Forecasting wind speeds using pattern recognition for North Cape, Prince Edward Island

This thesis determines the suitability of the phase space forecasting method as a medium time range wind speed prediction tool. Twenty-four hour wind speed forecasts are vitally important to electrical grids that contain a sizable amount of wind generation capacity because the wind speed changes significantly over time. Advanced knowledge of wind park behavior is essential in maintaining an electrical grid because the sum of the non-dispatchable energy from the wind and the dispatchable energy from conventional sources must always be greater than the electrical system load. Phase space methods are not currently utilized by any forecast service provider.; The forecasting ability of the phase space technique is tested using on-site, geographically separated and upper atmospheric meteorological data. Overall findings are that surface wind speed and knowledge of jet stream activity show promise for medium range site specific forecasting.