Flood control risk is one of the main risks affecting the safe operation of large-scale water transfer projects. Systematically identifying the flood control risk in the project and carrying out risk classification and hierarchical management are problems for project managers. Based on the theory of system and risk assessment, this paper starts with the various risk sources and risk events involved in the whole process of the flood disaster chain, the risk of flood disaster factors, the exposure of the disaster-bearing body, and the vulnerability of the disaster-originating environment are combined. Then, we systematically and comprehensively identify the flood control risks of a large-scale water transfer project, which are divided into four types of risk elements: rainfall–runoff; confluence and flow capacity; the geological characteristics of canal section; economic and social layouts. Specific risk factors are identified for each type of risk element, and a flood control risk evaluation index system for a water transfer project is proposed. According to the framework of the analytic hierarchy process (AHP), a quantitative assessment of comprehensive flood control for water transfer projects is carried out. Taking the middle route of the South-to-North Water Transfer Project in China as an example, this paper evaluates the integrated flood control risks of 39 engineering units, identifies six units with higher risk levels, analyzes the causes, and suggests engineering and non-engineering countermeasures to prevent and reduce the occurrence of risk accidents. This method is not only used for comprehensive flood control risk assessment and risk management in the operation and management stage of the large-scale inter-basin water transfer project, but also has a reference value in considering the optimal layout of the project water transmission line from the perspective of flood control in the planning and design stage.

Based on the regional disaster system theory, this study constructed a comprehensive flood risk indicator system for small reservoirs, covering the entire flood disaster process from three dimensions: hazard, vulnerability, and exposure. The analytic hierarchy process (AHP) and entropy weight method (EW) were used to determine indicator weights, and a risk assessment was conducted for small reservoirs in Huangshan City, Anhui Province, China. The results indicate that most reservoirs exhibit medium–low overall risk, yet distinct localized high-risk zones exist. High-economic-density areas such as Tunxi District, the central–eastern parts of Huangshan District, and the central and eastern parts of Qimen County have become high-risk clusters due to prominent exposure indicators (numbers of villages and medical facilities). Reservoirs in western and northern regions exhibit higher hazard levels, primarily driven by rainfall and catchment areas. Dam height and reservoir capacity are the main factors affecting vulnerability, with no significant spatial clustering for high-vulnerability reservoirs. Through the decoupling of three-dimensional indicators, this study reveals the differentiated driving mechanisms of “hazard–vulnerability–exposure,” providing a scientific basis for optimizing flood control engineering (e.g., reservoir capacity expansion, dam reinforcement) and risk zoning management (e.g., emergency evacuation planning in high-exposure areas) for small reservoirs.

The existing parameter calibration methods are based on the objective function surface to find the optimal value. This kind of method has the problems such as unstable result, poor convergence performance, low efficiency and failure to find the global optimal value. Based on the analysis of the objective function structure and the information that it provides for the parameter calibration, the essential problems existing in the present method was found in this paper. And the paper also found that the information provided by the parameter function surface is more direct and effective than by the objective function surface. Furthermore, the nonlinear model function can be linearized by the system response relationship between the increment of the dependent variables and the increment of parameters. Based on these researches, this paper proposed the system response parameter calibration method based on the parametric function surface. Firstly, the method is verified by an ideal model. The results showed that all the sensitive parameters could reach the real values not influenced by the different initial parameter values with higher convergence speed and accuracy, which verified that the method is feasible. Lastly, the XAJ Model parameters based on Shaowu basin measured data were also calibrated by the method. The results showed that the stable optimal parameter values could also be quickly got. So the system response parameter calibration method is an effective parameters optimization method.

High-resolution precipitation fields are essential for flash-flood forecasting and hydrological risk management, especially in small and medium-sized basins, yet single-source precipitation products often show limited accuracy over complex terrain. This study develops a two-stage machine-learning framework for 1 km/1 h multi-source precipitation fusion over Shaoxing, China, during the 2025 flood season. In the first stage, a machine-learning classifier identifies precipitation occurrence and reduces zero-inflated noise; in the second stage, an optimized tree-based residual-regression model corrects precipitation estimates for rainy samples. A 61-dimensional feature set was constructed by integrating satellite precipitation estimates, weather-radar precipitation estimates from the Zhejiang radar network, temporal-lag and accumulation statistics, neighborhood descriptors, cyclic time variables, and terrain-derived interaction features, with gauge observations used as the training target. After quality control, the dataset comprised 41,458 hourly station samples from 72 rain gauges. The stations were divided at the station level into a 57-station development set and a fixed 15-station held-out spatial test set containing 8637 hourly samples. Station-blocked fivefold cross-validation within the development set was used for model selection, hyperparameter tuning, and probability-threshold selection, whereas the held-out stations were used only for final performance evaluation. On the fixed held-out test set, the occurrence classifier achieved an overall accuracy of 0.947, with a probability of detection of 0.806, a false alarm ratio of 0.158, a critical success index of 0.700, and an F1 score of 0.823. For quantitative estimation, the two-stage fusion product reduced root mean square error from 2.342 mm for satellite precipitation estimates to 1.189 mm, corresponding to a 49.22% reduction, and decreased mean absolute error from 0.712 mm to 0.262 mm, while increasing the coefficient of determination to 0.685. The fused precipitation product also improved the detection of intense rainfall events, with probability of detection and critical success index reaching 0.511 and 0.442, respectively, for events exceeding 10 mm/h, while reducing false weak precipitation and showing closer agreement with observed station-level spatial variability. By separating precipitation-occurrence identification from rainfall-intensity correction, the framework reduces zero-inflated bias, improves heavy-rainfall representation, and demonstrates predictive skill at gauges excluded from model development during the 2025 flood season.