气候预测PS评分对业务影响
The Impact of PS Method on Operational Climate Forecast
-
摘要: 通过对中国气象局现行PS评分办法与理论PS评分的对比,发现现行评分办法对理论PS评分进行了两处修改。一处修改是扩大了预报正确的评定范围,但在PS评分办法实施后,在气候预测业务中,却出现了只预报2个等级的普遍现象,即在能获得高分的同时却降低了预报能力。另一处修改是按统一的要素距平划分等级,结果出现了预报对象的等级分布随着测站、月份变化而变化的现象。预报对象的等级分布是无技巧预报评分的决定因素,而预报技巧是由PS评分与无技巧预报评分之差决定的。在无技巧预报评分有差异的情况下,不同月份、不同区域之间的PS评分便失去了对比的基础。该文针对上述问题,对现行PS评分办法提出了修改建议。Abstract: In the operation of climate forecast, scoring methods of PS, SS, ACC are widely used. Among them, SS and ACC, known as technique scores, are defined as the similarity between forecast and realistic results when comparing with non technique forecasts or the expectations of non technique forecasts equal to zero. The technique scores are basically originated from the evaluations in the forecast results from different populations and thus provide a basis for impartial comparisons. However, PS is considered as the concordance of forecast grades and are not compared with technique scores, so PS is not technique scores. In China, PS is of greater importance in climate forecasts. The PS of short term climate forecasts by each province are performed every year in the China Meteorological Administration and the ranks of monthly precipitation, mean temperature and total forecasting ability are released. The impact of present PS method used on the climate forecast is analyzed. The results show that the theoretical PS is a score of consistence rate based on accurate forecasts on the grades; the foundational requirement of the PS inter comparison is that they share the same probability distribution in ranking. Through comparisons between the present and theoretical PS method, it's also found that the present scoring method is taken as the theoretical scoring method with two modifications which makes the present PS method actually become the concordance of weighted anomalies. The first revision has extended the range of forecast accuracy assessment, in which level 1, 2, 5 and 6 have extended to three grades, level 3 and 4 have extended to four grades, and weight coefficients are added for abnormal grades. However, only two grades are forecasted in climate operational forecast after PS method performs. Based on the examples of the annual precipitation extremes occur in Shaanxi Province, the random forecasts on level 2 and 6 are evaluated and compared. The results show that 2 level forecasts can score higher, while reducing the forecast capability. Another revision is to divide the grades based on the uniform anomalies of elements, which leads to the forecast grade distributions varying between the stations and months. The distribution of forecast grades of is the dominant factor of non technique forecasting scores. Moreover, the forecasting skill is actually determined by the differences between PS and non technique prediction scores. In the cases of different non technique forecasting scores, the foundation for comparisons becomes less stable to use PS in different months and stations. Some suggestions are introduced to the prevailing PS method to solve these problems.
-
表 1 三级评分制下随机预报准确率
Table 1 The precision of random forecast in 3-degree PS scheme
表 2 六级评分制趋势预报用语和分级标准
Table 2 The diction and grade of trend forecasting in the 6-degree PS scheme
表 3 六级评分制单站得分表
Table 3 The score of single station in the 6-degree PS scheme
表 4 六级评分制下的随机预报准确率
Table 4 The precision of random forecast in 6-degree scheme
表 5 只做3级和4级预报时六级评分制下的随机预报准确率
Table 5 The precision of random forecast of the 3rd degree and the 4th degree in the 6-degree PS scheme
表 6 全国160站月平均气温等级分布频数与频率比较
Table 6 The 6-degree frequency of monthly mean temperature on each level at 160 station
表 7 6月和7月不同测站间各等级的频率、概率分布
Table 7 The comparison of 6-degree frequency and probability between different stations in June and July
-
[1] 陈桂英, 赵振国.短期气候预报评估方法和业务初估.应用气象学报, 1998, (2):178-185. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=19980225&flag=1 [2] 王绍武, 朱锦红.短期气候预报的评估问题.应用气象学报, 2000, 11(增刊):1-10. http://www.cnki.com.cn/Article/CJFDTOTAL-YYQX2000S1000.htm [3] 王建国, 吴炜, 徐法彬.一种定量预报评分方案研究.气象, 2004, 30(10):27-29. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200410005.htm [4] 黄海洪, 郑凤琴, 孙崇智.一种新型降水预报评分方法.气象, 2004, 30(9):35-38. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200409008.htm [5] 王遂缠, 吉慧敏.重要灾害性天气评分系统.气象, 2004, 30(8):42-45. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200408010.htm [6] 王雨, 闫之辉.降水检验方案变化对降水检验评估效果的影响分析.气象, 2007, 33(12):53-61. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200712009.htm [7] 马振锋, 杨佑洪.成都区域气象中心短期气候预测业务评估.气象, 2001, 27(12):29-32. [8] 李辑, 金巍, 赵连伟.辽宁省近10年短期气候预测质量评估检验.气象, 2007, 33(4):82-87. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXX200704013.htm [9] 吉庭艳."八五"期间我省几种长期预报方法的评估检验.贵州气象, 1999, 23(3):36-42. http://www.cnki.com.cn/Article/CJFDTOTAL-GZQX199903010.htm [10] 陈桂英.我国现有短期气候业务预报方法综述.应用气象学报, 2000, 11(增刊):11-20. http://www.cnki.com.cn/Article/CJFDTOTAL-YYQX2000S1001.htm [11] 陈丽娟, 李维京.月动力延伸预报产品的评估和解释应用.应用气象学报, 1999, 10(4):486-490. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=199904101&flag=1 [12] 王会军, 陈丽娟, 李维京.中国区域月平均温度和降水的模式可预报性分析.气象学报, 2007, 65(5):725-732. http://www.cnki.com.cn/Article/CJFDTOTAL-QXXB200705007.htm [13] 陈丽娟, 李维京, 张培群.降尺度技术在月降水预报中的应用.应用气象学报, 2003, 14(6):648-655. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20030682&flag=1 [14] 柳艳香, 王凌, 赵振国.2006年中国夏季降水预测回顾.气候变化研究进展, 2007, 3(4):243-245. http://www.cnki.com.cn/Article/CJFDTOTAL-QHBH200704014.htm [15] 周家斌, 张海福.一种汛期降水分布的客观集成预报方法.应用气象学报, 2000, 11(增刊):93-97. http://www.cnki.com.cn/Article/CJFDTOTAL-YYQX2000S1011.htm [16] 黄嘉佑, 黄茂怡.主分量逐步筛选因子典型相关分析及其预报试验.应用气象学报, 2000, 11(增刊):72-78. http://www.cnki.com.cn/Article/CJFDTOTAL-YYQX2000S1008.htm
计量
- 摘要浏览量: 3347
- HTML全文浏览量: 923
- PDF下载量: 1673
- 被引次数: 0