MDA EarthSat Weather says its current wind-generation forecast product has now been expanded to include AESO (Alberta). Already providing accurate and reliable hourly forecasts for ERCOT, MISO, and BPA, PJM, CAISO and IESO, this web-based system uses state-of-the-art forecast methods to produce hourly forecasts for lead times of up to ten days. In the short-term forecast, statistical techniques learn from wind data and power variations and develop highly accurate predictive relationships between present wind and future conditions. In the day-ahead and longer term forecasts, method removes model biases, and weights models according to their skill, greatly improving forecast skill over single model technique.
In verifications, correlation coefficients are near 0.9 in the day-ahead market, with mean absolute errors of less than 10% of generation capacity. “We’re always on the lookout for improvements in our techniques that can squeeze every bit of predictive power from wind, wind-generation observations, and the dynamical forecast models. We’re especially excited by our improving skill in the 3 to 4 hour range, where a combination of dynamical and statistical modeling is required for an optimal forecast,” says Dr. Dan Kirk-Davidoff, Chief Scientist at MDA EarthSat.
MDA EarthSat Weather
Filed Under: News