基于深度学习灾害情景构建的热带气旋降雨－山洪风险地图生成研究

Research on the generation of tropical cyclone rainfall-induced flash flood risk maps based on disaster scenario construction using deep learning

摘要

热带气旋降雨引发的山洪灾害的历史气象事件观测数量有限，容易导致风险分布估算的偏差。为了更全面准确地评估热带气旋降雨－山洪灾害对建成环境的风险规律，本文提出了一种基于深度学习方法高效构建灾害情景，并结合10000a热带气旋事件大样本开展风险分析，以此得到完备的山洪风险数据。以杭州市岛石镇为例展示模型框架，基于大样本统计分析的风险分布绘制不同重现期的热带气旋降雨－山洪风险地图。研究结果表明，构建的深度学习代理模型能够高效模拟热带气旋降雨－山洪淹没，在保证精度的同时大幅提升了灾害情景生成效率，为生成大量灾害样本提供了可行方法。基于模拟结果绘制的风险地图揭示了研究区域内风险暴露的空间异质性，岛石村在中低重现期事件中表现出较高的建筑风险暴露，江川村则呈现平稳的风险特征，而新桥村和下许村则在百年一遇以上极端情景中表现出明显的风险跃升。这些特征反映出地形、建筑布局与热带气旋降雨－山洪灾害之间的复杂关系，为因地制宜制定防灾策略提供了科学依据。

Abstract

Limited historical meteorological observations of flash floods triggered by tropical cyclone rainfall can easily lead to biases in risk distribution estimation. To more comprehensively and accurately assess the risk patterns of tropical cyclone rainfall-flash flood disasters on the built environment, this paper proposes a surrogate modeling framework based on deep learning to efficiently construct disaster scenarios, conducts risk analysis using a large sample of 10000-year tropical cyclone events and obtains complete flash flood risk data. Taking Daoshi Town in Hangzhou City as an example, the model framework is demonstrated, and risk maps of tropical cyclone rainfall-flash floods with different return periods are drawn based on the risk distribution from large-sample statistical analysis. The results show that the constructed deep learning surrogate model can efficiently simulate tropical cyclone rainfall-flash flood inundation, significantly improve the efficiency of disaster scenario generation while maintaining accuracy, and provides a feasible method for generating a large number of disaster samples. The risk maps drawn based on the simulation results reveal the spatial heterogeneity of risk exposure in the study area. Daoshi Village shows high building exposure in low-to-medium return period events, Jiangchuan Village shows stable risk characteristics, while Xinqiao Village and Xiaxu Village show significant risk jumps in extreme scenarios with a return period of more than 100 years. These characteristics reflect the complex relationship between topography, building layout, and tropical cyclone rainfall-flash flood disasters, providing a scientific basis for developing disaster prevention strategies tailored to local conditions.

关键词

热带气旋降雨山洪 / 风险地图 / 深度学习 / 重现期 / 极值分析

Key words

tropical cyclone rainfall-induced flash flood / flood risk map / deep learning / return period / extreme value analysis

参考文献

[1] Liu Yesen, Yang Zhenshan, Huang Yaohuan, et al. Spatiotemporal evolution and driving factors of China’s flash flood disasters since 1949[J]. Science China Earth Sciences, 2018, 61(12): 1804-1817.

[2] 崔鹏, 邹强. 山洪泥石流风险评估与风险管理理论与方法[J]. 地理科学进展, 2016, 35(2): 137-147.

[3] Cui Peng, Zou Qiang. Theory and method of risk assessment and risk management of debris flows and flash floods[J]. Progress in Geography, 2016, 35(2): 137-147. (in Chinese)

[4] 刘芳. 洪涝灾害风险评估方法研究: 以浙江省台风暴雨为例[D]. 杭州: 浙江大学, 2021.

[5] Liu Fang. Research on flood disaster risk assessment method: a case study of typhoon rainstorm[D]. Hangzhou: Zhejiang University, 2021. (in Chinese)

[6] 帅嘉冰, 徐伟, 史培军. 长三角地区台风灾害链特征分析[J]. 自然灾害学报, 2012, 21(3): 36-42.

[7] Shuai Jiabing, Xu Wei, Shi Peijun. Characteristic analysis of typhoon disaster chains in the Yangtze River Delta region of China[J]. Journal of Natural Disasters, 2012, 21(3): 36-42. (in Chinese)

[8] 苏朝晖, 方伟华. 1980-2015年中国沿海地区热带气旋暴露度变化分析[J]. 自然灾害学报, 2023, 32(3): 102-117.

[9] Su Zhaohui, Fang Weihua. Analysis of tropical cyclone exposure changes in coastal areas of China from 1980 to 2015[J]. Journal of Natural Disasters, 2023, 32(3): 102-117. (in Chinese)

[10] 翁建中, 刘青. 浙江: 发展巨灾保险助力完善防灾减灾救灾体系[J]. 中国减灾, 2024(21): 38-39.

[11] Weng Jianzhong, Liu Qing. Zhejiang: developing catastrophe insurance to improve the disaster prevention, mitigation, and relief system[J]. Disaster Reduction in China, 2024(21): 38-39. (in Chinese)

[12] 李颖, 李琼, 邱静怡. 气象巨灾保险的现状、挑战、机遇与参与对策研究[J]. 浙江气象, 2020, 41(2): 30-35, 49.

[13] Li Ying, Li Qiong, Qiu Jingyi. Research on the current situation, challenges, opportunities, and participation strategies of meteorological catastrophe insurance[J]. Journal of Zhejiang Meteorology, 2020, 41(2): 30-35, 49. (in Chinese)

[14] 尹宜舟, Marco Gemmer, 苏布达, 等 . 台风灾害气象指数保险相关技术方法初探[J]. 自然灾害学报, 2012, 21(3): 28-35.

[15] Yin Yizhou, Gemmer Marco, Su Buda, et al. Preliminary study on technique for weather index insurance of typhoon disaster[J]. Journal of Natural Disasters, 2012, 21(3): 28-35. (in Chinese)

[16] 杨冰君. 浙江省台风保险的实施方案研究: 基于宁波巨灾保险试点的推广[D]. 杭州: 浙江工商大学, 2018.

[17] Yang Bingjun. Research on the plan of typhoon insurance in Zhejiang: based on the promotion of pilot[D]. Hangzhou: Zhejiang Gongshang University, 2018. (in Chinese)

[18] Chen Liutong, Yan Zhengtao, Li Qian, et al. Flash flood risk assessment and driving factors: a case study of the Yantanxi River Basin, southeastern China[J]. International Journal of Disaster Risk Science, 2022, 13(2): 291-304.

[19] Morss R E, Wilhelm O V, Downton M W, et al. Flood risk, uncertainty, and scientific information for decision making[J]. Bulletin of the American Meteorological Society, 2005, 86(11): 1593-1601.

[20] Merwade V, Olivera F, Arabi M, et al. Uncertainty in flood inundation mapping: current issues and future directions[J]. Journal of Hydrologic Engineering, 2008, 13(7): 608-620.

[21] Bevere L. Disaster loss data: the case of Swiss Re Institute’s sigma[C]// Geophysical Research Abstracts. Australia: EGU Publications, 2019: 21.

[22] Beguería S, Vicente-Serrano S M. Mapping the hazard of extreme rainfall by peaks over threshold extreme value analysis and spatial regression techniques[J]. Journal of Applied Meteorology and Climatology, 2006, 45(1): 108-124.

[23] Lumbroso D M, Boyce S, Bast H, et al. The challenges of developing rainfall intensity-duration-frequency curves and national flood hazard maps for the Caribbean[J]. Journal of Flood Risk Management, 2011, 4(1): 42-52.

[24] Zhou Y, Matyas C J. Regionalization of precipitation associated with tropical cyclones using spatial metrics and satellite precipitation[J]. GIScience & Remote Sensing, 2021, 58(4): 542-561.

[25] Chen Lingfang, Chen Yangbo, Zhang Yueyuan, et al. Spatial patterns of typhoon rainfall and associated flood characteristics over a mountainous watershed of a tropical island[J]. Journal of Hydrology, 2022, 613: 128421.

[26] Nabukulu C, Jetten V G, Ettema J, et al. Implications of tropical cyclone rainfall spatial-temporal variability on flood hazard assessments in the Caribbean Lesser Antilles[J]. GeoHazards, 2024, 5(4): 1275-1293.

[27] National Research Council, Water Science, Technology Board, et al. Mapping the zone: improving flood map accuracy[M]. Washington, D.C.: National Academies Press, 2009.

[28] Grey S, Liu Y, Simmons J. Hurricane surge and inundation in the Bahamas, Part 2: flood risk assessment[J]. Journal of Flood Risk Management, 2025, 18(1): e13022.

[29] Su Z Z, Ren F M, Wei J, et al. Changes in monsoon and tropical cyclone extreme precipitation in southeast China from 1960 to 2012[J]. Tropical Cyclone Research and Review, 2015, 4(1): 12-17.

[30] Apel H, Merz B, Thieken A H. Quantification of uncertainties in flood risk assessments[J]. International Journal of River Basin Management, 2008, 6(2): 149-162.

[31] Merz B, Thieken A H. Flood risk curves and uncertainty bounds[J]. Natural Hazards, 2009, 51(3): 437-458.

[32] Penning-Rowsell E, Viavattene C, Pardoe J, et al. The benefits of flood and coastal risk management: a handbook of techniques-2010[M]. London: Flood Hazard Research Centre, Middlesex University, 2010.

[33] Meyer V, Messner F, Penning-Rowsell E, et al. Evaluating flood damages: guidance and recommendations on principles and methods[J]. Integrated Flood Risk Analysis and Management Methodologies, 2007, 21(1): 9.

[34] Ward P J, De Moel H, Aerts J C J H. How are flood risk estimates affected by the choice of return-periods?[J]. Natural Hazards and Earth System Sciences, 2011, 11(12): 3181-3195.

[35] 李颖, 方伟华. 热带气旋降水重现期估算研究[J]. 自然灾害学报, 2014, 23(6): 58-69.

[36] Li Ying, Fang Weihua. Estimation on return period of tropical cyclone precipitation[J]. Journal of Natural Disasters, 2014, 23(6): 58-69. (in Chinese)

[37] Nofal O M, Van De Lindt J W. Understanding flood risk in the context of community resilience modeling for the built environment: Research needs and trends[J]. Sustainable and Resilient Infrastructure, 2022, 7(3): 171-187.

[38] 任晓航. 宁波市公共巨灾保险实施效果及其评价研究[D]. 宁波: 宁波大学, 2022.

[39] Ren Xiaohang. Study on the implementation effect and evaluation of public catastrophe insurance in Ningbo City[D]. Ningbo: Ningbo University, 2022. (in Chinese)

[40] 俞嘉庆, 俞哲农, 汪飞鹏. 基于洪水管理理念的防汛应急管理实践研究: 以浙江省杭州市临安区为例[J]. 中国减灾, 2021(17): 54-57.

[41] Yu Jiaqing, Yu Zhenong, Wang Feipeng. A study on flood emergency management practice based on flood management concepts: a case study of Lin’an District, Hangzhou City, Zhejiang Province[J]. Disaster Reduction in China, 2021(17): 54-57. (in Chinese)

[42] 临安区应急管理局. 杭州市临安区防汛防台抗旱应急预案[EB/OL]. (2015-07-31)[2026-06-18]. https://zfgb.hangzhou.gov.cn/10/105220253/t122220253054/519806.shtml.

[43] Lin’an District Emergency Management Bureau. Emergency plan for flood control, typhoon prevention, and drought relief in Lin’an district, Hangzhou[EB/OL]. (2015-07-31)[2026-06-18]. https://zfgb.hangzhou.gov.cn/10/105220253/t122220253054/519806.shtml. (in Chinese)

[44] Ying Ming, Zhang Wei, Yu Hui, et al. An overview of the China meteorological administration tropical cyclone database[J]. Journal of Atmospheric and Oceanic Technology, 2014, 31(2): 287-301.

[45] Liu Xiaoqin, Yu Hui, Ying Ming, et al. Western north Pacific tropical cyclone database created by the China Meteorological Administration[J]. Advances in Atmospheric Sciences, 2021, 38(4): 690-699.

[46] USDA-NRCS (United States Department of Agriculture-Natural Resources Conservation Service). Estimated direct runoff from storm rainfall[J]. National Engineering Handbook: Chapter 10, 2004, 10(6): 630.

[47] 任少华. 基于GIS和DEM的缺资料地区小流域分布式水文模型研究[D]. 郑州: 郑州大学, 2017.

[48] Ren Shaohua. Study on distributed hydrological model based on GIS and DEM in small basin of deficient data[D]. Zhengzhou: Zhengzhou University, 2017. (in Chinese)

[49] Che Yangzi, Li Xuecao, Liu Xiaoping, et al. Building height of Asia in 3D-GloBFP[DS/OL]. V1. Zenodo[ 2024-10-28]. https://zenodo.org/records/11397015.

[50] Vickery P J, Skerlj P F, Twisdale L A. Simulation of hurricane risk in the U.S. using empirical track model[J]. Journal of Structural Engineering, 2000, 126(10): 1222-1237.

[51] Vickery P J, Skerlj P F, Steckley A C, et al. Hurricane wind field model for use in hurricane simulations[J]. Journal of Structural Engineering, 2000, 126(10): 1203-1221.

[52] Vickery P J, Wadhera D, Twisdale L A Jr, et al. U.S. hurricane wind speed risk and uncertainty[J]. Journal of Structural Engineering, 2009, 135(3): 301-320.

[53] Darling R W R. Estimating probabilities of hurricane wind speeds using a large-scale empirical model[J]. Journal of Climate, 1991, 4(10): 1035-1046.

[54] Hong Hanping, Li Sihan, Duan Zhongdong. Typhoon wind hazard estimation and mapping for coastal region in the mainland of China[J]. Natural Hazards Review, 2016, 17(2): 04016001.

[55] Geoghegan K M, Fitzpatrick P, Kolar R L, et al. Evaluation of a synthetic rainfall model, P-CLIPER, for use in coastal flood modeling[J]. Natural Hazards, 2018, 92(2): 699-726.

[56] 杨勇川, 王俊彦, 文海家, 等 . 基于深度学习的山洪时空预测代理模型[J]. 自然灾害学报, 2024, 33(4): 164-175.

[57] Yang Yongchuan, Wang Junyan, Wen Haijia, et al. Surrogate model for spatio-temporal prediction of flash flooding based on deep learning[J]. Journal of Natural Disasters, 2024, 33(4): 164-175. (in Chinese)

[58] 彭清昭. 台风演化过程中面向山区村镇的应急疏散决策研究[D]. 杭州: 浙江大学, 2024.

[59] Peng Qingzhao. Research on emergency evacuation decision-making for mountainous villages and towns during typhoon evolution[D]. Hangzhou: Zhejiang University, 2024. (in Chinese)

[60] Do Lago C A F, Giacomoni M H, Bentivoglio R, et al. Generalizing rapid flood predictions to unseen urban catchments with conditional generative adversarial networks[J]. Journal of Hydrology, 2023, 618: 129276.

[61] Guo Z F, Moosavi V, Leitão J P. Data-driven rapid flood prediction mapping with catchment generalizability[J]. Journal of Hydrology, 2022, 609: 127726.

[62] Zhou Y R, Wu W Y, Natha N R, et al. Deep learning-based rapid flood inundation modeling for flat flood plains with complex flow paths[J]. Water Resources Research, 2022, 58(12): e2022WR033214.

[63] Löwe R, Böhm J, Jensen D G, et al. U-FLOOD-Topographic deep learning for predicting urban pluvial flood water depth[J]. Journal of Hydrology, 2021, 603: 126898.

[64] Hosseiny H. A deep learning model for predicting river flood depth and extent[J]. Environmental Modelling & Software, 2021, 145: 105186.

[65] 车子杰, 高飞, 吴兆福, 等 . 基于改进U-Net网络的多源遥感影像洪涝灾害信息提取与变化分析[J]. 测绘通报, 2022(1): 26-32.

[66] Che Zijie, Gao Fei, Wu Zhaofu, et al. Multi-source remote sensing image flood disaster information extraction and change analysis based on improved U-Net network[J]. Bulletin of Surveying and Mapping, 2022(1): 26-32. (in Chinese)

[67] Guo Z F, Leitão J P, Simões N E, et al. Data-driven flood emulation: speeding up urban flood predictions by deep convolutional neural networks[J]. Journal of Flood Risk Management, 2021, 14(1): e12684.

[68] He K, Zhang X, Ren S, et al. Delving deep into rectifiers: surpassing human-level performance on imagenet classification[C]//Computer Vision Foundation. Proceedings of the IEEE International Conference on Computer Vision, 2015: 1026-1034.

[69] 梁峰铭, 王洁, 林诚杰, 等 . 基于临界雨量不确定性的山洪概率预报研究[J]. 水力发电学报, 2024, 43(9): 35-46.

[70] Liang Fengming, Wang Jie, Lin Chengjie, et al. Study on probabilistic forecasting of flash flood events based on critical rainfall uncertainty[J]. Journal of Hydroelectric Engineering, 2024, 43(9): 35-46. (in Chinese)

[71] Neubert M, Naumann T, Hennersdorf J, et al. The geographic information system-based flood damage simulation model HOWAD[J]. Journal of Flood Risk Management, 2016, 9(1): 36-49.

[72] Zahura F T, Goodall J L, Sadler J M, et al. Training machine learning surrogate models from a high-fidelity physics-based model: application for real-time street-scale flood prediction in an urban coastal community[J]. Water Resources Research, 2020, 56(10): e2019WR027038.

[73] 高曹珀, 王自法, 周阳, 等 . 基于大样本模拟的巨灾损失计算方法[J]. 保险研究, 2023(3): 105-116.

[74] Gao Caopo, Wang Zifa, Zhou Yang, et al. The calculation method for catastrophe losses based on large-scale sampling[J]. Insurance Studies, 2023(3): 105-116. (in Chinese)

[75] Shen Z H, Wei K. Stochastic model of tropical cyclones along China coast including the effects of spatial heterogeneity and ocean feedback[J]. Reliability Engineering & System Safety, 2021, 216: 108000.

[76] Bates P D, Quinn N, Sampson C, et al. Combined modeling of US fluvial, pluvial, and coastal flood hazard under current and future climates[J]. Water Resources Research, 2021, 57(2): e2020WR028673.

[77] GB 50201- 2014 防洪标准[S]. 北京: 中国计划出版社, 2014.

[78] GB 50201- 2014 Standard for flood control[S]. Beijing: China Planning Press, 2014. (in Chinese)

[79] 赵庆良. 沿海山地丘陵型城市洪灾风险评估与区划研究[D]. 上海: 华东师范大学, 2010.

[80] Zhao Qingliang. Research on flood risk assessment and risk zoning in coastal mountainous and hilly city[D]. Shanghai: East China Normal University, 2010. (in Chinese)

基金资助

浙江省科学技术厅“尖兵领雁研发攻关计划项目”(2024C03255)