[Objective] Traditional flood risk assessment frameworks predominantly rely on static exposure indicators，such as population density and land use types，overlooking the tidal movement of people and vehicles between residential，working，and transportation areas. This limitation compromises the spatiotemporal accuracy of risk assessments. To address this，we integrate time-dependent traffic congestion data into the conventional urban flood risk assessment framework，aiming to expand the dimensionality of exposure characterization and provide a supplementary perspective for urban flood risk evaluation. [Methods] Taking the low-lying Future Sci-Tech City in Hangzhou as a case study，we developed a coupled 1D-2D hydrodynamic model （MIKE FLOOD） to simulate waterlogging scenarios under 1-，10-，and 50-year return periods. Building upon traditional static indicators，time-dependent traffic congestion data were introduced to construct a comprehensive flood risk assessment system based on the Hazard-Exposure-Vulnerability-Resilience （H-E-V-R） framework comprising eight indicators： maximum inundation depth，inundation duration，population density，traffic exposure，building density，greenland coverage，distribution of critical assets，and emergency rescue capacity （measured by euclidean distance）. The model was validated using observed inundation depths from actual rainfall events. [Results] Model validation demonstrated high accuracy，with simulation errors for inundation depth at waterlogging points controlled within 5%，and comprehensive runoff coefficient errors of 13.7% and 13.9%，respectively. The model effectively captured the flood evolution characteristics of the study area. As the design rainfall return period increased，both inundation depth and duration escalated，leading to a significant rise in overall flood hazard and comprehensive risk. Traffic exposure varied across different time periods. The morning peak exhibited higher maximum exposure values （0.693），whereas the evening peak showed a more spatially extensive distribution of highly exposed road segments. Under the 50-year return period scenario，the southern part of the study area remained at relatively low risk. In contrast，medium-to-high risk areas in the central and northern communities generally exceeded 22%，with Tanghe Community even experiencing a 2% area of extremely high risk. [Conclusion] The incorporation of time-dependent traffic congestion characteristics effectively identifies road segments with high exposure. Under the H-E-V-R framework，the comprehensive risk exhibits a distinct spatial pattern of “higher in the central-northern regions and lower in the south”，with high-risk zones concentrated in the urbanized core of the central-northern area. These findings provide a valuable reference for community-level urban flood control and traffic management during flood events.

[Objective] In urban flooding simulation，the loose coupling between surface flow and pipe networks fails to accurately describe bidirectional hydraulic exchange processes，and future urban underlying surface data cannot meet the fine-scale simulation requirements at the urban scale. To address these limitations，this study aims to simulate the pipe network overloading processes and surface inundation evolution under different rainfall scenarios and urbanization stages，providing a scientific basis for flood control planning and resilient city construction in riverside cities. [Methods] A tightly coupled 1D-2D hydrodynamic model was developed by integrating SWMM with LISFLOOD-FP，enabling real-time bidirectional data exchange between the surface and pipe network at a unified time step of 1 second. Three hydraulic exchange processes were considered： overflow from the pipe network to the surface，surface inflow into the pipe network via orifice flow，and weir flow. Design rainfall scenarios with return periods ranging from 1 to 100 years were generated using the Chicago rainfall pattern method with a duration of 120 minutes，and model parameters were calibrated using the comprehensive runoff coefficient as the calibration target. The Multi-engine Urban Expansion Simulator （MUSE） was introduced to simulate urban built-up area expansion from 2026 to 2040，adopting the Neighborhood-constrained Patch Growth Engine （Nei-PGE） and lognormal patch size distribution，with model accuracy evaluated using the Kappa coefficient，Figure of Merit （FoM），and Overall Accuracy （OA）. On this basis，the pipe network operating conditions and surface inundation response characteristics under different rainfall scenarios and future urbanization stages were systematically analyzed. [Results] （1） Model calibration results show that the coefficient of variation remains within ±10% across all return periods，confirming the model’s applicability. As the return period increases，the number of overflow nodes （NON） with overflow duration exceeding 2 hours rises continuously，reaching 79.17% of total nodes under the 100-year return period； the inundation area with water depth exceeding 1 m increases from 4.38 km² under the 1-year return period to 28.05 km² under the 100-year return period. （2） The urban expansion simulation achieves high accuracy，with a relative error of only 0.267% between the simulated and actual built-up area in 2024，a Kappa coefficient of 0.869，FoM of 0.573，and OA of 0.870； the urban built-up area is projected to increase from 95.485 km² to 131.453 km² during 2025-2040，with edge expansion as the dominant pattern alongside diversified internal infilling trends. （3） Under future urbanization scenarios，surface inundation risk intensifies progressively，with total inundation area under the 1-year return period increasing from 14.23 km² in 2026 to 22.98 km² in 2040，and inundation area with water depth exceeding 1 m under the 50-year return period increasing from 2.96 km² to 20.18 km². In contrast，the impact of urbanization on overflow node counts is relatively limited，with maximum NON variation not exceeding 4 nodes across urbanization stages under the same return period，indicating that the existing drainage network has approached saturation. [Conclusions] This study develops an integrated urban flood assessment framework coupling fine-scale hydrodynamic simulation with future urban expansion prediction for riverside cities. The tightly coupled hydrological-hydrodynamic model effectively captures the dynamic bidirectional hydraulic interaction between surface water and the drainage network，while the MUSE-based urban expansion simulation provides high-resolution future land use scenarios for flood modeling across multiple urbanization stages. The results demonstrate that as rainfall intensity increases，overflow nodes with shorter overflow duration progressively transition to longer-duration overflow. Urbanization significantly intensifies surface inundation risk，with shallow inundation areas gradually transitioning to deeper inundation zones as urbanization advances. In contrast，the impact of urbanization on overflow node counts is relatively limited，indicating that the existing drainage network has approached saturation，and that the effects of newly added impervious surfaces are primarily manifested as prolonged overflow duration and increased overflow volume rather than an expanded spatial distribution of overflow nodes.

[Objective] With urban flooding getting more frequently，assessing the resulting disruption of traffic networks is essential for urban resilience and emergency planning. This study develops a tightly coupled modeling and evaluation framework to quantify how urban flooding affects road network performance over time. The aim is to capture not only localized flooding impacts but also the system-wide and delayed effects on traffic efficiency and network structure. [Methods] An integrated framework was established by coupling a one-dimensional drainage model （SWMM），a two-dimensional surface flooding model based on Cellular Automata （CA），and the microscopic traffic simulation platform SUMO （Simulation of Urban Mobility）. The proposed Urban Road-masked Cellular Automata-SWMM coupled model （URCA） enables bidirectional interaction between underground pipe flow and surface road inundation at minute-level time steps. Exchange flows between the drainage system and road surface are calculated through manholes and gullies using modified hydraulic equations. Runoff from sub-catchments is simulated within SWMM，while road-related surface flow is represented in the CA model，which incorporates land-use-based infiltration，Manning-based routing，and adaptive time stepping for numerical stability. Simulated water depth on road grids is translated into traffic control rules in SUMO using predefined depth thresholds that trigger speed reduction or road closure，ensuring dynamic feedback between flood evolution and traffic redistribution. The road network is represented as a weighted directed graph integrating static road attributes and dynamic traffic indicators. Flood impacts are evaluated using average travel time increase （T_E%），OD-based time expansion rates，network coverage，and the 0th Betti number to quantify changes in connectivity under different time budgets. The framework is applied to a 3.89 km² urban area in Wuhan under a 20-year return-period rainfall event. [Results] Model calibration shows stable hydraulic performance，with a mass balance error of 2.97% and a Nash-Sutcliffe efficiency of 0.88. Under the 20-year rainfall scenario，flooding produces strong nonlinear traffic impacts. Although only 37 road segments （approximately 6% of 618 links） are significantly inundated，the average network travel time increases to 508% of the baseline level，equivalent to an effective network radius expansion of about 700 meters. This indicates that limited localized flooding can induce substantial system-wide degradation. Travel time exhibits multiple peaks during and after rainfall，with the maximum delay occurring after rainfall cessation，demonstrating a clear lag effect. Time-budget analysis shows that approximately 12 additional minutes are required to achieve baseline coverage，corresponding to a time expansion factor of 2.23. Network coverage declines sharply during critical periods，while Betti-0 analysis indicates increased fragmentation，especially within shorter time budgets. Trips within 40 minutes are the most affected. A small number of fully blocked or severely congested links account for most of the performance decline. [Conclusions] Urban flooding can cause significant increases in travel time，structural fragmentation of road networks，and delayed peak disruption after rainfall ends. A limited proportion of critical links can trigger large-scale performance degradation，and short-distance travel is particularly vulnerable. By tightly coupling drainage and traffic processes and integrating functional and structural network indicators，this study provides a comprehensive approach for assessing flood impacts and supporting resilience-oriented urban planning.

[Objective] Physically based hydrological-hydrodynamic models for urban pluvial flood simulation are computationally expensive. In contrast，deep learning-based approaches may produce spurious shallow water depths in non-inundated areas，leading to distorted inundation extents. To address this issue，this study develops MFGRU-Y，a classification and regression collaborative model that jointly predicts wet-dry status and maximum inundation depth from rainfall sequences to improve both depth accuracy and inundation-extent delineation while maintaining high simulation efficiency. [Methods] As a multi-task surrogate model，MFGRU-Y jointly predicts inundation occurrence （wet-dry status） and maximum inundation depth from single-event rainfall time series. To accommodate rainfall events with varying durations，each sample retains its effective sequence length，and variable-length sequence encoding is employed to ensure that the recurrent unit focuses on valid time steps rather than padded zeros. The architecture includes a GRU-based rainfall encoder，a shared representation layer，and two heads for wet-dry classification and depth regression，with a non-negativity constraint applied to depth outputs. A key feature of the proposed model is that classification confidence is explicitly used to guide depth regression，thereby improving inundation-boundary delineation. To address the problem of inundation extent distortion，classification confidence is used to constrain regression through a gating design that suppresses spurious shallow depths in predicted dry regions； during inference，a binary wet-dry mask is further applied to the depth field. Training uses an imbalance-aware classification loss，a wet-region-emphasized robust regression loss，and learnable loss weighting to adaptively balance the two tasks and reduce manual tuning. [Results] MFGRU-Y achieved the best overall performance among all evaluated models. For maximum inundation depth prediction，MFGRU-Y achieved an MAE of 0.0093 m，an RMSE of 0.05 m，and an R² of 0.99 on the test set，indicating close agreement with reference simulations across events. For inundation extent evaluation，MFGRU-Y reached an IoU of 0.987 3，while the wet-dry classification accuracy was 99.4%，showing that the predicted inundation footprint was highly consistent with the ground truth. In addition to these aggregate metrics，MFGRU-Y demonstrated clear practical advantages in terms of boundary interpretability. In comparative experiments，competing deep learning baselines exhibited severe false positives，including cases in which dry nodes were almost entirely misidentified as inundated，resulting in distorted inundation extents. MFGRU-Y substantially reduced such artifacts by using classification confidence to constrain depth prediction during training and by applying an explicit wet-dry mask during inference. As a result，the final depth fields showed fewer spurious shallow patches in non-inundated areas and a sharper，more coherent inundation boundary. Importantly，these gains in extent consistency and boundary clarity were achieved without degrading overall depth accuracy，supporting the effectiveness of collaborative wet-dry discrimination and depth regression for urban pluvial flood prediction. Furthermore，in terms of computational efficiency，MFGRU-Y significantly reduced the single-event simulation time from the minute scale required by the traditional mechanistic physically based model to the second scale. [Conclusions] The experimental results demonstrate that MFGRU-Y delivers both high depth accuracy and high inundation-extent fidelity and significantly improves boundary interpretability compared with alternative deep learning models that may overpredict inundation in dry areas. From an application perspective，MFGRU-Y provides more reliable inundation extents and maximum depth fields for risk assessment and emergency response，where boundary-related false inundation can otherwise mislead decision-making. Future work may focus on two aspects： model architecture refinement and input feature enrichment. On the one hand，the model's focus on deep-water zones should be strengthened. On the other hand，richer feature information，such as topography，hydraulic engineering operation schedules，and the initial water levels in river and drainage networks，should be incorporated，thereby further enhancing the reliability of engineering applications.

[Objective] As a key national urban agglomeration in the upper Yangtze River Basin，the Chengdu-Chongqing urban agglomeration features pronounced topographic contrasts，rapid urban expansion，and increasingly frequent extreme rainfall events，resulting in highly complex and multi-source flood impacts. [Methods] This study integrates passive microwave remote sensing precipitation data，multi-source rainfall datasets，land-use data，nighttime light imagery，and population data to construct a comprehensive indicator system comprising precipitation，waterlogging，impervious surface ratio，and development level. From the perspectives of hazard factors，rainfall triggering mechanisms，surface water accumulation processes，and urban expansion patterns，the spatiotemporal evolution of flood vulnerability in the Chengdu-Chongqing urban agglomeration from 1995 to 2020 is systematically analyzed. [Results] （1） Flood vulnerability exhibited a staged increasing trend during 1995-2020，with significant peak years in 2000 and 2020；（2） urban construction land expanded rapidly over the past three decades （an increase of 4，232 km²），mainly through the conversion of cropland，which significantly altered surface characteristics and intensified runoff concentration；（3） extreme rainfall and surface waterlogging intensity increased markedly in 2000 and 2020，forming a spatial pattern of “strong triggering in southwestern mountainous areas and easy waterlogging in eastern lowlands”；（4） flood vulnerability shows a non-linear relationship with urban development level—vulnerability decreased in Chongqing due to favorable terrain and enhanced flood control infrastructure，whereas it increased in Chengdu and surrounding plain cities due to rapid expansion and increased impervious surfaces；（5） approximately 42% of newly developed urban construction areas experienced waterlogging in 2020，with some plain cities exceeding 80%，indicating insufficient resilience in rapidly expanding areas. [Conclusions] Urban flood vulnerability is not solely determined by the level of development； rather，it is jointly influenced by topographical conditions，patterns of urban expansion，and engineering regulation capabilities. The Chengdu-Chongqing urban agglomeration faces compound flood processes involving both basin-wide floods and urban waterlogging. Therefore，measures should be taken to improve flood control and drainage standards，optimize urban spatial structures，strengthen sponge city initiatives，and enhance coordinated regulation between upstream and downstream areas within the urban agglomeration.These findings provide scientific support for flood disaster mitigation，vulnerability reduction，and resilient urban development in the Chengdu-Chongqing urban agglomeration.

[Objective] Conventional watershed delineation methods often fail to capture surface heterogeneity. To address this limitation，this study proposes a dynamic-scale zoning method for delineating catchment units and develops a flood risk identification framework integrating land cover and dynamic-scale zoning. This framework couples high-precision simulation with large-scale risk identification，aiming to provide a transferable technical pathway for flood risk management and early warning in large Chinese cities. [Methods] The established methods adopted in this study include：（1） a coupled 1D-2D hydrological and hydrodynamic model based on SWMM and LISFLOOD-FP；（2） parameter sensitivity analysis using the Morris screening method；and （3） spatial pattern evolution analysis of flood risk based on Moran’s I index. The novel approaches proposed in this study include：（1） a dynamic-scale zoning method that accounts for surface heterogeneity for precise catchment delineation； and （2） a road-network topology-based generalization strategy for underground drainage networks，providing reliable support for simulations lacking actual pipe network data.Based on the above SWMM 1D pipe network model and the LISFLOOD-FP 2D hydrodynamic model，a coupled 1D-2D urban surface flood risk assessment model was constructed，which not only accounts for underground drainage facilities but also fully simulates the flood evolution process.The coupled model was validated using three extreme storm events in the Wuhan Shahu small watershed. [Results] （1） The model constructed using the dynamic-scale zoning method exhibited significant performance，with over 85% agreement between simulated and observed inundation points，and good synchronization in rainfall-runoff trends. （2） Spatiotemporal analysis of flood risk across different years revealed a phased improvement in the risk pattern under the influence of changing underlying surface conditions： the area of low-risk clusters expanded from 6.2% to 26.4%，while the proportion of high-risk clusters decreased substantially and stabilized. The SWMM-LISFLOOD-FP coupled 1D-2D modeling framework based on the dynamic-scale zoning method demonstrated good effectiveness in identifying key high-risk areas and provided a reliable technical pathway for accurate flood risk identification in cities lacking underground pipe network data. （3） Parameter sensitivity analysis indicated that the sensitivity of all parameters gradually decreased with increasing storm intensity and slope. In highly urbanized areas，the surface runoff coefficient alone is insufficient to characterize surface ponding processes and must be calibrated in conjunction with physical hydrological parameters. [Conclusions] The dynamic-scale zoning method proposed in this study can effectively account for surface heterogeneity. Regional empirical results demonstrate the reliability of the coupled modeling framework in addressing long-term spatial changes and hydrodynamic coupling simulations，providing practical technical support for accurate flood risk assessment in highly urbanized areas. A limitation of this study is the lack of simulation for water accumulation at overpasses and subsurface flow in underground spaces. Future research should focus on more comprehensive and detailed validation of the applicability of this assessment framework in complex environments at larger scales.

[Objective] The Storm Water Management Model （SWMM） is widely used in urban flood simulation and prediction. Existing studies have primarily focused on drainage systems in plain urban areas，with relatively limited attention to the composite hydrological systems of coastal piedmont cities. This study aims to identify the core driving factors of model responses under different land use types and quantitatively analyze the variation characteristics of dominant parameters with increasing rainfall intensity，providing a scientific basis for constructing high-precision urban flood simulation models and parameter calibration. [Methods] Taking the Jiangbei District of Fuzhou as a case study，we investigated the dynamic evolution of parameter sensitivity within mountainous （Bayi Reservoir） and urban （Qinting Lake） catchments across varying return periods （5-50 a） using the SWMM model. [Results] （1） The modified Morris method revealed a sensitivity transition in the dominant parameters for peak flow in the mountainous catchment with increasing rainfall intensity. The Horton decay constant dominated under low rainfall intensity （5 a，SN=1.48），while the Manning’s roughness of pervious surfaces （N-Perv） dominated under high rainfall intensity （50 a，SN=0.88）. Conversely，urban areas exhibit a consistent land-surface control effect，primarily governed by the Manning’s roughness for impervious areas （N-Imperv） and depression storage. （2） Sobol global sensitivity analysis indicates that under extreme rainfall，the first-order and total sensitivity indices of mountainous parameters are highly convergent （difference<0.05），manifesting strong parameter independence. In urban areas，however，the total sensitivity significantly exceeds the first-order sensitivity （by 1.5 to 2.5 times），revealing intense nonlinear coupling and interactions between surface runoff and hydraulic processes in the drainage network driven by heavy rainfall. 3） The total runoff in mountainous regions shows extreme sensitivity to N-Perv reflecting the physical mechanism where surface resistance regulates runoff travel time and cumulative infiltration. In contrast，due to the constraints of highly impervious surfaces，the urban runoff response is characterized by multi-parameter joint driving. [Conclusions] Rainfall intensity significantly regulates the sensitivity structure of model parameters，and the response intensity of different physical parameters to model output exhibits a nonlinear evolutionary trend with changes in their own values. Extreme rainfall scenarios effectively amplify the dominant role of core driving factors，revealing the mechanism logic of system transition from runoff generation-dominated to runoff concentration-constrained. This finding provides a physical basis for refined calibration of mountain-urban composite hydrological models and spatiotemporally differentiated parameter validation，contributing to improved flood warning accuracy and the scientific design of waterlogging mitigation strategies in mountainous cities under multi-intensity rainfall conditions.

[Objective] Urban waterlogging has become an increasingly critical issue in highly urbanized coastal megacities，where flat terrain，high imperviousness，and limited drainage capacity amplify flood hazards under extreme rainfall. Quantitative comparisons of hazard evolution under multiple design rainfall scenarios remain limited. This study focuses on the central districts of Shanghai，a representative coastal megacity，and aims to （1） analyze the evolution characteristics of urban waterlogging under different rainfall return periods，and （2） identify the spatial distribution patterns of flood hazard based on hydrodynamic simulation results，providing a quantitative reference for urban flood risk management. [Methods] A two-dimensional hydrodynamic model based on TELEMAC-2D was applied to simulate urban waterlogging processes in the central districts of Shanghai，covering seven densely built districts with nearly 100% impervious surfaces. Design rainfall events corresponding to 20-，50-，and 100-year return periods were constructed according to local standards. Surface runoff generation was represented using the Horton infiltration model，while drainage processes were approximated through an equivalent parameterization approach due to the lack of detailed pipe network data. Model outputs included time-series water depth and flow velocity fields. Based on these，maximum inundation depth，peak flow velocity，inundation extent，and total water volume were derived. Flood hazard was evaluated using the Flood Hazard Rate （HR），which integrates water depth，flow velocity，and a depth-dependent adjustment factor. Hazard levels were further classified to enable spatial comparison across scenarios. Model performance was evaluated by comparing simulated inundation areas with historical waterlogging points and by assessing simulated water depths against observed data from a typical rainfall event. [Results] （1） Model performance： The comparison with historical waterlogging records shows that the model effectively captures the spatial distribution of inundation-prone areas. Simulated water depths are in good agreement with observed values for typical rainfall events，indicating that the model can well reproduce the main characteristics of urban waterlogging processes. （2） Hydrodynamic characteristics： Urban waterlogging in the study area is predominantly characterized by shallow water depths （<0.15 m），low flow velocities，and overall low hazard levels across all scenarios，reflecting the flat terrain and existing drainage capacity. （3） Response to rainfall intensity： As the rainfall return period increases from 20 to 100 years，the proportion of deep inundation areas （>0.5 m） increases from 1.63% to 4.06%，while high-velocity areas （>0.4 m/s） increase from 0.66% to 1.30%. The maximum inundation area expands by 15.8%，and total water volume increases by 33.8%，indicating that flood intensification is mainly reflected in increased water depth and accumulation rather than spatial expansion.（4） Hazard distribution： The proportion of medium，high，and above hazard zones increases from 17.35% to 30.77% with increasing rainfall intensity. High and very high hazard areas exhibit fragmented and patchy spatial patterns and are mainly concentrated in low-lying areas and regions with limited drainage capacity. （5） Spatial heterogeneity： Hazard escalation is highly heterogeneous. Instead of uniform expansion，localized amplification dominates，where minor topographic depressions and drainage constraints lead to disproportionately high hazard levels. [Conclusions] Urban waterlogging in central Shanghai is generally dominated by shallow depth and low hazard under typical rainfall conditions； however，increasing rainfall intensity leads to a clear amplification of high-hazard areas. This intensification is primarily driven by increased water depth and localized accumulation rather than uniform spatial expansion.The results highlight that flood hazard distribution is strongly influenced by local terrain and drainage conditions，and mitigation efforts should prioritize low-lying and drainage-limited areas. The TELEMAC-2D model can well reproduce both the spatial pattern and magnitude of urban waterlogging，supporting its applicability for scenario-based flood analysis in highly urbanized areas.

[Objective] Addressing the increasing demands for disaster prevention and mitigation，it is essential not only to enhance the accuracy of rainstorm forecasting but also to assess in advance the potential disaster-causing risks posed by rainstorms. In contrast to evaluation models that rely on uniform topographic factors，this study aims to reflect the distinct disaster mechanisms of mountain torrents and urban waterlogging，thereby accurately revealing the comprehensive risk patterns of regional flood disasters caused by rainstorms. [Methods] By assigning specific topographic factors to mountain torrents and urban waterlogging respectively，this study develops a refined disaster-causing risk model that integrates hazard-inducing factors and disaster-specific hazard-inducing environments. The model enables mechanism-based zonation of the hazard-inducing environments for both disaster types and the integration of comprehensive risks. Taking Qinhuangdao as the empirical area，the validity of the proposed model is verified. [Results] （1） Fine-scale modeling of the hazard-formative environment effectively distinguishes the disaster-prone contexts for flash floods and urban waterlogging. The flat central-southern plains，characterized by concentrated urban and agricultural land，show a high hazard-formative environmental index for urban waterlogging，while the mountainous northern and northeastern areas，with significant topographic relief and dense river networks，exhibit a high index for flash floods. （2） The comprehensive high-risk zones for torrential rain disasters in Qinhuangdao are mainly located in the central part of the city，representing areas where high hazard intensity overlaps with highly sensitive disaster-prone environments. （3） The comprehensive risk index of historical rainstorm events shows correlation coefficients R of 0.721，0.698，and 0.724 with direct economic loss （GDP），affected population，and affected crop area，respectively，all passing the significance test at α=0.05. The correlation coefficient R between the comprehensive disaster index and the comprehensive risk index across counties （districts） reaches 0.912，passing the significance test at α=0.01，indicating that the model has good indicative significance for actual disaster situations. [Conclusions] The refined comprehensive risk model developed in this study，which integrates hazard-inducing factors and disaster-specific hazard-forming environments，effectively captures the differential mechanisms of mountain torrents and urban waterlogging. This model not only offers a unified framework for accurately delineating the regional pattern of comprehensive rainstorm and flood risk，providing a paradigm for flood risk assessment in analogous regions，but also can be applied to dynamic risk assessment for single hazard-inducing processes.

[Objective] Accelerating urbanization has deteriorated the balance between water storage and discharge，leading to frequent extreme flood disasters in river basins. Currently，there is a lack of quantitative indicators to characterize the storage-discharge relationship in highly urbanized watersheds，and the underlying mechanisms driving its spatiotemporal evolution remain inadequately understood. This study analyzes the evolving characteristics and influencing mechanisms of the storage-discharge relationship in highly urbanized watersheds，aiming to reveal how extreme rainfall-runoff events impact this fundamental hydrological relationship. [Methods] Taking the Qinhuai River Basin as the study area，we constructed storage-discharge relationship curves and selected the curve slope as a characteristic indicator to quantitatively characterize the relationship and its spatiotemporal evolution. Furthermore，Granger causality tests and partial correlation analyses were employed to identify the influencing factors of the storage-discharge relationship and to reveal its response characteristics under extreme storm scenarios. [Results] （1） The dominant function of the Qinhuai River Basin shifted from water storage to drainage. While storage dominated from 1990 to 1999，the basin transitioned towards drainage after 2000，although dynamic storage volumes remained above 100 mm. Post-2010，intensified urbanization made the drainage-dominant pattern more pronounced. （2） The nonlinearity of the storage-discharge curve exhibited a “weakening-then-strengthening” trend，with urbanization and extreme storms significantly amplifying this nonlinearity. Hydrological responses differed between low- and high-flow regimes，with the latter likely influenced by anthropogenic regulations. （3） The influencing factors of the storage-discharge relationship varied dynamically with flow levels. At low flows，the relationship was primarily driven by extreme precipitation and drainage capacity. At medium flows，the midstream and upstream areas showed a pronounced response to extreme storm characteristics. At high flows，despite increased frequency and intensity of extreme storms，the interaction between the storage-discharge relationship and extreme storm indices weakened，necessitating anthropogenic regulation to ensure basin safety. （4） Urbanization significantly impacted the storage-discharge relationship，generally reducing basin sensitivity and weakening regulation capacity. The upstream Lishui River exhibited the highest sensitivity （0.005，0.013，and 0.03 mm^-1 under high，medium，and low flows，respectively） and the largest variation amplitude under extreme storms，marking it as a critical area for focused management. [Conclusion] The quantitative indicators and methodologies proposed in this study for analyzing storage-discharge relationships provide a valuable scientific reference for urban flood control planning and the sustainable development of river basins.

[Objective] This study aims to investigate the differential impacts of varying rainfall intensity and types on urban flooding in the Jianghan Plain，provide scientific guidance to enhance urban waterlogging prevention and control capacity and advance the construction of resilient cities. [Methods] The urban area of Jingzhou was taken as the research area. Thirty-five groups of precipitation scenarios involving seven rainfall patterns and five recurrence intervals were constructed by adopting the fuzzy recognition method for rainfall patterns and the generalized extreme value method.The construction is based on the measured data of minute-by-minute precipitation and waterlogging depth from 1957 to 2024，as well as the surrounding geographic information data.Combined with the SWMM and LISFLOOD-FP waterlogging coupling model，the impact characteristics of different precipitation combination scenarios on the scope，depth，and duration of waterlogging were analyzed.The results are further used to clarify the response characteristics of urban waterlogging to different precipitation conditions. [Results] （1） With respect to the effects of rainfall patterns，type I （single-peak early-type rainfall） is not only the most prevalent pattern in short-time heavy precipitation events across Jingzhou’s urban area，accounting for 27.5%，but also the primary driver of severe urban waterlogging. Compared to the other six rainfall patterns，this pattern is characterized by concentrated precipitation and high intensity in the early stage，which tends to overload drainage systems within a short period. Consequently，it leads to more extensive and severe waterlogging，with earlier onset and peak times of water accumulation，as well as a higher average water depth. The suddenness and severity of waterlogging disasters are thus more pronounced. （2） As for the effects of recurrence intervals，across the 2-year to 100-year recurrence interval spectrum，extremely severe waterlogging is the dominant waterlogging type in Jingzhou’s urban area，with Rainfall Pattern I accounting for 8.0%-15.9%—7.6%-8.8% percentage points higher than the area covered by the least prevalent rainfall pattern. Spatially，the belt extending from the urban southeast to northwest，marked by low-lying topography and sparse drainage pipe networks，constitutes a high-risk zone for extremely severe waterlogging. Given the highest risk of waterlogging under extreme precipitation，this area should be prioritized for prevention and control measures.（3） With regard to disparities across local urban blocks，variations in block attributes give rise to distinct differences in waterlogging signatures. Notably，the Xueyuan road area，characterized by suboptimal drainage capacity，is disproportionately vulnerable to precipitation impacts. Under extreme precipitation scenarios，this area exhibits a 2.26 m higher peak waterlogging depth compared with the Liujiatai intersection of Jingzhou avenue，a site with superior drainage performance. Additionally，it experiences a 0.27 m greater mean waterlogging depth，a prolonged inundation duration，and consequently，elevated challenges for waterlogging mitigation and control. [Conclusions] This study elucidates the holistic characteristics and localized disparities of urban waterlogging in Jingzhou under variable rainfall intensities and patterns，identifies the high-risk zones，key drivers，and evolutionary trends of waterlogging hazards，and fills the critical research gap in prior investigations on Jingzhou’s urban waterlogging，which inadequately addressed the coupled impacts of rainfall intensity and pattern. The findings offer robust theoretical foundations and technical underpinnings for the design of waterlogging mitigation projects and the optimal upgrading of drainage infrastructure in Jingzhou’s urban area. Furthermore，they provide a valuable reference framework for waterlogging research and hazard mitigation initiatives in cities across the Jianghan Plain.

[Objective] The aim of this study is to clarify the mechanism of waterlogging under heavy rainfall in the Lianhua Bridge area，analyze the operational characteristics of the drainage system and the causes of water accumulation，and develop a hindcast simulation model providing a quantitative basis for addressing drainage issues in the bridge area and other similar sunken bridge areas. [Methods] A coupled drainage network-river model was constructed using the MIKE+ platform，including the combined sewer network，stormwater system，and downstream river channels. The model was validated against measured river discharge and water depth data. Key hydraulic indicators，such as pipe fullness，node overflow，and river water levels，were analyzed. The influence of river backwater and backflow on network operation was assessed by combining river outfall discharges with river water level hydrographs. The formation mechanism of waterlogging was examined using node overflow hydrographs. [Results] （1） Simulated river discharge and water depth matched the measured data，with key evaluation indicators within reasonable ranges. Pipe fullness in the combined sewer system generally exceeded 0.8 during peak rainfall，with some sections under full-pipe or surcharged conditions. （2） No node overflow occurred in the stormwater system； overflow was confined to manholes of the DN1800 combined main pipe. （3） Upstream interception facilities diverted large volumes of rainwater into the combined main pipe，increasing hydraulic load，while rising downstream river levels caused reverse flow at the outfalls. The superposition of these factors triggered manhole overflow. （4） Maximum intercepted discharge of the four outfall interception facilities reached 2.41 m³/s； river backflow through Outfall E reached 1.80 m³/s. （5） Total inflow during node overflow was 3.11×10⁴ m³，with a maximum overflow depth of 0.41 m； 91.37% of inflow came from upstream interception，and 8.63% from river backflow. [Conclusions] This study identifies a coupled hazard mechanism where high-load upstream interception and downstream river backwater jointly trigger manhole surcharge. By quantifying the contribution proportions of these key factors，this research explains the phenomenon where waterlogging occurs even when rainfall intensity remains below the design standards of the stormwater system. The findings reveal that water accumulation is primarily triggered by manhole surcharge on the combined main pipe rather than insufficient capacity of the stormwater system. This study provides a scientific basis for outfall regulation，interception pipe optimization，and drainage retrofitting in similar concave overpasses.

[Objective] This study assesses the waterlogging risk of the bus system in Tianjin’s central urban area using a zonal layered numerical simulation model. The systematic assessment includes identifying risks at bus routes and stops，analyzing the spatial distribution of waterlogging risk，and quantifying the length of inundated road sections. [Methods] The waterlogging disaster risks on bus routes in Tianjin’s central urban area were assessed and analyzed under short-duration （3-hour） and long duration （24-hour） design storm hyetographs. The Pilgrim & Cordery hyetograph was used for the short-duration design storm，and a frequency-consistent hyetograph design method was used for the long-duration one. A zonal layered mathematical model for urban rainstorm waterlogging was established，incorporating both two-dimensional planar and three-dimensional spatial simulation approaches to account for the heterogeneity of the urban underlying surface. This model coupled hydrodynamic modules for communities，roads，rivers，and pipe networks，enabling refined simulation of the central urban area and the prediction of waterlogging risks on road sections. Using design storms of various return periods （for both 3-hour and 24-hour durations） as precipitation boundaries，and based on the current status of drainage facilities，the model simulated the waterlogging risks on bus routes within the area inside the Outer Ring Road under different return periods. [Results] 1） Case study simulations of waterlogging risks on bus routes were performed for two rainstorm events of different intensity. In terms of the simulation of key road sections，12 out of the 14 waterlogged and closed roads announced by the traffic management department during Event 1 were simulated to meet the traffic interruption criteria （i.e.，waterlogging risk at Level 3 or above）. During Event 2，all four simulated underpass interruption points met the criteria. The model demonstrated high accuracy in simulating road traffic interruption risks and could well match the actual waterlogging-induced traffic interruption situations. From the perspective of simulating the waterlogging risk distribution of individual bus routes，the model could clearly reflect the spatial distribution characteristics of waterlogging risks on a single bus route，providing customized support for the refined assessment and management of waterlogging risks on individual bus routes. 2） Under the scenarios of design storms with different return periods，the impacts of short-duration （3-hour） and long-duration （24-hour） rainfall on waterlogging risks of bus routes were simulated，and the waterlogging risks of bus routes under various return periods were further analyzed. The results indicated that the number of bus routes and stops with Level 1 waterlogging risk caused by short-duration rainstorms was larger than that caused by long-duration ones. In contrast，the number of bus routes and stops with waterlogging risk at Level 2 or above induced by long-duration rainfall was higher than that induced by short-duration rainfall. Under each return period，the maximum rainfall intensity of the 24-hour duration was greater than that of the 3-hour duration，which resulted in more prominent waterlogging risks at Level 2 or above under long-duration rainfall. Meanwhile，the relatively long duration of short-duration intense precipitation led to a wider impact range of Level 1 waterlogging risk. [Conclusion] Bus routes with high waterlogging risks in the central urban area are highly agglomerated in six old urban districts，where the public transit network is densely distributed. Once a waterlogging disaster occurs，it is prone to triggering severe public transportation paralysis. The simulation results from two rainstorm case studies validated the model’s effectiveness. It accurately identified key points of transit service interruption and mapped the risk distribution along individual routes，thereby providing a scientific research method for the flood control and dispatch of public transit systems. Restricted by the current status of built-up areas，it is recommended that the drainage capacity of waterlogging-prone road sections should be improved by constructing new pumping stations and adding temporary drainage measures. Meanwhile，traffic management department should establish a mechanism for waterlogging risk assessment and emergency response of public transit routes based on the early warning information from multiple departments，formulate transit flood response plans，and timely adjust or suspend the operation of transit routes to avoid losses.

[Objective] The integration of mechanism-based models and deep learning emerges as a pivotal direction in urban flood prediction. While mechanism-based models possess a robust physical foundation，their computational efficiency is often constrained under high-resolution，multi-scenario inference conditions. Conversely，purely data-driven models frequently suffer from instability in predicting deep-water inundation. Focusing on the compound scenario of urban rainfall and high Yangtze River water levels，this study aims to synergistically validate the prediction accuracy，physical rationality，and driving mechanisms of a deep learning model. It seeks to deepen the understanding of the contributions and interactions of rainfall and tidal factors across varying risk levels. [Methods] Taking the main urban area of Nanjing as the study site，this research proceeds along two main lines： “accelerated computation” and “mechanism interpretation.” First，a coupled hydrological-hydrodynamic mechanism model was constructed，calibrated，and validated to generate 72 scenarios of rainfall-tide combinations. Second，a Flood Long Short-Term Memory （Flood LSTM） network，incorporating a hybrid loss function and physical non-negative constraints，was proposed and systematically compared against various machine learning and deep learning baselines. Finally，the SHapley Additive exPlanations （SHAP） framework was introduced to identify the contributions of rainfall and tidal drivers and their evolutionary characteristics across different risk levels. [Results] The Flood LSTM model effectively reproduces spatial inundation patterns under typical rain-tide scenarios with significantly improved computational efficiency. Under identical hardware conditions，the average inference time was 9.31 s，representing only 0.50% of the computational cost of the mechanism model. The model demonstrated high accuracy，with test set RMSE，NSE，and Top-5 MAE values of 0.0108 m，0.9918 and 0.0197 m，respectively. The NSE for deep-water zones （≥1.00 m） reached 0.946 8，and the Critical Success Index （CSI） exceeded 0.85 in most scenarios. SHAP analysis revealed a distinct compound rain-tide driving characteristic in the formation of urban waterlogging. The maximum rainfall intensity change rate （RChg） and the average ebb tide rate （TFAv） were identified as key drivers. The evolution of waterlogging in Nanjing generally follows a three-stage stratified driving pattern of “Rain-Tide-Rain，” characterized by progressively increasing risk. [Conclusion] The proposed Flood LSTM model exhibits high precision in the rapid prediction of maximum urban flood inundation depths. These findings enrich the theoretical understanding of urban flood disaster mechanisms and provide a reference for rapid early warning，engineering dispatch，and refined risk management. While the model shows promise，limitations remain regarding its reliance on simulated training data and its focus on static maximum depth prediction. Future work will incorporate more field observations and dynamic process data to further enhance model applicability.

[Objective] This study proposes a dynamic flood risk assessment method for transmission lines in flood storage and retention areas by integrating multi-source satellite remote sensing. The aim is to enhance the disaster prevention and mitigation capabilities of power grids，as well as the efficiency of post-disaster emergency response. [Methods] The proposed framework consists of remote sensing monitoring of the flood process，inundation monitoring of transmission towers，and dynamic flood risk assessment. The method integrates Synthetic Aperture Radar （SAR） and optical satellite imagery. Specifically，the Sentinel-1 Dual-polarized Water Index （SDWI） and the Modified Normalized Difference Water Index （MNDWI），combined with Otsu’s thresholding algorithm，were employed to accurately extract flood extents. These were then fused temporally to reconstruct the flood evolution process. By incorporating the inundation dynamics of transmission towers，a risk assessment matrix centered on inundation duration and voltage level was constructed to achieve a dynamic quantitative evaluation. The method was validated through a case study of the catastrophic “23·7” Haihe River Basin flood in the Dongdian Flood Storage and Retention Area in 2023. [Results] The proposed method effectively characterizes the spatiotemporal patterns of flood risk for transmission lines within the study area： The assessment identified that 51.7% of the towers were at high risk，with this proportion reaching 92.2% for Ultra-High Voltage （UHV） lines. The dynamic evaluation revealed that risks escalate with the accumulation of inundation duration and converge towards downstream depressions. Overall，the risk distribution exhibited a strict dependence on voltage levels. High-risk zones were predominantly concentrated in core depressions with prolonged water stagnation，confirming that the ultimate pattern of flood risk is jointly determined by transmission line voltage levels and long-term inundation conditions. [Conclusion] From a methodological perspective，integrating SAR and optical imagery with targeted water extraction （SDWI-Otsu and MNDWI-Otsu） and temporal fusion strategies significantly improves the accuracy and continuity of flood extent extraction in complex scenarios，enabling dynamic reconstruction of the entire flood evolution process. The constructed two-dimensional risk matrix，based on inundation duration and voltage level，is physically interpretable with easily accessible parameters，allowing for rapid dynamic and comprehensive risk assessment. Future work will focus on improving the accuracy of tower inundation duration by incorporating space-air-ground integrated remote sensing and hydrological modeling. Additionally，more structural and environmental parameters，such as tower foundation types and soil conditions，will be integrated to build a more refined multi-dimensional risk assessment model.

[Objective] Traditional monitoring methods for flood disasters primarily rely on single ground-based stations，which often suffer from limited spatial coverage，vulnerability to extreme weather，and delayed response capabilities. This research aims to address these limitations by systematically reviewing the “Space-Air-Ground” integrated monitoring framework. The primary objective is to investigate the synergistic mechanisms and key emerging technologies—such as Multi-source Data Fusion，Geospatial Artificial Intelligence （GeoAI），and Digital Twins—that enable high-precision，real-time，and intelligent flood perception and decision support. [Methods] The study adopts a multi-dimensional perspective to analyze the integrated monitoring architecture consisting of three layers： Space （Satellite clusters），Air （UAVs and aviation platforms），and Ground （IoT sensors and hydrological stations）. It evaluates the synergy between these layers across the Data，Transmission，and Application tiers. Specifically，the research reviews three levels of data fusion： data-level，feature-level （utilizing Cross-modal Transformers），and decision-level. Furthermore，it explores advanced data assimilation techniques，comparing traditional variational and sequential methods with modern Deep Learning-based approaches like PINN （Physics-Informed Neural Networks） and End-to-End Neural Assimilation. The study also classifies GeoAI applications in flood monitoring into four categories： inundation extraction，water level inversion，discharge estimation，and vulnerability assessment. [Results] The integrated framework significantly enhances monitoring reliability by achieving multi-layered spatio-temporal complementarity. Key results include：（1） Synergistic Perception： Ground-based sensors provide high-confidence “truth” data to calibrate satellite and UAV inversions，while satellite signals can trigger UAV swarms for targeted detailed inspections in a “discovery-tracking-verification” loop. （2） GeoAI Advancement： Modern models like Segment Anything Model （SAM） and Bitemporal Image Transformer （BIT） have improved water body segmentation accuracy under complex conditions，such as urban shadows or cloud cover，by utilizing cross-modal feature reconstruction. （3） Digital Twin Evolution： The technology has evolved from geometric visualization （L1） and physical mechanism simulation （L2） to logic-driven intelligent prediction （L3）. The integration of Reduced-Order Models （ROM） and GeoAI allows for “second-level” decision feedback by shifting computational burdens to the pre-training phase. （4） Operational Efficacy： Pilot applications in the Changjiang and Pearl River basins demonstrate that digital twin systems can support the “Four Pre-s” （Forecasting，Warning，Rehearsal，and Planning），providing critical technical support during major flood events. [Conclusions] This research concludes that the “Space-Air-Ground” integrated network represents the future of intelligent flood disaster management. The study’s innovation lies in proposing a vertically integrated “Perception-Model-Intelligence” architecture that moves beyond simple multi-source observation to a hybrid synergistic mechanism driven by both physical laws and data. However，several challenges remain：（1） technical barriers in high-precision cross-modal spatio-temporal alignment；（2） the lack of physical interpretability in “black-box” GeoAI models； and （3） the need for deeper coupling between digital twin systems and actual flood control business workflows. Future research should focus on：（1） Edge Intelligence： Developing lightweight algorithms for on-orbit real-time data processing. （2） Physics-Driven Sensing： Embedding hydrological mechanisms into neural networks to ensure consistency with physical laws. （3） Standardization： Establishing unified technical standards and interface specifications for digital twin flood systems to enhance their engineering applicability and cross-regional generalization.

[Objective] In existing urban flood resilience assessment frameworks，flood stress is typically quantified using precipitation data or hazard characteristics derived from hydrological and hydrodynamic models. This study utilizes Sentinel-1 SAR data to extract urban inundation extents and establishes a comprehensive indicator system to evaluate the flood resilience of cities within the water source area of the Middle Route of the South-to-North Water Diversion Project. This research aims to objectively reflect the actual coping capacities and vulnerabilities of these cities under extreme hydrological events，providing a scientific basis for flood risk management and resilience enhancement in the water source region. [Methods] Based on Sentinel-1 dual-polarization SAR satellite data，the SDWI （Sentinel-1 Dual-Polarized Water Index） for the study area was calculated，and water body was extracted using an empirical threshold. U-Net deep learning model was employed to improve the accuracy of water extraction from optical images. Urban flood resilience was assessed using the PSR （Pressure-State-Response） model from the three dimensions of pressure，state，and response. A total of nine indicators were adopted： water expansion area，river network density，average slope，per capita GDP，population density，green coverage rate of built-up areas，road network density，drainage pipeline density in built-up areas，and hospital accessibility. The EWM-TOPSIS method （Entropy Weight Method - Technique for Order Preference by Similarity to an Ideal Solution） was used to determine the indicator weights and assess urban flood resilience. [Results] Cross-validation between the pre-flood water extraction results and optical imagery achieved an overall accuracy of 94.6%. Based on SAR imagery，the continuous surface water extent dynamics in the water source area during the 2025 autumn flood were obtained for the period from July 15 to November 2. The results indicate that the increase in surface water area in and around the Danjiangkou Reservoir was initially concentrated in the Danjiang section of the reservoir； by September 26，the Hanjiang section also began to expand rapidly，reaching its maximum on October 21，when the reservoir was impounded to its normal storage level of 170 m. The inundated areas in the water source area during the flood period were primarily concentrated in the middle and lower reaches of the basin. Xichuan County exhibited the largest water expansion area （222.633 0 km²），followed by Danjiangkou City （104.566 5 km²），while Baihe County showed the smallest water expansion area （0.098 4 km²）. 2） Urban flood resilience assessment results reveal that water expansion area and per capita GDP had the highest indicator weights，at 0.345 and 0.228，respectively，whereas average slope and road network density had the lowest weights，at 0.015 and 0.013，respectively. Maojian District and Zhangwan District in Shiyan City，Danfeng County and Shangzhou District in Shangluo City，Mian County，Hantai District，and Liuba County in Hanzhong City，as well as Taibai County in Baoji City，exhibited high resilience. Shiquan County，Ziyang County，Langao County，and Ningshan County in Ankang City，Zhashui County in Shangluo City，and Foping County in Hanzhong City exhibited low resilience. [Conclusion] The inundated areas were concentrated in the middle and lower reaches of the water source area，and regions with larger water expansion extents were all located near the reservoir，indicating that the arrival of the autumn flood posed challenges to reservoir flood control regulation. Inundation was more pronounced at the confluences of main streams and tributaries，while areas at higher elevations experienced less inundation； overall，the inundated areas were concentrated in flat terrain. Areas with higher resilience in the water source area were generally distributed in the upper and lower reaches and in city center areas，whereas lower-resilience areas were located in the middle reaches and rural areas. Water expansion area and per capita GDP are important factors influencing urban flood resilience. To enhance the urban flood resilience of the water source area of the middle route，efforts should focus on improving the level of urban economic development and the capacity for reservoir flood control regulation.

[Objective] Existing research on digital twin-based urban flood control and drainage dispatch predominantly focuses on forward simulation，while inverse deduction from reservoirs to downstream river networks remains insufficient. This study develops a digital twin platform to conduct scenario simulations across the entire chain from reservoir forward dispatch，inverse calculations for discharge-limited control at downstream sections，to iterative scheme optimization，aiming to provide more precise technical support for urban flood control decision-making. [Methods] Taking Shenzhen Reservoir and its downstream Shenzhen River as the study area，we constructed a coupled hydrological-hydrodynamic model system covering the entire process of rainfall，runoff，dispatch，and evolution. Leveraging the JavaScript engine，hot-reloading，and XML batch processing technologies，the combined simulation workflow was decoupled into standardized functional units. Through the encapsulation of these units，independent configuration and flexible combination of hydrological，hydrodynamic，and dispatch rule models were achieved. Consequently，a digital twin flood dispatch platform for urban reservoirs was developed，featuring robust “forward-inverse-forward” iterative optimization capabilities. [Results and Conclusion] The platform successfully enables forward reservoir dispatch calculations，inverse calculations for discharge-limited control at downstream sections，iterative scheme optimization，and the decoupling and modular batch processing of flood simulation processes. This significantly enhances the operational iteration efficiency of the “forward-inverse-forward” simulation. Future research will introduce parallel computing and reduced-order modeling （ROM） techniques to improve hydraulic response speeds under complex dispatch scenarios.

[Objective] This study aims to systematically analyze the evolution law of water environment under the overlapping impacts of non-point source pollution （NPSP） and rain-sewage pipe network misconnections，particularly within the unique context of mountainous cities. Given the steep terrain，complex drainage topologies，and fragile ecological conditions in the upper reaches of the Yangtze River，traditional management models developed for plain cities are inadequate in addressing the compound hydrological and pollution dynamics in this region. Therefore，the primary objective of this study is to quantify the specific pollution contributions of these overlapping factors and develop an adaptive，full-process collaborative control framework—covering source，pipe network，and terminal—that is tailored to the topographical and hydrological characteristics of mountainous cities，thereby providing a scientific paradigm for regional water environment management. [Methods] A typical mountainous city in the upper reaches of the Yangtze River was selected as the research case，and an integrated methodological approach was adopted. The empirical coefficient method was employed to systematically trace and quantify pollution loads from various sources，distinguishing between point-source and non-point source contributions. Subsequently，the Storm Water Management Model （SWMM） was established and calibrated to reflect the specific terrain-driven hydrological characteristics of the study area，such as rapid runoff convergence and steep pipe slopes. Based on these foundational analyses，multi-scenario simulations were designed and implemented，progressively comparing the effectiveness of isolated interventions with integrated strategies. This process ultimately led to the formulation and simulation of a collaborative control scheme integrating source （Low Impact Development，LID），pipe network （renovation），and terminal （storage tanks）. The SWMM model was then used to dynamically quantify the hydrological responses and pollution reduction effects under each scenario. [Results] The results provide critical insights into the pollution dynamics and control effectiveness in mountainous cities.（1） NPSP from urban runoff was identified as the dominant contributor to water environment degradation，with contribution rates of chemical oxygen demand （COD_Cr），ammonia nitrogen （${\mathit{NH}}_{4}^{+}$），and total phosphorus （TP） reaching 93.5%，74.6%，and 82.7%，respectively.（2） An assessment of the current drainage infrastructure revealed severe deficiencies： the existing annual runoff control rate was only 20%，and rain-sewage pipe network misconnections further exacerbated pollution，resulting in ammonia nitrogen concentrations at multiple overflow outlets that far exceeded the prescribed water quality standards.（3） Multi-scenario simulations demonstrated that isolated control measures are insufficient for mountainous terrains. In contrast，the implementation of the proposed coupled control scheme—integrating source LID facilities，pipe network renovation，and terminal storage tanks—achieved significant synergistic effects. This collaborative scheme increased the annual total runoff control rate from 20% to 60.5% and achieved a high annual NPSP reduction rate of 68.5% for total suspended solids （TSS）. Most notably，regarding the critical issue of overflow pollution，the compliance rate of ammonia nitrogen concentrations at overflow outlets reached an unprecedented 99.7%，effectively mitigating the severe contamination previously caused by rain-sewage mixing during storm events. [Conclusions] This study highlights the dominant role of NPSP and the critical vulnerability of misconnected rain-sewage pipe networks in shaping the water environment of mountainous cities. The core innovation of this research lies in the conceptualization and validation of a full-process “source-pipe network-terminal” collaborative control scheme specifically adapted to the terrain-driven characteristics of mountainous hydrology. Unlike conventional end-of-pipe treatments or isolated LID implementations，this integrated framework synergistically addresses the dual challenges of rapid runoff convergence and combined sewer overflows. The proven effectiveness of this coupled scheme—evidenced by substantial improvements in runoff control，TSS reduction，and near-complete ${\mathit{NH}}_{4}^{+}$ compliance—demonstrates that coordinated intervention throughout the entire drainage continuum is essential. Consequently，this full-process collaborative strategy effectively resolves the complex dilemma of NPSP control in mountainous cities，providing a transferable scientific paradigm and robust technical support for water environment governance in similar mountainous cities in the upper reaches of the Yangtze River and other regions.

[Objective] The static design of urban drainage infrastructure，which separates green-gray infrastructure （GGI） planning from dynamic operational strategies，limits its effectiveness in mitigating flood peaks under extreme rainfall events. This study develops a multi-objective optimization model that integrates real-time control （RTC） with GGI to enhance both peak shaving capacity and cost-effectiveness. The primary objective is to systematically evaluate how RTC influences the cost-benefit relationship and optimal configuration of GGI under various climate scenarios. [Methods] A representative urban drainage catchment in Shenzhen，China，covering an area of 85.6 hectares，is selected as the case study. A coupled TVGM-SWMM hydrological model is established to simulate rainfall-runoff and pipe network processes. Storage tanks are real-time controlled through a predictive fuzzy logic control （PFLC） method combined with a target flow allocation strategy，enabling coordinated dynamic regulation. The Non-dominated Sorting Genetic Algorithm II （NSGA-II） is applied to minimize life cycle cost and maximize life cycle comprehensive environmental benefit. Decision variables include： ratio of green infrastructure coverage area，total storage volume of gray infrastructures，and RTC parameters. Future climate scenarios are generated using the change factor methodology applied to ten CMIP6 GCMs dataset，producing design storms for three shared socioeconomic pathways （SSP126，SSP245，SSP585） at three future time horizons （2030，2040，2050）. This comprehensive framework allows systematic assessment of RTC impacts on GGI performance across current and future climate conditions. [Results] （1） The integration of RTC significantly enhances the peak shaving capability of GGI under extreme rainfall conditions. For the 100-year design storm with 24-hour duration，RTC achieves a peak flow reduction of 38.5% at the catchment outlet，reducing discharge from 24.98 to 15.32 m³/s. The advantage of RTC becomes increasingly pronounced as rainfall intensity and duration increase. （2） From an economic perspective，coupling RTC with GGI substantially improves cost-effectiveness. To achieve equivalent peak reduction targets of 20%，40%，and 60%，the life cycle cost of the system is reduced by 32%，47%，and 39%，respectively，compared to static control scenarios. The investment threshold required to generate positive environmental returns is markedly lowered，and the diminishing marginal utility effect is mitigated. As rainfall intensity increases across SSP scenarios （SSP585>SSP245>SSP126），environmental benefits naturally decline，but RTC achieves incremental benefit improvements of 6%-13% at equivalent cost levels compared to static control. （3） RTC also alters the optimal configurations of infrastructures and their investment structure. Under low-cost constraints （<0.25 million CNY/ha），RTC enables gray infrastructure volumes up to 8×10³ m³ with green infrastructure coverage below 1%，whereas static control relies predominantly on green infrastructure （coverage 17%-28%） with gray volumes below 1×10³ m³. Analysis of contribution ratios confirms that under RTC，gray infrastructure's benefit contribution consistently exceeds its cost contribution，whereas under static control the opposite pattern prevails. Despite this shift，green infrastructure retains significant investment share （i.e.，25%-75%） across all climate scenarios，indicating a complementary synergy rather than substitution between green and gray components. [Conclusions] Real-time control significantly enhances both the flood mitigation performance and economic efficiency of green-gray infrastructure systems under current and future climate conditions. By dynamically coordinating distributed storage facilities，RTC strengthens the marginal contribution of gray infrastructure，reduces the investment threshold for achieving environmental gains，and optimizes the synergy between green and gray components. The finding that green infrastructure maintains substantial investment share under RTC underscores its continuing importance in source control and runoff reduction，which alleviates downstream storage demand and indirectly enhances RTC effectiveness. This complementary relationship suggests that RTC does not replace green infrastructure but rather enables more efficient utilization of the integrated system. These findings demonstrate that integrating RTC with GGI offers a promising pathway to improve urban flood resilience and investment returns. The proposed modeling framework，which couples hydrological simulation，multi-objective optimization，and climate scenario analysis，provides a valuable decision-support tool for climate-adaptive design and operation of urban drainage systems. Future research should explore the application of this framework to larger spatial scales，incorporate water quality objectives，and investigate real-time implementation challenges including sensor networks，communication systems，and control reliability.

[Objective] Model Predictive Control （MPC） has been increasingly applied to the coordinated regulation of sluices and pumps in urban drainage systems to maximize the utilization of existing storage capacity and mitigate waterlogging risks. However，research on leveraging observational data to improve model prediction performance remains limited. This study focuses on measurement uncertainty in real-time control of urban drainage systems and proposes an Unscented Kalman Filter-based Model Predictive Control （UKF-MPC） method to enhance system adaptability under complex conditions. [Methods] By integrating model predictions with sensor observations，the proposed method incorporates data assimilation into the error feedback strategy and state update of the MPC prediction model，establishing a real-time data-driven MPC framework. The approach was validated in the Doumen area of Fuzhou City. [Results] （1） UKF-MPC significantly improved flow prediction accuracy by dynamically fusing observational information with model outputs，increasing the Nash-Sutcliffe Efficiency （NSE） from 0.39-0.58 to 0.56-0.79，with a mean Relative Improvement Percentage （RIP） of 36.4%. （2） Through data assimilation，the method upgraded traditional error feedback to state-space correction. Under 50-year and 100-year return period storm scenarios，UKF-MPC consistently generated more robust control strategies，enhancing system adaptability in complex environments and providing a framework capable of real-time system state perception and dynamic control optimization. [Conclusions] Compared with conventional MPC，the control strategies derived from UKF-MPC exhibit superior applicability and robustness under uncertain conditions. Future research will quantitatively analyze the influence of UKF parameter settings on correction performance and conduct systematic validation across multiple watersheds and diverse rainfall events using richer observational data，to further clarify the applicability boundaries and parameter sensitivity mechanisms of UKF-MPC. On this basis，further exploration will focus on incorporating state covariance information directly into the optimization problem to develop a risk-aware model predictive control method.

[Objective] Conventional grey pipe-network-dominated drainage systems exhibit limited adaptability when confronted with beyond-design storm events. Existing optimization studies of green-grey infrastructure are predominantly conducted under historical rainfall conditions and insufficiently account for future climate change scenarios and their associated uncertainties，particularly with respect to the systematic selection and integration of multiple climate models. To address these gaps，this study constructs a multi-objective optimization framework coupling an XGBoost surrogate model with the NSGA-III algorithm，driven by CMIP6 multi-model climate projections. [Methods] Dahongmen Area in the Liangshui River Basin of Beijing was taken as a case study. We first evaluated eight CMIP6 global climate models （GCMs） that have shown relatively strong performance in northern China. Using historical rainfall observations from 1982-2014 as the benchmark，spatial downscaling was conducted via linear interpolation，followed by bias correction using the Delta method. A comprehensive assessment framework was then established by integrating the Taylor Score （TS） and the Interannual Variability Score （IVS）. Based on this framework，three models—EC-Earth3，ACCESS-CM2，and IPSL-CM6A-LR—were identified as the best-performing candidates. These selected models were subsequently combined using a weighted ensemble approach to construct future rainfall sequences under SSP1-2.6，SSP2-4.5，and SSP5-8.5. Second，to improve computational efficiency for optimization，an XGBoost （XGB） surrogate model was developed to characterize the nonlinear response relationships among rainfall characteristics，the deployment scale of green-grey infrastructure，and the resulting total runoff and cumulative overflow. Finally，with the minimization of annualized cost，total runoff，and total overflow as the objective functions，the XGBoost was coupled with the NSGA-III algorithm for multi-objective optimization. This produced Pareto-optimal solution sets under each scenario. Representative designs—including cost-optimal，compromise，and benefit-optimal solutions—were then selected to systematically analyze the configuration structure and evolutionary patterns of green-grey infrastructure across varying investment levels. [Conclusions] （1） EC-Earth3，ACCESS-CM2，and IPSL-CM6A-LR show relatively good performance in precipitation simulation for the Dahongmen area of Beijing. Compared with single models，the weighted multi-model ensemble improves the simulation accuracy of historical precipitation and reduces the uncertainty caused by biases of individual models. （2） Under different future emission scenarios，extreme rainfall intensity shows an overall increasing trend. Under the SSP5-8.5 scenario，the 10-year return-period rainfall exceeds the historical 50-year level，and the 100-year return-period rainfall intensity increases by 43.9%，indicating a higher risk of exceedance for urban drainage systems. （3） Optimization of green-grey infrastructure effectively reduces runoff and overflow risks. Under the low-emission scenario （SSP1-2.6），overflow can be completely controlled. Under the high-emission scenario （SSP5-8.5），a certain overflow risk still exists even at high investment levels，and the investment cost required to achieve the same control target increases with the intensification of emission scenarios. （4） The allocation of green-grey infrastructure presents obvious staged evolutionary characteristics. In the early stage of optimization，centralized grey detention facilities are dominant，which enhances the basic regulation capacity of the drainage system. With increasing investment，the optimization strategy gradually shifts toward distributed green infrastructure. Within green infrastructure measures，the priority changes from permeable pavement to green roofs，reflecting a structural transition from centralized detention to source control. This study reveals the adaptive evolution mechanism of urban green-grey infrastructure configuration under climate change，and provides a scientific basis and technical support for the phased construction and investment decision-making of urban drainage systems under intensified extreme rainfall conditions.

[Objective] To address the insufficient physical representation and poor adaptability to complex boundaries of the traditional Preissmann slot method in simulating open channel-to-pressurized flow transitions in drainage networks，this study develops a one-dimensional hydrodynamic model using the Godunov-type finite volume method. [Methods] The model employs two sets of governing equations for free-surface and pressurized flows，coupling the Two-Component Pressure Approach （TPA） with an improved Harten-Lax-van Leer with Source term （HLLS） Riemann solver. This establishes a unified simulation framework for open-channel flow，pressurized flow，and mixed open channel-pressurized flow. Accordingly，the model proposes treatment methods for various typical boundaries，including junction nodes，storage tanks，control gates，pump station outlets，and river-lake systems，based on the characteristic line equations and actual pipeline geometric parameters，significantly enhancing its applicability in pipe network systems. The model is validated through classical test cases such as tree-like and looped pipe networks，showing excellent agreement between simulated and expected results. [Result] The study reveals that the widely used SWMM model，which is based on the Preissmann slot method and the “pipe-node” conceptualization approach，produces unrealistic physical phenomena—such as synchronized water level variations at adjacent nodes—when simulating unsteady flow propagation in pipe networks. In contrast，the proposed model accurately captures non-uniform water level rises and transient fluctuation characteristics along pipelines，providing a more realistic representation of flow propagation. [Conclusion] The proposed model is suitable for detailed dynamic simulations of mixed free-surface and pressurized flows in urban drainage networks.

[Objective] Frequent urban waterlogging caused by insufficient flow capacity of stormwater drainage systems under extreme weather conditions has become a pressing issue. This study aims to investigate the variation in flow capacity of a stormwater drainage system with rectangular box culverts，considering the geometric parameters of an L-shaped sediment barrier at the inlet and the backwater effect （downstream water-level-induced backwater）. [Methods] A dual-method approach was adopted，combining hydraulic model tests with 3D computational fluid dynamics （CFD） simulations. The hydraulic model，based on a gravity-driven similarity criterion （λ_L=20），simulated a typical stormwater drainage system in Eastern China consisting of rectangular box culverts and multiple manholes. Flow rates were measured using standard thin-walled triangular weirs，and downstream water levels were precisely regulated to simulate backwater conditions. Hydraulic model tests were conducted to analyze the flow capacity of stormwater drainage system inlet structures under conditions with and without L-shaped sediment barriers and backwater effects. Following the hydraulic model tests，numerical simulations were conducted using ANSYS Fluent. The SST k-ω model was selected，and the numerical model was validated against experimental data，with deviations in total head and piezometric head kept within 10%. To quantify the impact of the L-shaped sediment barrier，three primary geometric dimensions were systematically varied： the opening size in the flow direction （a），the opening size perpendicular to the flow direction （c），and the vertical height of the barrier （e）. [Results] Sediment barriers and water level differences significantly dictate the system’s drainage efficiency. Hydraulic model tests showed that removing the sediment barrier entirely could increase flow capacity by up to 64.69% compared to the test with the barrier in place. The numerical analysis of the barrier’s geometric parameters provided deeper insights into structural optimization. Among the studied variables，the dimension perpendicular to the flow direction （c） was found to be the most influential factor，resulting in a 7.68% increase in flow rate. In contrast，the dimension in the flow direction （a） showed a moderate impact，resulting in a 4.98% improvement in drainage efficiency. The vertical height （e） proved to be the least sensitive parameter，with reducing it from 650 mm to 350 mm yielding only a 2.27% gain in capacity. Furthermore，the study highlighted a strong positive correlation between the inlet-outlet water level differential and the overall flow rate. When the downstream backwater effect was minimized by lowering the water level to 86.00 m，the flow capacity showed a dramatic increase even without any structural changes. Crucially，the benefits of geometric optimization became more pronounced under low water levels. Specifically，with the high head difference provided by the 86.00 m downstream level，optimizing the perpendicular dimension （c） yielded a 17.54% improvement rate，significantly outperforming the gains observed at higher downstream levels. [Conclusions] While structural optimization of inlet components is beneficial，addressing downstream backwater effects in the primary lever for enhancing stormwater drainage performance.Among the geometric parameters of L-shaped sediment barriers，widening the opening perpendicular to the flow direction is identified as the most effective structural modification for inlets. Accordingly，a dual-stage strategy combining water level regulation and structural optimization is proposed.The study provides a scientific basis for urban waterlogging mitigation under extreme rainfall scenarios and offers novel insights into the optimized design of stormwater drainage systems，demonstrating substantial engineering application value.

[Objective] Sedimentation at the downstream of sluice gates severely constraints the drainage capacity of pumping sluices in the tidal reaches of the Yangtze River，and also raises dredging costs. This study seeks to elucidate the causes and find solutions for the sedimentation of pumping sluices by optimizing their layouts to improve regional drainage capacity and river hydrodynamic condition. [Methods] The Beiheng River Pumping Station and Sluice Project in Pudong New Area，Shanghai was taken as a case study. Numerical simulations and analyses of the engineering effects under different layout schemes were conducted using measured data analysis and a three-dimensional hydrodynamic-sediment mathematical model. By comprehensively comparing the flow patterns，velocity distributions，and erosion-deposition results of each layout scheme，the planar layout of the Beiheng River project was evaluated. [Results and Conclusion] （1） Different planar layouts of pumping station and sluice projects in the tidal reaches of the Yangtze River result in significant differences in hydrodynamic characteristics and sedimentation control effectiveness. （2） The length of the straight flow section upstream of the sluice and the coverage area of the high-velocity zone downstream of the sluice are the core factors controlling sedimentation downstream of the sluice. （3） By adapting the planar layout of the pumping station and sluice to local topography and boundary conditions，hydrodynamic-sediment simulations can provide universal design guidelines for the construction and renovation of similar river-connected pumping stations and sluices in the Yangtze River Delta region. This is of great significance for enhancing the flood control resilience of coastal urban clusters and advancing the modernization of water management systems.