日益严峻的城市洪涝灾害对城市安全和可持续发展带来了巨大挑战，全球气候变暖和城市化也加剧了城市洪涝风险。城市洪涝过程模拟与风险评估是减少洪涝灾害中人员伤亡和财产损失的有效手段之一，在城市发展和防灾减灾中发挥着十分重要的作用。深圳河流域近年来屡遭暴雨洪涝侵袭，选取简单、容易获取且更加精细化的指标来量化洪涝灾害危险性和脆弱性，从而科学准确地识别和评估流域内洪涝灾害综合风险，对于及时有效的防汛应急管理工作具有重要的现实意义。

本文以深圳河流域为研究区，基于元胞自动机模型构建二维城市洪涝淹没仿真CAflood模型，模拟不同暴雨重现期下的洪涝过程，识别流域内最大淹没范围和淹没水深。重点关注精细化经济指标空间分布并将多源数据融合，对流域内跨行政区单元社会经济进行空间精细化模拟，在暴雨情景模拟结果的基础上结合指标体系法，基于“危险性-脆弱性”的框架结构，构建适用于沿海城市的洪涝灾害风险评估指标体系，识别和预测不同暴雨情景下的洪涝灾害高风险区，绘制深圳河流域城市洪涝灾害风险区划图，以期为变化环境下城市风险管理和防汛救灾应急预案的制定提供科学依据。论文的研究内容与主要结论如下：

（1）构建元胞自动机淹没仿真模拟模型模拟洪涝淹没水深和淹没范围

基于DEM、土地利用和土地覆盖数据构建深圳河流域CAflood模型，选择流域内66个关键的积水点位置对模型参数进行率定和验证，率定结果与验证结果基本一致，充分证明构建的CAflood模型具有较好的适用性，能有效地模拟深圳河流域在不同暴雨情景下的淹没状态。将2年、10年、50年和100年一遇四种不同暴雨情景输入率定验证好的CAflood模型中，得到淹没水深和淹没范围分布图，模拟结果表明随着暴雨重现期的增加，流域内淹没点数量没有明显变化，淹没面积明显增大，最大淹没水深大于2 m的区域主要集中在福田区、罗湖区和龙岗区南部的流域干流及支流沿河低洼地带，淹没状态大多呈密集点状或条带状分布，局域性较强。

（2）聚焦精细化经济空间布局并融合多源数据进行GDP空间化模拟

基于第一产业GDP统计数据和土地利用数据构建第一产业GDP空间化模型，基于第二、三产业GDP统计数据耦合夜间灯光数据和9类POI兴趣点数据构建第二、三产业GDP空间化模型，模拟出分辨率为500 m的精细化GDP密度空间网格图。结果表明单一的土地利用数据或夜间灯光数据在模拟流域GDP密度时存在一定的局限性，结合多源数据能更好地揭示研究区域内部的经济多样性与细节特征。流域GDP产值密度空间差异显著，深圳侧GDP密度显著高于香港侧，并且与第二、三产业空间化结果高度吻合，深圳福田区、罗湖区、龙岗区以及香港北部经济高度发达，GDP产值密度最高达617214万元/km²。流域内GDP主要源自第二、三产业的发展，第一产业的占比较低。

（3）构建新的适用于沿海城市的洪涝灾害风险评估指标体系，识别和预测灾害高风险区

选择“危险性­脆弱性”的框架结构，结合致灾因子危险性指标（最大淹没水深、距河口距离）、孕灾环境危险性指标（地面高程、坡度、土地利用）、承灾体脆弱性指标（人口密度、精细化GDP密度和建筑物密度）来直观反映研究区的洪涝风险等级，加入距河口距离这一指标来体现深圳河感潮的特殊性和代表性，构建新的适用于沿海城市的洪涝灾害风险评估指标体系，同时选用结合AHP法和CRITIC法的博弈论组合赋权结果来综合主客观理论分析优势，赋权结果比单一赋权方法更为科学合理。研究结果表明深圳河流域绝大部分地区在4种暴雨情景下始终表现为较为稳定的中低风险状态，深圳侧洪涝灾害风险明显高于香港侧。高风险区分布表现出显著的时空差异和区域特征，重点分布在深圳福田区、罗湖区以及龙岗区南部，随着暴雨重现期的增加，点状高风险区不断扩张为面状或条带状，面积占比从2年一遇的1.2%、10年一遇的2.1%、50年一遇的3.3%增加到100年一遇的3.8%。高风险区域与深圳市各区政府印发的防汛应急预案中的洪涝灾害高风险区高度吻合，表明本文构建的城市洪涝灾害风险评估体系可以有效揭示需重点关注的高风险区域，是针对性制定防汛减灾管理措施的重要参考依据。

Increasingly severe urban flooding poses a great challenge to urban safety and sustainable development, and global warming and urbanization have exacerbated the risk of urban flooding. Urban flooding/waterlogging simulation and risk assessment is one of the important ways to reduce casualties and property losses in floods, and plays a crucial role in urban development as well as disaster prevention and mitigation. The Shenzhen River basin has been frequently hit by heavy rainfall and flooding in recent years, and it is important to select simple, easily accessible and more refined indicators to quantify flood hazard and vulnerability, so as to scientifically and accurately identify and assess the comprehensive risk of flooding/waterlogging in the basin for timely and effective flood control and emergency management.

In this study, the Shenzhen River basin was selected as the study area, and a two-dimensional urban flooding/waterlogging simulation CAflood model was developed based on cellular automata to simulate the flooding/waterlogging processes under different rainstorm return periods, and to identify the inundation range and inundation depth in the basin. Focusing on the precise spatial patterns of economic indicator and fusing multi-source data, the spatial refinement of socio-economics of inter-administrative units in the basin was simulated. Based on the simulation results of heavy rainstorm scenarios and the indicator system method, a new flood risk assessment index system for coastal cities was proposed based on the "Hazard-Vulnerability" assessment framework. The high-risk areas of floods under different rainstorm scenarios were identified and predicted, and the risk zoning map of urban floods in the Shenzhen River basin was drawn, so as to provide scientific references for urban flooding/waterlogging risk management as well as flood prevention and mitigation contingency plans under the changing environment. The research content and main conclusions of the thesis are as follows:

（1） Development of a cellular automata flooding/waterlogging simulation model to simulate the inundation depth and inundation range

Based on the DEM, and land use/cover data the CAflood model was developed for the Shenzhen River basin, the 66 measured important flood-prone sites in the study area were selected for model calibration and validation. The calibration results were similar with the results of the validation. The results showed that the CAflood model had reliable applicability, and could be used for inundation simulation in the Shenzhen River basin. Simulation was made for four different rainstorm scenarios with the return periods of 2 years, 10 years, 50 years, and 100 years. The designed precipitation were inputted into the CAflood model, and the inundation depth and inundation area were obtained. The simulation results showed that the number of inundation sites in the basin did not change significantly with the increase of the rainstorm return period, while the inundation area increased significantly. The areas with maximum inundation depth of greater than 2 m were mainly located in Futian District, Luohu District, and the southern part of Longgang District, and the inundation state was mostly distributed in the form of dense sites or long strips, and the localization was obvious.

（2） Focusing on refined economic spatial layout and integrating multi-source data for GDP spatialization simulation

The GDP spatialization model of primary industry was developed based on primary industry GDP statistics, and land use data. The GDP spatialization model of secondary and tertiary industry was developed based on secondary and tertiary industry GDP statistics coupled with nighttime lighting data and 9 types of points of interest, and the spatial distribution map of refined GDP density with a resolution of 500 m was obtained. The results showed that single landuse data or nighttime light data has limitations in modeling GDP density, and combining multi-source data was a better approach to carry out the spatial refinement of GDP. The GDP density showed obvious spatial differences, and the GDP density of the Shenzhen side was higher than that of the Hong Kong side as a whole. The results of the spatialization with the secondary and tertiary industries showed high consistency. Central areas of Futian Distracts, Luohu Districts, the southern part of Longgang District, and the northern part of Hong Kong were highly developed economically, with the highest density of output value of 617,214,000 yuan/km². The GDP of the central area of each district was mainly from the secondary and tertiary industries, and the contribution of the primary industry was relatively small, while the hilly and mountainous areas around the watershed had developed the secondary and tertiary industries as well as the primary industry.

（3） A new flood risk assessment index system was proposed for coastal cities, and high-risk areas were identified and predicted.

Based on the framework of "Hazard - Vulnerability", in terms of maximum inundation depth, distance from the estuary, elevation, slope, landuse, population density, refined GDP density and building density, the flooding/waterlogging risk level in the study area was identified. The distance from the estuary as an indicator was seected to highlight the specificity and representativeness of the sensory tides in the Shenzhen River. A new assessment index system of flood risk applicable to coastal cities was proposed to identify and predict high-risk areas.  Both AHP method and CRITIC method of game theory combined with the results of empowerment were selected to synthesize the advantages of subjective and objective theoretical analysis, the results of empowerment were more scientific and reasonable than a single empowerment method. The results of the study showed that the vast majority of the Shenzhen River basin was always a medium-low risk area under different rainstorm scenarios, and the risk of flooding on the Shenzhen side was significantly higher than that on the Hong Kong side. The high-risk area showed typical spatial and temporal differences and regional characteristics, especially in Futian District, Luohu District and the southern part of Longgang District. With the increase of the rainstorm return period, the point-shaped high-risk area expanded to a face-shaped or long strip. The high-risk area increased from 1.2% for the 2-years rainstorm, 2.1% for the 10-years rainstorm, 3.3% for the 50-years rainstorm, to 3.8% for the 100-years rainstorm. The high-risk areas highly coincided with the high-risk areas of flooding/waterlogging in the flood control emergency plans issued by the Shenzhen municipal governments, indicating that the urban flood risk assessment system proposed in this study could effectively reveal the high-risk areas that need to pay attention , and it was an effective way to formulate targeted flood control and disaster reduction management measures.