在全球气候变化和城市化进程逐渐加快的背景下，高强度降雨事件的发生频率正在逐年上升，导致由此引发的城市内涝积水事件频繁发生，道路交通成为了城市内涝最主要的承灾体之一，行驶安全、通行能力受到内涝积水的严重影响，而应急服务会受到道路交通与城市内涝交互作用的影响，使其响应能力受到严重破坏，对城市居民的生命和财产安全以及社会经济活动的正常开展造成不良影响。本研究主要研究内容及结论如下：

（1）本研究基于研究区内排水管网、土地利用分布及高程等基础数据，细致量化降雨事件中产汇流及积水过程，构建北京市中心城区包含一维管网、二维淹没及地表产汇流的耦合内涝模型并率定及验证，之后运用内涝模型识别了不同重现期设计降雨下的内涝积水空间分布特征。结果发现：研究区中部聚集了淹没区，但研究区北部、西部和南部的淹没程度最为严重，各重现期下淹没范围及相同位置积水深度呈现随重现期增加而增加的趋势。

（2）本研究基于道路矢量数据，结合城市内涝模型构建不同重现期降雨事件下道路淹没数据集，结合安全风险判别标准构建评估框架并揭示各路段因城市内涝对行人及车辆所产生的安全风险，分析安全风险空间分布特征，指出高风险路段位置。之后，通过100-year的评估结果提取不同等级水平安全风险点位，利用地理探测器结合影响因子数据库探寻道路交通安全风险的驱动因子。结果发现：1）在相同重现期暴雨事件下，大部分中风险和高风险路段出现在研究区域的中部和西部以及北部。随着暴雨重现期的增加，对三类不同的交通参与者产生中风险和高风险的路段占路段总长度的比例均呈单调上升趋势。2）在随后的地理探测器驱动因子分析的结果显示，坡度与排水管网规格的交互作用解释力最大，q值为0.6478，说明在排水管网规格低、坡度大的路段更容易产生严重的道路交通安全风险。

（3）本研究利用Python语言从交通大数据开放平台获取不同交通时段的实际路况，并将其与各重现期淹没面积最大时刻内涝模拟数据交叉构筑出共计24个复合情景。之后，基于ArcGIS软件，结合所构筑情景及水深-交通流速度衰减模型，搭建研究区交通出行模拟模型，对各情景下居民出行路径进行模拟，统计各情景下居民行程数据、出行时间及燃料成本，构建行程损失率及广义出行成本指标，评估居民出行难度及其空间分布特征，揭示各情景下出行困难区。结果发现：1）当降雨重现期从5-year逐步上升至100-year时，因内涝而无法出行的人口从254.86万人逐步增长至297.97万人、298.75万人、384.18万人以及416.84万人。2）通过广义出行成本这一指标可以得知，晴天状态下区域内交通可达性的整体情况遵循“自由流时段优于平峰优于早晚高峰”，但在城市内涝事件中，交通时段的不同对广义出行成本的影响随着重现期增大而逐渐减弱。在城市内涝事件中研究区内各交通小区的广义出行成本开始大幅度上涨，道路交通可达性将会遭到严重破坏，随着城市内涝重现期的增加，部分交通小区的广义出行成本的上涨幅度会进一步增加，这些区域主要分布在研究区北部，中部和南部，其中的居民出行相比其他地区将会更加困难，与之相反，研究区西南部的广义出行成本的上涨幅度较小，说明这部分区域的道路交通可达性在城市内涝事件中的韧性较强。

（4）本研究结合已构筑的情景及交通出行模拟模型，以研究区内急救站及消防站为设施点，探究交通时段的不同与城市内涝事件对应急服务响应范围的影响。之后，加入对供给规模的考虑，利用以修正高斯函数为衰减函数的改进后两步移动搜索法评估各情景下居民的应急服务可达性，描述其空间分布特征，并利用洛伦兹曲线及基尼系数评估各情景下应急服务的公平性。最后，通过构建道路救援服务区，细致分析城市内涝事件对高安全风险路段道路救援能力的影响。结果发现：1）城市内涝事件会令研究区急救及消防服务响应范围受到严重破坏，响应面积随着降雨事件重现期的增大而以非线性的形式急剧缩小，并有收缩至应急站点附近的趋势。2）城市内涝事件会令研究区内应急服务可达性恶化并呈现“马太效应”，应急服务可达性高的区域其可达性会不降反升，与应急服务可达性低的区域间的可达性差距会随着重现期的增大而越来越大，且其更倾向于聚集在供给规模比较大的站点附近。3）进一步的公平性研究显示，当内涝来临时，所有重现期下急救服务与消防服务的基尼系数都上升到了0.9左右，这表示城市内涝极大程度影响了应急服务的公平性。4）对于应急服务道路救援能力的研究结果显示，随着城市内涝重现期的上升，可救援及不可救援的路段总长占比之间差距逐渐拉大，研究区内应急服务道路救援能力会逐渐弱化。

In the context of global climate change and the gradual acceleration of urbanisation, the frequency of high-intensity rainfall events is increasing year by year, resulting in frequent urban waterlogging events, road traffic has become one of the most important bearers of urban waterlogging, driving safety, accessibility by the serious impact of waterlogging, and the emergency services will be subject to the interaction between road traffic and urban waterlogging, causing adverse impacts on the life and property safety of urban residents and the normal development of socio-economic activities. response capacity is seriously damaged, which adversely affects the life and property safety of urban residents and the normal development of socio-economic activities. The main research contents and conclusions of this study are as follows:

(1) Based on the basic data of drainage pipe network, land use distribution and elevation in the study area, this study quantified in detail the process of runoff-producing, converging and waterlogging during rainfall events, and constructed a coupled urban waterlogging model with one-dimensional pipe network, two-dimensional inundation, surface runoff-producing and converging in the downtown area of Beijing which was calibrated and validated. Then the urban waterlogging model was used to identify the characteristics of spatial distribution of urban waterlogging in the rainfall events designed for different return periods. It was found that the inundation area was concentrated in the central part of the study area, but the northern, western and southern parts of the study area had the most serious inundation, and the extent of inundation and the depth of waterlogging at the same location under each return period showed a tendency to increase with the increase of the return period.

(2) Based on the road vector data, this study combined the urban waterlogging simulation model to construct the road inundation dataset under different return period rainfall events, combined the safety risk discrimination criteria to construct the assessment framework and reveal the safety risk of pedestrians and vehicles due to urban waterlogging in each road section, analysed the spatial distribution characteristics of the safety risk, and pointed out the location of high-risk road sections. After that, the safety risk locations of different grade levels were extracted from the 100-year assessment results, and the driving factors of road traffic safety risk were explored by using Geodetector combined with the influence factor database. The results found that 1) under the same return period rainstorm events, most of the medium and high risk road sections appeared in the central and western parts of the study area as well as in the northern part. As the return period of the rainstorm increased, the proportion of road sections that generated medium and high risk for the three different types of traffic participants as a percentage of the total length of the road section showed a monotonically increasing trend.2) The results of the subsequent Geodetector driving factors analysis showed that the interaction of slope and drainage network specification had the highest explanatory power with a q-value of 0.6478, which suggested that serious road safety risk were more likely to occur on road sections with low drainage network specification and high slopes.

(3) In this study, Python was used to obtain the actual road conditions during different traffic periods from the open platform of traffic big data, and constructed a total of 24 composite scenarios by intersecting them with the simulated data of urban waterlogging at the moment of the largest inundation area in each return period. After that, based on ArcGIS software, by combining the constructed scenarios and the bathymetry-traffic velocity attenuation model to build a traffic travel simulation model for the study area, the residents ‘travel paths were simulated under the 24 scenarios and the residents’ trip data, travel time and energy cost under each scenario were counted. The trip loss rate and generalised travel cost indicators was constructed, the residents' travel difficulty and its spatial distribution characteristics were assessed and the difficult travel zones under each scenario were revealed. The results show that: 1) when the rainfall return period gradually increases from 5-year to 100-year, the population unable to travel due to waterlogging gradually increases from 2,548,600 to 2,979,700, 2,987,500, 3,841,800, and 4,168,400; 2) the indicator of generalized travel cost shows that the overall situation of intra-regional accessibility in a sunny day follows the pattern of ‘free-flow’, which is the same as that in a sunny day. The overall situation followed the principle of ‘free-flow hours are better than Off-peak and better than morning and evening peaks’, but in urban waterlogging events, the effect of different traffic hours on generalized travel cost gradually decreases with the increase of return period. During urban waterlogging events, the generalized travel costs of the traffic districts in the study area started to rise significantly, and road traffic accessibility was seriously damaged. With the increase of the return period of urban waterlogging, the increase of the generalized travel costs of some of the traffic districts will be further increased, and these districts are mainly located in the northern, central and southern parts of the study area, where the travelling of the inhabitants is more difficult compared to the other districts. In contrast, the south-western part of the study area will experience a smaller increase in generalized travel costs, suggesting that road traffic accessibility in this part of the study area is more resilient under urban waterlogging events.

(4) This study combined the constructed scenarios and traffic and travel simulation models to investigate the impact of different traffic hours and urban waterlogging events on the response range of emergency services, using the emergency and fire stations in the study area as the facility points. After that, the scale of supply was taken into account, and the improved two-step moving search method with modified Gaussian function as the decay function was used to evaluate the accessibility of emergency services for residents under each scenario, described its spatial distribution characteristics, and used the Lorenz curve and the Gini coefficient to evaluate the fairness of emergency services under each scenario. Finally, the impact of urban waterlogging events on the road rescue capacity of high safety risk road sections was carefully analysed by constructing a road rescue service area. The results show that: 1) urban waterlogging events cause serious damage to the response area of emergency and fire services in the study area, and the response area shrinks drastically in a non-linear manner with the increase of the return period of rainfall events, and tends to shrink to the vicinity of the emergency station. 2) urban waterlogging events cause deterioration in the accessibility of emergency services in the study area with the accessibility of emergency services showing the ‘Matthew effect’. The impact of urban waterlogging events on the accessibility of emergency services on the high safety-risk road sections was assessed. The accessibility gap between areas with high emergency service accessibility and areas with low emergency service accessibility increases with the return period, and tend to cluster near the stations with larger supply sizes.3) Further equity studies show that when urban waterlogging events occurs, the Gini coefficients of emergency services and firefighting services for all the return periods The Gini coefficient of emergency services and fire services rises to about 0.9 in all return periods, which indicates that urban waterlogging greatly affects the equity of emergency services.4) The results of the study on the roadside rescue capacity of emergency services show that as the return period of urban waterlogging rises, the gap between the total lengths of rescuable and non-rescuable road sections gradually widened, and the roadside rescue capacity of the emergency services in the study area gradually weakened.