Research Abstracts 1994-1995: Report No. 95-02

Barbour, P.L. and S. N. Walker, "Individual Site Evaluation of Model Output Statistics (MOS) Wind Forecasts," OSU Wind Research Cooperative, WRC Report No. 95-02, March 25, 1995, 60 pp., $15.00

Wind forecasting has the potential for reducing the uncertainty involved in using a variable resource such as wind energy. In the work presented here, an evaluation is performed to determine the accuracy and types of problems involved in wind forecasts suitable for use by utility schedulers (for forecast projection times of six to twenty four hours). Wind forecasts prepared by the National Weather Service (NWS) using the Model Output Statistics (MOS) technique are evaluated for two inland sites and six coastal (Coastal Marine Automated Network (CMAN) ) sites. Forecast accuracy is examined with an emphasis on identifying the error characteristics of individual sites. Errors in forecasts for wind speed, direction and power output from a single hypothetical wind turbine are studied. Results indicate that forecast accuracy depends largely on the site and it's wind characteristics and on the time of day for which the forecast is made. Larger wind speed forecast errors and lower direction errors were observed in forecasts for afternoon hours when the winds at most sites were at their diurnal peak. Forecast of power output from a single turbine also showed that errors are larger during afternoon hours. The errors in both wind and power forecasts followed similar diurnal patterns. The errors associated with forecasts of power output were found to be quite large when separated into separate wind speed categories.