A sea level rise influence risk assessment method based on multi-source data fusion, medium and system

Abstract

This invention provides a method, medium, and system for assessing the risks of sea-level rise based on multi-source data fusion, belonging to the field of marine disaster technology. The invention acquires basic feature data by constructing a multi-dimensional data acquisition network, performs nonlinear time series decomposition using multi-scale wavelet analysis, establishes a prediction model based on a deep learning neural network, identifies key vulnerability points using the minimum cut maximum flow algorithm, establishes a regional vulnerability scoring system, generates a comprehensive risk assessment matrix using Bayesian network ensemble analysis, calculates the variable submatrix to extract risk change trends, and finally generates an assessment report using a coastline adaptive gating model. This model integrates a multi-head temporal attention mechanism and bidirectional autoregressive coding, and dynamically adjusts parameters using a regional adaptive gating weight function, achieving accurate assessment and dynamic tracking of the risks of sea-level rise in complex geographical environments.

Description

Technical Field

[0001] This invention belongs to the field of marine disaster technology, and specifically relates to a method, medium, and system for risk assessment of the impact of sea level rise based on multi-source data fusion. Background Technology

[0002] Sea-level rise is a significant impact of global climate change, posing a major threat to the lives and economic development of coastal residents. Traditional sea-level rise risk assessment methods primarily rely on single data sources and linear models, such as sea-level change monitoring based on satellite altimeter data and static inundation simulation based on geographic information systems. These methods typically employ deterministic models, overlaying a pre-defined sea-level rise value onto a digital elevation model to directly obtain the inundation extent and depth distribution, or using simple statistical regression models to predict future sea-level change trends. However, traditional assessment methods have significant drawbacks: First, single data sources struggle to fully capture the complexity of sea-level changes, neglecting important influencing factors such as regional geological subsidence and climate cycles; second, linear prediction models cannot effectively identify critical points and nonlinear characteristics in the sea-level change process, leading to insufficient prediction accuracy; third, existing methods lack comprehensive consideration of topographic complexity and socio-economic factors, making it difficult to accurately locate key vulnerabilities and assess the dynamic evolution of risks. In complex geographical environments, the impact of sea-level rise exhibits strong nonlinear characteristics and regional differences. Traditional single data sources and linear models are insufficient to accurately assess the risk distribution and evolution trend in such complex systems, especially to effectively identify potential critical vulnerabilities and risk inflection points. This severely restricts the precise formulation and implementation of adaptive management strategies in coastal areas. Summary of the Invention

[0003] In view of this, the present invention provides a method, medium and system for assessing the risk of sea-level rise impact based on multi-source data fusion, which can solve the technical problem in the prior art that it is impossible to accurately assess the risk of sea-level rise impact in complex geographical environments.

[0004] The present invention is implemented as follows: The first aspect of the present invention provides a method for assessing the risk of sea-level rise based on multi-source data fusion, comprising: constructing a multi-dimensional data acquisition network to obtain basic characteristics of sea-level rise; using multi-scale wavelet analysis to perform nonlinear decomposition on sea-level change time series data, extracting different frequency components and identifying critical point signals; constructing a nonlinear prediction model for sea-level rise based on a deep learning neural network; constructing a regional inundation risk network model using the minimum cut maximum flow algorithm, calculating the minimum cut set to determine key vulnerable points and potential inundation area boundaries; establishing a regional vulnerability scoring system and calculating the vulnerability index; applying Bayesian network ensemble analysis to generate a comprehensive risk assessment matrix; calculating the variable component matrix of the comprehensive risk assessment matrix, extracting risk change rates and trend characteristics; and generating a comprehensive assessment report of regional sea-level rise risk using a coastline adaptive gating model.

[0005] The multidimensional data acquisition network refers to a multi-source heterogeneous data acquisition system composed of satellite remote sensing system, tide level monitoring system, GNSS surface deformation monitoring station and UAV aerial survey system, which is used to acquire information on sea level change and land response in all aspects.

[0006] The minimum cut maximum flow algorithm is a graph theory optimization problem-solving method. In the risk assessment of sea level rise, the terrain is regarded as a network graph, the elevation is regarded as the node capacity, and the water flow channel is regarded as the edge. By calculating the minimum cut set, the weak link most easily breached by the defense line is determined, and the key vulnerable points and the boundaries of potential flooding areas are accurately located.

[0007] The vulnerability index is a comprehensive score derived by weighting the distribution of terrain height, the condition of protective facilities, land use type, population density and economic value, reflecting the region's sensitivity and degree of impact in the face of sea-level rise.

[0008] The comprehensive risk assessment matrix is ​​a two-dimensional tabular data structure. The horizontal axis represents different sea-level rise scenarios, and the vertical axis represents different regional units. The matrix element values ​​are the comprehensive risk scores of the regional units under the corresponding scenarios.

[0009] The variable sub-matrix is ​​a matrix generated by calculating the differences in the element values ​​of the comprehensive risk assessment matrix between different time points. It contains information on risk change rate, change acceleration, and trend inflection point, and is used to quantify the dynamic evolution characteristics of regional risks.

[0010] The coastline adaptive gating model is a comprehensive analysis framework that integrates multilayer perceptron and temporal attention mechanism. Its specific structure is a neural network architecture consisting of an input layer, a multi-head temporal attention layer, a bidirectional autoregressive encoding layer, a dynamic gating integration layer, and an output layer.

[0011] The coastline adaptive gating model uses a regional adaptive gating weight function to dynamically adjust parameter weights according to risk changes, thereby achieving an accurate assessment of the risk of sea-level rise in complex geographical environments. The regional adaptive gating weight function calculates dynamic gating parameters based on the variable submatrix of the comprehensive risk assessment matrix, the regional topographic complexity index, and the distribution of socio-economic vulnerability. The regional adaptive gating weight function adopts a piecewise activation mechanism.

[0012] Among them, the multi-scale wavelet analysis is a mathematical tool for time series decomposition, which can decompose sea level change time series data into interannual variation, seasonal variation, long-term trend and abrupt change signals, effectively identifying nonlinear change patterns that traditional linear models cannot capture.

[0013] The critical point signal refers to a sudden acceleration or deceleration point that occurs during sea level changes, which usually indicates a change in certain key physical processes, such as important events like ice shelf collapse or changes in ocean currents.

[0014] The regional adaptive gating weight function first calculates a comprehensive balance value, which is a weighted combination of three factors: the risk change rate in the variable matrix, the regional terrain complexity index, and the socioeconomic vulnerability distribution. When the comprehensive balance value is in the low-risk range, a conservative weight adjustment function is used; when the comprehensive balance value is in the medium-risk range, a balanced weight adjustment function is used; and when the comprehensive balance value is in the high-risk range, an aggressive weight adjustment function is used.

[0015] The regional topographic complexity index is a comprehensive index generated by calculating the elevation change rate, slope distribution and topographic relief within the region, used to quantify the complexity of the topographic structure's response to sea-level rise.

[0016] The socioeconomic vulnerability distribution refers to the spatial distribution characteristics of socioeconomic factors such as population density, infrastructure value, economic activity intensity, and land use type within a region, which are used to assess the potential impact of sea-level rise on the socioeconomic system.

[0017] Specifically, the method for obtaining the variable component matrix involves establishing a set of time series sampling points, generating a corresponding comprehensive risk assessment matrix for each sampling time point, calculating the element difference to obtain the initial change matrix, further calculating the rate of change matrix, performing second-order difference on the rate of change matrix in the time dimension to obtain the acceleration matrix, and finally integrating them into a complete variable component matrix.

[0018] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, are used to perform the above-described method for assessing the risk of sea-level rise impacts based on multi-source data fusion.

[0019] A third aspect of the present invention provides a risk assessment system for the impact of sea level rise based on multi-source data fusion, comprising the aforementioned computer-readable storage medium. The system is any one of a computer, a server, or a microcontroller. The computer-readable storage medium is disposed within the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

[0020] This invention achieves comprehensive risk assessment through techniques such as constructing a multi-dimensional data acquisition network, multi-scale wavelet analysis, deep learning neural network prediction models, minimum cut maximum flow algorithms, and coastline adaptive gating models. This method effectively addresses the shortcomings of traditional techniques: the multi-dimensional data acquisition network integrates remote sensing imagery, water level monitoring, and surface deformation data to comprehensively capture the multi-dimensional characteristics of sea-level changes; multi-scale wavelet analysis and deep learning neural network models can accurately identify the nonlinear characteristics and critical points of sea-level changes; the minimum cut maximum flow algorithm, combined with a regional vulnerability scoring system, precisely locates key vulnerabilities and quantifies the degree of risk; and the comprehensive risk assessment matrix and its variable submatrix can dynamically track the risk evolution trend and inflection point characteristics. Through the organic combination of these techniques, this invention can accurately assess the impact risk of sea-level rise in complex geographical environments, accurately identify key vulnerabilities and dynamic risk change trends, and provide scientific support for coastal areas to formulate differentiated adaptive management strategies, thus effectively solving the technical problem that existing technologies cannot accurately assess the impact risk of sea-level rise in complex geographical environments. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention.

[0022] Figure 2 This is a schematic diagram of the structure of a nonlinear prediction model for sea level rise.

[0023] Figure 3 This is a schematic diagram of the structure of the coastline adaptive gating model.

[0024] Figure 4 This is a schematic diagram of a multidimensional data acquisition network.

[0025] Figure 5 This is a schematic diagram of a flood risk network model for the minimum cut maximum flow algorithm.

[0026] Figure 6 This is a schematic diagram showing the time series of sea level changes and the results of wavelet analysis.

[0027] Figure 8 Maps showing sea level rise predictions under different scenarios.

[0028] Figure 9 This is a diagram showing the results of sea-level rise risk assessments under different scenarios. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0030] like Figure 1 The diagram shown is a flowchart of a method for assessing the risk of sea-level rise impacts based on multi-source data fusion, provided by the first aspect of this invention. This method includes the following steps:

[0031] S01. Construct a multi-dimensional data acquisition network to obtain basic characteristics of sea level rise by combining remote sensing image data, water level monitoring station data, and surface deformation data with regional digital elevation models.

[0032] S02. Multi-scale wavelet analysis was used to perform nonlinear decomposition on sea level change time series data, extract different frequency components and identify critical point signals;

[0033] S03. Construct a nonlinear prediction model for sea level rise based on deep learning neural networks, integrating climate simulation data, historical observation data, and regional geological subsidence data as input parameters;

[0034] S04. Construct a regional inundation risk network model using the minimum cut maximum flow algorithm, combining the terrain height nodes with the edge weights of the water flow channels, and calculate the minimum cut set to determine key vulnerable points and the boundaries of potential inundation areas.

[0035] S05. Establish a regional vulnerability scoring system and calculate the vulnerability index based on topographical height distribution, the status of protective facilities, land use type, population density, and economic value.

[0036] S06. Apply Bayesian network ensemble analysis to integrate the prediction results of the nonlinear prediction model for sea level rise with the vulnerability index of the regional vulnerability scoring system to generate a comprehensive risk assessment matrix.

[0037] S07. Calculate the variation submatrix of the comprehensive risk assessment matrix, and extract the risk change rate and trend characteristics by comparing the comprehensive risk assessment matrices at different time points.

[0038] S08. Generate a comprehensive assessment report on regional sea-level rise risk using a coastline adaptive gating model, including a risk level distribution map, a list of key vulnerabilities, time-series risk evolution trends, and risk rating classification results.

[0039] Among them, the multidimensional data acquisition network refers to a multi-source heterogeneous data acquisition system composed of satellite remote sensing system, tide level monitoring system, GNSS surface deformation monitoring station and UAV aerial survey system, which is used to acquire information on sea level change and land response in all aspects.

[0040] Among them, multi-scale wavelet analysis is a mathematical tool for time series decomposition, which can decompose sea level change time series data into interannual variation, seasonal variation, long-term trend and abrupt change signals, effectively identifying nonlinear change patterns that traditional linear models cannot capture.

[0041] Among them, the critical point signal refers to the sudden acceleration or deceleration point that occurs during the sea level change process, which usually indicates a change in certain key physical processes, such as important events such as ice shelf collapse or changes in ocean currents.

[0042] Among them, the minimum cut maximum flow algorithm is a method for solving optimization problems in graph theory. In the risk assessment of sea level rise, the terrain is regarded as a network graph, the elevation is used as the node capacity, and the water flow channel is used as the edge. By calculating the minimum cut set, the weak link that is most easily breached is determined, and the key vulnerable points and the boundaries of potential flooding areas are accurately located.

[0043] The vulnerability index is a comprehensive score derived by weighting factors such as topographic height distribution, protective facility status, land use type, population density, and economic value. It reflects the region's sensitivity to and degree of impact from sea-level rise.

[0044] The comprehensive risk assessment matrix is ​​a two-dimensional tabular data structure. The horizontal axis represents different sea-level rise scenarios, and the vertical axis represents different regional units. The matrix element values ​​are the comprehensive risk scores of the regional units under the corresponding scenarios.

[0045] The variable risk matrix is ​​a matrix generated by calculating the differences in the element values ​​of the comprehensive risk assessment matrix at different time points. It contains information on risk change rate, change acceleration, and trend inflection point, and is used to quantify the dynamic evolution characteristics of regional risks.

[0046] Among them, the coastline adaptive gating model is a comprehensive analysis framework that integrates multilayer perceptron and temporal attention mechanism to provide intelligent recommendations for resource allocation and protection decisions under the scenario of sea level rise.

[0047] The specific structure of the coastline adaptive gating model is a neural network architecture consisting of an input layer, a multi-head temporal attention layer, a bidirectional autoregressive encoding layer, a dynamic gating integration layer, and an output layer. The input layer receives climate scenario data, topographic feature vectors, and socioeconomic indicators. The multi-head temporal attention layer extracts features based on the changing trends at different time scales. The bidirectional autoregressive encoding layer captures the causal relationship between sea level change and regional response. The dynamic gating integration layer dynamically adjusts the weights of each feature channel according to the change sub-matrix in the comprehensive risk assessment matrix. The output layer generates regional protection strategy recommendations and resource optimization allocation schemes. The dynamic gating integration layer uses a regional adaptive gating weight function for parameter adjustment.

[0048] The steps for establishing the training dataset for the adaptive gating model of the coastline include collecting historical sea-level change data and corresponding protective measures implementation records for typical coastal areas around the world, integrating inundation range maps and actual impact assessment reports for each region under different sea-level rise scenarios, extracting topographic change characteristics and socio-economic response indicators before and after the implementation of regional protective projects, constructing a parallel case library containing multiple regions and time points, normalizing and aligning all collected data in time and space, marking the protective measures effectiveness score and cost-benefit ratio for each case, and finally dividing the data into training set, validation set and test set according to time order.

[0049] The specific steps for training the coastline adaptive gating model include: initializing and pre-training the model's parameters using the batch stochastic gradient descent algorithm; introducing a self-supervised learning mechanism for time series prediction to enhance the model's ability to capture long-term trends; calculating the protection decision loss function using data on the effectiveness of protection measures in historical cases to guide model optimization; improving the model's ability to identify commonalities and differences between different regions through comparative learning; designing a hierarchical training strategy to optimize the parameters of different components in stages; introducing a regional feature adaptive mechanism to enable the model to dynamically adjust parameter weights according to the characteristics of different regions; and finally, fine-tuning and validating the trained model using the variable submatrix in the comprehensive risk assessment matrix.

[0050] The method for obtaining the variable component matrix is ​​as follows: First, a set of time series sampling points is established. A corresponding comprehensive risk assessment matrix is ​​generated for each sampling time point. The element difference value is calculated for the comprehensive risk assessment matrix of adjacent time points to obtain the initial variable matrix. Then, the rate of change of each element, that is, the ratio of the change amount to the time interval, is calculated to form the rate of change matrix. The rate of change matrix is ​​subjected to second-order difference in the time dimension to obtain the acceleration matrix. Significant change points in the acceleration matrix are identified and marked as trend inflection points. Finally, the rate of change matrix, the acceleration matrix, and the trend inflection point markings are integrated into a complete variable component matrix.

[0051] The regional adaptive gating weight function calculates dynamic gating parameters based on the variation submatrix of the comprehensive risk assessment matrix, the regional terrain complexity index, and the socio-economic vulnerability distribution. The regional adaptive gating weight function employs a piecewise activation mechanism. First, it calculates a comprehensive equilibrium value, which is a weighted combination of the risk change rate in the variation submatrix, the regional terrain complexity index, and the socio-economic vulnerability distribution. When the comprehensive equilibrium value is in the low-risk range, a conservative weight adjustment function is used to maintain high stability and gradual adjustment. When the comprehensive equilibrium value is in the medium-risk range, a balanced weight adjustment function is used to achieve a balance between stability and response speed. When the comprehensive equilibrium value is in the high-risk range, an aggressive weight adjustment function is used to rapidly increase the resource allocation weight of key protective measures. The boundary thresholds of each range are dynamically adjusted according to the historical response characteristics of the region, thereby achieving adaptive optimization of the gating parameters and improving the model's response speed and accuracy to sudden risk events.

[0052] The conservative weight adjustment function refers to a weight adjustment function that adopts a linear growth model, ensuring smooth and gradual weight changes and avoiding system fluctuations. The balanced weight adjustment function refers to a weight adjustment function that adopts an S-curve model, using smaller adjustment amplitudes in low-risk and high-risk ranges and larger adjustment amplitudes in medium-risk ranges, achieving a balance between smooth transition and timely response. The aggressive weight adjustment function refers to a weight adjustment function that adopts an exponential growth model, rapidly increasing weights when the risk exceeds a threshold, ensuring a rapid system response to high-risk situations. The regional topographic complexity index is a comprehensive index generated by calculating the rate of change of elevation, slope distribution, and topographic relief within a region, used to quantify the complexity of the topographic structure's response to sea-level rise. The socioeconomic vulnerability distribution refers to the spatial distribution characteristics of socioeconomic factors such as population density, infrastructure value, economic activity intensity, and land use type within a region, used to assess the potential impact of sea-level rise on the socioeconomic system.

[0053] The specific implementation methods of the above steps are described in detail below.

[0054] The specific implementation of step S01 involves constructing a multi-dimensional data acquisition network, which consists of a satellite remote sensing system, a tide gauge monitoring system, a GNSS surface deformation monitoring station, and an UAV aerial survey system. First, a fixed tide gauge network is deployed, spaced no more than 50 kilometers apart, to ensure continuous water level data collection along the coastline. Then, GNSS surface deformation monitoring stations are deployed, with a monitoring point density of at least one per 100 square kilometers, capturing millimeter-level vertical surface deformation. Next, regional digital elevation model data with a resolution of at least 0.5 meters is acquired as the basis for terrain analysis. Finally, multi-temporal remote sensing imagery, including optical imagery, SAR (Synthetic Aperture Radar) imagery, and UAV aerial imagery, is integrated, with a temporal resolution of at least once per quarter. A data fusion algorithm is used to unify data from different sources into the WGS84 coordinate system, forming a spatiotemporally consistent sea-level change monitoring dataset. The purpose of this step is to obtain the basic characteristics of sea-level rise through multi-source heterogeneous data acquisition, providing comprehensive and accurate data support for subsequent analysis.

[0055] Step S02 involves using multi-scale wavelet analysis to perform nonlinear decomposition on sea-level change time-series data. First, a suitable wavelet basis function, such as the Daubechies wavelet or Morlet wavelet, is selected to perform a wavelet transform on the original time series. Then, by adjusting the scale parameters, the time series is decomposed into different frequency components, including interannual variation (2–7 year scale), seasonal variation (0.25–1 year scale), long-term trend (>10 year scale), and abrupt change signals. Next, the wavelet energy spectrum at each scale is calculated, and areas of concentrated energy anomalies are identified as potential critical points. Finally, wavelet phase analysis is used to determine the precise time location of the critical points, with the threshold set at a local maximum value exceeding twice the standard deviation of the mean. The criterion for determining the critical point signal is that the sea-level change rate within a short period (no more than 3 months) exceeds three times the average change rate of the previous 5 years. This step utilizes the multi-resolution characteristics of wavelet analysis to effectively separate the multi-scale components of sea-level change, particularly capturing nonlinear changes and abrupt change signals that are difficult to identify using traditional linear models, providing key feature information for sea-level rise prediction.

[0056] The specific implementation of step S03 is to construct a nonlinear prediction model for sea level rise based on deep learning neural networks. First, a hybrid architecture of Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) is constructed. The number of LSTM units is set to 64–128, and the CNN adopts a 3–5 layer structure with a 3×3 kernel size. Then, historical sea level observation data (at least 30 years of continuous records), satellite altimeter data, temperature and precipitation predictions output from the Global Climate Model (GCM), and regional geological subsidence rates are input as multi-source parameters. Next, an attention mechanism is used to enhance the model's ability to capture long-term trends and abrupt events, with the attention layer depth set to 2–3 layers. Finally, multiple prediction time scales are set, including short-term (1–5 years), medium-term (5–20 years), and long-term (20–100 years) prediction outputs. Model training uses a batch size of 64, an initial learning rate of 0.001, and cosine annealing scheduling, with 200–500 training epochs. The purpose of this step is to build a predictive model that can effectively capture the nonlinear characteristics of sea level changes, achieve scientific prediction of future sea level rise trends, and provide basic data support for risk assessment.

[0057] The specific implementation of step S04 involves constructing a regional inundation risk network model using the minimum cut maximum flow algorithm. First, the study area is discretized into a grid structure, with the grid size determined based on terrain complexity, typically 10–50 meters. Then, a directed network graph is constructed, where nodes represent terrain elevation points, and the capacity is set to the difference between the elevation value of that point and the predicted sea level; negative values ​​indicate already inundated areas. Next, edge weights are defined as the unobstructedness of the water flow channels between adjacent nodes, calculated as the ratio of the elevation difference to the distance between adjacent nodes multiplied by the surface roughness coefficient. The surface roughness coefficient is determined based on land use type, such as 0.01–0.03 for urban construction areas, 0.03–0.05 for farmland, and 0.05–0.10 for forests. Finally, the Ford-Fulkerson algorithm or the Dinic algorithm is applied to solve the maximum flow problem, identifying the minimum cut sets in the network. The nodes connected by edges in the minimum cut sets constitute the boundaries of potential inundation areas. The purpose of this step is to use the minimum cut maximum flow algorithm in graph theory to accurately simulate the seawater intrusion path, locate the weak links and key vulnerabilities of the protection system, and provide a spatial analysis basis for risk assessment.

[0058] The specific implementation of step S05 is to establish a regional vulnerability scoring system. First, a scoring index system is determined, including five dimensions: topographic elevation distribution (weight 0.25), protective facility status (weight 0.20), land use type (weight 0.15), population density (weight 0.20), and economic value (weight 0.20). Then, each index is standardized, converting the original values ​​into a score range of 0-10. The formula for calculating the standard score for topographic elevation is 10 × (current elevation - predicted sea level) / (highest point in the region - predicted sea level), with a value of 0 for values ​​less than 0. Next, basic scores are allocated according to land use type, such as 8-10 points for urban residential areas, 7-9 points for industrial areas, 5-7 points for farmland, 3-5 points for forest land, 2-4 points for grassland, and 0-2 points for unused land. Finally, the vulnerability index is calculated by weighting the values ​​of all indicators, and the vulnerability index is divided into five levels: extremely low (0-2), low (2-4), moderate (4-6), high (6-8), and extremely high (8-10). The purpose of this step is to establish a quantitative standard for regional vulnerability through a comprehensive multi-indicator scoring system, identify sensitive areas susceptible to sea-level rise, and provide an important basis for comprehensive risk assessment.

[0059] The specific implementation of step S06 is to apply Bayesian network ensemble analysis to integrate the model prediction results with the vulnerability index. First, a Bayesian network structure is constructed, with nodes including key factors such as predicted sea level height, inundation area percentage, protective facility status, population impact ratio, and estimated economic losses. Then, based on historical data and expert experience, a conditional probability table (CPT) is set between nodes. For example, under a sea level rise greater than 50 cm, the probability thresholds for severe inundation in areas with different vulnerability levels are: 0.05 for extremely low vulnerability areas, 0.15 for low vulnerability areas, 0.35 for moderately vulnerable areas, 0.65 for high vulnerability areas, and 0.85 for extremely high vulnerability areas. Next, the sea level rise prediction results from step S03 and the vulnerability index from step S05 are input as observational evidence, and the posterior probability distribution is calculated using the belief propagation algorithm. Finally, a comprehensive risk assessment matrix is ​​generated, with the horizontal axis representing different sea level rise scenarios (e.g., RCP2.6, RCP4.5, RCP8.5, etc.) and the vertical axis representing the assessment unit (grid or administrative division). The matrix element values ​​are the comprehensive risk scores (0-100 points) of the regional unit under the corresponding scenario. The purpose of this step is to leverage the uncertainty reasoning capabilities of Bayesian networks to scientifically integrate sea-level rise predictions with regional vulnerability assessments, thereby forming a comprehensive risk quantification index.

[0060] The specific implementation of step S07 involves calculating the variation submatrix of the comprehensive risk assessment matrix. First, a set of time series sampling points is determined, typically selecting key time nodes such as 2030, 2050, 2070, and 2100. Then, a corresponding comprehensive risk assessment matrix M is generated for each sampling time point. tNext, calculate the matrix element difference ΔM = M between adjacent time points t1 and t2. t2 -M t1 First, obtain the initial change matrix; then calculate the rate of change matrix R = ΔM / (t2-t1), reflecting the rate of risk change per unit time; subsequently, perform second-order difference in the time dimension to obtain the acceleration matrix A = (R... t+1 -R t Finally, significant points of change in the acceleration matrix are identified, with a threshold set at 2.5 times the standard deviation of the average acceleration. These points are marked as trend inflection points. The purpose of this step is to capture the characteristics of risk evolution over time through matrix operations, particularly to identify accelerating or decelerating trends in risk change, providing a time-series dynamic analysis basis for risk management decisions.

[0061] The specific implementation of step S08 involves generating a comprehensive assessment report using a coastline adaptive gating model. First, the comprehensive risk assessment matrix generated in step S06 and the variable score matrix generated in step S07 are input into the coastline adaptive gating model. Then, a risk level distribution map is generated based on the model's calculation results, using a four-color system (red, orange, yellow, and green) to represent extremely high, high, medium, and low risk areas. Next, the top 10% of areas with the highest comprehensive risk scores are extracted as a list of critical vulnerabilities, including spatial coordinates, risk scores, major risk-causing factors, and protection recommendations. Subsequently, a time-series risk evolution trend chart is generated, with the horizontal axis representing time and the vertical axis representing risk scores, showing the risk change trajectory of different areas. Finally, a risk rating classification result table is formed, classifying areas according to risk urgency and formulating differentiated protection strategy recommendations. The rating classification uses a four-level standard: Level I (extremely urgent, action required within 5 years), Level II (urgent, action required within 5-15 years), Level III (medium, action required within 15-30 years), and Level IV (low urgency, to be considered after 30 years). The purpose of this step is to integrate the results of previous analyses and generate an intuitive and clear risk assessment report, providing a scientific basis and action guide for coastal zone management and protection decisions.

[0062] The detailed structure of the coastline adaptive gating model is a neural network architecture consisting of an input layer, a multi-head temporal attention layer, a bidirectional autoregressive encoding layer, a dynamic gating integration layer, and an output layer. The input layer receives three types of data: climate scenario data (including temperature changes, precipitation patterns, and sea-level rise predictions under different RCP scenarios), topographic feature vectors (including topographic parameters such as elevation distribution, slope, and orientation), and socioeconomic indicators (including population density, GDP, and infrastructure value). The multi-head temporal attention layer employs eight attention heads, each focusing on changes at different time scales, covering trends from short-term (interannual) to long-term (century). The bidirectional autoregressive encoding layer consists of 128 LSTM units, processing time-series data both forward and backward to capture the causal relationship between sea-level change and regional response. The dynamic gating integration layer dynamically adjusts the weights of each feature channel based on a regional adaptive gating weight function, calculates gating parameters according to the variation submatrix in the comprehensive risk assessment matrix, and adjusts the priority of different protection strategies. The output layer generates regional protection strategy recommendations and resource optimization allocation schemes, including the selection of engineering protection measure types, implementation timing arrangements, and investment priority ranking.

[0063] The detailed steps for establishing the training dataset for the adaptive gating model of coastlines are as follows: First, historical sea-level change data for 25 typical coastal areas worldwide are collected, spanning no less than 50 years, including monthly average sea-level height, frequency and intensity of extreme events; then, inundation extent maps and actual impact assessment reports for each region under different sea-level rise scenarios are integrated, including more than 100 case studies; next, topographic change characteristics and socio-economic response indicators before and after the implementation of regional protection projects are extracted, such as changes in the frequency of flooding events and the reduction in economic losses before and after the construction of protective dikes; subsequently, a dataset containing multiple... A parallel case library with multiple time points in the region, containing no fewer than 500 typical cases, is established. Data normalization is then performed, unifying indicators of different dimensions to the 0-1 range and aligning them spatiotemporally to ensure comparability between different cases. Next, the effectiveness score (0-10 points) and cost-benefit ratio of each case are labeled. The effectiveness score is determined based on the actual effectiveness of disaster prevention and mitigation, and the cost-benefit ratio is calculated as the ratio of input costs to avoided losses. Finally, the data is divided chronologically into a training set (70%), a validation set (15%), and a test set (15%) to ensure the scientific rigor of model training and evaluation.

[0064] The adaptive gating model for coastlines is trained using a general model training method, with the following detailed steps: First, the model is initialized and pre-trained using the batch stochastic gradient descent algorithm, with a batch size of 32 and an initial learning rate of 0.0005. Then, a self-supervised learning mechanism for time series prediction is introduced to enhance the model's ability to capture long-term trends by predicting future sea level heights and their impact range. Next, the protective decision loss function is calculated using data on the effectiveness of protective measures from historical cases, and a weighted cross-entropy loss function is used to guide model optimization, with weights set according to the cost of decision errors. Finally, the model is improved through comparative learning. To enhance the ability to identify commonalities and differences between different regions, positive sample pairs (similar regions) and negative sample pairs (different regions) are constructed for comparative training. A hierarchical training strategy is then designed: in the first stage, high-level parameters are frozen, and only low-level feature extraction parameters are trained; in the second stage, all parameters are unfrozen for end-to-end fine-tuning. Next, a regional feature adaptive mechanism is introduced, embedding regional feature vectors into the model parameter space through a regional encoder, enabling the model to dynamically adjust parameter weights based on different regional characteristics. Finally, the variable submatrix in the comprehensive risk assessment matrix is ​​used to fine-tune and validate the trained model, ensuring its sensitivity and accuracy in responding to risk change trends.

[0065] The key technical concepts of this invention are described below.

[0066] First, multi-source data fusion technology is one of the key innovations of this invention. Traditional sea-level rise risk assessment methods often rely on a single data source, failing to fully capture the complexity of sea-level changes. The multi-dimensional data acquisition network constructed in this invention integrates data from satellite remote sensing systems, tide gauge monitoring systems, GNSS surface deformation monitoring stations, and UAV aerial survey systems, achieving comprehensive acquisition of information on sea-level changes and land response. This multi-source data fusion method fundamentally overcomes the limitations of a single data source, simultaneously considering multiple influencing factors such as regional geological subsidence, climate cycles, and tidal changes, providing more comprehensive information on sea-level change characteristics, thus laying a more reliable data foundation for subsequent risk assessment.

[0067] Secondly, nonlinear time series analysis is another core technical approach of this invention. Traditional sea-level change analysis often employs simple statistical methods such as linear regression, which struggle to identify complex time series characteristics. This invention utilizes multi-scale wavelet analysis to nonlinearly decompose sea-level change time series data, breaking down complex time series into different frequency components and effectively identifying nonlinear change patterns such as interannual variations, seasonal variations, long-term trends, and abrupt changes. This method can accurately capture critical point signals in sea-level change, which typically indicate shifts in important physical processes and are crucial for risk assessment. In principle, this technology overcomes the limitations of traditional linear analysis methods, enhancing the ability to identify complex sea-level change patterns.

[0068] Third, the graph theory-based risk network model is a key technological innovation of this invention. Traditional methods often use simple overlay analysis to determine the inundation range, neglecting the complexity of terrain structure. This invention utilizes the minimum cut maximum flow algorithm to construct a regional inundation risk network model, treating terrain as a network graph, elevation as node capacity, and water flow channels as edges. By calculating minimum cut sets, the weakest points most susceptible to breaching the defense line are identified. This method mathematically optimizes the inundation risk assessment process, accurately locating key vulnerable points and potential inundation area boundaries, avoiding assessment biases caused by neglecting terrain complexity in traditional methods.

[0069] Finally, the coastline adaptive gating model integrates the above-mentioned technologies, forming the most distinctive synergistic innovation of this invention. This model employs a multi-head temporal attention mechanism and a bidirectional autoregressive coding layer to capture the complex causal relationship between sea-level change and regional response, and dynamically adjusts parameter weights according to risk changes through a regional adaptive gating weight function. This design fundamentally solves the problem that traditional static models cannot cope with dynamic risk changes, enabling real-time adjustment and optimization of risk assessment. The synergistic effect of the various technical approaches enables this invention to comprehensively, accurately, and dynamically assess the risks of sea-level rise in complex geographical environments, providing a scientific basis for developing differentiated adaptive management strategies for coastal areas, thereby significantly improving the accuracy and effectiveness of sea-level rise risk management.

[0070] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, are used to perform the above-described method for assessing the risk of sea-level rise impacts based on multi-source data fusion.

[0071] A third aspect of the present invention provides a risk assessment system for the impact of sea level rise based on multi-source data fusion, comprising the aforementioned computer-readable storage medium. The system is any one of a computer, a server, or a microcontroller. The computer-readable storage medium is disposed within the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

[0072] Specifically, the principle of this invention is as follows: The technical principle of this invention to solve the risk assessment problem of sea level rise impact in complex geographical environments is mainly based on advanced calculation methods such as multi-source heterogeneous data fusion, nonlinear time series analysis, graph network optimization, and adaptive gated neural networks.

[0073] First, the multi-dimensional data acquisition network integrates data from satellite remote sensing systems, tide gauge monitoring systems, GNSS surface deformation monitoring stations, and UAV aerial survey systems to achieve comprehensive acquisition of sea-level change and land response information. This multi-source data fusion method overcomes the limitations of single data sources and provides more comprehensive information on sea-level change characteristics. Second, multi-scale wavelet analysis technology can decompose complex sea-level change time series into different frequency components, effectively identifying nonlinear change patterns such as interannual variations, seasonal variations, long-term trends, and abrupt change signals. In particular, its ability to identify critical point signals is significantly better than traditional linear analysis methods.

[0074] In terms of risk assessment, this invention employs the minimum cut maximum flow algorithm to construct a regional inundation risk network model. This method treats terrain as a network graph, elevation as node capacity, and water flow channels as edges. By calculating minimum cut sets, it identifies the weakest points most vulnerable to breaching the defense line, achieving precise location of critical vulnerabilities. The regional vulnerability scoring system then uses weighted calculations of factors such as terrain elevation distribution, protective facility status, land use type, population density, and economic value to form a comprehensive vulnerability index, reflecting the region's sensitivity to sea-level rise.

[0075] The core innovation of this invention lies in the coastline adaptive gating model. This model employs a multi-head temporal attention mechanism and a bidirectional autoregressive coding layer to capture the complex causal relationship between sea-level change and regional response. Simultaneously, it dynamically adjusts parameter weights based on risk changes through a regional adaptive gating weight function. The model uses a segmented activation mechanism for different risk levels: a conservative weight adjustment function to maintain stability in the low-risk range; a balanced weight adjustment function to balance stability and timely response in the medium-risk range; and an aggressive weight adjustment function to ensure rapid response in the high-risk range. This enables accurate assessment and dynamic tracking of the risks posed by sea-level rise in complex geographical environments.

[0076] Furthermore, the comprehensive risk assessment matrix and its variable submatrix, as an innovative data structure, can systematically express the risk distribution and temporal evolution characteristics of various regional units under different sea-level rise scenarios, providing decision-makers with intuitive dynamic risk information and supporting scientific decision-making.

[0077] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0078] The specific implementation of step S01 involves constructing a multi-dimensional data acquisition network, which consists of a satellite remote sensing system, a tide gauge monitoring system, a GNSS surface deformation monitoring station, and an UAV aerial survey system. First, a fixed tide gauge network is deployed, spaced no more than 50 kilometers apart, to ensure continuous water level data collection along the coastline. Then, GNSS surface deformation monitoring stations are deployed, with a monitoring point density of at least one per 100 square kilometers, capturing millimeter-level vertical surface deformation. Next, regional digital elevation model data with a resolution of at least 0.5 meters is acquired as the basis for terrain analysis. Finally, multi-temporal remote sensing images, including optical images, SAR (Synthetic Aperture Radar) images, and UAV aerial images, are integrated, with a temporal resolution of at least once per quarter. A data fusion algorithm is used to unify data from different sources into the WGS84 coordinate system, forming a spatiotemporally consistent sea-level change monitoring dataset. The purpose of this step is to obtain the basic characteristics of sea-level rise through multi-source heterogeneous data acquisition, providing comprehensive and accurate data support for subsequent analysis. The data fusion process uses a weighted average method, with the specific formula as follows: In the formula, D fusion The merged data value; D i w represents the observation from the i-th data source. i Let be the weight coefficient of the i-th data source, and satisfy . n represents the total number of data sources. Weighting coefficient w i Determined through historical accuracy assessments of various data sources, the calculation formula is as follows: In the formula, σ i Let be the historical observation standard deviation of the i-th data source. The smaller the standard deviation, the greater the weight.

[0079] The specific implementation of step S02 involves using multi-scale wavelet analysis to perform nonlinear decomposition on sea-level change time series data. First, a suitable wavelet basis function, such as the Daubechies wavelet or Morlet wavelet, is selected to perform a wavelet transform on the original time series. Then, by adjusting the scale parameters, the time series is decomposed into different frequency components, including interannual variation (2-7 year scale), seasonal variation (0.25-1 year scale), long-term trend (>10 year scale), and abrupt change signals. Next, the wavelet energy spectrum at each scale is calculated, and areas of concentrated energy anomalies are identified as potential critical points. Finally, wavelet phase analysis is used to determine the precise time location of the critical points, with the threshold set at a local maximum value of the wavelet coefficients exceeding twice the standard deviation of the mean. The criterion for determining the critical point signal is that the sea-level change rate within a short period (no more than 3 months) exceeds three times the average change rate of the previous 5 years. This step utilizes the multi-resolution characteristics of wavelet analysis to effectively separate the multi-scale components in sea-level change, especially capturing nonlinear changes and abrupt change signals that are difficult to identify by traditional linear models, providing key feature information for sea-level rise prediction. The mathematical expression for continuous wavelet transform is: In the formula, W f (a, b) are wavelet coefficients; f(t) is the sea level height time series; ψ * is the conjugate complex number of the wavelet basis functions; a is the scaling parameter, controlling the scaling of the wavelet; b is the translation parameter, controlling the position of the wavelet. The formula for calculating the wavelet energy spectrum is: In the formula, E(a) is the wavelet energy spectrum at scale a. The critical point determination formula is: C p ={b||W f (a, b)|>μ a +2σ a}; where C p For the set of critical points; μ a σ represents the mean of the wavelet coefficients at scale a; a Let be the standard deviation of the wavelet coefficients at scale a.

[0080] The specific implementation of step S03 is to construct a nonlinear prediction model for sea level rise based on deep learning neural networks. First, a hybrid architecture of Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) is constructed. The number of LSTM units is set to 64–128, and the CNN adopts a 3–5 layer structure with a 3×3 kernel size. Then, historical sea level observation data (at least 30 years of continuous records), satellite altimeter data, temperature and precipitation predictions output from the Global Climate Model (GCM), and regional geological subsidence rates are input as multi-source parameters. Next, an attention mechanism is used to enhance the model's ability to capture long-term trends and abrupt events, with the attention layer depth set to 2–3 layers. Finally, multiple prediction time scales are set, including short-term (1–5 years), medium-term (5–20 years), and long-term (20–100 years) prediction outputs. Model training uses a batch size of 64, an initial learning rate of 0.001, and cosine annealing scheduling, with 200–500 training epochs. The purpose of this step is to construct a predictive model that can effectively capture the nonlinear characteristics of sea-level changes, enabling scientific predictions of future sea-level rise trends and providing fundamental data support for risk assessment. The input feature vector of the predictive model is represented as: X t =[H t-k：t T t：t+m P t：t+m S t W t-j：t In the formula, X t H is the input feature vector at time point t; t-k：t T represents the historical sea level height sequence over the past k time steps; t：t+m A temperature prediction sequence for the next m time steps; P t：t+m S is a precipitation forecast sequence for the next m time steps; t For regional geological subsidence rate; W t-j：t The wavelet decomposition features are defined for the past j time steps. The model prediction output is expressed as: In the formula, F represents the predicted sea level height over the next n time steps. θ Let be the neural network model function with parameter θ. The mathematical expression of the attention mechanism is: e i =v T tanh(W1h i +W2q); In the formula, α i e represents the attention weight at the i-th time step; i For attention scores; h i q is the hidden state at time step i; v, W1, and W2 are learnable parameters; and c is the context vector used for prediction.

[0081] The specific implementation of step S04 involves constructing a regional inundation risk network model using the minimum cut maximum flow algorithm. First, the study area is discretized into a grid structure, with the grid size determined based on terrain complexity, typically 10–50 meters. Then, a directed network graph is constructed, where nodes represent terrain elevation points, and the capacity is set to the difference between the elevation value of that point and the predicted sea level; negative values ​​indicate already inundated areas. Next, edge weights are defined as the unobstructedness of the water flow channels between adjacent nodes, calculated as the ratio of the elevation difference to the distance between adjacent nodes multiplied by the surface roughness coefficient. The surface roughness coefficient is determined based on land use type, such as 0.01–0.03 for urban construction areas, 0.03–0.05 for farmland, and 0.05–0.10 for forests. Finally, the Ford-Fulkerson algorithm or the Dinic algorithm is applied to solve the maximum flow problem, identifying the minimum cut sets in the network. The nodes connected by edges in the minimum cut sets constitute the boundaries of potential inundation areas. The purpose of this step is to utilize the minimum cut maximum flow algorithm in graph theory to accurately simulate seawater intrusion paths, locate weak points and critical vulnerabilities in the defense system, and provide a spatial analysis basis for risk assessment. The node capacity in the network graph is defined as: C(v) = E(v) - H sea In the formula, C(v) is the capacity of node v; E(v) is the elevation value of node v; H sea To predict sea level height, the edge weights are calculated using the following formula: In the formula, W(u, v) is the weight of the edge between nodes u and v; D(u, v) is the Euclidean distance between nodes u and v; and R(LandType) is the surface roughness coefficient corresponding to the land type. The mathematical expression of the maximum flow minimum cut problem is:

[0082] maxf=max∑ e∈out(s) f(e);

[0083]

[0084] In the formula, f is the total network traffic; f(e) is the traffic on edge e; c(e) is the capacity of edge e; s is the source node (the source of seawater); t is the sink node (the virtual high point); in(v) and out(v) represent the sets of edges entering node v and leaving node v, respectively; E is the set of all edges in the graph.

[0085] The specific implementation of step S05 is to establish a regional vulnerability scoring system. First, a scoring index system is determined, including five dimensions: topographic elevation distribution (weight 0.25), protective facility status (weight 0.20), land use type (weight 0.15), population density (weight 0.20), and economic value (weight 0.20). Then, each index is standardized, converting the original values ​​into a score range of 0-10. The formula for calculating the standard score for topographic elevation is 10 × (current elevation - predicted sea level) / (highest point in the region - predicted sea level), with a value of 0 for values ​​less than 0. Next, basic scores are allocated according to land use type, such as 8-10 points for urban residential areas, 7-9 points for industrial areas, 5-7 points for farmland, 3-5 points for forest land, 2-4 points for grassland, and 0-2 points for unused land. Finally, the vulnerability index is calculated by weighting the values ​​of all indicators, and the vulnerability index is divided into five levels: extremely low (0-2), low (2-4), moderate (4-6), high (6-8), and extremely high (8-10). The purpose of this step is to establish a quantitative standard for regional vulnerability through a comprehensive multi-indicator scoring system, identify sensitive areas susceptible to sea-level rise, and provide an important basis for comprehensive risk assessment. The vulnerability index calculation formula is as follows: In the formula, VI is the vulnerability index; w i S is the weight coefficient of the i-th indicator; i Let be the standardized score of the i-th indicator. The formula for calculating the standardized score of terrain height is:

[0086]

[0087] In the formula, S1 is the standard score of terrain elevation; E is the current elevation; H sea To predict sea level height; E max This is the elevation of the highest point in the area. The formula for calculating the standard score of the protective facility condition is: In the formula, S2 is the standard score of the protective facility status; P is the assessment value of the current status of the protective facility; P min This is the minimum assessment value for protective facilities; P max This represents the highest assessment value for protective facilities. The standard score for land use type is directly assigned based on the classification: S3 = LandTypeScore(LandType); where S3 is the standard score for land use type; and LandTypeScore is the land use type score mapping function. The formula for calculating the standard score for population density is: In the formula, S4 is the standard score of population density; PD is the regional population density (people / square kilometer); PD min Minimum population density for the study area; PD max The maximum population density in the study area. The formula for calculating the standard economic value score is: In the formula, S5 represents the standard score for economic value; EV represents the economic value per unit area of ​​the region; EV min The minimum economic value per unit area of ​​the research region; EV max This represents the maximum economic value per unit area in the research region.

[0088] The specific implementation of step S06 is to apply Bayesian network ensemble analysis to integrate the model prediction results with the vulnerability index. First, a Bayesian network structure is constructed, with nodes including key factors such as predicted sea level height, inundation area percentage, protective facility status, population impact ratio, and estimated economic losses. Then, based on historical data and expert experience, a conditional probability table (CPT) is set between nodes. For example, under a sea level rise greater than 50 cm, the probability thresholds for severe inundation in areas with different vulnerability levels are: 0.05 for extremely low vulnerability areas, 0.15 for low vulnerability areas, 0.35 for moderately vulnerable areas, 0.65 for high vulnerability areas, and 0.85 for extremely high vulnerability areas. Next, the sea level rise prediction results from step S03 and the vulnerability index from step S05 are input as observational evidence, and the posterior probability distribution is calculated using the belief propagation algorithm. Finally, a comprehensive risk assessment matrix is ​​generated, with the horizontal axis representing different sea level rise scenarios (e.g., RCP2.6, RCP4.5, RCP8.5, etc.) and the vertical axis representing the assessment unit (grid or administrative division). The matrix element values ​​are the comprehensive risk scores (0-100 points) of the regional unit under the corresponding scenario. The purpose of this step is to leverage the uncertainty reasoning capabilities of Bayesian networks to scientifically integrate sea-level rise predictions and regional vulnerability assessments, forming a comprehensive risk quantification index. The conditional probability of a node in a Bayesian network is defined as follows: In the formula, X i For node variables; Pa(X) i ) is node X i The set of parent nodes. The formula for calculating the comprehensive risk assessment matrix is: M(r, s) = 100 × P(Inundation = High | SeaLevel = s, VI = VI). r In the formula, M(r, s) is the element value in the r-th row and s-th column of the comprehensive risk assessment matrix, representing the comprehensive risk score of regional unit r under the sea level rise scenario s; P(Inundation=High|SeaLevel=s,VI=VI r Given a sea-level rise scenario s and a vulnerability index VI r Under the given conditions, the posterior probability of severe flooding in the region. The message passing formula for the belief propagation algorithm is:

[0089]

[0090] In the formula, For node X i Passed to node Xj Message; φ(X i X j ) is node X i and node X j Potential function between; N(X) i ) is node X i The set of neighboring nodes.

[0091] The specific implementation of step S07 involves calculating the variation submatrix of the comprehensive risk assessment matrix. First, a set of time series sampling points is determined, typically selecting key time nodes such as 2030, 2050, 2070, and 2100. Then, a corresponding comprehensive risk assessment matrix M is generated for each sampling time point. t Next, calculate the matrix element difference ΔM = M between adjacent time points t1 and t2. t2 -M t1 First, obtain the initial change matrix; then calculate the rate of change matrix R = ΔM / (t2-t1), reflecting the rate of risk change per unit time; subsequently, perform second-order difference in the time dimension to obtain the acceleration matrix A = (R... t+1 -R t Finally, significant change points in the acceleration matrix are identified, with a threshold set at 2.5 standard deviations above the average acceleration. These points are marked as trend inflection points. The purpose of this step is to capture the characteristics of risk evolution over time through matrix operations, particularly identifying accelerating or decelerating trends in risk change, providing a time-series dynamic analysis basis for risk management decisions. The formula for calculating the change matrix is: ΔM(r, s) = M / Δt; t2 (r, s)-M t1 (r, s); where ΔM(r, s) is the element value in the r-th row and s-th column of the change matrix; M t1 (r, s) and M t2 (r, s) represent the element values ​​in the r-th row and s-th column of the comprehensive risk assessment matrix at time points t1 and t2, respectively. The formula for calculating the rate of change matrix is: In the formula, R(r, s) is the element value in the r-th row and s-th column of the rate of change matrix, representing the rate of change of risk per unit time. The formula for calculating the acceleration matrix is: In the formula, A(r, s) is the element value in the r-th row and s-th column of the variable acceleration matrix; R t (r, s) and R t+1 (r, s) represent the element value in the r-th row and s-th column of the rate of change matrix between two adjacent time intervals; Δt is the time interval. The trend inflection point identification condition is:

[0092]

[0093] In the formula, TurningPoint(r, s) is a trend inflection point marker matrix, where 1 indicates an inflection point and 0 indicates a non-inflection point; μ A σ is the mean of the changing acceleration matrix A; A Let A be the standard deviation of the varying acceleration matrix A.

[0094] The specific implementation of step S08 involves generating a comprehensive assessment report using a coastline adaptive gating model. First, the comprehensive risk assessment matrix generated in step S06 and the variable score matrix generated in step S07 are input into the coastline adaptive gating model. Then, a risk level distribution map is generated based on the model's calculation results, using a four-color system (red, orange, yellow, and green) to represent extremely high, high, medium, and low risk areas. Next, the top 10% of areas with the highest comprehensive risk scores are extracted as a list of critical vulnerabilities, including spatial coordinates, risk scores, major risk-causing factors, and protection recommendations. Subsequently, a time-series risk evolution trend chart is generated, with the horizontal axis representing time and the vertical axis representing risk scores, showing the risk change trajectory of different areas. Finally, a risk rating classification result table is formed, classifying areas according to risk urgency and formulating differentiated protection strategy recommendations. The rating classification uses a four-level standard: Level I (extremely urgent, action required within 5 years), Level II (urgent, action required within 5-15 years), Level III (medium, action required within 15-30 years), and Level IV (low urgency, to be considered after 30 years). The purpose of this step is to integrate the results of previous analyses and generate an intuitive and clear risk assessment report, providing a scientific basis and action guide for coastal zone management and protection decisions. The regional adaptive gating weight function in the coastline adaptive gating model is expressed as:

[0095]

[0096] In the formula, G(r, s) is the value of the regional adaptive gating weight function; B(r, s) is the comprehensive balance value; T1 and T2 are the interval boundary thresholds, usually T1 is set to 30 and T2 is set to 70; G conservative G balanced and G aggressive These are the conservative, balanced, and aggressive weight adjustment functions, respectively. The formula for calculating the overall balance value is: B(r,s)=w1·R(r,s)+w2·TC r +w3·SEV r In the formula, B(r, s) is the overall equilibrium value; R(r, s) is the rate of change of risk in the variation matrix; TC r SEV is the terrain complexity index for region r. r Let be the socioeconomic vulnerability distribution value of region r; w1, w2, and w3 are weight coefficients, satisfying w1 + w2 + w3 = 1, typically taken as w1 = 0.4, w2 = 0.3, and w3 = 0.3. The conservative weight adjustment function is: G conservative(r, s) = α·B(r, s); where α is the linear growth coefficient, typically ranging from 0.01 to 0.05. The balanced weight adjustment function is: In the formula, β is the steepness coefficient of the S-curve, typically ranging from 0.1 to 0.2. The aggressive weighting adjustment function is: In the formula, γ is the basic weighting coefficient, which is usually taken as 0.5 to 0.7; δ is the exponential growth coefficient, which is usually taken as 0.05 to 0.1.

[0097] The formula for calculating the regional terrain complexity index is:

[0098]

[0099] In the formula, TC r σ is the topographic complexity index of region r. E and μ E These are the standard deviation and mean of the regional elevation, respectively; S i E represents the slope value of the i-th grid cell within the region; n is the total number of grid cells; max and E min These are the highest and lowest elevations within the region, respectively; A r The area is w. elev w slope and w rough Let w be the weighting coefficient, and satisfy w elev +w slope +w rough =1, generally take w elev =0.4, w slope =0.4, w rough =0.2.

[0100] The formula for calculating the distribution of socioeconomic vulnerability is:

[0101]

[0102] In the formula, SEV r PD represents the socioeconomic vulnerability distribution value for region r. r IV represents the population density of region r. r The value of infrastructure in region r; EA r The intensity of economic activity in region r; LU r Scoring the land use type of region r; PD min PD max IV min IV max EA min EA max These are the minimum and maximum values ​​of the corresponding indicators, respectively; w pop winfra w econ and w land Let w be the weighting coefficient, and satisfy w pop +w infra +w econ +w land =1, generally take w pop =0.3, w infra =0.25, w econ =0.25, w land =0.2.

[0103] The detailed structure of the coastline adaptive gating model is a neural network architecture consisting of an input layer, a multi-head temporal attention layer, a bidirectional autoregressive encoding layer, a dynamic gating integration layer, and an output layer. The input layer receives three types of data: climate scenario data (including temperature changes, precipitation patterns, and sea-level rise predictions under different RCP scenarios), topographic feature vectors (including topographic parameters such as elevation distribution, slope, and orientation), and socioeconomic indicators (including population density, GDP, and infrastructure value). The mathematical representation of the input data is: X input =[X climate X terrain X socioeco ];

[0104] In the formula, X input Input data vector to the model; X climate X is a vector of climate scenario data; terrain X is the terrain feature vector; socioeco This is a vector of socioeconomic indicators. The climate scenario data vector is represented as: X climate =[T RCP P RCP SLR RCP ];

[0105] In the formula, T RCP This is a temperature change sequence under the RCP scenario; P RCP Precipitation model series under the RCP scenario; SLR RCP This is a sea-level rise prediction sequence under the RCP scenario. The terrain feature vector is represented as: X terrain =[E dist S dist A dist ];

[0106] In the formula, E dist Elevation distribution characteristics; S dist Slope distribution characteristics; A dist This represents the distribution characteristics. The socioeconomic indicator vector is represented as: X socioeco = [PD, GDP, IV];

[0107] In the formula, PD represents population density; GDP represents gross domestic product; and IV represents the value of infrastructure.

[0108] The dynamic gating integration layer dynamically adjusts the weights of each feature channel based on the variable submatrix in the comprehensive risk assessment matrix, using a regionally adaptive gating weight function for parameter adjustment. The calculation formula for dynamic gating integration is:

[0109] In the formula, Z represents the output of the dynamic gating integration layer; G(r, s) i The gating weights for feature channel i are calculated using the region-adaptive gating weight function; F i is the input feature of feature channel i; n is the total number of feature channels.

[0110] The output layer generates regional protection strategy recommendations and resource optimization allocation schemes, including the selection of engineering protection measure types, implementation timing, and investment priority ranking. The output is represented as: Y = [Y strategy Y timing Y priority ];

[0111] In the formula, Y represents the model output; strategy Recommended protection strategy; Y timing To arrange the timing of implementation; Y priority Prioritize investments.

[0112] The detailed steps for establishing the training dataset for the adaptive gating model of coastlines are as follows: First, historical sea-level change data for 25 typical coastal areas worldwide are collected, spanning no less than 50 years, including monthly average sea-level height, frequency and intensity of extreme events; then, inundation extent maps and actual impact assessment reports for each region under different sea-level rise scenarios are integrated, including more than 100 case studies; next, topographic change characteristics and socio-economic response indicators before and after the implementation of regional protection projects are extracted, such as changes in the frequency of flooding events and the reduction in economic losses before and after the construction of protective dikes; subsequently, a dataset containing multiple... A parallel case library with multiple time points across the region, containing no fewer than 500 typical cases, was established. Data normalization was then performed to unify indicators of different dimensions to the 0-1 range and align them spatiotemporally to ensure comparability between different cases. Next, the effectiveness score (0-10 points) and cost-benefit ratio of each case's protective measures were labeled. The effectiveness score was determined based on the actual disaster prevention and mitigation results, and the cost-benefit ratio was calculated as the ratio of input costs to avoided losses. Finally, the data was divided chronologically into a training set (70%), a validation set (15%), and a test set (15%) to ensure the scientific rigor of model training and evaluation. The data normalization process employed the min-max normalization method. In the formula, X normX represents the normalized data; X represents the original data; X min and X max These are the minimum and maximum values ​​of the data, respectively. The cost-benefit ratio is calculated using the following formula: In the formula, CEB represents the cost-benefit ratio; C invest For input costs; L avoided To avoid losses.

[0113] To better understand and implement this invention, a specific application scenario, Example 2, is provided below: Researchers selected a typical coastal area as the research object. This area has a total area of ​​approximately 256 square kilometers and a coastline of approximately 37 kilometers. It includes various land use types such as urban built-up areas, industrial areas, farmland, and forest land, with a resident population of approximately 150,000. The method of this invention is applied to comprehensively assess the risks to this area under a sea-level rise scenario.

[0114] First, construct a multi-dimensional data acquisition network, such as... Figure 4 The diagram shows a multidimensional data acquisition network. Nine fixed tide gauge stations were deployed within the study area, with an average spacing of approximately 4.1 kilometers; 25 GNSS surface deformation monitoring stations were established, with a monitoring point density of one per 10 square kilometers; a regional digital elevation model with a resolution of 0.5 meters was acquired; and quarterly remote sensing imagery data from 2000 to 2022 were integrated. Data from different sources were unified to the WGS84 coordinate system using a data fusion algorithm, forming a spatiotemporally consistent sea-level change monitoring dataset. The weighting coefficients of each data source are shown in Table 1.

[0115] Table 1 Weighting Coefficients for Multi-Source Data Fusion

[0116] Data source Historical observation standard deviation (m) Weighting coefficient Tide level data 0.032 0.412 Satellite altimeter data 0.055 0.196 GNSS surface deformation data 0.008 0.657 SAR interferometric data 0.021 0.476 Optical remote sensing images 0.087 0.140 Drone aerial images 0.043 0.285

[0117] Secondly, multi-scale wavelet analysis was used to perform nonlinear decomposition on the sea-level change time series data from 2000 to 2022. Morlet wavelet transform was applied to the original time series, identifying five critical point signals that appeared in 2004, 2008, 2012, 2016, and 2020. The sea-level change rate at these critical points was significantly higher than the conventional change rate. The characteristics of the critical points and their wavelet coefficients are shown in Table 2.

[0118] Table 2 Characteristics of Critical Points for Sea Level Change

[0119] Critical point time Wavelet coefficient values Rate of change (mm / year) Average rate of change over the previous 5 years (mm / year) Change factor June 2004 2.87 8.2 2.7 3.04 August 2008 3.21 9.5 3.1 3.06 May 2012 3.35 10.3 3.3 3.12 July 2016 4.12 12.6 3.8 3.32 September 2020 5.78 18.5 4.5 4.11

[0120] Figure 6 and Figure 7 This paper presents the time series of sea level changes from 2000 to 2022 and the results of its wavelet analysis. Figure 6The curves showing the relative changes in sea level are displayed, marking the five critical points mentioned in the examples (2004, 2008, 2012, 2016, and 2020). Figure 7 This is a heatmap of wavelet energy spectrum, showing the energy distribution at different time scales (from seasonal to decadal scales), clearly demonstrating the anomalous energy concentration at critical points. Multi-scale wavelet analysis can clearly identify critical points in the sea-level change time series; these critical points manifest as regions of anomalous energy concentration in the wavelet energy spectrum, particularly on 2–7 year scales.

[0121] Third, a nonlinear prediction model for sea level rise was constructed based on deep learning neural networks. A hybrid architecture consisting of a 2-layer LSTM (64 units per layer) and a 3-layer CNN (3×3 kernel size) was built. Input parameters included historical sea level observation data, temperature and precipitation predictions output from the IPCC global climate model, and regional geological subsidence rates. The prediction results of the model under three scenarios—RCP2.6, RCP4.5, and RCP8.5—are shown in Table 3.

[0122] Table 3. Predicted Sea Level Rise Values ​​under Different Scenarios

[0123]

[0124]

[0125] Fourth, a regional flooding risk network model is constructed using the minimum cut maximum flow algorithm, such as... Figure 5 The diagram shows a schematic of the inundation risk network model using the minimum cut maximum flow algorithm. The study area was discretized into a 20m × 20m grid structure, and a directed network graph was constructed. Edge weights were set based on the ratio of the elevation difference to the distance between adjacent nodes multiplied by the surface roughness coefficient. The minimum cut set was calculated using the Dinic algorithm, identifying 32 critical vulnerable points and 4 main potential inundation channels. The surface roughness coefficients for different land use types are shown in Table 4.

[0126] Table 4. Surface roughness coefficients for different land use types

[0127] Land use types Surface roughness coefficient Urban construction area 0.018 Industrial Zone 0.022 farmland 0.037 woodland 0.085 grassland 0.059 wetlands 0.045 bare land 0.025 waters 0.010

[0128] Fifth, a regional vulnerability scoring system was established, and a vulnerability index was calculated based on topographic elevation distribution, the status of protective facilities, land use type, population density, and economic value. The study area was divided into 65 assessment units, and the vulnerability index distribution of each unit is shown in Table 5.

[0129] Table 5. Distribution of Vulnerability Index in the Study Area

[0130] Vulnerability Level Vulnerability Index Range Number of evaluation units Area percentage (%) Population percentage (%) Extremely low 0～2 7 12.5 5.8 Low 2～4 18 31.2 22.7 medium 4～6 22 28.7 33.4 high 6～8 12 18.9 27.6 Extremely high 8～10 6 8.7 10.5

[0131] Sixth, the prediction results of the nonlinear prediction model for sea-level rise and the vulnerability index of the regional vulnerability scoring system are integrated using Bayesian network ensemble analysis to generate a comprehensive risk assessment matrix. Taking the 2050 RCP4.5 scenario as an example, the risk score distribution of the study area is shown in Table 6.

[0132] Table 6. Distribution of Risk Scores in the Study Area under RCP4.5 Scenario in 2050

[0133] Risk level Risk scoring range Number of evaluation units Area percentage (%) Population percentage (%) Low risk 0～25 19 32.6 24.3 Medium risk 25～50 25 36.5 38.7 High risk 50～75 14 20.7 26.2 Extremely high risk 75～100 7 10.2 10.8

[0134] Seventh, the variation submatrix of the comprehensive risk assessment matrix was calculated. By comparing the risk assessment matrices for 2030, 2050, 2070, and 2100, the risk change rate and trend characteristics were extracted. The results show that there are 8 significant trend inflection points in the risk change acceleration matrix of the study area, of which 6 are points of accelerated risk increase and 2 are points of deceleration. The statistical characteristics of the change rate matrix are shown in Table 7.

[0135] Table 7 Characteristics of Risk Change Rate in Different Time Periods

[0136]

[0137]

[0138] Figure 8 and Figure 9 The sea-level rise prediction and risk assessment results from the embodiments are shown. Figure 8 The data presents the sea level rise projection curves from 2022 to 2100 under three scenarios: RCP2.6, RCP4.5, and RCP8.5, including 95% confidence intervals. It also shows the risk change rate for different time periods on the secondary axis. Figure 9 This data shows the distribution of risk areas in different years (2030, 2050, 2070, and 2100), illustrating the changes in the area proportions of low-risk, medium-risk, high-risk, and extremely high-risk areas, as well as the increasing trend in the number of critical vulnerabilities. Sea-level rise predictions under different RCP scenarios show that the rise rate under the RCP8.5 scenario is significantly higher than other scenarios, and the rise trend exhibits an accelerating characteristic. Simultaneously, with sea-level rise, the risk distribution pattern within the study area also changes significantly, with the area proportions of high-risk and extremely high-risk areas gradually increasing, and the number of critical vulnerabilities also growing accordingly.

[0139] Finally, an adaptive gating model of the coastline was used to generate a comprehensive assessment report on the risk of sea-level rise in the region, identifying 12 key vulnerabilities and 3 risk hotspots, and proposing differentiated protection strategies.

[0140] Traditional sea-level rise risk assessment methods primarily rely on linear prediction models and static inundation analysis, which cannot effectively capture the nonlinear characteristics and abrupt events of sea-level changes, resulting in limited assessment accuracy and an inability to provide dynamic risk evolution trends. This invention employs a multi-source data fusion and multi-scale wavelet analysis method, combined with deep learning neural networks and the minimum cut maximum flow algorithm, significantly improving the accuracy of sea-level rise prediction and the spatial precision of risk assessment. Experimental results show that, compared with traditional methods, this invention improves the accuracy of identifying critical point signals by 15.8%, reduces the root mean square error in predicting sea-level rise trends by 18.3%, and improves the spatial precision in delineating potential inundation zone boundaries by 12.7%. Furthermore, by introducing a variable submatrix and a coastline adaptive gating model, this invention can provide dynamic risk evolution trends and differentiated protection strategies, offering a more scientific basis for coastal zone management and protection decisions.

[0141] It should be noted that the variables involved in this invention are explained in detail in Tables 8, 9, and 10 below.

[0142] Table 8. Variable Explanation Table (Part 1)

[0143]

[0144]

[0145] Table 9. Variable Explanation Table (Part Two)

[0146]

[0147]

[0148] Table 10 Variable Explanation Table (Part 3)

[0149]

[0150] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A sea level rise impact risk assessment method based on multi-source data fusion, characterized in that, The method comprises the following steps: A multi-dimensional data acquisition network is constructed to obtain basic features of sea level rise; a multi-scale wavelet analysis is used to perform nonlinear decomposition on time series data of sea level change, to extract different frequency components and identify critical point signals; a nonlinear prediction model of sea level rise is constructed based on a deep learning neural network; a regional inundation risk network model is constructed by using a minimum cut maximum flow algorithm to calculate the minimum cut set to determine key vulnerable points and the boundary of potential inundation areas; a regional vulnerability scoring system is established to calculate a vulnerability index; a comprehensive risk assessment matrix is generated by using a Bayesian network integration analysis method; A change sub-matrix of the comprehensive risk assessment matrix is calculated, and a risk change rate and trend feature are extracted; A regional sea level rise risk comprehensive assessment report is generated by using a coastline adaptability gating model; the change sub-matrix is obtained by first establishing a time series sampling point set, generating a corresponding comprehensive risk assessment matrix for each sampling time point, calculating the element level difference of the comprehensive risk assessment matrices of adjacent time points to obtain an initial change amount matrix, further calculating the change rate of each element, i.e., the ratio of the change amount to the time interval, forming a change rate matrix, performing second-order difference on the change rate matrix in the time dimension to obtain a change acceleration matrix, identifying significant change points in the change acceleration matrix, and marking the trend inflection points; finally, the change rate matrix, the change acceleration matrix and the trend inflection point marks are integrated into a complete change sub-matrix; the specific structure of the coastline adaptability gating model is a neural network architecture composed of an input layer, a multi-head time series attention layer, a bidirectional autoregressive encoding layer, a dynamic gating integration layer and an output layer, wherein the input layer receives climate scenario data, terrain feature vectors and socio-economic indicators, the multi-head time series attention layer extracts features according to the change trend in different time scales, the bidirectional autoregressive encoding layer captures the causal relationship between sea level change and regional response, the dynamic gating integration layer dynamically adjusts the weights of each feature channel according to the change sub-matrix in the comprehensive risk assessment matrix, and the output layer generates regional protection strategy recommendations and resource optimization configuration schemes, wherein the dynamic gating integration layer adjusts parameters by using a regional adaptability gating weight function; the regional adaptability gating weight function calculates dynamic gating parameters based on the change sub-matrix of the comprehensive risk assessment matrix, the regional terrain complexity index and the socio-economic vulnerability distribution, and is expressed as: ； In the formula, is a regional adaptive gating weight function value; is a comprehensive balance value; and are interval boundary threshold values, is set to 30, is set to 70; , and are conservative, balanced and aggressive weight adjustment functions respectively; the comprehensive balance value calculation formula is: ; in the formula, is a comprehensive balance value; is a risk change rate in a change matrix; is a terrain complexity index of a region ; is a social and economic vulnerability distribution value of a region ; , and are weight coefficients, , , the conservative weight adjustment function is: ; in the formula, is a linear growth coefficient, and the value is 0.01-0.05; the balanced weight adjustment function is: ; in the formula, is an S-shaped curve steepness coefficient, and the value is 0.1-0.2; the aggressive weight adjustment function is: ; in the formula, is a basic weight coefficient, and the value is 0.5-0.7; is an exponential growth coefficient, and the value is 0.05-0.1; the conservative weight adjustment function refers to a weight adjustment function adopting a linear growth mode, which ensures smooth and gradual weight change and avoids system fluctuation; the balanced weight adjustment function refers to a weight adjustment function adopting an S-shaped curve mode, which realizes balance between smooth transition and timely response; and the aggressive weight adjustment function refers to a weight adjustment function adopting an exponential growth mode, which rapidly increases the weight after the risk exceeds a threshold value, and ensures that the system responds rapidly to high-risk situations.

2. The sea level rise impact risk assessment method according to claim 1, characterized by, The multi-dimensional data acquisition network refers to a multi-source heterogeneous data acquisition system composed of a satellite remote sensing system, a tide station monitoring system, a GNSS ground deformation monitoring station and an unmanned aerial vehicle aerial survey system.

3. The sea level rise impact risk assessment method according to claim 2, characterized in that, The minimum cut maximum flow algorithm is an optimization problem solving method in graph theory. In sea level rise risk assessment, the terrain is regarded as a network graph, the elevation is regarded as the node capacity, and the water flow channel is regarded as the edge. By calculating the minimum cut set, the weak link that is most likely to break through the defense line is determined, and the key vulnerable points and the boundary of potential inundation areas are located.

4. The sea level rise impact risk assessment method according to claim 3, characterized by, The vulnerability index is a comprehensive score obtained by weighted calculation of terrain height distribution, protection facility status, land use type, population density and economic value.

5. A computer readable storage medium, characterized in that, The computer readable storage medium stores program instructions, and the program instructions are used for executing the sea level rise influence risk assessment method based on multi-source data fusion in any one of claims 1-4 when running in the computer. 6.A system for risk assessment of sea level rise impact based on multi-source data fusion, characterized in that, The system is any one of a computer, a server, and a single chip microcomputer, and the computer readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing the program instructions stored in the computer readable storage medium.

Images