A case study of the use of statistical models in forecast verification: Precipitation probability forecasts

被引：0
|
作者：
Murphy, Allan H. [1 ]
Wilks, Daniel S. [2 ,3 ]
机构：
[1] Prediction and Evaluation Systems, Corvallis, OR, United States
[2] Atmospheric Science Group, Cornell University, Ithaca, NY, United States
[3] Atmospheric Science Group, Cornell University, 1113 Bradfield Hall, Ithaca, NY 14853, United States
来源：
Weather and Forecasting | 1998年 / 13卷 / 3 PART 2期
关键词：
Mathematical models - Precipitation (meteorology) - Probability - Regression analysis;
D O I：
暂无
中图分类号：
学科分类号：
摘要：
Selected samples of probability of precipitation (PoP) forecasts and the matching binary observations were evaluated to determine the use of statistical models for forecast verification. A linear regression model was fitted to the conditional distributions of the observations given the forecasts and a beta distribution was fitted to the frequencies of use of the allowable probabilities. These two models described the joint distribution of the forecasts and observations and reduced a 21-dimensional verification problem to 4 dimensions. The selected PoP forecasts was compared across forecast type, location and lead time in terms of the four parameters.
引用
收藏
页码：795 / 810
相关论文
共 50 条
12345