Probabilistic approach in comparitive verification of high resolution models
Kees KOK
Royal Netherlands Meteorological Institute (KNMI), Netherlands
Abstract
High resolution forecasting is particularly aiming at predicting the smaller scale atmospheric phenomena. However, the exact location and amplitude of small-scale features are difficult to predict. Moreover, the resulting errors may deteriorate the deterministic verification scores of high resolution models in such a way that they sometimes can't beat the scores of lower resolution models that do not produce these small scales. In this presentation a probabilistic approach is suggested which offers the possibility not only to objectively assess the skill of small scale information but also to quantify the additional value of high resolution models over lower resolution ones.