长江流域资源与环境 >> 2023, Vol. 32 >> Issue (8): 1710-1723.doi: 10.11870/cjlyzyyhj202308014

• 生态环境 • 上一篇    下一篇

基于机器学习算法的洪涝灾害风险评估——以宜昌市为例

王德运1,2 ,张露丹1,吴祈1,郭海湘1,2,柯小玲1,吕新彪3   

  1. (1.中国地质大学(武汉)经济管理学院,湖北 武汉 430074;2.中国地质大学(武汉)自然灾害风险防控与应急管理实验室,湖北 武汉 430074;3.中国地质大学(武汉)高等研究院,湖北 武汉 430074)

  • 出版日期:2023-08-20 发布日期:2023-08-23

Flood Risk Assessment Based on Machine Learning Algorithms:A Case Study of Yichang City

WANG De-yun1,2,ZHANG Lu-dan1,WU Qi1,GUO Hai-xiang1,2,KE Xiao-ling1,LV Xin-biao3    

  1. (1. School of Economics and Management,China University of Geosciences,Wuhan 430074,China;2. Laboratory of Natural Disaster Risk Prevention and Emergency Management, China University of Geosciences,Wuhan 430074,China;3. Institute for Advanced Study, China University of Geosciences,Wuhan 430074,China)
  • Online:2023-08-20 Published:2023-08-23

摘要:  近年来,城市洪涝事件频发,对人民的生命和财产安全造成了严重影响。客观、准确的风险评估对于提升城市洪涝灾害防控水平至关重要,以宜昌市2020年6月末遭受的洪涝灾害为例进行城市洪涝灾害影响因子分析及风险评估。首先,基于RS遥感技术利用哨兵二号雷达影像提取和对比灾前、灾后水体形成淹没区范围并通过随机采样获取淹没点数据;然后,从致涝、孕涝、承涝和恢复能力4个方面选取16个指标,使用随机森林模型计算各指标贡献率并依据排序结果优化指标体系;最后,使用XGBoost模型对优化后的指标体系赋权并对宜昌市的洪涝灾害风险进行评估。结果显示:(1)在宜昌市洪涝灾害影响因素中,地形地貌及河流分布的影响>社会经济因素>气象因素;(2)高风险地区的范围与长江、清江、沮漳河、黄柏河、渔洋河等主要河系的分布关系十分密切,对于以上主要河系的水位线监测应保持高度敏感并制定针对性的应急管理措施;(3)低风险、较低风险和中风险地区占研究区域总面积的71.8%,但只包含8%的淹没点;而高风险地区仅占研究区域总面积的7.32%,但却包含81.33%的淹没点,表明高风险区域内洪涝灾害事件集中;(4)使用小尺度历史灾害事件对评估模型的验证结果显示,72%的验证点落在高风险及较高风险区,高达92%的验证点落在中高风险区。上述验证结果显示了该评估模型的有效性,其评估结果与宜昌市实际情况相符,研究解决了城市洪涝灾害风险难以精细化、定量化评估的部分问题并为城市洪涝风险管理、防灾减灾和区域规划提供科学参考。

Abstract: In recent years, flooding events happened frequently and severely affected people’s lives and property. Objective and accurate risk assessment is vital for urban flood risk prevention and emergency management. This paper takes the flood event happened in Yichang City in late June 2020 as an example to analyze the influencing factors of urban flooding and conduct risk assessment. Firstly, based on RS remote sensing technology, we extracted the inundation area of water bodies before and after flooding using Sentinel II radar images and conducted random sampling. Then, initially selected 16 basic indicators from four different perspectives: flood-causing, flood-pregnant, flood-bearing and recovery capability. Finally, the XGBoost model was used to assign weights to the optimized indicators and carry out risk assessment. The assessment results show that: (1) among the factors influencing flood risk in Yichang, the influence of topography and river distribution > socio-economic factors > meteorological factors; (2) the scope of high-risk areas is closely related to the distribution of major river systems such as the Yangtze River, Qingjiang River, Fuzhan River, Huangbai River and Yuyang River. The relative departments in Yichang should be highly sensitive to the water level of these river systems and make the essential emergency management measures; (3) the low-risk to medium-risk areas account for 71.8% of the total area of the region, but only contain 8% of the flood hazard sites; while the high-risk area only accounts for 7.32% of the total area, but contains 81.33% of the flood hazard sites, indicating a high flooding intensity in this area; (4) the assessment results of the model are verified to be consistent with the actual situation in Yichang City by using small-scale historical disaster data, the verification results show that 72% of the validation points fall into the high to higher-risk areas, and up to 92% of the validation points fall into the medium-high risk areas. This study solves the difficulty of quantitative assessment of flood risk at fine scales, and provides a useful reference for urban flood risk management, disaster prevention and mitigation efforts and regional planning.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 王锐, 余涛, 杨忠芳, 侯青叶, 曾庆良, 马宏宏. 富硒土壤硒生物有效性及影响因素研究[J]. 长江流域资源与环境, 2018, 27(07): 1460 .
[2] 付书科, 李小帆, 彭甲超, 张子悦, 陈 思. 基于动态因子分析法的长江三角洲城市群城镇化质量测度研究[J]. 长江流域资源与环境, 2018, 27(09): 1909 -1918 .
[3] 吴殿鸣, 邵大伟. 基于最佳尺度的城市扩张特征研究——以苏州中心城区为例[J]. 长江流域资源与环境, 2018, 27(09): 1937 -1946 .
[4] 阮甜 查芊郁 杨茹 高超. 全球升温1.5℃和2.0℃对长江寸滩站以上流域径流的影响[J]. 长江流域资源与环境, , (): 0 .
[5] 沈雪, 张露, 张俊飚, 骆兰翎. 稻农低碳生产行为影响因素与引导策略——基于人际行为改进理论的多组比较分析[J]. 长江流域资源与环境, 2018, 27(09): 2042 -2052 .
[6] 马坤, 唐晓岚, 刘思源, 王奕文, 任宇杰, 刘小涵. 长江流域国家级保护地空间分布特征及其国家公园廊道空间策略研究[J]. 长江流域资源与环境, 2018, 27(09): 2053 -2069 .
[7] 曾 刚, 杨舒婷, 王丰龙, . 长江经济带城市协同发展能力研究[J]. 长江流域资源与环境, 2018, 27(12): 2641 -2650 .
[8] 查凯丽, 刘艳芳, 孔雪松, 田雅丝, 刘耀林, . 村镇路网通达性与空间出行研究——以武汉市李集镇为例[J]. 长江流域资源与环境, 2018, 27(12): 2663 -2672 .
[9] 起永东, 何明琼, 郑永宏, 高 洁, 王 丹, 孔繁希. 汉江流域降水结构时空特征及影响因素分析[J]. 长江流域资源与环境, 2018, 27(12): 2830 -2838 .
[10] 梁志会, 张露, 张俊飚, Jorge Ruiz Menjivar, 刘勇. 小农发展气候智慧型农业的效率与成本改进:倡导农地流转还是发展社会服务?[J]. 长江流域资源与环境, 2019, 28(05): 1164 -1175 .