数值天气预报检验方法邻域法的研究进展

Research Progress of Neighborhood Methods for Numerical Weather Forecast Verification

邻域检验方法被广泛应用于数值天气预报的检验,随着高分辨率数值模式的不断发展,新的邻域检验方法不断涌现,邻域法的应用也逐渐从确定性预报扩展到集合概率预报。首先对不同邻域法进行了回顾,并按照对预报和观测场处理方式的不同,将邻域法分为升尺度法、邻域最大值法、格点概率法和其他方法。而后基于一套观测和预报资料,对不同邻域法进行比较。不同邻域法适用情形不同,在对预报进行检验时应根据需要选择合适的方法。在全面回顾的基础上,指出邻域法的优势和不足,并对未来发展方向进行展望。

Neighborhood method has been widely used in numerical weather forecast verification.With the development of high-resolution numerical model,there have emerged new neighborhood methods,and the application of neighborhood method has been extended from deterministic forecasts to ensemble forecasts.Firstly,different neighborhood methods are reviewed.According to different processing methods for forecast and observation field,neighborhood methods are categorized into four types,upscaling method,neighborhood maximum method,grid probability method and other methods.Then different neighborhood methods are compared based on a set of observation and forecast data.The application of different neighborhood methods should be chosen according to the actual requirements.The advantages and disadvantages of neighborhood methods are pointed out and the future development direction is prospected based on a comprehensive review.