Abstract:Model forecast performance and forewarning service validity of disastrous high wind have been among the chief concerns in industrial meteorological services. In view of these concerns, evaluation of the power grid system oriented meteorological forewarning for a high wind event was chosen as an example in order to demonstrate the applicability of Method for Object based Diagnostic Evaluation (MODE) for high wind events. Through the determination of the MODE crucial parameters and the object matching process, we found that (1) convolution radius, pairwise attributes weight and matching threshold are crucial to the identification of interest objects and the objectivity of error analysis. As a result, comprehensive considerations should be paid to the spatial resolution of forecast/observation grid data and user specified spatial and temporal error redundancy. (2) General forecast performance can be depicted quantitatively by MODE, including the deviation of high wind location, range extent and temporal phase, so as to detect the area of false alarm and missing report every hour, and assess the forecast performance of moving speed and life cycle of high wind objects.