气候变化研究进展 ›› 2020, Vol. 16 ›› Issue (6): 714-724.doi: 10.12006/j.issn.1673-1719.2020.061

• 气候系统变化 • 上一篇    下一篇

BCC_CSM1.1m模式对福建前汛期降水预测的误差订正

池艳珍1,2(), 梁潇云3(), 何芬4, 吴伟杰2, 唐振飞4   

  1. 1 福建省灾害天气重点实验室,福州 350001
    2 海峡气象开放实验室,厦门 361012
    3 中国气象局国家气候中心,北京 100081
    4 福建省气候中心,福州 350001
  • 收稿日期:2020-03-30 修回日期:2020-05-14 出版日期:2020-11-30 发布日期:2020-12-03
  • 通讯作者: 梁潇云
  • 作者简介:池艳珍,女,正研级高工,cyz0605@126.com
  • 基金资助:
    国家重点研发计划项目(2018YFC1505805);国家重点研发计划项目(2016YFA0602100);国家重点研发计划项目(2018YFE0196000);福建省气象局开放式基金(2018K03)

Verification and preliminary correction of the precipitation prediction in the pre-flood season over Fujian province by BCC_CSM1.1m climate model

CHI Yan-Zhen1,2(), LIANG Xiao-Yun3(), HE Fen4, WU Wei-Jie2, TANG Zhen-Fei4   

  1. 1 Disasters Weather Key Laboratory of Fujian Province, Fuzhou 350001, China
    2 Laboratory of Straits Meteorology, Xiamen 361012, China
    3 National Climate Center, China Meteorological Administration, Beijng 100081, China
    4 Fujian Climate Center, Fuzhou 350001, China
  • Received:2020-03-30 Revised:2020-05-14 Online:2020-11-30 Published:2020-12-03
  • Contact: LIANG Xiao-Yun

摘要:

采用1991—2017年BCC_CSM1.1m季节预测模式的月降水预测数据及福建省前汛期(4—6月)66个国家气象站降水资料,利用距平相关系数(ACC)、时间相关系数(TCC)、平均方差技巧评分(MSSS)和趋势异常综合评分(Ps)等评估方法,检验评估了提前0、1、2和5个月模式对福建省前汛期降水的预测能力。采用系统偏差、一元线性回归和EOF-相似误差(EOFL和EOFNL)等4种统计方法对回报结果进行订正,并进行效果检验。BCC_CSM1.1m在不同起报时间对福建省前汛期降水的预测均能抓住降水的前两个主模态:全省一致和南北反向分布的空间特征,但预测的气候平均值较实况存在负偏差。模式在不同起报时间对前汛期降水预测的TCC高技巧区主要位于福建省北部,ACC技巧和Ps评分存在比较大的年际差异,负系统偏差的存在使得MSSS技巧不高。经订正后,模式的预测能力得到明显提升。系统偏差、线性回归、EOF相似误差线性和非线性订正方法提前2个月起报的2011—2017年平均Ps评分分别提高5.9、3.5、6.7和7.8分;不同起报时间线性回归订正的2011—2017年平均ACC技巧分别提高0.02、0.21、0.12和0.11;上述4种方法订正的MSSS评分均有了显著提高,其中系统偏差和线性回归订正后达正技巧。综合而言,线性回归订正较其他3种订正方法表现出更为稳定的订正技巧。

关键词: 降水预测, 检验, 统计订正, BCC_CSM1.1m

Abstract:

Based on the re-forecast and operational data from the second-generation seasonal prediction model of Beijing Climate Center (BCC_CSM1.1m), and the monthly observational precipitation of 66 stations over Fujian province in 1991-2017, the precipitation prediction ability of the Model during the pre-flood season at different lead time was assessed. The metrics of verification used in this study were anomaly correlation coefficient (ACC), temporal correlation coefficient (TCC), mean square skill score (MSSS) and the prediction score (Ps). The system bias correction (BC), the linear regression (LR) and Empirical Orthogonal Function-analogue correction were used to revise the forecast. Results show that: (1) Although there is always a systematic negative bias in the prediction of the climatological precipitation for the pre-flood season over Fujian province at different leading time, the Model can predict the first and second typical modes of the precipitation in the pre-flood period: the uniform distribution in the whole province and the decrease from south to north; (2) The inter-annual variation of ACC skill and Ps scores of the precipitation prediction are prominent, and the MSSS scores are negative due to the systematic negative bias. The high TCC skill can be found in the northern Fujian province; (3) The prediction ability of the model for precipitation improved significantly after being corrected. The average Ps scores in 2011-2017 are 5.9, 3.5, 6.7 and 7.8 points higher than the raw ensemble at LM2 (leading two months) after the BC, LR, EOFL and EOFNL correction, respectively. The average ACC skill scores in 2011-2017 are 0.02, 0.21, 0.12 and 0.11 points higher than raw predictions after the LR correction. There are significant improvements of MSSS scores for the four correction methods, of which the bias correction and the linear regression correction receive positive techniques; (4) In general, the linear regression correction shows more advantages than the other three correction methods.

Key words: Precipitation prediction, Verification, Statistical correction, BCC_CSM1.1m

京ICP备11008704号-4
版权所有 © 《气候变化研究进展》编辑部
地址:北京市海淀区中关村南大街46号 邮编:100081 电话/传真:(010)58995171 E-mail:accr@cma.gov.cn
本系统由北京玛格泰克科技发展有限公司设计开发 技术支持:support@magtech.com.cn