This presentation examines the precipitation and temperature forecast performance of a number of major forecast providers for the City of Winnipeg, Manitoba, Canada. Their performance is weighed against the expectations of the public as indicated in public surveys.
The commercial providers chosen were picked from a larger group. The final group of providers assessed represented the better performers and encompassed a broad range of NWP input.
The temperature forecasts were assessed for the day-time high. The primary assessment used was Mean Absolute Error, since the end user normally compares the predicted values versus actual values to ascertain error.
The precipitation forecasts from each provider were categorized and these categories were assessed.
Based upon the results, the presentation examines one approach that could significantly improve the utility of medium to long range temperature forecasts.
Finally, this presentation will explore the implications of these results to high-impact weather forecasting.