基于LightGBM的台风直接经济损失评估与预测——以福建省为例
张之夏(2004—),女,辽宁辽阳人,本科生,研究方向为灾害经济损失,(E-mail)zzx4712@163.com; |
收稿日期: 2025-02-11
修回日期: 2025-03-21
网络出版日期: 2025-05-07
基金资助
国家重点研发计划(2024YFC3016800)
国家资助博士后研究人员计划B档(GZB20230966)
Assessment and Prediction of Typhoon-Related Direct Economic Loss in Fujian Province Based on LightGBM
Received date: 2025-02-11
Revised date: 2025-03-21
Online published: 2025-05-07
台风是中国沿海地区最严重的自然灾害之一,常导致重大经济损失。准确评估和预测台风直接经济损失对于提升防灾减灾能力和优化资源配置至关重要。文章以福建省84个区县为研究对象,基于2009—2021年影响福建省的30场台风灾害数据,结合致灾因子、孕灾环境因子和承灾体暴露度因子,共计20个关键影响因子,采用LightGBM方法构建台风直接经济损失预测模型,对台风风险进行定量评估并通过实际案例探讨模型在台风实际直接经济损失动态预测中的适用性。模型重要性分析表明,日最大风速、河网密度、日最大降水、累计降水、单位面积GDP和城市化率是影响福建省台风直接经济损失的主要因素。文章构建的模型在训练集上Pearson相关系数R达到0.836、可决系数R 2达到0.66,通过4个超强台风案例验证模型性能,预测直接经济损失与实际的损失相关系数在0.6~0.71,表明模型具有较好的应用潜力。以超强台风“莫兰蒂”为例,利用所建模型开展动态预测应用,结果显示模型较好地模拟了台风动态变化过程中直接经济损失的动态分布变化,可为福建省及其他沿海地区的台风灾害损失评估和应急管理提供科学支持。
张之夏 , 杨剑 , 陈思孝 , 林森 . 基于LightGBM的台风直接经济损失评估与预测——以福建省为例[J]. 热带地理, 2025 , 45(4) : 648 -659 . DOI: 10.13284/j.cnki.rddl.20250080
Typhoons are among the most destructive natural disasters affecting China's coastal regions, often resulting in substantial economic loss and casualties. The annual average Direct Economic Loss (DEL) caused by typhoon disasters in China exceeds 60 billion yuan, accounting for 10%-30% of the DEL caused by all disasters each year. Consequently, the accurate assessment and prediction of typhoon-induced DEL are essential for improving disaster mitigation strategies and optimizing resource allocation. Rapid development of artificial intelligence and the growth of multi-source spatiotemporal big data have introduced data-driven methods for assessing disaster losses. These methods have the advantage of using large samples to improve adaptability and consider more risk factors. In this study, DELs of 30 typhoon events in Fujian Province at the county level and a total of 911 samples were collected from 2009 to 2021 to establish an assessment model. Owing to the large range of the DEL in different districts and counties during the same typhoon, the logarithm of the DEL was used as the model output. This study included three steps for constructing the model. First, 24 influencing factors of typhoons, including disaster-inducing factors, disaster-forming environmental factors, and disaster-bearing body exposure factors, were calculated using the Pearson correlation coefficient and variance inflation coefficient to analyze the multicollinearity effect, and 20 key factors were selected to assess the DEL. Second, a LightGBM-based model is developed using the selected indicator factors as model inputs. Of the 911 samples, 734 were used to train the model, and 177 were used for validation. Finally, Super Typhoon Meranti was used as a case study to evaluate the applicability of the model in the dynamic DEL assessment of a typhoon. This study evaluated predictive performance of the model using five indicators: the Pearson correlation coefficient (R), coefficient of determination (R 2), mean squared error, mean absolute error, and median absolute error. The importance of LightGBM factors shows that the maximum daily wind speed, river network density, maximum daily precipitation, cumulative precipitation, and GDP per unit area are the primary determinants of typhoon-induced economic losses in Fujian Province. In the training set, R between the predicted results of the model and the actual loss was 0.836, and R 2 was 0.66, indicating good fitting ability. In real-world applications, the proposed model effectively captured the spatial distribution of losses from Typhoon Meranti, demonstrating its potential for disaster loss prediction. This study provides valuable insights into typhoon risk assessment and emergency management in Fujian Province and other coastal areas. We sorted the relevant research literature and found that economic loss assessment is more difficult than population, housing, and other loss assessments because economic loss is a comprehensive statistical indicator in China. Therefore, we drew on the method of processing DEL as logarithms from the literature. By comparing with other studies, the results of this study can improve model performance in terms of data quality inspection and sample size.
表1 本研究使用数据描述及其来源Table 1 Description and sources of the data used in this study |
影响因子 | 数据描述 | 数据来源 |
---|---|---|
致灾因子 | 日最大风速/(m·s-1) | 欧洲中期天气预报中心ERA5历史再分析数据 (https://cds.climate.copernicus.eu/datasets/reanalysis-era5-pressure-levels) |
日最大降水/mm | ||
日累计降水/mm | ||
孕灾环境 | 地形高程/m | ASTER GDEM全球30 m分辨率数字高程数据(https://search.earthdata.nasa.gov/search) |
坡度/(°) | ||
河流强度指数 | ||
地形湿度指数 | ||
公路密度/(km·km-2) | 全国地理信息资源目标服务系统的1∶100万基础地理信息数据 (https://www.webmap.cn/commres.do?method=result100W) | |
河网密度/(km·km-2) | ||
地貌类别(6个) | 全国30 m分辨率土地利用遥感监测数据集(https://www.resdc.cn/DOI/DOI.aspx?DOIID=54) | |
土壤类别(6个) | 1:100万全国土壤类别空间分布数据集(https://www.resdc.cn/data.aspx?DATAID=145) | |
承灾体暴露度 | 人均GDP/元 | 福建省统计局(https://tjj.fujian.gov.cn/xxgk/njgb/) |
单位面积GDP/(元·km-2) | ||
城市化率/% | ||
灾害损失 | 福建省台风灾害区县级 直接经济损失/元 | 自然灾害灾情管理系统(https://www.nndims.com) |
表2 各影响因子的方差膨胀系数VIFTable 2 Variance inflation factors (VIFs) of different impact factor |
影响因子 | VIF | 影响因子 | VIF |
---|---|---|---|
最大风速 | 1.60 | 地形湿度指数 | 2.56 |
最大降水 | 2.36 | 公路密度 | 4.61 |
累计降水 | 2.35 | 河网密度 | 2.61 |
地形高度 | 6.40 | 人均GDP | 1.69 |
坡度 | 16.62 | 单位面积GDP | 2.60 |
河流强度指数 | 3.61 | 城市化率 | 2.29 |
|
表3 模型最优参数设置Table 3 Optimal parameters of the model |
模型参数 | 最优参数设置 |
---|---|
叶节点数 | 31 |
树的数量 | 300 |
最大深度 | 7 |
学习率 | 0.01 |
L1正则化 | 0.3 |
L2正则化 | 0.5 |
采样部分特征 | 0.9 |
采样部分数据 | 0.7 |
表4 模型性能评估指标Table 4 Model performance evaluation indicators |
性能评估指标 | 训练集 | 测试集 |
---|---|---|
R | 0.836 | 0.654 |
R 2 | 0.660 | 0.427 |
MSE | 1.158 | 1.425 |
MAE | 1.538 | 1.863 |
MedAE | 0.889 | 1.203 |
张之夏:负责数据处理、实验结果分析、论文撰写和修改;
杨 剑:负责模型训练,参与论文撰写和修改;
陈思孝:参与模型训练,参与论文修改;
林 森:负责研究数据收集和技术路线,参与文章修改。
包文轩,宫阿都,徐澯,李博艺,黄智卿. 2023. 基于区域灾害系统论的广东省台风灾害风险评估——以“山竹”台风为例. 北京师范大学学报(自然科学版),59(1):94-103.
Bao Wenxuan, Gong Adu, Xu Can, Li Boyi, and Huang Zhiqing. 2023. Typhoon Disaster Risk Assessment in Guangdong Province by Regional Disaster System Theory: Ase of "Mangkhut". Journal of Beijing Normal University(Natural Science), 59(1): 94-103.
|
Burton I, Kates R W, and White G F. 1993. The Environment as Hazard. New York: The Guilford Press.
|
陈佩燕,杨玉华,雷小途,钱燕珍. 2009. 我国台风灾害成因分析及灾情预估. 自然灾害学报,18(1):64-73.
Chen Peiyan, Yang Yuhua, Lei Xiaotu, and Qian Yanzhen. 2009. Cause Analysis and Preliminary Hazard Estimate of Typhoon Disaster in China. Jounal of Natural Disasters, 18(1): 64-73.
|
葛全胜,邹铭,郑景云. 2008. 中国自然灾害风险综合评估初步研究. 北京:科学出版社.
Ge Quansheng, Zou Ming, and Zheng Jingyun. 2008. Integrated Assessment of Natural Disaster Risks in China. Beijing: Science Press.
|
Guo Guizhen, Liu Lulu, Li Yuqing, Gao Jiangbo, Lin Sen, and Wu Shaohong. 2021. A Vulnerability Curve Method to Assess Risks of Climate-Related Hazards at County Level. Atmosphere, 12(8): 1022.
|
Hersbach H, Bell B, Berrisford P, Biavati G, Horányi A, Muñoz Sabater J, Nicolas J, Peubey C, Radu R, Rozum I, Schepers D, Simmons A, Soci C, Dee D, and Thépaut J N. 2018. ERA5 Hourly Data on Pressure Levels from 1940 to Present. (2025-01-10) [2025-04-29]. https://cds.climate.copernicus.eu/datasets/reanalysis-era5-pressure-levels.
|
Huang M, Wang Q, Liu M, Lin N, Wang Y, Jing R, Sun J, Murakami H, and Lou W. 2022. Increasing Typhoon Impact and Economic Losses Due to Anthropogenic Warming in Southeast China. Scientific Reports, 12(1): 14048.
|
IPCC. 2022. AR6 Climate Change 2022: Mitigation of Climate Change. Cambridge: Cambridge University Press.
|
Ke Guolin, Meng Qi, Finley T, Wang Taifeng, Chen Wei, Ma Weidong, Ye Qiwei, and Liu Tieyan. 2017. LightGBM: A Highly Efficient Gradient Boosting Decision Tree. Neural Information Processing Systems. Long Beach, USA: 31st Conference on Neural Information Processing Systems.
|
李宁,张正涛,陈曦,冯介玲. 2017. 论自然灾害经济损失评估研究的重要性. 地理科学进展,36(2):256-263.
Li Ning, Zhang Zhengtao, Chen Xi, and Feng Jieling. 2017. Importance of Economic Loss Evaluation in Natural Hazard and Disaster Research. Progress in Geography, 36(2): 256-263.
|
刘蓓蓓,赵飞,王曦,闫雪,林森. 2024. 多源数据驱动的台风动态风险评估. 热带地理,44(6):1102-1112.
Liu Beibei, Zhao Fei, Wang Xi, Yan Xue, and Lin Sen. 2024. Multi-Source Data-Driven Modeling of Typhoon Dynamic Risk Assessment. Tropical Geography, 44(6): 1102-1112.
|
Lin Peihui and Wang Naiyu. 2024. A Data-Driven Approach for Regional-Scale Fine-Resolution Disaster Impact Prediction under Tropical Cyclones. Natural Hazards, 120: 7461-7479.
|
林森,刘蓓蓓,闫雪,孙宁,郭桂祯. 2022. 灾害大数据驱动的县域重大洪涝过程灾害风险评估. 灾害学,37(4):166-172.
Lin Sen, Liu Beibei, Yan Xue, Sun Ning, and Guo Guizhen. 2022. Disaster Risk Assessment at County Level of a Heavy Flooding Driven by Disaster Big Data. Journal of Catastrophology, 37(4): 166-172.
|
邵婧妍,方伟华. 2024. 基于机器学习算法的热带气旋灾害县级直接经济损失等级评估:以福建省为例. 热带地理,44(6):1064-1078.
Shao Jingyan and Fang Weihua. 2024. Assessment of Direct Economic Loss Levels Caused by Tropical Cyclone Disasters in County-Level Using Machine Learning: A Case Study of Fujian Province. Tropical Geography, 44(6): 1064-1078.
|
史培军. 1991. 灾害研究的理论与实践. 南京大学学报(自然科学版),(11):37-42. [Shi Peijun. 1991. On the Theory of Disaster Research and Its Practice. Journal of Nanjing University, (11): 37-42. ]
|
薛建军,李佳英,张立生,王秀荣,许映龙. 2012. 我国台风灾害特征及风险防范策略. 气象与减灾研究,35(1):59-64.
Xue Jianjun, Li Jiaying, Zhang Lisheng, Wang Xiurong, and Xu Yinglong. 2012. Characteristics of Typhoon Disasters in China and Risk Prevention Strategies. Meteorology and Disaster Reduction Research, 35(1): 59-64.
|
Wang Zhenkang, Xia Nan, Zhao Xin, Ji XianKai, and Wang Jiechen. 2024. Comprehensive Risk Assessment of Typhoon Disasters in China's Coastal Areas Based on Multi-Source Geographic Big Data. Science of the Total Environment, 926: 171815.
|
王军,谭金凯. 2021. 气候变化背景下中国沿海地区灾害风险研究与应对思考. 地理科学进展,40(5):870-882.
Wang Jun and Tan Jinkai. 2021. Understanding the Climate Change and Disaster Risks in Coastal Areas of China to Develop Coping Strategies. Progress in Geography, 40(5): 870-882.
|
Yang Jian, Chen Sixiao, Tang Yanan, Lu Ping, Lin Sen, Duan Zhongdong, and Ou Jinping. 2024. A Tropical Cyclone Risk Prediction Framework Using Flood Susceptibility and Tree-Based Machine Learning Models: County-Level Direct Economic Loss Prediction in Guangdong Province. International Journal of Disaster Risk Reduction, 114: 104955.
|
杨奕杰. 2022. 基于近20年历史数据的中国沿海城市台风灾害风险评估. 上海:上海海洋大学.
Yang Yijie. 2022. Risk Assessment of Typhoon Disasters in Coastal Cities in China Based on Historical Data in the Past 20 Years. Shanghai: Shanghai Ocean University.
|
殷洁,戴尔阜,吴绍洪. 2013. 中国台风灾害综合风险评估与区划. 地理科学,33(11):1370-1376.
Yin Jie, Dai Erfu, and Wu Shaohong. 2013. Integrated Risk Assessment and Zoning of Typhoon Disasters in China. Scientia Geographic Sinica, 33(11): 1370-1376.
|
袁艺. 2010. 自然灾害灾情评估研究与实践进展. 地球科学进展,25(1):22-32. [Yuan Yi. 2010. Advances in the Assessment of Natural Disaster Situation. Advances in Earth Science, 25(1): 22-32. ]
|
曾鹏,苏朝晖,方伟华,张海霞,余璐.2022. 基于高精度房屋类型数据的海口市台风次生洪涝灾害损失评估.灾害学,37(4):155-165.
Zeng Peng, Su Zhaohui, Fang Weihua, Zhang Haixia, and Yu Lu. 2022. Typhoon Flooding Loss Assessment in Haikou City Based on High Precision Building Type Data. Journal of Catastrophology, 37(4): 155-165.
|
张宝军. 2015. 我国自然灾害情况统计制度与标准化进展. 灾害学,30(3):150-155.
Zhang Baojun. 2015. Advance in System and Standardization of Natural Disasters Information Statistics in China. Journal of Catastrophology, 30(3) : 150-155.
|
中国气象局. 2018. 中国气象灾害年鉴:2017. 北京:气象出版社.
China Meteorological Administration. 2018. Yearbook of Meteorological Disaster in China: 2017. Beijing: China Meteorological Press.
|
中华人民共和国应急管理部. 2024. 自然灾害情况统计调查制度. 北京:应急管理出版社.
Ministry of Emergency Management of the People's Republic of China. 2024. Natural Disasters Statistical Investigation System. Beijing: Emergency Management Press.
|
/
〈 |
|
〉 |