|
Abstract:
|
With the advent of increasingly higher resolution weather forecasts came the need for new and improved methods for evaluating those forecasts. Traditional methods often indicated that coarser resolution models were superior because of nuances with the statistical summary measures rather than any tangible evidence; e.g., the forecaster would often subjectively find the high-resolution model to be vastly superior. Moreover, the traditional methods, which tally performance on a grid-point by grid-point basis, do not provide any much needed diagnostic information about forecast performance in order to know in what ways the forecast performed well or poorly. For these reasons, and because of existing methods in other research areas (e.g, computer vision, image analysis, etc.), a vast many new methods for evaluating high-resolution weather forecast performance were rapidly introduced, which led to the need for meta-verification methods intercomparison project. Now, a second such project is under way to further investigate the various new spatial verification methods.
|
