P1.5
On the use of neural networks and conditional climatology to predict peak wind speed at Cape Canaveral's Atlas launch pad
Kenneth P. Cloys, Air Force Institute of Technology, 28th Operational Weather Squadron, Shaw Air Force Base, SC; and M. K. Walters, L. K. Coleman, and W. P. Roeder
Neural networks and conditional climatology were investigated for their applicability in providing the 45th Weather Squadron (45 WS) with advance warning of wintertime (November-March) peak wind speeds. The 45 WS provides weather support to the United States space program at Cape Canaveral Air Station (CCAS), NASA's Kennedy Space Center (KSC), and Patrick Air Force Base. Both methods used data from the Weather Information Network Display System, a collection of 47 meteorological towers located throughout the KSC and CCAS. Conditional probabilities of meeting or exceeding a given threshold speed during 8 consecutive one-hour periods were calculated using the current wind direction and peak wind speed as inputs. Accuracy was measured by constructing contingency tables and calculating various measures of accuracy. Results were tested for significance by calculating p-values for the chi-square test. This method showed very little skill in forecasting maximum wind speeds. The neural network forecasts were valid for 16 30-minute intervals, for a total forecast period of 8 hours. After the 6-hour forecast, neural network performance showed skill over persistence, climatology, and random wind speeds selected from a climatologically based distribution.
Poster Session 1, Aviation Range and Aerospace Meteorology: Formal Viewing
Tuesday, 12 September 2000, 5:30 PM-7:00 PM
Previous paper Next paper