth winds, temperature and humidity are also computed from the downscaled fields.

In the past year, the system has been unified to more easily run for a user defined grid domain as well as ensuring consistent downscaling approaches for all domains. A more representative way to represent terrain effects on winds could be accomplished for the case when the NAM first guess topography lie below the output analysis topography (2b). The spatially constat γ_ terrain scaling term could be replaced by using high resolution analysis land-use roughness (e.g.: De Rooy and KOK, 2004, W&F) or similarity theory (Troen and Mahrt, 1986). Also, simple diagnostic mass field codes are available to include terrain effects on winds. Winds are adjusted by computing terrain gradients and then velocity potential using the Poisson equation (Sherman, 1977; Calmet, Scire, 1986; Ratto, 1996). In this study, the impact of these approaches for improving the downscaled winds are evaluated against observations.