PDF(11188 KB)
PDF(11188 KB)
PDF(11188 KB)
基于GEE与机器学习的长沙市洪涝灾害风险评估
Flood Risk Assessment in Changsha Using Google Earth Engine and Machine Learning
极端降雨引发的城市洪涝灾害频发,开展客观、精准的风险评估对提升灾害预警与应急管理能力至关重要。以2024年6月长沙地区典型洪涝灾害为例,基于谷歌地球引擎(GEE)搭建了Sentinel-1/2协同水体提取模型,实现了洪涝淹没范围的高精度、自动化高效提取;经筛选关键因子后,集成XGBoost、Light GBM、优化MLP和XGBoost-RF四种算法搭建了风险预测模型并进行对比评估。结果表明:NDVI和地表起伏度是驱动长沙市洪涝发生的显著因子;XGBoost-RF模型综合表现最优,其准确率达79.19%,精准率和特异性分别为83.02%和85.00%,AUC值为0.849 7;该模型精准刻画了洪涝风险在缓坡区、河网邻近区及高强度人类活动区的聚集特征,预测结果也与实际淹没范围高度吻合。研究成果发挥多源遥感协同监测优势,确保了极端天气下洪涝评估的时效性,由此构建了XGBoost-RF风险预测框架,为城市灾害预警与动态防控提供了高效的技术支撑和科学参考。
[Objective] This study develops an automated and reproducible framework for dynamic urban flood risk assessment by integrating multi-source satellite imagery with machine learning algorithms. It aims to: (1) establish a high-precision flood extraction methodology utilizing Google Earth Engine (GEE); (2) evaluate the performance of an optimized XGBoost-RF ensemble model against traditional machine learning algorithms; and (3) quantify the primary driving mechanisms of urban flooding. Innovation Statement:The primary innovation of this research lies in the synergistic integration of Sentinel-1 (SAR) and Sentinel-2 (Optical) data to overcome cloud interference and urban spectral noise. This is coupled with a robust three-step factor screening protocol and a weighted XGBoost-RF ensemble architecture specifically designed to address the non-linear complexities and class imbalance inherent in urban hydrological systems. [Methods] The extreme flood event of June 2024 in Changsha, China was taken as a case study. The GEE cloud platform was utilized to develop a collaborative water extraction model. Sentinel-1 VV-polarization SAR imagery (Lee-filtered) was fused with Sentinel-2 MNDWI-processed optical data. An OTSU adaptive thresholding algorithm was applied for initial water body extraction, followed by SRTM DEM-based slope filtering (<5°) to eliminate terrain shadows and connected component analysis for noise reduction. For risk modeling, 14 initial hazard factors were selected across four dimensions: topography, human activity, natural conditions, and historical disaster records. To ensure model robustness, a three-step screening protocol was implemented: (1) Pearson correlation analysis (∣r∣<0.7); (2) multicollinearity diagnosis (VIF<10); and (3) ReliefF algorithm for variable importance quantification. Four models were constructed and compared: MLP (Adam-optimized), LightGBM (with SMOTE oversampling), XGBoost, and a weighted XGBoost-RF ensemble (0.6∶0.4 ratio). Model training was conducted on a 2017 historical dataset (600 samples), with validation performed against an independent 2024 dataset (400 samples) using AUC, accuracy, and precision metrics. [Results] The GEE-based collaborative model effectively mitigated the spectral limitations of single-sensor data, enabling the high-precision, automated mapping of flood inundation during the June 2024 event. This provided a reliable empirical sample set for the risk assessment framework. Factor importance analysis identified NDVI and surface relief as the most significant drivers of flood occurrence in Changsha (Variable Importance >0.036). Comparative evaluation of the four algorithms revealed that theXGBoost-RF ensemble modelachieved the highest predictive accuracy (79.19%) and AUC (0.849 7), outperforming LightGBM, optimized MLP, and standalone XGBoost. The model demonstrated high reliability, with a precision of 83.02% and a specificity of 85.00%. Spatial analysis further revealed that flood risk is primarily clustered in gentle slope zones (0-10°), areas adjacent to river networks, and regions with high-intensity human activity. These predicted high-risk zones showed strong alignment with the actual inundation extent, confirming the model’s ability to capture the spatial heterogeneity of urban floods. [Conclusions] This research validates a synergistic framework for dynamic urban flood risk mapping, yielding three primary insights for disaster management: (1) The integration of Sentinel-1 and Sentinel-2 data on the GEE platform provides a robust, automated solution for rapid flood monitoring. By overcoming “cloud contamination” and urban spectral noise, this approach ensures that “ground truth” data remains accessible even under extreme weather conditions.(2) The XGBoost-RF ensemble architecture represents a significant advancement in risk prediction. By balancing the strengths of boosting and bagging techniques, the model effectively addresses the non-linear complexities and class imbalances of urban hydrological systems, offering superior stability and generalization compared to single-algorithm models.(3) The identification of NDVI and surface relief as dominant drivers challenges conventional risk assumptions. The high vulnerability of high-vegetation areas during extreme rainfall suggests that “saturation-excess” runoff is a critical factor in urban flood evolution.
洪涝灾害 / 风险评估 / Google Earth Engine(GEE) / 机器学习 / 长沙市
flood disaster / risk assessment / Google Earth Engine(GEE) / machine learning / Changsha city
| [1] |
李雯, 姜仁贵, 解建仓, 等. 基于文献计量学的城市洪涝灾害研究可视化知识图谱分析[J]. 西安理工大学学报, 2020, 36(4): 523-529.
(
|
| [2] |
Risk management has reduced vulnerability to floods and droughts globally 1,2, yet their impacts are still increasing 3. An improved understanding of the causes of changing impacts is therefore needed, but has been hampered by a lack of empirical data 4,5. On the basis of a global dataset of 45 pairs of events that occurred within the same area, we show that risk management generally reduces the impacts of floods and droughts but faces difficulties in reducing the impacts of unprecedented events of a magnitude not previously experienced. If the second event was much more hazardous than the first, its impact was almost always higher. This is because management was not designed to deal with such extreme events: for example, they exceeded the design levels of levees and reservoirs. In two success stories, the impact of the second, more hazardous, event was lower, as a result of improved risk management governance and high investment in integrated management. The observed difficulty of managing unprecedented events is alarming, given that more extreme hydrological events are projected owing to climate change 3.\n
|
| [3] |
|
| [4] |
中华人民共和国水利部. 中国水旱灾害防御公报2020[R]. 北京: 中国水利水电出版社, 2021.
(Ministry of Water Resources of the People’s Republic of China. China water and drought disaster defense bulletin 2020[R]. Beijing: China Water Resources and Hydropower Press, 2021.) (in Chinese)
|
| [5] |
国家防灾减灾救灾委员会办公室, 应急管理部.2024年全国自然灾害基本情况[R]. 北京: 应急管理部,2025-01-17.
(Office of the State Council for Disaster Prevention, Reduction and Relief, Ministry of Emergency Management. Basic situation of national natural disasters in 2024[R]. Beijing: Ministry of Emergency Management, 2025.(in Chinese))
|
| [6] |
曾照洋, 王兆礼, 吴旭树, 等. 基于SWMM和LISFLOOD模型的暴雨内涝模拟研究[J]. 水力发电学报, 2017, 36(5):68-77.
(
|
| [7] |
程麒铭, 刘霖皓, 刘非, 等. 基于MIKE FLOOD耦合模型模拟的城市内涝对道路安全风险的影响评估[J]. 中国农村水利水电, 2024(8): 94-104.
城市内涝风险评估对预防洪涝灾害有重要作用，但现有研究很少针对道路安全风险评估。以重庆市秀山县为研究对象，通过构建MIKE FLOOD耦合模型对道路安全风险进行评估。结果表明：地面积水深度主要集中在 0.05~0.15 m，不同重现期（P=1%~20%）降雨下的水深在空间分布上大体呈一致状态，积水时间主要集中在60~90 min，积水流速主要大于0.8 m/s。其中，老城区部分区域的基础设施安全风险以极高风险（水深>0.4 m）为主，其余区域为轻微风险（水深<0.5 m）为主，而行人安全风险以Ⅲ级（危险性指数1.25~2）为主，其余区域则以Ⅰ级（危险性指数<0.75）为主。大部分路段的交通运行状况以畅通等级（保留系数>0.7）为主，但部分路段遇暴雨时中断（保留系数=0）。研究结果可为城市内涝发生前的交通管控和通行预警提供科学依据。
(
|
| [8] |
夏铭辉, 尹世洋. 基于SWMM耦合HEC-RAS的人工建设水网防洪排涝模拟研究[J]. 人民珠江, 2024, 45(6):30-38.
(
|
| [9] |
王德运, 张露丹, 吴祈, 等. 基于机器学习算法的洪涝灾害风险评估: 以宜昌市为例[J]. 长江流域资源与环境, 2023, 32(8): 1710-1723.
(
|
| [10] |
|
| [11] |
徐韧, 吉阳光, 赵东儒, 等. 基于遥感与GIS技术的洪水淹没状况分析: 以安徽省安庆市望江县为例[J]. 水土保持通报, 2018, 38(5): 282-287.
(
|
| [12] |
吴馨悦. 集成机器学习模型下的短期暴雨型城市洪涝灾害风险评估分析[D]. 廊坊: 防灾科技学院, 2025.
(
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
At present, urban flood risk analysis and forecasting and early warning mainly use numerical models for simulation and analysis, which are more accurate and can reflect urban flood risk well. However, the calculation speed of numerical models is slow and it is difficult to meet the needs of daily flood control and emergency. How to use artificial intelligence technology to quickly predict urban flooding is a key concern and a problem that needs to be solved. Therefore, this paper combines a numerical model with good computational accuracy and an LSTM artificial neural network model with high computational efficiency to propose a new method for fast prediction of urban flooding risk. The method uses the simulation results of the numerical model of urban flooding as the data driver to construct the LSTM neural network prediction model of each waterlogging point. The results show that the method has a high prediction accuracy and fast calculation speed, which can meet the needs of daily flood control and emergency response, and provides a new idea for the application of artificial intelligence technology in the direction of flood prevention and mitigation.
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
邢子瑶, 董芯蕊, 昝糈莉, 等. 融合VGI和遥感等多源数据的洪涝范围提取与模拟方法[J]. 地球信息科学学报, 2023, 25(9): 1869-1881.
及时准确的洪涝范围提取可以提高应急管理部门对于洪涝灾害的响应能力，减轻灾害影响。SAR遥感不受云雨影响，是洪涝灾害监测的有效工具。然而，由于卫星重返周期的限制，基于SAR的洪涝连续观测较难实现。在应急背景下，如何快速、实时进行洪涝范围提取是急需解决的问题。本文提出了一种结合遥感、VGI等多源数据的洪涝范围提取与模拟方法： ① 构建Albert+CNN的文本分类模型提取社交媒体洪涝信息；② 基于异常值剔除方法利用社交媒体和OSM等VGI数据代替人工采样对Sentinel-1 SAR数据进行分类，提取洪涝范围；③ 结合社交媒体、水位数据等多源数据基于SNIC分割和成本距离等方法模拟无可用SAR数据时的洪涝淹没情况，提高洪涝的淹没范围提取频次。研究表明，本文基于Albert+CNN与异常值剔除的样本自动生成方法，可以有效辅助SAR数据的洪涝分类；利用VGI数据结合水情、DEM等多源数据进行了洪涝范围的模拟可以增加洪涝监测的时间分辨率。本研究有助于提高洪涝信息提取能力，为VGI支持洪涝灾害的应急管理提供参考。
(
|
| [21] |
刘小燕, 崔耀平, 史志方, 等. GEE平台下多源遥感影像对洪灾的监测[J]. 遥感学报, 2023, 27(9):2179-2190.
(
|
| [22] |
崔佳琪. 基于多源遥感影像和Google Earth Engine的农用地洪水信息范围提取与应用[D]. 郑州: 河南农业大学, 2023.
(
|
| [23] |
李健锋. 基于多源遥感影像和Google Earth Engine的斯里兰卡地表水体及洪水信息快速提取研究[D]. 西安: 长安大学, 2020.
(
|
| [24] |
刘朝君, 李丹. 基于多源数据的长沙市半城市化地区识别与演变研究[J]. 地理科学研究, 2025, 14(3):645-656.
(
|
| [25] |
唐小晴. 长沙暴雨小时雨量破纪录多路段内涝地铁停运[EB/OL].(2024-06-24) [2024-11-19]. EB/OL].(2024-06-24)[2024-11-19]. in Chinese)
|
| [26] |
奉永成, 林城. 湖南降雨较历年同期偏多近五成长沙单小时雨量破纪录[EB/OL].(2024-06-25),[2024-11-19]. EB/OL].(2024-06-25)[2024-11-19]. in Chinese)
|
| [27] |
|
| [28] |
Land cover (LC) determines the energy exchange, water and carbon cycle between Earth's spheres. Accurate LC information is a fundamental parameter for the environment and climate studies. Considering that the LC in China has been altered dramatically with the economic development in the past few decades, sequential and fine-scale LC monitoring is in urgent need. However, currently, fine-resolution annual LC dataset produced by the observational images is generally unavailable for China due to the lack of sufficient training samples and computational capabilities. To deal with this issue, we produced the first Landsat-derived annual China land cover dataset (CLCD) on the Google Earth Engine (GEE) platform, which contains 30m annual LC and its dynamics in China from 1990 to 2019. We first collected the training samples by combining stable samples extracted from China's land-use/cover datasets (CLUDs) and visually interpreted samples from satellite time-series data, Google Earth and Google Maps. Using 335 709 Landsat images on the GEE, several temporal metrics were constructed and fed to the random forest classifier to obtain classification results. We then proposed a post-processing method incorporating spatial-temporal filtering and logical reasoning to further improve the spatial-temporal consistency of CLCD. Finally, the overall accuracy of CLCD reached 79.31% based on 5463 visually interpreted samples. A further assessment based on 5131 third-party test samples showed that the overall accuracy of CLCD outperforms that of MCD12Q1, ESACCI_LC, FROM_GLC and GlobeLand30. Besides, we intercompared the CLCD with several Landsat-derived thematic products, which exhibited good consistencies with the Global Forest Change, the Global Surface Water, and three impervious surface products. Based on the CLCD, the trends and patterns of China's LC changes during 1985 and 2019 were revealed, such as expansion of impervious surface (+148.71 %) and water (+18.39 %), decrease in cropland (-4.85 %) and grassland (-3.29 %), and increase in forest (+4.34 %). In general, CLCD reflected the rapid urbanization and a series of ecological projects (e.g. Gain for Green) in China and revealed the anthropogenic implications on LC under the condition of climate change, signifying its potential application in the global change research.
|
| [29] |
邓启睿, 张英, 刘佳, 等. 基于Sentinel-1/2数据的洪水淹没范围提取模型[J]. 人民长江, 2024, 55(9):71-77.
(
|
| [30] |
|
| [31] |
|
| [32] |
|
| [33] |
王泉, 柳德江, 杨海兰, 等. MNDWI和NDWI指数提取富营养化湖泊水体边界的对比研究[J]. 地球科学前沿, 2017, 7(6): 731-738.
(
|
| [34] |
|
| [35] |
|
| [36] |
|
| [37] |
The Longnan mountainous area, characterized by its complex geological structure and fragile geological environment, is one of the four major regions in China prone to geological disasters. Previous studies have employed traditional evaluation methods to assess landslide susceptibility in the Longnan mountainous area. However, these traditional methods are often subjective, and their accuracy and efficiency are difficult to guarantee. This study, supported by GIS technology, focuses on Wen County in Longnan City, a region frequently affected by landslide disasters. Based on 260 collected landslide disaster points, the study combines the ReliefF model to evaluate and zone landslide susceptibility in Wen County, Longnan City, based on feature contribution values. The lithology and rainfall factors have significant impacts on geological disasters, respectively. Areas along rivers and roads, with loose soil, heavy rainfall, steep slopes, and dense vegetation, are more prone to landslide disasters due to the combined effects of natural factors and human activities. This study also uses the receiver operating characteristic (ROC) curve to validate the accuracy of the evaluation results. The area under the curve (AUC) for the ReliefF feature fusion method is 0.853, which is higher than the 0.838 obtained from the information value method. The ReliefF method demonstrates excellent performance in landslide susceptibility evaluation, offering better predictive capability at a lower computational cost, thus achieving a balance between accuracy and efficiency. This approach can provide valuable references for rapid decision-making by relevant geological disaster prevention and management departments.
|
| [38] |
|
| [39] |
|
| [40] |
|
| [41] |
Flooding is a very common natural hazard that causes catastrophic effects worldwide. Recently, ensemble-based techniques have become popular in flood susceptibility modelling due to their greater strength and efficiency in the prediction of flood locations. Thus, the aim of this study was to employ machine learning-based Reduced-error pruning trees (REPTree) with Bagging (Bag-REPTree) and Random subspace (RS-REPTree) ensemble frameworks for spatial prediction of flood susceptibility using a geographic information system (GIS). First, a flood spatial database was constructed with 363 flood locations and thirteen flood influencing factors, namely altitude, slope angle, slope aspect, curvature, stream power index (SPI), sediment transport index (STI), topographic wetness index (TWI), distance to rivers, normalized difference vegetation index (NDVI), soil, land use, lithology, and rainfall. Subsequently, correlation attribute evaluation (CAE) was used as the factor selection method for optimization of input factors. Finally, the receiver operating characteristic (ROC) curve, standard error (SE), confidence interval (CI) at 95%, and Wilcoxon signed-rank test were used to validate and compare the performance of the models. Results show that the RS-REPTree model has the highest prediction capability for flood susceptibility assessment, with the highest area under (the ROC) curve (AUC) value (0.949, 0.907), the smallest SE (0.011, 0.023), and the narrowest CI (95%) (0.928-0.970, 0.863-0.952) for the training and validation datasets. It was followed by the Bag-REPTree and REPTree models, respectively. The results also proved the superiority of the ensemble method over using these methods individually.
|
| [42] |
|
| [43] |
魏盛宇, 翟越, 赵廿. 基于多源数据的西安市暴雨洪涝脆弱性评估及其空间分异特性[J]. 清华大学学报(自然科学版), 2025, 65(1): 174-185.
可靠的脆弱性评估模型及空间分异特性研究有助于提升灾害应急处置能力。针对传统脆弱性评估只考虑基础设施的抵御能力且权重计算方法单一等问题, 该文围绕暴露度、敏感性、应对能力将城市基础条件与人的主动应对能力纳入指标体系, 构建洪涝脆弱性评估模型, 并采用博弈论组合优化主客观权重值。以洪涝风险较高的西安市为例, 借助多源数据在1 km栅格尺度中进行模型应用, 并通过全局Moran's I和局部空间关联指数(LISA)分析脆弱性空间集聚特征。结果表明：西安市暴雨洪涝脆弱性呈现出显著的自相关与空间分异特性, 整体脆弱性评估等级偏低, 其主要致脆因子为暴露度较高且应对能力较弱, 在高与较高脆弱性等级区间中, 敏感-应对能力不足致脆型和强综合脆弱型居多。该模型可为城市暴雨洪涝脆弱性的减缓与治理等提供启示和借鉴。
(
|
| [44] |
刘业森, 刘媛媛, 李敏, 等. 降雨数据空间分辨率在城市流域洪峰变化分析中的影响[J]. 地球信息科学学报, 2022, 24(7): 1326-1336.
在气候变化和城市化背景下,城市暴雨的局部特征愈发明显,而降雨监测和预报数据空间分辨率难以满足空间分布描述要求,造成城市洪涝模拟和预测结果存在不确定性。本文选择深圳市城区的3个流域,基于118场实测降雨数据,通过逐步增大网格尺寸以达到降低降雨空间分辨率的效果,同时以流域边界作为限制,保证流域内面平均雨量不变,然后将不同空间分辨率的降雨数据分别作为城市洪涝模型的输入条件,对流域出口断面的洪水过程进行模拟计算,每场降雨得到6种不同空间分辨率下的洪峰流量。结合降雨中心、流域几何中心等空间特征参数,分析降雨空间分辨率对流域洪峰的影响。研究结果表明,针对118场实测降雨,随着降雨数据空间分辨率逐渐降低,3个流域的出口断面的洪峰变化幅度逐渐增大;如果忽略了流域内部降雨数据的空间分布差异,洪峰变化幅度和方向与降雨中心和流域几何中心的空间关系具有相关性,降雨中心在流域几何中心上游方向的距离越大,洪峰降低的幅度越大。本文基于3个城市流域得到了相似的研究结论,研究方法可为城市流域洪涝模拟提供借鉴,研究结果可为河道洪水风险评价提供参考。
(
|
| [45] |
杨林, 赵广举, 穆兴民, 等. 基于Budyko假设的洮河与大夏河径流变化归因识别[J]. 生态学报, 2021, 41(21): 8421-8429.
(
|
| [46] |
|
| [47] |
黄国如, 罗海婉, 卢鑫祥, 等. 城市洪涝灾害风险分析与区划方法综述[J]. 水资源保护, 2020, 36(6): 1-6, 17.
(
|
| [48] |
杨柳, 冯畅, 郭哿, 等. 基于四维度模型的湖南省湘江流域洪涝灾害风险评估[J]. 水资源与水工程学报, 2024, 35(1):26-36.
(
|
| [49] |
焦胜, 操婷婷, 牛彦合, 等. 基于耦合PLUS-SCS模型的城市用地内涝风险预测及优化管控: 以湖南省长沙市为例[J]. 水土保持通报, 2023, 43(5): 195-202.
(
|
/
| 〈 |
|
〉 |
