基于MODE对模式预报强风风场的检验分析

Evaluation and Analysis of Model Forecast Performance of High Wind Based on MODE Method

基于专业气象服务领域十分关注的模式强风预报性能的客观检验和致灾性强风预警服务效果的合理性判别问题,利用一次强风过程对电网专业气象预警服务效果的检验示范,探讨基于目标(面向对象)的诊断检验方法(Method for ObjectBased Diagnostic Evaluation,MODE)对强风事件检验的适用性。通过对强风空间检验的各个关键参数的选择确定、匹配分析,发现:(1)MODE检验的卷积半径、诊断量权重和匹配阈值等参数的选择,将影响检验目标的适度聚焦和误差表达的客观性,因此需要综合考虑预报和实况场的水平分辨率、用户对该类强天气预报误差的空间和时间冗余度等;(2)MODE检验可较好地量化给出模式预报的整体效果,包括:强风落区的范围大小和位置偏差、强风过程的时间相位差等,从而可量化判别出模式在各个时刻的空报和漏报区域以及对强风过程移动速度和生命周期长度的预报性能。

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 considera- tions 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 cy-cle of high wind objects.