Space-time coupling feature-oriented drainage basin flood disaster analysis system

Abstract

The invention belongs to the technical field of flood disaster monitoring and analysis, and particularly relates to a space-time coupling feature-oriented drainage basin flood disaster analysis system, which comprises a space-time data fusion module, a coupling process simulation module, a disaster chain deduction module and a dynamic decision support module, and is characterized in that the space-time coupling feature-oriented drainage basin flood disaster analysis system is obtained through multi-source data assimilation, space-time coupling modeling and dynamic risk deduction. And accurate simulation of the flood process, chain-type development analysis of secondary disasters and optimal generation of emergency schemes are realized. By constructing a unified spatio-temporal data fusion framework and a coupling process simulation mechanism, the technical bottlenecks that multi-source heterogeneous data is difficult to integrate and spatio-temporal coupling feature expression is insufficient are effectively solved; and the physical authenticity and prediction accuracy of flood process simulation are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of hydrological disaster monitoring and analysis technology, specifically a watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics. Background Technology

[0002] In the field of natural disaster monitoring and risk assessment, flood disasters have become a crucial global research topic due to their suddenness, wide impact, and enormous socio-economic losses. Watershed flood disaster analysis systems, as a core technological tool in this field, aim to simulate, predict, and assess flood processes by integrating multi-source data and model algorithms, providing a scientific basis for disaster prevention and mitigation decision-making.

[0003] Among these, basin flood disaster analysis oriented towards spatiotemporal coupling characteristics is a current research focus. The core of this technical direction lies in accurately characterizing the dynamic interaction and feedback mechanisms of hydrological and hydraulic elements in the temporal and spatial dimensions during flood evolution, in order to reveal the inherent laws of disaster formation and improve the accuracy of forecasting and early warning.

[0004] Existing technologies typically employ independent time-series analysis or static spatial interpolation methods to process flood data, which struggles to effectively represent the nonlinear coupling relationships of flood elements across time and space. Traditional models are insufficiently responsive to dynamic factors such as changes in underlying surface conditions and human activity disturbances, leading to significant discrepancies between simulation results and actual conditions. Furthermore, existing systems lack a unified spatiotemporal coupling framework when integrating multi-scale and multi-temporal remote sensing monitoring data with ground observation data, resulting in low data utilization and difficulties in model parameterization. In terms of disaster chain analysis, existing methods have limited ability to identify the spatiotemporal correlation of secondary disasters during flood evolution, making it difficult to support risk assessment and emergency response across the entire chain.

[0005] Therefore, this invention provides a watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0007] The technical solution adopted by the present invention to solve its technical problem is: a watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics, comprising: a spatiotemporal data fusion module, a coupling process simulation module, a disaster chain inference module, and a dynamic decision support module; The spatiotemporal data fusion module receives and processes multi-source heterogeneous data from remote sensing satellites, ground monitoring stations, meteorological radars, and hydrological stations, achieving seamless integration of multi-scale data by constructing a unified spatiotemporal reference framework. This module further employs an adaptive interpolation algorithm to repair missing data and removes outlier observations based on spatiotemporal consistency verification criteria, thus forming a watershed data cube with spatiotemporal continuity. The coupling process simulation module, connected to the spatiotemporal data fusion module, drives a distributed hydrological and hydrodynamic model with a spatiotemporal coupling mechanism based on the data cube. The distributed hydrological and hydrodynamic model describes surface runoff and soil water movement through a coupled form of solving the Saint-Venant and Richards equations. It embeds spatiotemporal convolution operators during the equation solving process to accurately enhance the characterization of the spatial heterogeneity of precipitation runoff processes and its temporal lag effects. Simultaneously, it uses a two-way coupling method to simulate the dynamic interaction between surface water and groundwater. The distributed hydrological and hydrodynamic model also incorporates human activity influencing factors, reflecting the interference of land use change and water conservancy project scheduling on flood processes by real-time correction of underlying surface parameters. The disaster chain simulation module is connected to the coupled process simulation module, used for spatiotemporal correlation analysis of secondary disasters based on simulated flood inundation results. This module first identifies the spatial distribution of disaster-bearing bodies within the flood's impact area, then constructs a multi-hazard coupled probabilistic network model. By calculating disaster triggering thresholds and propagation paths, it simulates the chain-like development process of secondary disasters such as landslides, debris flows, and urban flooding. The dynamic decision support module is connected to the disaster chain simulation module, used to generate risk assessment maps and emergency response plans based on the simulation results. This module dynamically calculates the risk level at different spatiotemporal locations by overlaying vulnerability curves of disaster-bearing bodies with real-time flood elements, and outputs the optimal evacuation routes and rescue dispatch plans based on a preset emergency resource distribution and traffic accessibility model.

[0008] Preferably, the data processing procedure of the spatiotemporal data fusion module includes the following steps. First, spatiotemporal registration is performed on the multi-source data, transforming data with different resolutions and projected coordinate systems to a standard spatiotemporal grid. Then, Kriging interpolation is used to fill in spatially missing data, while time series analysis algorithms are applied to repair time-series fragmented data. Finally, by calculating the data gradient and statistical distribution characteristics of adjacent spatiotemporal units, outliers deviating from the normal range are identified and removed, thereby ensuring the integrity and reliability of the data cube.

[0009] Preferably, the distributed hydrological and hydrodynamic model of the coupled process simulation module uses a grid discretization method to divide the watershed into several computational units. Each unit includes soil moisture content, surface roughness, permeability coefficient, and river cross-sectional morphology parameters. This distributed hydrological and hydrodynamic model describes surface runoff and soil water movement through a coupled solution of the Saint-Venant and Richards equations. During the equation solving process, a spatiotemporal convolution operator is embedded to accurately enhance the characterization of the spatial heterogeneity of precipitation runoff and its temporal lag effects. Simultaneously, a two-way coupling method is used to simulate the dynamic interaction between surface water and groundwater. Human activity impact factors are embedded in the model as dynamic parameter correction terms, automatically adjusting the hydrological parameters of the corresponding units based on real-time land use change data and reservoir scheduling instructions.

[0010] Preferably, the construction process of the multi-hazard coupling probability network model in the disaster chain inference module is as follows: Using flood inundation depth and flow velocity as initial disaster-causing factors, the slope stability index and sediment initiation critical conditions are calculated based on a historical disaster case database and a mechanical mechanism model. When the index or condition exceeds a preset threshold, the corresponding secondary disaster node is triggered. The model establishes a directed graph network for disaster propagation by analyzing the spatial proximity and temporal sequence between disaster nodes, and uses Monte Carlo simulation to statistically analyze the trigger probability and impact range of each link.

[0011] Preferably, the risk assessment map generation method of the dynamic decision support module includes the following steps. First, the flood inundation map and the disaster-bearing body distribution map are spatially overlaid. Then, according to the disaster-bearing body type, the corresponding vulnerability curve function is called to convert the flood intensity index into economic loss rate and probability of casualties. Finally, by combining real-time population thermal data and building structure information, the comprehensive risk index of each grid unit is calculated, and a dynamic risk map is generated according to a preset level classification standard.

[0012] Preferably, the emergency response plan generation process of the dynamic decision support module also includes the optimization calculation of evacuation routes. This process, based on real-time road network capacity and flood evolution prediction results, uses an improved Dijkstra algorithm to search for the optimal route from the risk area to the safe area. It also considers the capacity limitations of path nodes and the spatiotemporal variations in crowd evacuation speed to ensure that the generated route plan is feasible and efficient in actual evacuation.

[0013] Preferably, the system further includes a model parameter adaptive calibration module, which is connected to the coupled process simulation module and is used to dynamically adjust the model parameters based on real-time observation data. The calibration module employs an ensemble Kalman filter algorithm to update the model's state variables and parameter fields by assimilating the latest water level and flow rate observations, thereby adjusting the simulation accuracy and reducing prediction uncertainty.

[0014] Preferably, the system adopts a multi-layered software architecture, including a data access layer, a business logic layer, and a user interaction layer. The data access layer is responsible for the collection and storage management of multi-source data, the business logic layer encapsulates all computational models and algorithm cores, and the user interaction layer provides a visual display and solution configuration interface. Data exchange and function calls between the layers are conducted through standard application programming interfaces (APIs), ensuring the system's scalability and maintainability.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: By constructing a unified spatiotemporal data fusion framework and coupled process simulation mechanism, the technical bottlenecks of difficulty in integrating multi-source heterogeneous data and insufficient expression of spatiotemporal coupling features are effectively solved. By introducing a distributed hydrological and hydrodynamic model with spatiotemporal convolution operators, the physical realism and prediction accuracy of flood process simulation are significantly improved. The established disaster chain inference module can systematically identify the spatiotemporal correlation of secondary disasters caused by floods, providing a scientific basis for full-chain risk management. The dynamic decision support module generates risk maps and emergency plans based on real-time inference results, which have high timeliness and pertinence, greatly enhancing the decision-making efficiency of disaster emergency response. The introduction of the adaptive calibration module for model parameters further ensures the long-term reliability and accuracy stability of the system. The overall system architecture is rationally designed, and the modules work together efficiently, meeting the practical needs of flood disaster analysis in complex watershed environments. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 This is a schematic diagram of the architecture of the watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the distributed hydrological and hydrodynamic model with a spatiotemporal coupling mechanism in this invention; Figure 3 This is a logical flowchart of the multi-source data assimilation and spatiotemporal data fusion in this invention; Figure 4 This is a schematic diagram illustrating the construction process of the multi-hazard coupling probability network model in the disaster chain inference module of this invention; Figure 5 This is a logical framework diagram of risk assessment map generation and emergency plan optimization in the dynamic decision support module of this invention; Figure 6 This is a schematic diagram illustrating the interaction between the model parameter adaptive calibration module and the system's multi-layer software architecture in this invention. Detailed Implementation

[0018] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0019] Example 1

[0020] This embodiment details a specific implementation scheme for a watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics. Please refer to the appendix. Figure 1 The system consists of a spatiotemporal data fusion module, a coupled process simulation module, a disaster chain simulation module, a dynamic decision support module, and a model parameter adaptive calibration module. These modules operate collaboratively via a data bus and control logic. The system employs a multi-layered software architecture, including a data access layer, a business logic layer, and a user interaction layer, with inter-layer communication and module interaction achieved through standard application programming interfaces (APIs).

[0021] The specific implementation process of the spatiotemporal data fusion module is as follows. This module is responsible for accessing multi-source heterogeneous data, including remote sensing satellite imagery, meteorological elements collected by ground monitoring stations, flow and water level data recorded by hydrological stations, and meteorological radar reflectivity information; As mentioned above, please refer to the appendix. Figure 3 The module first performs spatiotemporal registration on the input data, transforming data with different spatial resolutions and projection coordinate systems into a standard spatiotemporal grid. Continuing from the above, the Gauss-Krüger projection is used as the spatial reference, and the International Standard Time System is used as the time dimension to generate regular three-dimensional grid cells covering the entire watershed. Each grid cell contains geographic coordinates, timestamps, and multiple observation attribute fields. For spatially missing data, the module uses the Kriging interpolation algorithm for repair. Its core principle is to calculate spatial autocorrelation based on the semi-variogram function and estimate the value of unknown points through weighted averaging. For temporally fragmented data, an autoregressive integral moving average model is used for imputation. This model predicts missing values ​​through the trend and periodic components of historical sequences. In the data quality control stage, the module calculates the data gradient and statistical distribution characteristics of adjacent spatiotemporal units, and initially uses a deviation from the mean of the same region exceeding three standard deviations as the threshold for outlier screening. Then, a second judgment is made by combining the differences in data types and the spatiotemporal correlation. For extreme data of natural hydrological elements such as rainfall and flow, the data of surrounding monitoring units in the same period, historical extreme value records and weather process background are further compared. Only when it is confirmed that there is no real physical cause to support it is it judged as an outlier and removed. Extreme values ​​with real causes are retained to ensure data integrity and rationality. The module outputs a watershed data cube with spatiotemporal continuity. Its data structure includes more than 20 fields such as latitude and longitude, elevation, time, precipitation, soil moisture, and river flow. Each field is accompanied by a data quality identifier and uncertainty measure.

[0022] In this embodiment, the coupling process simulation module is connected to the spatiotemporal data fusion module and is used to drive a distributed hydrological and hydrodynamic model with a spatiotemporal coupling mechanism based on the data cube. The distributed hydrological and hydrodynamic model accurately describes the heterogeneity of runoff generation and confluence in different spatial units of the watershed, as well as the transmission lag and cumulative effect of runoff in the time dimension, by introducing spatiotemporal convolution operators, and restores the spatiotemporal evolution law of flood formation. The model adopts a two-way coupling method between surface water and groundwater to construct a hydraulic interaction mechanism between the two. By setting the hydraulic boundary conditions between the subsurface layer and the unsaturated and saturated zones, it simulates in real time the infiltration and recharge of surface water to groundwater, the discharge and return of groundwater to surface water through the rise of the water table, and the water volume transfer and momentum exchange process between the two at the interaction interface, so as to realize the quantitative simulation of the dynamic interaction between surface water and groundwater. The model incorporates human activity influencing factors and dynamically corrects key parameters such as underlying surface permeability, roughness, and reservoir capacity based on real-time land use and water conservancy project scheduling data output by the spatiotemporal data fusion module. This reflects the interference of land use change on watershed runoff generation and confluence conditions and water conservancy project scheduling on the process of regulating surface runoff on flood evolution. By synergistically simulating surface water, groundwater, and the impact of human activities, the realism, continuity, and prediction accuracy of distributed hydrological and hydrodynamic models in simulating watershed flood processes can be improved.

[0023] The coupling process simulation module is connected to the spatiotemporal data fusion module and receives a data cube as input. Please refer to the attached document. Figure 2 This module simulates flood evolution based on a distributed hydrological and hydrodynamic model. The model discretizes the watershed into several square grid cells, and classifies them into two types according to the underlying surface type: channel-type grid cells and slope-type grid cells. Both types of grid cells include soil moisture content, Manning coefficient of surface roughness, saturated hydraulic conductivity and slope parameters, while only the river type grid cell additionally includes the river cross-section width parameter. The model describes surface runoff and soil water movement processes by solving the Saint-Venant equations and the Richards equations in a coupled manner. It should be noted that the Saint-Venant equations govern the flow movement in the river channel and on the slope, while the Richards equations characterize soil moisture migration. To reflect the spatiotemporal coupling characteristics, a spatiotemporal convolution operator is embedded in the solution process of the Saint-Venant equations. This operator achieves its function through weighted integration in both time and space dimensions: in the time dimension, it incorporates all historical hydrological data from the initial moment to the current moment, quantifying the impact of the upstream water inflow confluence time delay on the current grid flow state; In the spatial dimension, it covers all computational grids across the entire watershed, assigning differentiated influence weights to inflow and precipitation at different locations through a spatial response function. The core of the operator is the spatiotemporal convolution kernel function, which is jointly constructed by the actual confluence time parameters of the watershed and the spatial response function, and can accurately describe the spatiotemporal superposition weights of upstream inflow and local precipitation; It should be clarified that this operator does not replace the physical solution process of the Saint-Venant equations, but rather improves the accuracy of the equations in describing the spatiotemporal coupling effect of water flow through weighted optimization, making the simulation results more consistent with the actual hydrological process in the watershed. By embedding a spatiotemporal convolution operator in the solution of the Saint-Venant equations, a combination of physical mechanisms and numerical optimization is achieved, which not only ensures the physical realism of the flood simulation, but also improves the accuracy of the spatiotemporal coupling characteristics.

[0024] Human activity impact factors are embedded in the model as dynamic correction terms, and a differentiated and timely parameter adjustment strategy is adopted: by periodically accessing land use change data and real-time accessing reservoir scheduling instructions, the underlying surface parameters such as Manning coefficient and permeability parameters of the corresponding grid and engineering control parameters and reservoir scheduling instructions are adjusted respectively, and the Manning coefficient and permeability parameters of the corresponding grid are dynamically adjusted. The disaster chain simulation module is connected to the coupled process simulation module and receives flood inundation results as input. Please refer to the appendix. Figure 4 This module first overlays disaster-bearing body distribution data through a geographic information system, including the spatial boundaries of residential areas, transportation networks, power facilities, and farmland. Then, it constructs a multi-hazard coupled probabilistic network model, with network nodes representing four types of hazards: floods, landslides, debris flows, and urban flooding, and edges representing the triggering relationships between hazards. For landslide disasters, the module calculates the slope stability index based on the Mohr-Coulomb criterion, and triggers a landslide node when the index is below 1.2; For debris flow disasters, a critical velocity model for sediment activation is used to determine when a debris flow node is activated when the flood velocity exceeds 1.5 meters per second and the slope is greater than 25 degrees. The model uses a directed graph structure to describe the disaster propagation path, and 1000 random samples are performed through Monte Carlo simulation to statistically analyze the trigger probability and impact range of each link. For example, in a typical mountainous watershed scenario, the module may calculate that the probability of a flood triggering a landslide is 0.32, and the conditional probability of the landslide further triggering a debris flow is 0.18, and finally output a disaster chain development sequence with spatiotemporal markers.

[0025] The dynamic decision support module connects to the disaster chain simulation module and generates decision products based on the simulation results. Please refer to the appendix. Figure 5 The risk assessment map generation process of this module includes three steps: First, the flood inundation map and the disaster-bearing body distribution map are spatially overlaid. Second, the vulnerability curve function is called according to the disaster-bearing body type. Finally, the comprehensive risk index is calculated by combining real-time population density data. Specifically, for residential buildings, a piecewise linear vulnerability curve is used to convert inundation depth into an economic loss rate. The loss rate is 5% when the inundation depth is below 0.5 meters, rises to 25% between 0.5 and 1 meter, and reaches 60% when the depth exceeds 1 meter. Regarding the risk of casualties, the logistic function is used to convert flow velocity and water depth into the probability of injury or death. The calculation principle is as follows: ; in: Indicates the risk of casualties; water depth With flow rate The data is expressed in meters and meters per second, respectively. The module updates the risk map every 10 minutes, dividing the watershed into 100-meter by 100-meter grids, with each grid labeled as high-risk, medium-risk, and low-risk. During the emergency response plan generation phase, the module uses a modified Dijkstra algorithm to calculate the optimal evacuation route based on real-time road network data and flood evolution prediction. The algorithm considers a road capacity decay model; when the flood depth of a road segment exceeds 0.3 meters, the capacity drops to zero. It also introduces node capacity constraints to prevent route congestion. The generated plan includes evacuation direction, alternative routes, and estimated arrival time, and is pushed to emergency command personnel in a visual format through the user interaction layer.

[0026] The model parameter adaptive calibration module is connected to the coupled process simulation module to achieve dynamic optimization of model parameters. Please refer to the appendix. Figure 6 This module employs an ensemble Kalman filter algorithm, assimilating real-time water level and flow observation data every 30 minutes. The algorithm maintains 50 model state sets, each containing a complete set of hydrological and hydrodynamic parameters. By calculating the deviation between observed and simulated values, the mean and covariance of the parameter distribution are adjusted inversely. For example, when the observed water level at a hydrological station remains 0.2 meters higher than the simulated value, the module automatically lowers the saturated hydraulic conductivity parameter of the upstream soil by 12% and increases the Manning coefficient by 8%. The calibrated parameters are immediately fed back to the coupled process simulation module, forming a closed-loop optimization.

[0027] The system achieves unified management of multi-source data through a data access layer, employing a spatiotemporal database to store historical and real-time data, supporting millisecond-level concurrent queries. The business logic layer encapsulates all computational models and uses a parallel computing framework to accelerate large-scale grid computing. The user interaction layer provides access to both web and mobile platforms, supporting dynamic rendering of risk maps, emergency response simulation, and historical case retrospective analysis. Data in JSON format is transmitted between layers via a RESTful application programming interface, ensuring system scalability and cross-platform compatibility.

[0028] Example 2

[0029] This embodiment focuses on the special implementation method of the system in watersheds with strong human activity interference. For areas with dense water conservancy projects, the coupled process simulation module adds a joint scheduling interface for reservoir groups to receive flood control scheduling instructions in real time as model boundary conditions; Specifically, the module incorporates a reservoir flood control operator model, which calculates the discharge flow based on the inflow process curve and the scheduling rule curve, and couples it to the main stream model through river connection conditions. When a pre-discharge command is received, the module lowers the reservoir water level 6 hours in advance and adjusts the initial conditions of the downstream river accordingly.

[0030] The disaster chain simulation module enhances the coupling analysis function of underground pipe networks for urban flooding scenarios. The module integrates municipal drainage network data to construct a one-dimensional and two-dimensional coupled hydrodynamic model. When the surface inundation depth exceeds the elevation of the storm drain grates, pipe network runoff calculations are initiated, simultaneously considering pump station start-up and shutdown strategies and gate control logic. A surface water accumulation time factor has been added to the flooding triggering conditions; when the water accumulation lasts for more than 2 hours and the depth is greater than 0.4 meters, it is automatically marked as a flooding disaster node.

[0031] The dynamic decision support module extends the multi-objective emergency resource scheduling model. In addition to evacuation routes, the module integrates spatial databases of emergency supplies depots, medical points, and shelters, establishing a spatiotemporal matching algorithm for resource demand and supply. By solving the vehicle routing problem with time windows, it simultaneously outputs relief supply delivery plans and personnel transfer plans. The model considers real-time road traffic conditions, automatically switching to the rural road network when major arterial roads are interrupted, and updates the population distribution heatmap in real time using mobile signaling data.

[0032] The adaptive calibration module for model parameters incorporates a dynamic learning mechanism for underlying surface parameters, tailored to the characteristics of urbanized watersheds. By comparing remotely sensed surface temperature with model simulations, the module automatically corrects parameters such as the proportion of impervious areas and green coverage. Simultaneously, a transfer learning framework is introduced, using calibrated parameters from adjacent watersheds as prior knowledge to accelerate the convergence process of new watershed models. The system performs rolling parameter optimization every 15 minutes to ensure simulation accuracy is maintained in rapidly urbanizing areas.

[0033] The user interaction layer features a dedicated console for emergency command scenarios, supporting multi-screen interaction and scheme comparison analysis. Commanders can adjust emergency resource deployment via drag-and-drop, and the system recalculates scheme effectiveness indicators in real time. It also integrates a voice command recognition module, enabling rapid retrieval of key information through natural language interaction in emergency situations. All operation logs and decision-making processes are recorded throughout for subsequent review and model optimization.

[0034] Example 3

[0035] This embodiment details the adaptive implementation plan of the system in high-altitude mountainous watersheds. For glacial meltwater-fed floods, the spatiotemporal data fusion module adds a snow and ice remote sensing data processing channel. The module integrates data from microwave radiometers and optical sensors, inverts snow water equivalent through brightness and temperature, and constructs a snow and ice ablation model by combining it with time-series snow cover data. During the data processing stage, a terrain correction algorithm is employed to eliminate the impact of mountain shadows on the accuracy of remote sensing inversion.

[0036] The coupled process simulation module expands the hydrological process components in cold regions, adding dynamic simulation capabilities for the permafrost impermeable layer. The model calculates changes in the active layer thickness based on temperature data and automatically adjusts infiltration parameters when the permafrost thawing depth exceeds a critical value. Simultaneously, a snow and ice meltwater runoff module is introduced, calculating glacier ablation based on the degree-day factor method and incorporating it as a source term into the water balance equation. The spatiotemporal convolution operator is optimized for the runoff characteristics of mountainous areas, adding slope factors and river channel curvature correction coefficients.

[0037] The disaster chain simulation module enhances the ability to analyze the chain reaction of glacial lake outbursts and avalanches. The module automatically identifies glacial lake boundaries using a digital elevation model and calculates dam stability based on lake water level monitoring data. When the dam's safety factor falls below 1.0, an outburst warning is triggered, and the peak flow propagation process is calculated using empirical formulas. The avalanche disaster node incorporates terrain curvature triggering conditions; when the slope is greater than 35 degrees and the curvature is negative, the avalanche risk probability is assessed.

[0038] The dynamic decision support module develops a dedicated emergency response plan library for high-altitude and cold-weather environments. Considering constraints such as high-altitude hypoxia and inconvenient transportation, the model is optimized to include helicopter rescue route planning functionality. By integrating digital elevation models and weather forecast data, it automatically avoids areas with severe convection and steep terrain along flight routes. The material distribution model considers the special resource needs such as thermal equipment and medical oxygen, establishing a categorized distribution system for emergency resources in cold regions.

[0039] The adaptive calibration module for model parameters develops a multi-source assimilation strategy for data-scarce regions. In addition to hydrological station data, the module simultaneously assimilates remote sensing soil moisture products and gravity satellite-monitored water storage change data. A multi-objective optimization function is constructed to balance the contribution weights of different types of observation data to parameter adjustment. In data-free areas, a parameter regionalization method constrained by physical laws is employed, transplanting known watershed parameters based on the principle of watershed similarity.

[0040] The system hardware architecture is reinforced for harsh environments, deploying edge computing nodes for rapid localized response. Data transmission reliability is ensured through dual links of 5G private network and satellite communication. A degradation processing mechanism is added to the business logic layer, automatically switching to locally cached data to continue running core algorithms when communication is interrupted, ensuring that basic system functions are unaffected under extreme conditions.

[0041] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics, characterized in that, include: The spatiotemporal data fusion module is used to receive and process multi-source heterogeneous data from remote sensing satellites, ground monitoring stations, meteorological radars and hydrological stations. It achieves seamless integration of multi-scale data by constructing a unified spatiotemporal reference framework, and uses an adaptive interpolation algorithm to repair missing data. It also removes abnormal observations based on spatiotemporal consistency verification criteria, thereby forming a watershed data cube with spatiotemporal continuity. The coupling process simulation module, connected to the spatiotemporal data fusion module, is used to drive a distributed hydrological and hydrodynamic model with a spatiotemporal coupling mechanism based on a data cube. This distributed hydrological and hydrodynamic model describes surface runoff and soil water movement by solving the Saint-Venant equations and the Richards equations in a coupled form. In the process of solving the equations, a spatiotemporal convolution operator is embedded to accurately enhance the characterization of the spatial heterogeneity of precipitation runoff and its temporal lag effect. At the same time, a two-way coupling method is used to realize the dynamic interaction simulation of surface water and groundwater. The disaster chain deduction module, connected to the coupled process simulation module, is used to perform spatiotemporal correlation analysis of secondary disasters based on the simulated flood inundation results. This module first identifies the spatial distribution of disaster-bearing bodies within the flood's impact range, then constructs a probability network model of flood disasters, and deduces the chain development process of secondary disasters by calculating disaster triggering thresholds and propagation paths. The dynamic decision support module, connected to the disaster chain simulation module, is used to generate risk assessment maps and emergency plans based on the simulation results. This module dynamically calculates the risk level of different time and space locations by overlaying the vulnerability curve of the disaster-bearing body and real-time flood elements, and outputs the optimal evacuation route and rescue dispatch plan based on the preset emergency resource distribution and traffic accessibility model.

2. The watershed flood disaster analysis system based on spatiotemporal coupling characteristics according to claim 1, characterized in that, The data processing procedure of the spatiotemporal data fusion module includes: First, spatiotemporal registration is performed on the multi-source data to uniformly transform data with different resolutions and projection coordinate systems to a standard spatiotemporal grid. Then, the Kriging interpolation algorithm was used to fill in the spatially missing data, and the time series analysis algorithm was used to repair the time series fragmented data. Finally, by calculating the data gradient and statistical distribution characteristics of adjacent spatiotemporal units, outliers that deviate from the normal range are identified and eliminated.

3. The watershed flood disaster analysis system based on spatiotemporal coupling characteristics according to claim 1, characterized in that, The distributed hydrological and hydrodynamic model of the coupling process simulation module uses a grid discretization method to divide the watershed into several computational units. Each unit contains soil moisture content, surface roughness, permeability coefficient and river cross-sectional morphology parameters. Distributed hydrodynamic models simulate surface runoff and soil water movement by solving the coupled form of the Saint-Venant equations and the Richards equations. Spatiotemporal convolution operators are embedded in the solution of the spatial and temporal terms of the equation system. The spatiotemporal weights of upstream water and local precipitation are quantified by the convolution kernel function to adjust the simulation accuracy of water flow propagation time delay and water volume superposition effect.

4. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics according to claim 3, characterized in that, Human activity impact factors are embedded in the model as dynamic parameter correction terms, and the hydrological parameters of the corresponding units are automatically adjusted based on real-time land use change data and reservoir scheduling instructions.

5. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics according to claim 1, characterized in that, The construction process of the multi-hazard coupled probabilistic network model of the disaster chain inference module includes: Using flood inundation depth and flow velocity as initial disaster-causing factors, the slope stability index and sediment initiation critical conditions are calculated based on a historical disaster case database and a mechanical mechanism model. When the index or condition exceeds the preset threshold, the corresponding secondary disaster node is triggered; The model establishes a directed graph network of disaster propagation by analyzing the spatial proximity and temporal sequence between disaster nodes, and uses Monte Carlo simulation to statistically analyze the trigger probability and impact range of each link.

6. The watershed flood disaster analysis system based on spatiotemporal coupling characteristics according to claim 1, characterized in that, The risk assessment map generation method of the dynamic decision support module includes: First, the flood inundation map and the disaster-bearing body distribution map are spatially overlaid; Then, based on the type of disaster-bearing body, the corresponding vulnerability curve function is called to convert the flood intensity index into the economic loss rate and the probability of casualties; Finally, by combining real-time population thermal data and building structure information, the comprehensive risk index of each grid unit is calculated, and a dynamic risk map is generated according to the preset level classification standard.

7. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics according to claim 1, characterized in that, The emergency response plan generation process of the dynamic decision support module also includes the optimization calculation of evacuation routes; Based on real-time road network capacity and flood evolution prediction results, the process uses an improved Dijkstra algorithm to search for the optimal path from the risk area to the safe area, while taking into account the capacity constraints of the path nodes and the spatiotemporal variations of the evacuation speed.

8. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics according to claim 1, characterized in that, It also includes a model parameter adaptive calibration module, which is connected to the coupled process simulation module and is used to dynamically adjust the model parameters based on real-time observation data. The calibration module uses an ensemble Kalman filter algorithm to update the model's state variables and parameter fields by assimilating the latest water level and flow rate observations.

9. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics according to claim 8, characterized in that, The model parameter adaptive calibration module performs a parameter assimilation and update process once per time period. It calculates the deviation between observed and simulated values ​​by maintaining several model state sets, and adjusts the mean and covariance of the parameter distribution in reverse.

10. A watershed flood disaster analysis system oriented towards spatiotemporal coupling characteristics according to claim 1, characterized in that, The system is implemented using a multi-layer software architecture, including a data access layer, a business logic layer, and a user interaction layer. The data access layer is responsible for the collection and storage management of multi-source data; The business logic layer encapsulates all computational models and algorithm cores; The user interaction layer provides a visual display and solution configuration interface; Data exchange and function calls are performed between different layers through standard application programming interfaces.

Images