Citation

McKittrick L, Adams E, Gauer P, Mewes J, Curran A. Transp. Res. Circular 2004; (E-C063): 95-108.

Copyright

(Copyright © 2004, U.S. National Academy of Sciences Transportation Research Board)

DOI

unavailable

PMID

unavailable

Abstract

To forecast terrain-dependent weather conditions with a focus on hazardous road conditions (such as icing and high winds), a model chain has been developed. In this chain, a series of three weather modeling programs have been linked together, successively nesting the mesh of one regional model inside another. The three programs in use are Eta, with computational analysis provided by the National Center for Environmental Prediction (NCEP); ARPS (Advanced Regional Prediction System), with computational analysis provided by Meridian Environmental Technology; and RadTherm/RT, with computational analysis provided by Montana State University (MSU). Once Eta data are available, they are downloaded from NCEP servers to Meridian servers and used to run a sequence of ARPS models, to represent terrain-dependent changes in the meteorological conditions down to resolutions of 1 km. As each analysis progresses, successive forecast conditions are written to files and passed to computational servers at MSU. At MSU, these meteorological files are used to define environment conditions (such as long- and short-wave radiation, cloud cover, precipitation, and convective effects) in RadTherm. Successive downloads are then used to run RadTherm models somewhat in parallel with, though slightly behind, each ARPS forecast. As successive forecast models become complete, forecast data are graphed alongside real-time measured data and posted on the Internet. To allow viewers to evaluate the validity of the current forecast, graphs corresponding to weather station sites are updated hourly, typically by extracting measured data from websites associated with the given station. These same data sets can then be compared on a weekly basis for long-term performance evaluation and improvement.