Python在短临气象预报检验中的应用

Application of Python in Test of Nowcasting

基于机器学习方法的短临多气象要素预报系统(Weather Elements Nowcasting based on machine learning,简称WEN)具有高发布频次、高时间分辨率、基于候和时辰的复杂预报模型等特点。应用多维标签数组、机器学习工具、并行计算框架等Python库,以快速计算为目标,建立以预报模型覆盖时间范围为统计检验时间边界的检验子系统,客观给出预报性能,为模型调优效果评估、产品业务化运行提供依据。

Weather Elements Nowcasting based on machine learning(WEN)has the characteristics of high release frequency,high time resolution,and complex forecast model based on climate and time.Using Python libraries such as multidimensional tag array,machine learning tools,parallel computing framework aiming at"fast computing",a testing subsystem is established.It takes the time range covered by the"prediction model"as the statistical test time boundary.It objectively gives the prediction performance,which provides a basis for evaluating model optimization effect and the operation of products.