|
Activity Number:
|
190
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308880 |
|
Title:
|
Probabilistic Weather Forecasting for Winter Road Maintenance
|
|
Author(s):
|
Veronica J. Berrocal*+ and Adrian E. Raftery and Tilmann Gneiting
|
|
Companies:
|
University of Washington and University of Washington and University of Washington
|
|
Address:
|
Box 354322, Seattle, WA, 98195,
|
|
Keywords:
|
spatial hierarchical model ; probabilistic weather forecasting ; predictive probability distributions
|
|
Abstract:
|
Road maintenance is a critical issue during winter time. In order for anti-icing operations to be efficient, predictions of road ice need to be accurate and reliable. Probabilistic forecasts take forecast uncertainty into account and allow decision makers to make informed, flexible judgments. In this paper, we propose a spatial hierarchical model that post-processes numerical weather forecasts and yields joint predictive probability distributions of precipitation occurrence and temperature. These in turn lead forecasts of the probability of ice on the road. In predictions of ice formation along Washington State Interstate 90 Mountains to Sound Greenway for the 2003--2004 and 2004--2005 winter seasons, the statistically post-processed probabilistic forecasts were sharp and calibrated, and had higher economic value than the deterministic forecasts from a numerical weather prediction model.
|
