[1]李海宏,吴吉东,王强,等.基于机器学习方法的上海市暴雨内涝灾情预测模型研究[J].自然灾害学报,2021,30(01):191-200.[doi:10.13577/j.jnd.2021.0120]
 LI Haihong,WU Jidong,WANG Qiang,et al.A study on rainstorm waterlogging disaster prediction models in Shanghai based on machine learning[J].,2021,30(01):191-200.[doi:10.13577/j.jnd.2021.0120]
点击复制

基于机器学习方法的上海市暴雨内涝灾情预测模型研究
分享到:

《自然灾害学报》[ISSN:/CN:23-1324/X]

卷:
30
期数:
2021年01期
页码:
191-200
栏目:
出版日期:
2021-02-28

文章信息/Info

Title:
A study on rainstorm waterlogging disaster prediction models in Shanghai based on machine learning
作者:
李海宏12 吴吉东2 王强1 杨辰1 潘顺1
1. 上海市气象灾害防御技术中心, 上海 200030;
2. 北京师范大学地理科学学部 环境演变与自然灾害教育部重点实验室, 北京 100875
Author(s):
LI Haihong12 WU Jidong2 WANG Qiang1 YANG Chen1 PAN Shun1
1. Shanghai Center for Meteorological Disaster Prevention Technology, Shanghai 200030, China;
2. Key Laboratory of Environmental Change and Natural Disaster of Ministry of Education, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China
关键词:
BP模型XGBoost模型机器学习暴雨内涝灾情灾情阈值上海
Keywords:
BP modelXGBoost modelmachine learningrainstorm waterlogging disasterdisaster thresholdShanghai
分类号:
P429;X43
DOI:
10.13577/j.jnd.2021.0120
摘要:
本文基于区域灾害系统理论,综合考虑致灾因子、暴露度和脆弱性,提出了一套暴雨内涝灾情预测指标体系;在此基础上利用上海市应急联动中心110接报暴雨内涝灾情数据,构建了暴雨内涝灾情预测BP模型(Back Propagation Model)和XGBoost模型(Extreme Gradient Boosting Model),并对比分析了预测模型效果,实现对上海市暴雨过程内涝灾情数量预测;最后对内涝灾情影响等级进行阈值划分,以期为暴雨内涝影响预报与风险预警业务、服务及灾害管理提供技术支撑。结果表明:1)综合考虑致灾因子、暴露度、脆弱性指标且不经主成分分析降维的指标组合作为暴雨内涝灾情预测指标体系时,BP模型和XGBoost模型的预测精度最优;2)全量样本XGBoost模型总体表现最优,暴雨内涝灾情的右偏分布和内涝灾情的异常高值均对XGBoost模型预测误差有不同程度的贡献;3)综合评价法在历史灾情百分位法和模拟灾情百分位法基础上,结合多年业务实践经验和用户对于110灾情的处置承受力对暴雨内涝灾情进行阈值划分,在实际应用中具有一定参考意义。
Abstract:
Based on the regional disaster system theory, this paper comprehensively considers the hazard, exposure and vulnerability, and puts forward a set of rainstorm waterlogging disaster forecasting index system. On this basis, it constructs the BP model (Back Propagation Model) and XGBoost model (Extreme Gradient Boosting Model) for the rainstorm waterlogging disaster prediction, comparatively analyzes the prediction models effect by using the rainstorm waterlogging disaster reports which originate from Shanghai Emergency Linkage Platform 110 alert phone calls, and realizes the quantitative prediction of the rainstorm waterlogging disaster. Finally, it classifies the impact-level of the rainstorm waterlogging with a view to provide technical support for rainstorm waterlogging impact-based forecast and risk-based warning services and disaster management. The results indicate that: 1) Comprehensively considering the hazard, exposure and vulnerability as the rainstorm waterlogging disaster prediction index system without principal component analysis, the prediction accuracy of BP model and XGBoost model is optimal;2) The overall performance of the XGBoost model based on the full sample is the best, the right-skewed distribution and the abnormally high value of the rainstorm waterlogging disaster numbers have important contribution to the prediction error of the XGBoost model;3) The comprehensive evaluation method divides the rainstorm waterlogging disaster threshold based on the historical disaster percentile and the simulated disaster percentile, and combines with professional practical experience for the 110 alert phone calls from the Shanghai Emergency Linkage Platform, it has certain reference significance in practical application.

参考文献/References:

[1] The Intergovernmental Panel on Climate Change.IPCC Fifth Assessment Synthesis Report:Climate Change 2014[R]. Geneva:IPCC, 2015.
[2] World Economic Forum. The Global Risks Report 2020(15 Edition)[R].Geneva:WEF, 2020.
[3] Christopher B F, Vicente B, Thomas F S, et al. Managing the risks of extreme events and disasters to advance climate change adaptation:special report of the intergovernmental panel on climate change[J]. Journal of Clinical Endocrinology & Metabolism, 2012, 18(6):586-599.
[4] 吴吉东,傅宇,张洁, 等. 1949-2013年中国气象灾害灾情变化趋势分析[J]. 自然资源学报, 2014, 29(9):1520-1530. WU Jidong, FU Yu, ZHANG Jie, et al. Meteorological disaster trend analysis in China:1949-2013[J]. Journal of Natural Resources, 2014, 29(9):1520-1530.(in Chinese)
[5] 刘敏, 王军, 殷杰, 等. 上海城市安全与综合防灾系统研究[J]. 上海城市规划, 2016, (1):1-8. LIU Min, WANG Jun, YIN Jie, et al. Study on the urban safety and comprehensive disaster prevention in Shanghai[J]. Shanghai Urban Planning Review,2016, (1):1-8.(in Chinese)
[6] 黄清雨, 董军刚, 李梦雅, 等. 暴雨内涝危险性情景模拟方法研究-以上海中心城区为例[J]. 地球信息科学学报,2016, 18(04):506-513. HUANG Qingyu, DONG Jungang, LI Mengya, et al. Research on the scenario simulation method of rainstorm waterlogging hazard:A case study in the central urban area of Shanghai[J]. Journal of Geo-information Science, 2016, 18(4):506-513.(in Chinese)
[7] 滕五晓, 夏剑霺, 万蓓蕾. 城市社区暴雨脆弱性评估研究-以上海市杨浦区为例[J]. 广州大学学报(社会科学版), 2018, 17(2):20-26. TENG Wuxiao, XIA Jianwei, WAN Beilei. On rainstorm vulnerability assessment of urban community:A case study on Yangpu District in Shanghai[J]. Journal of Guangzhou University (Social Science Edition), 2018, 17(2):20-26.(in Chinese)
[8] 徐艺扬, 李昆, 谢玉静, 等. 基于GIS的城市内涝影响因素及多元回归模型研究-以上海为例[J]. 复旦学报(自然科学版), 2018, 57(2):182-198. XU Yiyang, LI Kun, XIE Yujing, et al. Studies on influencing factors and multiple regression model of urban waterlogging based on GIS- A case of Shanghai, China[J]. Journal of Fudan University (Natural Science), 2018, 57(2):182-198.(in Chinese)
[9] 尹占娥, 暴丽杰, 殷杰. 基于GIS的上海浦东暴雨内涝灾害脆弱性研究[J]. 自然灾害学报, 2011, 20(2):29-35. YIN Zhane, BAO Lijie, YIN Jie. GIS-based study on vulnerability to rainstorm inundation in pudong of Shanghai[J]. Journal of Natural Disasters, 2011, 20(2):29-35.(in Chinese)
[10] 耿焕同, 吴正雪, 计浩军, 等. 基于GIS的上海市嘉定区暴雨积涝灾害风险区划研究[J]. 灾害学, 2015, 30(1):96-101. GENG Huantong, WU Zhengxue, JI Haojun, et al. Risk zonation and assessment of flood disaster in Jiading District of Shanghai[J]. Journal of Catastrophology, 2015, 30(1):96-101.(in Chinese)
[11] 张振国, 温家洪. 基于PGIS的城市社区居民暴雨内涝灾害风险认知-以上海市普陀区金沙居委地区为例[J]. 大连民族学院学报, 2015, 17(5):491-494. ZHANG Zhenguo, WEN Jiahong. Risk perception of urban community for rainstorm waterlogging based on PGIS-Taking the Jinsha community in Shanghai as a Case[J]. Journal of Dalian Nationalities University, 2015, 17(5):491-494.(in Chinese)
[12] 陆敏, 刘敏, 权瑞松, 等. 上海暴雨灾害的系统特征与脆弱性分析[J]. 华东师范大学学报(自然科学版), 2010, (2):9-15. LU Min, LIU Min, QUAN Ruisong, et al. Analysis of system characteristics and vulnerability of rainstorm hazards in Shanghai[J]. Journalof East China Normal University (Natural Science), 2010, (2):9-15.(in Chinese)
[13] 黄曦涛, 李怀恩, 张瑜, 等.基于PSR和AHP方法的西安市城市内涝脆弱性评价体系构建与脆弱度评估[J].自然灾害学报, 2019, 28(6):167-175. HUANG Xitao, LI Huaien, ZHANG Yu, et al. Construction of urban waterlogging vulnerability assessment system and vulnerability assessment based on PSR & AHP Method in Xi’an city[J]. Journal of Natural Disasters, 2019, 28(6):167-175.(in Chinese)
[14] 贺芳芳, 杨涵洧, 穆海振, 等. 上海地区短历时强降水致灾阈值探索[J]. 高原气象, 2017, 36(6):1567-1575. HE Fangfang, YANG Hanwei, MU Haizhen, et al. The exploratory research of thresholds value causing urban waterlogging of short diachronic strong precipitation in Shanghai[J]. Plateau Meteorology, 2017, 36(6):1567-1575.(in Chinese)
[15] 汪松年. 上海地区洪涝灾害的特点和防御对策(上)[J]. 上海建设科技, 2007, (5):24-25. WANG Songnian. Characteristics and preventive countermeasures of flood disasters in Shanghai (Part 1)[J]. Shanghai Construction Science & Technology, 2007, (5):24-25.(in Chinese)
[16] 汪松年. 上海地区洪涝灾害的特点和防御对策(下)[J]. 上海建设科技, 2007,(6):25-27. WANG Songnian. Characteristics and preventive countermeasures of flood disasters in Shanghai (Part 2)[J]. Shanghai Construction Science &Technology, 2007,(6):25-27.(in Chinese)
[17] 张炜,李思敏,时真男.我国城市暴雨内涝的成因及其应对策略[J].自然灾害学报,2012, 21(5):180-184. ZHANG Wei, LI Simin, SHI Zhennan.Formation causes and coping strategies of urban rainstorm waterlogging in China[J]. Journal of Natural Disasters, 2012, 21(5):180-184.(in Chinese)
[18] 杨辰, 王强, 顾宇丹. 上海市城市暴雨内涝评估建模及模拟研究[J]. 气象, 2017, 43(7):879-886. YANG Chen, WANG Qiang, GU Yudan. Study of Shanghai urban flooding assessment modeling and simulation[J]. Meteorology Monthly, 2017, 43(7):879-886.(in Chinese)
[19] 麻蓉,白涛,黄强,等.MIKE 21模型及其在城市内涝模拟中的应用[J].自然灾害学报, 2017, 26(4):172-179. MA Rong, BAI Tao, HUANG Qiang, et al.MIKE 21 model and its application on urban waterlogging simulation[J]. Journal of Natural Disasters, 2017, 26(4):172-179.(in Chinese)
[20] 史军, 穆海振, 杨涵洧, 等. 上海中心城区暴雨内涝阈值研究[J]. 暴雨灾害, 2016, 35(4):344-350. SHI Jun, MU Haizhen, YANG Hanwei, et al. Study on the threshold of rainstorm waterlogging in the urban areas of Shanghai[J]. Torrential Rain and Disaster, 2016, 35(4):344-350.(in Chinese)
[21] 赵思健,陈志远,熊利亚. 利用空间分析建立简化的城市内涝模型[J]. 自然灾害学报, 2004, 13(6):8-14. ZHAO Sijian, CHEN Zhiyuan, XIONG Liya. Establishment of simplified urban waterlogging model using spatial analysis[J]. Journal of Natural Disasters, 2004, 13(6):8-14.(in Chinese)
[22] 孙阿丽. 基于情景模拟的城市暴雨内涝风险评估[D]. 华东师范大学, 2011. SUN Ali. Risk Assessment of Rainstorm Waterlogging Based on Scenario Simulation in Cities[D]. East China Normal University, 2011.(in Chinese)
[23] 沈澄,孙燕,尹东屏,等. 江苏省暴雨洪涝灾害特征分析[J]. 自然灾害学报, 2015, 24(2):203-212. SHEN Cheng, SUN Yan, YIN Dongping, et al. Characteristic analysis of rainstorm-induced flood disaster in Jiangsu Province[J]. Journal of Natural Disasters,2015, 24(2):203-212.(in Chinese)
[24] Liu W H, Wu J D, Tang R M, et al. Daily precipitation threshold for rainstorm and flood disaster in the mainland of China:An economic loss perspective[J]. Sustainability, 2020, 12(1):407-421.
[25] 刘建华,张正,吴洁明. 基于BP神经网络的城市水灾灾情预测模型[J]. 计算机工程与设计, 2005, 26(3):699-701. LIU Jianhua, ZHANG Zheng, WU Jieming. City flood forecast model based on BP network[J]. Computer Engineering and Design, 2005, 26(3):699-701.(in Chinese)
[26] 卢宇丹. 基于混合优化的神经网络在气象灾情评估中的应用研究[D]. 南京信息工程大学, 2012. LU Yudan. Study of Meteorological Disaster Assessment Based on Hybrid Optimization Neural Network[D]. Nanjing University of Information Science & Technology, 2012.(in Chinese)
[27] 史培军.再论灾害研究的理论与实践[J].自然灾害学报, 1996, 5(4):8-19. SHI Peijun. Theory and practice of disaster study[J]. Journal of Natural Disasters, 1996, 5(4):8-19.(in Chinese)
[28] 李海宏, 吴吉东. 2007-2016年上海市暴雨特征及其与内涝灾情关系分析[J]. 自然资源学报, 2018, 33(12):2136-2148. LI Haihong, WU Jidong. Rainstorm characteristics and its relationship with waterlogging disaster in Shanghai during 2007-2016[J]. Journal of Natural Resources, 2018, 33(12):2136-2148.(in Chinese)
[29] Tianqi C, Carlos G. XGBoost. A Scalable Tree Boosting System[A].Proceedings of the 22nd ACM SIGKDD International Conference on knowledge Discovery and Data Mining[C]. New York:Association for Computing Machinery, 2016.

备注/Memo

备注/Memo:
收稿日期:2020-07-17;改回日期:2020-09-12。
基金项目:国家重点研发计划课题(2018YFC1508903);国家自然科学基金项目(42077437);上海市2020年度"科技创新行动计划"社会发展科技攻关定向项目(20DZ1200403)
作者简介:李海宏(1989-),女,工程师,硕士,主要从事暴雨内涝风险评估研究.E-mail:l.haihong@outlook.com
通讯作者:吴吉东(1981-),男,教授,博士,主要从事灾害经济学、自然灾害风险管理研究.E-mail:wujidong@bnu.edu.cn
更新日期/Last Update: 1900-01-01