###
DOI:
本文二维码信息
决策树模型辅助下极端暴雨天气的气象服务过程及决策分析——以北京“7.21”特大暴雨为个例*
扈海波1, 梁旭东2, 王瑛3,4, 张西雅1
(1.北京城市气象研究院;2.中国气象科学研究院;3.北京师范大学地表过程与资源生态国家重点实验室,;4.北京师范大学地表过程与资源生态国家重点实验室)
Decision making on the meteological services under extreme thunderstorm condition supported by Bayesian model: a case study of the Beijing ‘7.21’heavy rain.
摘要
相似文献
本文已被:浏览 10次   下载 11
投稿时间:2019-12-10    修订日期:2021-06-07
中文摘要: 本文主要从预报员的审视角度分析及回顾2012年北京“7.21”极端暴雨天气过程中的气象服务过程,探讨这种天气形势下,是否或者如何应用最优决策模型辅助决策服务,从而更好地做出气象服务决策,降低灾害风险。研究基于决策树模型及贝叶斯方法对“7.21”极端暴雨天气气象服务决策过程进行情景模拟及分析。通过对这一典型暴雨灾害过程中的气象服务决策进行诊断性分析,揭示在现有技术条件下,面对不确定性决策问题时,能否借助决策模型方法,在依据现有数据资料(中尺度数值预报结果、观测资料等),得出最优决策。模拟利用“7.21”前一天的数值预报结果,做了事前的暴雨洪涝灾害及暴雨泥石流灾害风险评估,评估结果显示“7.21”当天这两种灾害风险较大。结合历史灾情及“7.21”的预报雨量,同样可判定“7.21”存在较大的暴雨灾害风险。而选用贝叶斯方程算出的“7.21”当天有灾的后验概率仅为23.1%。但是进一步分析有灾及无灾可导致的期望损失,发现无灾预报可导致的期望损失较大。分析判定“7.21”当天还是应选择有灾的决策判断,并采取对应的“保守型气象服务方案”,才可达到最优决策效果。决策模型分析留下的启示是,为消除气象部门在重大气象服务过程中有担心虚报等的压力,在高影响天气服务过程中,除了需要一定的误报容忍度外,还应该提供天气预报及预警的不确定性信息,以供用户做实际决策判断选择,变确定性预报及预警模式为不确定性模式。
中文关键词: 贝叶斯  决策支持  雷暴  极端天气预警
Abstract:Based on the models of decision tree and Bayesian, the progressing of decision making in the meteorological service for the extreme thunderstorm in Beijing on 21 July 2012 has been simulated and analyzed in scenario. Thorough the diagnosis of decision makings in the typical thunderstorm, it will prove whether it is possible of applying for the supporting decision making model to attain the optimal decision in solving the uncertainties problems of decision making, under the current condition of data resources( e.g., the mesoscale NWP system and observation data) and technology standards. Using the grid data of NWP on that day, the risks of floods and debris flow have been assessed to be high, respectively. Considering the threshold rainfalls of causing floods in the historical flooding events and the predicted rainfall distribution on that day, the high risk of thunderstorms also could be recognized simultaneously. The posteriori probability deduced using the Bayesian model is only 23.1%. However, considering the difference of expected losses in predicting severe weather and non-severe weather, the expected losses of predicting non-severe weather can be obviously greater than that of predicting severe weather. Therefore, it would be advisable of predicting severe weather and pick the pessimistic scheme in the meteorological service, which was the optimal decision making in that situation. The simulation of the meteorological service during the thunderstorm on that day reveals that (1) it should be more tolerable to the uncertainty in severe weather forecasting and warning to relieve the necessary pressure to forecasters risen from the afraid of false forecasting and warning, and (2) it should provide the public with the uncertainty information, which will be useful in the decision making of the actual meteorological information users, in the weather forecasting.
文章编号:201912100426     中图分类号:    文献标志码:
基金项目:国家重点研发计划(2017YFC1502505)、国家自然基金(41875125)、北京师范大学重点实验室项目开发课题(2017-KF-23)
引用文本:
扈海波,梁旭东,王瑛,张西雅,.Decision making on the meteological services under extreme thunderstorm condition supported by Bayesian model: a case study of the Beijing ‘7.21’heavy rain.[J].Meteor Mon,():-.
Haibo HU,Xudong LIANG,Ying WANG,Xiya ZHANG,.Decision making on the meteological services under extreme thunderstorm condition supported by Bayesian model: a case study of the Beijing ‘7.21’heavy rain.[J].Meteor Mon,():-.