Post-event debris-flow gullies frequently retain substantial loose material, yet the residual risk they pose is difficult to evaluate due to a scale mismatch: catchment-scale runout models lack the resolution to assess local structural impacts, while structural-scale models typically rely on idealized inflow conditions. To address this, we develop a cross-scale analytical framework applied to Baima Gully along the Jiumian Expressway, which retained abundant loose debris following the 16 August 2020 event. Residual source materials were quantified through field investigations, UAV surveys, high-resolution topographic data, and remote-sensing interpretation. The 2020 event was back-analyzed using MassFlow software to calibrate dynamic parameters, and the validated model was then employed to predict potential residual debris-flow behavior. Resulting hydrographs at the dam site were subsequently used as inputs for fluid–structure interaction simulations in ANSYS CFX and complementary physical flume tests. MassFlow predictions indicate that a future residual flow could attain a peak velocity of 4.11 m·s−1 near the gully outlet and form a fan-shaped deposit with a maximum flow depth of approximately 5.91 m. Implementation of a proposed solid gravity check dam reduces outlet impact velocity by 49

Flood disasters seriously affect the safe development of human society. Climate change and urbanization have led to an increase in flood susceptibility, and the future development model is full of challenges. However, floods in existing research results are often static and lack dynamic change analysis. This study systematically proposes a new comprehensive framework. Based on the background of CMIP6, the PLUS model and the spatiotemporal dual attention network (STDAN) are coupled to predict the dynamic changes of flood susceptibility under the background of future climate change. The interpretability module is connected to the network to explain the contribution of flood factors to flood susceptibility. The experiment predicts the distribution of land use types and the spatial state of flood susceptibility under different climate scenarios in 2030. A case study is conducted in Shenzhen to verify the feasibility of this method. The following results are obtained from the experiment: the degree of land use change in 2030 is consistent, the expansion of construction land is the most obvious in the SSP585 scenario, the protection of ecological land is the best in the SSP119 scenario, and the development of various types of land in the SSP245 scenario is balanced. The STDAN model has higher accuracy for flood sensitivity, and the proposed method shows higher stability and enhanced goodness of fit compared with traditional machine learning methods and deep learning architectures established under different experimental conditions. Compared with 2020, the flood susceptibility of each scenario in 2030 has increased, with SSP585 being the highest, followed by SSP245, and SSP119 being the lowest. The analytical model found that elevation, precipitation, and land use type are the primary influencing factors in the study area. The experimental results can provide a scientific reference for regional flood management to reduce the uncertainty brought about by future climate change.

Abstract Flash floods are sudden flood events triggered by intense rainfall, and often exacerbated by mountainous terrain that accelerates surface runoff. To support disaster mitigation and management, deep learning (DL) models have been widely applied to flash flood susceptibility (FFS) modeling. However, traditional deep learning (DL) models overlook both the intra‐annual temporal variations and the spatial interactions between nearby catchments. To address these, this study proposes a graph‐based DL model (named LTG model) for spatiotemporal FFS simulation at a daily scale in China considering catchment topology. The proposed LTG model mainly integrates three components: Long Short‐Term Memory Networks, Temporal Graph Convolutional Networks, and Graph Convolutional Networks aiming to capture temporal dependencies, spatio‐temporal interactions, and spatial dependencies between catchments, respectively. This enables the LTG model to perform spatiotemporal dynamic FFS simulation as well as incorporating catchment topology information. We demonstrated the proposed LTG model in China and found that the proposed model outperforms the baseline models with the highest Area Under the ROC Curve (AUC) of 0.911 and Critical Success Index of 0.719. Compared to yearly‐scale modeling, the daily‐scale simulation generated by LTG model exhibits a higher ability to capture seasonal variations, with a significant intra‐annual standard deviation of 0.263. By conducting a detailed analysis of FFS changes along river networks and establishing biased rainfall scenarios, we found that the proposed model not only considering the spatial clustering along river networks, and upstream–downstream dependence, but also enhances its inferential ability by leveraging information from nearby catchments.

This study presents a multimode coupled nonlinear flutter approach in terms of the rational function (RF) approximation technique and amplitude-dependent flutter derivatives, to address the underdevelopment of three-dimensional (3D) nonlinear flutter analysis for long-span suspension bridges. A high-order RF is invoked to characterize the 3D inhomogeneous distributed nonlinear self-excited forces resulting from spanwise variated flutter derivatives and multi-mode coupling. An iterative approach is utilized to determine the spanwise amplitude distributions under specific cycle steps, while a matrix optimization procedure is introduced to least squares identifying the coefficient set of the RF relevant to 3D aerodynamics. The frequency-independent complex eigenvalue method combined with cycle calculations is used to determine the 3D distributed modal characteristics as functions of wind speed and amplitude. By utilizing the existing double-layer iterative method along with a numerical example, the precision and robustness of the proposed method are validated from various perspectives, including flutter response, modal characteristics, spanwise limit cycle oscillation amplitude, and 3D nonlinear vibration characteristics. In conclusion, the method demonstrates good precision, robustness, and the ability to automatically search for multi-mode coupling characteristics and shows fewer limitations on the number of modes.

Long pipelines installed overhead in mountain tunnels crossing fault zones are particularly vulnerable to failure under strong earthquake excitation. Therefore, the seismic performance of pipeline bearings requires careful attention. To improve the seismic resilience of pipeline-tunnel structure in these high intensity earthquake areas, a novel buffer pipeline bearing design has been proposed. The damping effect of this new design was validated using numerical simulation and large-scale shaking table testing. First, the failure modes of typical pipeline bearings were analyzed, with a focus on the characteristics and mechanisms of damage observed in past earthquake events. Based on these findings, a novel buffer bearing design was developed, and its core structure was introduced. Subsequently, the dynamic responses, including stress, strain, and pipeline displacement at various monitoring points of the pipeline-tunnel structure under seismic excitation, were investigated. The results indicate that the novel buffer pipeline bearing increases the peak value of tensile stress at the vault and invert regions of the tunnel structure, while reducing the peak value of compressive stress at the arch waist. More importantly, it significantly reduces the axial tensile and compressive stresses, as well as the displacement deformation of the pipeline structure under strong seismic excitation, allowing the pipeline structure to undergo overall longitudinal displacement. Analysis of the shaking table test data further shows that the strain value at the tunnel arch shoulder in each monitoring section is approximately 1.6 times that at the arch foot, and that the arrangement of different pipeline bearings exerts minimal impact on the tunnel lining. The seismic isolation measures of the new buffer bearing significantly reduce the strain response of the pipeline structure, with strain values at the pipeline bottom and waist reduced by 36.9% and 66.5%, respectively, compared with the original design bearing, and the peak value of x-direction displacement reduced by 35.2%. Therefore, incorporation of the novel buffer bearing into the structural design of pipeline-tunnel crossing active fault fracture zones is recommended to enhance the seismic performance of pipeline structure.

To enhance wind resistance safety for construction personnel and structural integrity, this study investigates the buffeting response and vibration damping measures of a steel truss stiffened arch bridge with a main span of 400 meters during its maximum cantilever construction state. A finite element model was developed, and a three-dimensional pulsating wind field was simulated using the harmonic synthesis method. Time-domain analysis was applied to compute buffeting displacement responses at the cantilever end of the arch rib and the top of the construction buckle tower. Numerical results were compared with wind tunnel tests of a full-bridge model under varying wind speeds, revealing similar buffeting response patterns. At the bridge reference wind speed, predicted peak buffeting displacements were 22.157 and 21.778 cm in the lateral and vertical directions of the arch rib cantilever end, and 16.994 cm laterally at the buckle tower top, with deviations from wind tunnel tests of 13.2%, 10.2%, and 6.9%, respectively. To mitigate these displacements, lateral wind-resistant cables and flexible connections were analyzed. Lateral cables reduced displacements by up to 84.8% at the arch rib cantilever end and 61.0% at the buckle tower top, while flexible cables further reduced responses by up to 76.8%, ensuring enhanced construction safety. The agreement between numerical and experimental results validates the proposed methods, providing a strong basis for wind-resistant design and vibration damping strategies in similar large-span bridges.

During construction, long-span steel truss arch bridges exhibit complex structural behaviors and are vulnerable to wind-induced vibrations. Effective evaluation of wind resistance during critical construction stages remains challenging. The objective of this research is to investigate the wind-resistant performance of a large-scale steel truss arch bridge during construction using experimental and numerical approaches, with insights into recent relevant patents on wind-resistant measures for bridge construction. A large-scale aeroelastic full-bridge wind tunnel test was performed alongside refined finite element (FE) analyses of different construction stages. FE models were established to characterize the dynamic behavior evolution throughout construction. Wind-induced responses under critical construction scenarios, particularly at the maximum cantilever state, were experimentally measured and numerically validated. Results indicated significant variations in structural dynamic characteristics throughout different construction stages. Among these, the maximum cantilever stage exhibited the most pronounced wind-induced response, with increased sensitivity to wind loading. Combining wind tunnel experiments with FE analysis provided an accurate assessment of the bridge';s wind resistance. The effectiveness of proposed wind-resistant measures specifically designed for construction phases, aligned with recent patented technologies, was demonstrated through substantial reductions in wind-induced vibrations. The comparative study between experimental and numerical methods highlights the complementary strengths of both approaches in capturing wind-induced responses during construction. The proposed mitigation strategies not only align with patented wind control technologies but also provide practical insight into construction-stage wind safety for similar bridge types. The integrated methodology employing wind tunnel tests and finite element simulations effectively assessed and enhanced the wind resistance performance during the construction of long-span steel truss arch bridges. Findings of this study offer valuable guidance for wind-resistant design and construction management of similar large-scale bridge projects, contributing to advancements outlined in recent patents related to wind mitigation technologies for bridge engineering.

Long-span bridges are increasingly vulnerable to flutter instability due to their reduced stiffness and natural frequencies, which amplify their sensitivity to wind-induced vibrations. While traditional analyses focus on uniform flow conditions, real-world atmospheric wind is predominantly turbulent, making it critical to evaluate the impact of turbulence on bridge aerodynamic stability. This study investigates the flutter characteristics of streamlined box girders under the influence of grid-generated turbulence. Using wind tunnel tests, flutter derivatives were identified through forced vibration experiments conducted in both uniform and turbulent flow fields, and free vibration tests were performed to determine flutter critical wind speeds under varying turbulence conditions. The results reveal that turbulence significantly alters the flutter behavior of streamlined box girders. In uniform flow, flutter is characterized by abrupt vibration divergence at a critical wind speed, while in turbulent flow, the response transitions to gradual amplitude growth without a clear divergence point. Turbulence parameters, including intensity and integral scale, influence aerodynamic damping and modify flutter derivatives, leading to changes in flutter critical wind speeds. A comparison of theoretical and experimental flutter critical wind speeds demonstrates the reliability of flutter derivatives identified using the forced vibration method. This research highlights the complex role of turbulence in bridge aerodynamics, providing theoretical and experimental insights into the effects of turbulent flow on flutter stability. The findings contribute to improved predictive capabilities for aerodynamic performance and offer guidance for the design and safety assessment of long-span bridges in turbulent wind environments.

The net primary productivity (NPP) is an important indicator for assessing the carbon sequestration capacities of different ecosystems and plays a crucial role in the global biosphere carbon cycle. However, in the context of the increasing frequency, intensity, and duration of global extreme climate events, the impacts of extreme climate and vegetation phenology on NPP are still unclear, especially on the Qinghai-Xizang Plateau (QXP), China. In this study, we used a new data fusion method based on the MOD13A2 normalized difference vegetation index (NDVI) and the Global Inventory Modeling and Mapping Studies (GIMMS) NDVI3g datasets to obtain a NDVI dataset (1982–2020) on the QXP. Then, we developed a NPP dataset across the QXP using the Carnegie-Ames-Stanford Approach (CASA) model and validated its applicability based on gauged NPP data. Subsequently, we calculated 18 extreme climate indices based on the CN05.1 dataset, and extracted the length of vegetation growing season using the threshold method and double logistic model based on the annual NDVI time series. Finally, we explored the spatiotemporal patterns of NPP on the QXP and the impact mechanisms of extreme climate and the length of vegetation growing season on NPP. The results indicated that the estimated NPP exhibited good applicability. Specifically, the correlation coefficient, relative bias, mean error, and root mean square error between the estimated NPP and gauged NPP were 0.76, 0.17, 52.89 g C/(m2·a), and 217.52 g C/(m2·a), respectively. The NPP of alpine meadow, alpine steppe, forest, and main ecosystem on the QXP mainly exhibited an increasing trend during 1982–2020, with rates of 0.35, 0.38, 1.40, and 0.48 g C/(m2·a), respectively. Spatially, the NPP gradually decreased from southeast to northwest across the QXP. Extreme climate had greater impact on NPP than the length of vegetation growing season on the QXP. Specifically, the increase in extremely-wet-day precipitation (R99p), simple daily intensity index (SDII), and hottest day (TXx) increased the NPP in different ecosystems across the QXP, while the increases in the cold spell duration index (CSDI) and warm spell duration index (WSDI) decreased the NPP in these ecosystems. The results of this study provide a scientific basis for relevant departments to formulate future policies addressing the impact of extreme climate on vegetation in different ecosystems on the QXP.

To gain an in-depth understanding of the failure characteristics and reinforcement mechanisms of the anchorage layer in tunnel, a series of loading failure experiments were conducted. By isolating the anchorage layer, its bearing capacity was effectively quantified. This study systematically examined the stress, deformation, and failure characteristics of the surrounding rock under unsupported conditions (surrounding rock layer, SRL), anchor bolt support (anchorage layer, AL), and pre-tensioned anchor bolt support (pre-tensioned anchorage layer, PAL). Furthermore, the incorporation of P-wave velocity (Vp) and PIV testing provided a robust framework for elucidating the reinforcement mechanisms of the anchorage layer. The failure of SRL initiated at the tunnel shoulders, whereas for AL and PAL, failure originated at the tunnel crown. The Vp at the tunnel crown of PAL and AL exhibited increases of approximately 11.0

Flood is one of the most destructive natural disasters occurring across the globe. Employing machine learning models to construct flood susceptibility maps has emerged as an effective strategy in disaster prevention and management. Sample size is one of the primary sources of uncertainty in machine learning model, posing significant challenges to the flood susceptibility in data-scarce regions. However, the understanding of uncertainty patterns and effective methods to improve modeling accuracy under limited sample conditions are still evolving. Here, we applied uncertainties analysis theory to clarify this pattern for seven base machine learning models. Further, an integration strategy was developed by coupling geographical similarity, semi-supervised learning and active learning method. The analysis of uncertainty pattern indicates that each base machine learning model exhibits varying degrees of tolerance to changes in sample size. Specifically, a threshold exists below which the accuracy of model declines sharply, leading to significant changes in the distribution patterns of predicted flood susceptibility maps. The proposed integration strategy can enhance the accuracy and stability of models operating with limited sample sizes. Applying the ensemble strategy and increasing the number of labeled samples from 10 to 500, the average AUC values for the models improved as follows: RF ranged from 0.76 to 0.85, SVM from 0.46 to 0.86, MLP from 0.77 to 0.86, NB from 0.75 to 0.86, KNN from 0.72 to 0.83, DT from 0.65 to 0.78, and LR from 0.70 to 0.86.The insights into uncertainty pattern derived from this study can help guide the balancing of sample collection costs with model accuracy. Moreover, the proposed integration strategy is expected to improve flood susceptibility prediction in areas with limited samples.

With the high incidence of extreme rainfall, the differences in flooding brought about by the urban-rural gradient (URG) have led to unexpected damage, and rainfall patterns may affect the extent of flooding differences on the URG. Currently, the patterns of dynamic responses of flood risk (FR) along the URG to different rainfall patterns remain poorly understood. In this study, a framework combining coupled hydrodynamic modeling and multi-indicator decision analysis was developed to quantify the characteristics of the dynamic response of flood risk to different rainfall patterns (RFRRP) and analyze its trends along the URG, with key data sourced from local hydrological and meteorological departments. Results indicated that the further back the rainfall peak was during the rainfall period, the longer the flood risk response time was, which could be shortened by increased rainfall. The time of peak occurrence of FR in the lower reaches of the river may be outside of the rainfall period: FR in these areas was lower during rainfall and starts to increase after rainfall stops, and the higher the rainfall, the more significant the increase in FR was. The RFRRP pattern remained across the URG. FR showed a decreasing and then increasing trend on the URG (Urban -- Sub-urban - Middle -- Sub-rural -- Rural). In general, FR was significantly larger in areas close to the countryside than in those close to the city, and FR in the rural area was about three times as large as that in the sub-urban area and about 1.5 times as large as that in the urban area under the 50-year return period, with differences worsening under heavier rainfall. The results of the study can guide planning and design of urban-rural integration and flood mitigation in areas with similar urban-rural gradient distribution.

Debris flow is one of the most devastating natural hazards. Identifying the dynamic changes and driving factors of debris flow risk can enhance hazard mitigation and prevention. It is not clear what factors can mostly lead to debris flow risk change in mountainous areas, particularly some of these areas in the context of intense earthquakes, rapid urbanization, and climate change. To address these questions, an ensemble learning model was constructed to estimate the debris flow risk of the baseline period (2000) and the current period (2020) in the upper reach of the Min River. The study found that the areas with extremely high debris flow risk decreased by 18.57%, while the areas with moderate and high risk levels increased by 8% and 14% respectively. With this trend of overall risk increasing, the population and buildings affected by extremely high debris flow risk have increased by 20% and 30% respectively. Based on the interpretable learning model of SHAP (The Shapley Additive Explanations value), the mechanisms by driven factors that lead to changes in risk were explored. Population, elevation and NDVI are the most influential factors leading to changes in risk. Specifically, the increase in risk in the low elevation area is due to the rapid urbanization caused by the increase of population and GDP. While the risk change in higher elevation areas contributes to the variation of vegetation and precipitation. These findings have implications for debris flow mitigation and contribute to the understanding of the multiple factors that impact debris flow risk.

Flash floods frequently occur in China, seriously threatening the ecological environment and human safety. To effectively prevent flood disasters, assessing flash flood susceptibility (FFS) using a hybrid model that integrates statistical and machine learning methods is feasible. Before constructing this hybrid model, classifying the flash flood conditioning factors is a key step. However, the impact of the classification number for these factors on hybrid model performance is still indistinct, and there is a lack of a method to determine the optimal classification number. Therefore, this study proposes an innovative framework to improve FFS modeling. Within this framework, we introduce the Optimal Parameter Geographic Detector (OPGD) method to perform optimal discretization of factors triggering flash floods. This method primarily utilizes the q statistic to automatically select discretization schemes with optimal classification performance. The classification results are then applied to construct three hybrid models: a hybrid model combining Information Value (IV) with Random Forest (RF), Support Vector Machine (SVM), and Lightweight Gradient Boosting Machine (LGBM). Taking Shannan City in the Tibet Autonomous Region as an example, the performance of the hybrid models before and after applying OPGD was evaluated using ROC curves and confusion matrices, and FFS map was generated. Multiple experimental results indicate that applying the OPGD method to the three hybrid models enhances their performance. For example, in the IV-RF hybrid model, accuracy improved by 0.02, 0.06, and 0.01, respectively, after introducing OPGD, and the AUC values also increased by 0.003, 0.014, and 0.003, respectively; in the IV-SVM hybrid model, the accuracy improved by 0.04 and 0.01, respectively, and the AUC values increased by 0.027 and 0.007, respectively; for the IV-LGBM hybrid model, the accuracy improved by 0.02 and 0.02, respectively, and the AUC values increased by 0.015 and 0.005, respectively. Although the improvement in AUC values is limited, even minor performance improvements hold significant practical implications for flash flood prevention and FFS assessment. This study validated the effectiveness of OPGD in enhancing the performance of FFS modeling, providing new insights for improving the accuracy of FFS modeling. (c) 2025 Published by Elsevier B.V. on behalf of COSPAR.

Long-span bridges are vulnerable to flutter instability, which can lead to catastrophic failure if not properly assessed. Traditional analyses have focused on smooth flow conditions, which do not fully reflect real-world aerodynamic conditions where boundary layer turbulence plays a significant role. This study delves into the flutter characteristics of streamlined box girder bridge decks, focusing on the effects of boundary layer turbulence. A novel analytical approach is introduced, incorporating spanwise correlation of self-excited aerodynamic forces into flutter analysis. Initially, wind tunnel tests involving forced vibration of segmental models were conducted in both smooth and turbulent flows to determine the flutter derivatives of the bridge deck. This was followed by an investigation of flutter critical wind speeds under varying conditions using taut strip model free vibration tests. The highlight of this research is the development of a comprehensive three-dimensional flutter analysis method that integrates the spanwise correlation effect. Findings indicate a significant influence of boundary layer turbulence on the flutter derivatives, with the observed flutter critical wind speeds in turbulent conditions surpassing those in smooth flow. The study also notes a decrease in flutter critical wind speeds with increasing turbulence intensity and integral scale. Importantly, the incorporation of spanwise correlation effects into the analysis yields theoretical flutter critical wind speeds that closely match those observed in wind tunnel experiments. This research contributes to a deeper understanding of aerodynamic behavior in bridge decks under turbulent conditions and enhances predictive capabilities in bridge aerodynamics.

Flood is one of the most devastating natural hazards. Employing machine learning models to construct flood susceptibility maps has become a pivotal step for decision-makers in disaster prevention and management. Existing flood conditioning factors inadequately account for regional characteristics of flood in the depiction of topography, potentially leading to an overestimation of flood susceptibility in flat areas. Addressing this gap, this study proposes a novel flood conditioning factor, local convexity factor (LCF), to enhance the accuracy of flood susceptibility modeling. Initially, LCF is computed based on a standard normal Gaussian surface to highlight elevation variations in local terrain. Subsequently, LCF is applied to flood susceptibility modeling using seven machine learning models across four distinct basins. Comparative analysis is conducted between flood susceptibility maps with and without the application of LCF to evaluate its impact on flood susceptibility modeling. The results demonstrate that the proposed LCF can enhance the accuracy of flood susceptibility modeling to varying degrees, across the four basins investigated. The Fujiang basin exhibited the most substantial improvement, with its AUC improved from 0.861 to 0.886, Producer’s Agreement improved from 0.869 to 0.899, and Overall Agreement improved from 0.778 to 0.811. Comparation with hydrodynamic inundation maps shows that particularly in relatively flat terrain areas, flood susceptibility maps incorporating LCF offer more precise delineation between flood-prone and non-flood-prone zones. This research holds potential for widespread application in the prediction of flood susceptibility using machine learning models, providing a novel perspective for enhancing their accuracy

Flooding has emerged as a critical global issue exacerbated by climate change and urban development. This study addresses the significant spatial heterogeneity in flood severity along the urban-rural gradient, which is influenced by variations in ecohydrological and geochemical structures. Despite the importance of understanding the dynamic evolution of flood risk under different disaster scenarios, its influence by urban-rural gradients and response to future uncertainty remain inadequately explored. By developing a comprehensive vulnerability and hazard index framework integrated with a hydrodynamic model, this study not only simulated the dynamic evolution of Urban Vulnerability Value (UV) and Township Hazard Value (TH) during heavy rainfall events but also highlighted their wider implications for flood risk management under future uncertainty. The findings revealed that urban vulnerability exhibited a stable and concentrated distribution, while township hazard values demonstrated significant sensitivity and variability. Notably, the study underscored that under future uncertainty, flood risk responses are likely to intensify (UV and TH increasing by 2.4 and 2.7 times, respectively), with township areas becoming more sensitive to flooding. These insights contributed to a deeper understanding of how urban-rural gradient influences dynamic flood risk, offering valuable guidance for effective flood management strategies in similar regions.

This investigation meticulously explores the vortex-induced vibration (VIV) characteristics in Pi-shaped girders within long-span cable-stayed bridges, presenting novel VIV mitigation techniques. Employing detailed wind tunnel evaluations of a 1:50 scale model alongside computational fluid dynamics (CFD) analyses to model the two-dimensional flow around the girder, this study reveals the formation and shedding of vortices behind the railings and main steel girder, which significantly contribute to VIV in such structures. The introduction of horizontal splitter plates, one m in width, in conjunction with two-m-wide V-shaped guide vanes, was found to substantially reduce the VIV amplitude of the Pi-shaped girder, minimizing vortex generation on the deck's sides and dramatically reducing the maximum vertical VIV amplitude from 208.75 mm to only 5.56 mm. CFD simulations confirm the effectiveness of these measures in suppressing Karman vortex street formation, thereby significantly improving downstream airflow characteristics. These findings offer pivotal insights for the advancement of bridge design and the proactive management of VIV, showcasing the potential of integrated aerodynamic modifications to enhance structural resilience and performance.

Geological disasters in the Emei to Mianning section of the Chengdu–Kunming Railway located in southwestern China occur frequently, and many debris flow gullies are distributed along the line. Debris flows often cause damage to the railway and highway that threatens the safe operation of the original line and the double-track line. Along the railway, the debris flow outbreak risks are analyzed for 53 debris flow gullies using field survey, remote sensing interpretation, and numerical simulation. Their developmental characteristics were judged according to the code. The results are used for numerical simulation of debris flow to further determine the movement characteristics and the accumulation range of debris flow. The results are combined with high-definition satellite images to analyze the form of the railway passing from the mouth of the ditch and the simulation results to determine the impact on the safe operation of the Chengdu–Kunming Railway when the debris flow erupts. Finally, disaster prevention and mitigation projects are proposed, and the prevention and control effects of the proposed works are validated.

Abstract. In the past decades, the world has experienced rapid urbanization and observed the appearances of large amount urbanizing watersheds with enhanced flooding, which has a constant changing land use/cover(LUC) types as the most significant feature. Simulating and forecasting urbanizing watershed flood processes faces great challenges, one is how to relate model parameters with the changing LUCs to secure an accurately and reliable simulation and forecasting results. In this study, a methodology for simulating and forecasting urbanizing watershed flood processes is proposed, which employs Liuxihe model as the watershed hydrological model. This methodology sets up the Liuxihe model with latest terrain properties, then derives initial parameter look-up table based on terrain properties, and optimizes it if there is observed hydrological data. If there is LUC changes, then the parameters are updated with the changed LUCs based on the optimized parameter look-up tables. Case study in a highly urbanizing watershed in the Pearl River Delta Area in southern China has shown that this method acquires accurate and reliable flood processes simulation results. Further more, this study has proven an assumption that the hydrological model parameters are LUC stationary, i.e., with the LUC changes, the parameter look-up table will not change, parameter look-up table optimized in a specific time with current LUCs will not change even the LUCs changed. With this assumption, the parameter look-up table only needs to be optimized once. This is a science question that has not been not well answered yet by the scientific communities.