A method for flood and drought disaster dynamic risk assessment by fusing multi-source land information

Abstract

The application discloses a kind of flood and drought dynamic risk assessment methods of fusion multi-source land information, it is related to disaster risk assessment technical field.The method first obtains multidimensional land information source data, constructs unified standard fusion dataset by geographic registration, abnormal value processing and feature level fusion;Again, based on the dataset, respectively construct flood and drought hazard assessment model containing flood and drought sub-model, disaster environment exposure assessment model and disaster body vulnerability assessment model;Through risk coupling matrix and multivariate nonlinear coupling algorithm, the coupling operation of three kinds of models is realized, and the dynamic risk assessment result is obtained;Finally, establish risk dynamic updating mechanism, based on time series data iterative optimization evaluation result, and output spatial distribution diagram, time series curve and picture-text integrated report and other visual achievements.The method realizes multi-source data depth fusion and multi-factor coupling analysis, improves the accuracy and timeliness of evaluation, and provides scientific support for disaster prevention and mitigation decision-making.

Description

Technical Field

[0001] This invention relates to the field of disaster risk assessment technology, specifically to a dynamic risk assessment method for floods and droughts that integrates multi-source land information. Background Technology

[0002] Floods and droughts, as natural disasters with a wide impact and high frequency, pose a serious threat to the ecological environment, agricultural production, socio-economic development, and the safety of people's lives and property. With the intensification of global climate change and the rapid transformation of land use patterns, the suddenness of disasters and the uncertainty of their impact have significantly increased. Traditional assessment methods based on static data are no longer adequate to meet the complex and ever-changing needs of disaster prevention and control. There is an urgent need to build a dynamic assessment technology system that can accurately capture the spatiotemporal evolution characteristics of disasters.

[0003] Land information serves as the core carrier of the process of flood and drought disaster formation, and its multi-dimensional attributes (such as topography, soil characteristics, and land use types) directly affect the probability and severity of disasters. Existing assessment methods mostly rely on land data from single sources or limited types, resulting in low data fusion and insufficient spatiotemporal resolution. This makes it difficult for assessment results to fully reflect the true state of disaster risk. At the same time, there are significant shortcomings in dynamic risk updates and multi-factor coupling analysis, which restricts the scientific nature and timeliness of disaster prevention and mitigation decisions.

[0004] Chinese patent (patent number: CN112819340A) discloses a dynamic assessment method for urban flood disasters based on multi-source data. This technology overcomes the shortcomings of traditional methods, such as difficulty in data acquisition and poor spatiotemporal resolution, by collecting real-time data, constructing a disaster-bearing body dataset, and combining it with an urban stormwater model. This enables dynamic assessment of flood disaster losses and promotes the cross-application of big data technology and the hydrological field. However, this patent only focuses on the single disaster of urban flooding and does not cover the assessment scenario of drought disasters. Furthermore, the fusion of multi-source land information is limited to the data level and does not delve into the intrinsic relationship between core attributes such as land physicochemical properties and topographic features and disaster risks. The assessment model lacks specificity and comprehensiveness.

[0005] Therefore, it is urgent to construct a heterogeneous data adaptation and fusion mechanism, deeply explore the correlation between land information and disaster risk, optimize the multi-factor coupled assessment model and dynamic update mechanism, and achieve accurate assessment of flood and drought disaster risks. Summary of the Invention

[0006] Based on the aforementioned technical problems, this application discloses a dynamic risk assessment method for flood and drought disasters that integrates multi-source land information, specifically including:

[0007] Collect multi-dimensional land information source data, which includes remote sensing inversion data, land use status survey data, soil physicochemical properties data, topographic and geomorphological features data, and socio-economic statistics data;

[0008] Heterogeneous data adaptation processing is performed on land information source data from various dimensions to construct a fusion dataset with unified spatial benchmarks and data standards.

[0009] Based on the fused dataset, a flood and drought disaster risk assessment model, a disaster-prone environment exposure assessment model, and a disaster-bearing body vulnerability assessment model were constructed respectively.

[0010] The dynamic risk assessment results of floods and droughts are obtained by coupling the aforementioned flood and drought disaster risk assessment model, disaster-prone environment exposure assessment model, and disaster-bearing body vulnerability assessment model.

[0011] Establish a dynamic risk update mechanism, iteratively optimize the dynamic risk assessment results of floods and droughts based on time-updated land information data, and output visualized assessment results.

[0012] Preferably, the heterogeneous data adaptation processing of land information source data of various dimensions includes:

[0013] A georeferencing model is used to spatially correct remote sensing inversion data, and a coordinate transformation algorithm is used to unify land information data from different projection coordinate systems to the target coordinate system.

[0014] Anomaly detection algorithms are used to identify and remove outlier samples from soil physicochemical properties data and socioeconomic statistics data, and data interpolation methods are used to fill in missing values.

[0015] The planar fitting method is used to calculate topographic factors such as slope, aspect, and roughness based on the digital elevation model. After geometric registration with the geomorphological data, topographic and geomorphological feature data are formed.

[0016] A multi-source data weight allocation model is constructed, and the fusion weight of each data dimension is determined by combining the information contribution and reliability indicators of each data dimension. A fused dataset is generated through a feature-level fusion algorithm.

[0017] Preferably, the flood and drought hazard assessment model includes a flood hazard assessment sub-model and a drought hazard assessment sub-model, and the construction process of the flood hazard assessment sub-model is as follows:

[0018] Extreme rainfall, topographic slope, topographic aspect, river network density, soil saturated moisture content, and land use type were selected as indicators for flood risk assessment.

[0019] The subjective weights of each evaluation indicator are determined based on the analytic hierarchy process (AHP), the objective weights of each indicator are calculated using the entropy weight method, and the comprehensive weights are determined by the coupled weight formula.

[0020] A flood hazard index calculation model is constructed. Through index standardization, weighted summation, and index hierarchical mapping, the flood hazard assessment results are obtained. The formula for the flood hazard index calculation model is as follows: ,in, The flood risk index, For the first The combined weight of each indicator For the first The standardized value of each indicator, These are the model correction coefficients. To assess the number of indicators.

[0021] Preferably, the construction process of the drought risk assessment sub-model is as follows:

[0022] The vegetation water supply and demand index, soil moisture anomaly, potential evapotranspiration to precipitation ratio, water availability and irrigation guarantee rate were selected as indicators for drought risk assessment.

[0023] A correlation matrix of indicators was constructed, and the grey relational analysis method was used to quantify the correlation between each indicator and drought disaster, and to screen the core influencing indicators.

[0024] A drought hazard classification model is constructed based on the support vector machine algorithm. A high-dimensional feature space is mapped using a radial basis function kernel. The drought hazard classification model satisfies the following: ,in, The output results are the drought risk level. For the Lagrange multipliers of support vector machines, The class labels for the training samples. For radial basis kernel functions, , For kernel function parameters, Let be the feature vector of the sample to be evaluated. For the first Each training sample feature vector For the number of support vectors, For model bias terms;

[0025] The model parameters are optimized using a training sample set, and standardized assessment index data is input to output drought risk level assessment results.

[0026] Preferably, the construction process of the environmental exposure assessment model for pregnancy-related disasters is as follows:

[0027] The types of exposure carriers are classified, including natural ecosystems, agricultural production land, urban built-up areas, and infrastructure;

[0028] For different types of exposure carriers, exposure area, spatial distribution density, and distance from the disaster source are selected as exposure assessment factors;

[0029] An exposure quantification model is constructed, and the exposure intensity of each exposure carrier is calculated through spatial overlay analysis. Combined with the exposure sensitivity coefficient, the exposure assessment results of the disaster-prone environment are obtained. The formula for the exposure quantification model is as follows: ,in, For environmental exposure during pregnancy, For the first The area percentage of the surface area exposed to the carrier. For the first Distribution density of the type of exposure vector, For the first Sensitivity coefficient of the type of exposure vector, For the first The average distance between the exposed carrier and the disaster source. This is the distance attenuation coefficient. Number of exposed carrier types.

[0030] Preferably, the process of constructing the vulnerability assessment model for the disaster-bearing body is as follows:

[0031] Vulnerability assessment indicators were selected from three dimensions: disaster resistance, resilience, and vulnerability, including population density, economic development level, infrastructure completeness, emergency response capability, and ecological resilience index.

[0032] The weights of each evaluation index are determined by fuzzy hierarchical analysis, and the index data are fuzzified by membership function to construct a fuzzy evaluation matrix.

[0033] A comprehensive vulnerability assessment model is constructed to calculate the correlation degree and determine the vulnerability level. The formula for the comprehensive vulnerability assessment model is as follows: ,in, As a vulnerability index for disaster-bearing bodies, For the first The weight of each indicator, For the first The standardized value of each indicator, To assess the number of indicators, we obtain the vulnerability assessment results of the disaster-bearing body.

[0034] Preferably, the process of obtaining the dynamic risk assessment results of floods and droughts through the coupled operation of three models includes:

[0035] Construct a risk coupling matrix to clarify the interaction among the three factors: the risk of water and drought disasters, the exposure of disaster-prone environments, and the vulnerability of disaster-bearing bodies, and quantify the degree of coupling between each pair.

[0036] A risk assessment model is established using a multivariate nonlinear coupling algorithm, which introduces a coupling coordination coefficient to adjust the influence weights of the three factors. The formula for the multivariate nonlinear coupling algorithm is as follows: ,in, This represents the dynamic risk value for floods and droughts. The results of the flood and drought disaster risk assessment The results of the environmental exposure assessment for pregnancy-related disasters. The results of the vulnerability assessment of disaster-bearing bodies, , , These are the coefficients of the interaction term.

[0037] Preferably, the process of establishing the risk dynamic update mechanism is as follows:

[0038] A data update cycle is set, and based on remote sensing image time series data, land use change data and real-time monitoring data, an incremental update algorithm is used to periodically update the fused dataset, and only the data of the changed areas are reprocessed.

[0039] A risk change detection model was constructed, and the spatial migration characteristics and temporal evolution patterns of flood and drought disaster risks were identified through a spatiotemporal difference analysis algorithm.

[0040] Based on the characteristics of risk changes, an incremental learning algorithm is used to iteratively optimize the parameters of the dynamic risk assessment model for floods and droughts, without retraining the entire model, thus enabling rapid dynamic updates of risk assessment results.

[0041] Preferably, the formula for setting the data update cycle is: ,in, This refers to the actual update cycle. As the baseline update cycle, For the first The rate of change of land-related information data For the first Weights of class data For the first Sensitivity of data type to risk assessment For the number of data types, For environmental stability coefficient, This represents the intensity coefficient of regional disaster activity.

[0042] Preferably, the output of the visualization evaluation results includes:

[0043] Geographic Information System (GIS) technology is used to generate a spatial distribution map of flood and drought disaster risk levels. Combined with a three-dimensional terrain model, the risk distribution is visualized in three dimensions. The spatial distribution differences of risks are presented intuitively through a hierarchical color scheme.

[0044] We construct time-series risk change curves, introduce a sliding window algorithm to smooth data fluctuations, and clearly show the dynamic evolution trend of flood and drought disaster risks in different periods.

[0045] Generate risk assessment reports, including risk level statistics, high-risk area identification, risk driver contribution rate analysis, and targeted disaster prevention and mitigation recommendations, and support integrated text and graphic output.

[0046] Compared with the prior art, the technical solution of this application has the following technical effects:

[0047] This invention achieves standardized fusion of multi-dimensional land information source data by constructing a heterogeneous data adaptation and processing mechanism. Through geographic registration, outlier processing, and feature-level fusion algorithms, it solves the compatibility problem of land data from different sources, scales, and formats, forming a fusion dataset with a unified spatial benchmark and data standards. This provides comprehensive and high-quality data support for subsequent evaluation models, significantly improving the reliability and completeness of the evaluation data.

[0048] This invention utilizes a specially constructed sub-model for assessing flood and drought risks, combined with a multivariate nonlinear coupling algorithm, to achieve a deep coupling analysis of flood and drought disaster risks, exposure to disaster-prone environments, and vulnerability of disaster-bearing bodies. By scientifically selecting assessment indicators, optimizing weight allocation, and designing algorithms, the model accurately captures the intrinsic relationship between multidimensional attributes of land information and disaster risks, significantly improving the accuracy and relevance of risk assessment results and overcoming the limitations of traditional assessment methods that are singular and superficial.

[0049] The risk dynamic update mechanism established in this invention achieves rapid iterative optimization of risk assessment results through the synergistic application of incremental update and incremental learning algorithms. It eliminates the need for full data reprocessing and model retraining, updating only the changed areas and key parameters. This efficiently responds to the impact of changes in underlying surface conditions on flood and drought disaster risks, forming a closed loop of assessment, update, and reassessment. This completely overcomes the shortcomings of traditional static assessments, which cannot adapt to the real-time evolution of disasters.

[0050] This invention clearly demonstrates the spatial distribution differences and dynamic evolution trends of flood and drought disaster risks, accurately identifies high-risk areas and analyzes risk-driving factors, providing comprehensive and intuitive technical support for disaster prevention and mitigation decision-making, effectively reducing the difficulty of decision-making, and improving the scientific nature and timeliness of disaster prevention and mitigation work.

[0051] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the preferred embodiments of this application are described in detail below with reference to the accompanying drawings.

[0052] The above and other objects, advantages and features of this application will become more apparent to those skilled in the art from the following detailed description of specific embodiments in conjunction with the accompanying drawings. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0054] Based on the description of the figures and their corresponding technical content in the document, the titles of the figures are as follows:

[0055] Figure 1 A schematic diagram illustrating the overall process steps of the dynamic risk assessment method for floods and droughts;

[0056] Figure 2 A schematic diagram illustrating the breakdown of the flood and drought sub-models in the flood and drought hazard assessment model;

[0057] Figure 3 A schematic diagram illustrating the specific steps of a multi-model coupled computation model;

[0058] Figure 4 Comparison of three-dimensional surface graphs showing the correlation between the weights of flood and drought risk indicators and the indices in the Ganjiang River Basin;

[0059] Figure 5 Scatter plot of correlation between exposure to disaster-prone environments and vulnerability of disaster-bearing bodies for four types of exposed carriers in the Ganjiang River Basin;

[0060] Figure 6 Comparison curves of the coupling degree of the three factors of flood and drought disasters and the dynamic risk value response in the Ganjiang River Basin;

[0061] Figure 7 Comparative analysis curves of the changing characteristics of multi-dimensional indicators of the dynamic risk update effectiveness in the Ganjiang River Basin;

[0062] Figure 8 Stacked curves showing the weighting of driving factors for floods and droughts at different risk levels in the Ganjiang River Basin;

[0063] Figure 9 A three-dimensional comparison curve of the time-series evolution of dynamic risk values ​​for floods and droughts in the Ganjiang River Basin and the percentage of each level. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. In the following description, specific details such as specific configurations and components are provided merely to help fully understand the embodiments of this application. Therefore, those skilled in the art should understand that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. In addition, for clarity and brevity, descriptions of known functions and structures are omitted in the embodiments.

[0065] It should be understood that the phrase "an embodiment" or "this embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "an embodiment" or "this embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0066] Furthermore, reference numerals and / or letters may be repeated in different examples within this application. Such repetition is for the purpose of simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or settings discussed.

[0067] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" in this article describes another type of relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the related objects before and after it are in an "or" relationship.

[0068] In this article, the term "at least one" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, "at least one of A and B" can mean: A exists alone, A and B exist simultaneously, or B exists alone.

[0069] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion.

[0070] Example 1

[0071] This embodiment mainly describes a method for dynamic risk assessment of floods and droughts that integrates multi-source land information, such as... Figure 1 As shown, it specifically includes:

[0072] Collect multi-dimensional land information source data, which includes remote sensing inversion data, land use status data, soil physicochemical properties data, topographic data, and socio-economic statistics.

[0073] Heterogeneous data adaptation processing is performed on land information source data from various dimensions to construct a fusion dataset with unified spatial benchmarks and data standards.

[0074] Based on the fused dataset, a flood and drought disaster risk assessment model, a disaster-prone environment exposure assessment model, and a disaster-bearing body vulnerability assessment model were constructed respectively.

[0075] The dynamic risk assessment results of floods and droughts are obtained by coupling the aforementioned flood and drought disaster risk assessment model, disaster-prone environment exposure assessment model, and disaster-bearing body vulnerability assessment model.

[0076] Establish a dynamic risk update mechanism, iteratively optimize the dynamic risk assessment results of floods and droughts based on time-updated land information data, and output visualized assessment results.

[0077] Furthermore, the collection of multi-dimensional land information source data needs to cover the data required for the entire chain of flood and drought disaster assessment, and each type of data needs to meet the preset accuracy and spatiotemporal resolution requirements; among them, remote sensing inversion data is obtained through inversion of multispectral remote sensing images and synthetic aperture radar images, including parameters such as vegetation cover, surface temperature, and precipitation distribution, with a spatial resolution of no less than 30 meters and a temporal resolution of no less than 16 days for remote sensing images; land use status data adopts the national land use classification standard, clearly distinguishing types such as cultivated land, forest land, grassland, urban construction land, and water areas, and the data format is vector or raster surface data; soil physicochemical property data Indicators including soil texture, soil organic matter content, soil saturated water content, and soil permeability coefficient are obtained through a combination of soil sampling and testing and soil database queries. Topographic and geomorphological feature data are extracted using a digital elevation model (DEM), covering topological parameters such as slope, aspect, altitude, topographic relief, and river network density, with a DEM resolution of no less than 30 meters. Socioeconomic statistics include population density, GDP spatial distribution, agricultural output share, infrastructure distribution, and irrigation facility coverage. Spatial interpolation methods are used to convert statistical data into raster data, with the raster size consistent with other data. All collected data must record metadata information such as data source, collection time, accuracy indicators, and data format to ensure data traceability.

[0078] Furthermore, the heterogeneous data adaptation process is carried out according to the workflow of spatial benchmark unification, data quality optimization, and feature-level fusion. In the spatial benchmark unification stage, a geographic registration algorithm is used to perform geometric fine correction on the remote sensing inversion data to eliminate image distortion errors. A seven-parameter coordinate transformation model is used to unify all land information data from different projection coordinate systems (such as Gauss-Kruger projection and UTM projection) to the WGS-84 coordinate system, ensuring the consistency of the data's spatial location. In the data quality optimization stage, based on the Laida criterion (3... (Guidelines) An outlier detection algorithm is constructed to identify and remove abnormal samples from soil physicochemical properties data and socioeconomic statistics. To address the issue of missing data, different interpolation methods are selected according to the data type: Kriging interpolation is used for continuous data, and nearest neighbor interpolation is used for discrete data to fill in missing values ​​and ensure the integrity of the dataset. In the feature-level fusion stage, a multi-source data weight allocation model is constructed. The information contribution of each data dimension is calculated by the information gain rate. The reliability index is determined by combining the data collection accuracy and timeliness. The fusion weight of each dimension of data is determined by the linear weighting method. The weights meet the normalization condition. Then, the principal component analysis algorithm is used to extract and fuse features of each dimension of data to generate a fused dataset with a unified spatial benchmark, unified data format, and unified resolution.

[0079] Furthermore, such as Figure 2 As shown, the flood and drought disaster risk assessment model includes a flood risk assessment sub-model and a drought risk assessment sub-model, which are constructed to address the disaster-causing mechanisms of floods and droughts, respectively.

[0080] The flood hazard assessment sub-model selects extreme rainfall (using the annual maximum 24-hour rainfall), topographic slope, river network density, soil saturated moisture content, and land use type as core assessment indicators. In determining the weights, a judgment matrix is ​​constructed based on the Analytic Hierarchy Process (AHP), and the subjective weights of each indicator are obtained after consistency checks. ; Calculate the objective weights of each indicator using the entropy weight method. The entropy weight method reflects the degree of data dispersion by calculating the information entropy of the index; the smaller the information entropy, the greater the weight. This is achieved through coupling the weight formula. Determine the overall weight, where The weighting balancing coefficients (ranging from 0.4 to 0.6) are used. The index standardization process employs extreme value standardization, with positive indices (river network density, extreme rainfall) using... The conversion is applied to negative indicators (terrain slope). The conversion ensures that all indicators have a uniform value range of [0,1]. A logistic regression model for calculating the flood hazard index is constructed, with the following formula: ,in, The flood risk index, For the first The combined weight of each indicator For the first The standardized value of each indicator, These are the model correction coefficients (obtained through calibration using historical flood disaster data). To assess the number of indicators, the flood hazard index was divided into four levels: low hazard, medium hazard, high hazard, and extremely high hazard using the natural discontinuity method, thus completing the flood hazard assessment.

[0081] The drought risk assessment sub-model selects the vegetation water supply and demand index (VSWI), soil moisture anomaly, potential evapotranspiration to precipitation ratio (PET / P), topographic relief, and irrigation guarantee rate as core assessment indicators. First, an indicator correlation matrix is ​​constructed, and grey relational analysis is used to calculate the correlation between each indicator and the frequency and extent of drought disasters. Filter the correlation The core impact indicator is ≥0.6. A drought hazard classification model is constructed based on the Support Vector Machine (SVM) algorithm. The low-dimensional feature space is mapped to a high-dimensional feature space using the Radial Basis Function (RBF) to solve the nonlinear classification problem. The model formula is as follows: ,in, Output results for drought risk level (values ​​are 1, 2, 3, and 4, corresponding to low, medium, high, and extremely high risk levels, respectively). For the Lagrange multipliers of support vector machines, The class labels for the training samples. It is a radial basis kernel function, specifically in the form of: , The kernel function parameters are determined through grid search optimization. Let be the feature vector of the sample to be evaluated. For the first Each training sample feature vector For the number of support vectors, This is the model bias term. During model training, a training sample set is constructed using historical drought disaster data and corresponding indicator data. The model parameters are optimized through cross-validation. Standardized assessment indicator data is input, and drought risk level assessment results are output.

[0082] Furthermore, the disaster-prone environment exposure assessment model is conducted through three stages: exposure carrier classification, assessment factor selection, and quantitative model construction. Exposure carrier types are categorized into four main types: natural ecosystems (forests, grasslands, wetlands, etc.), agricultural production land (cultivated land, orchards, etc.), urban built-up areas (residential areas, industrial and commercial land, etc.), and infrastructure (transportation arteries, water conservancy facilities, power facilities, etc.). The spatial boundaries and attribute information of each type of carrier are clearly defined. For different types of exposure carriers, core assessment factors are selected: for natural ecosystems, exposure area, topographic position index, and distance from river networks; for agricultural production land, exposure area, spatial distribution density, soil fertility level, and distance from disaster sources; for urban built-up areas, exposure area, population density, building density, and distance from flood risk areas; and for infrastructure, exposure length (linear facilities) or exposure area (area facilities), importance level, and distance from disaster sources. A quantitative exposure model is constructed, calculating the exposure intensity of each exposure carrier through spatial overlay analysis, combined with the exposure sensitivity coefficient and distance attenuation effect. The formula is: ,in, For environmental exposure during pregnancy, For the first The area percentage of the surface area exposed to the carrier. For the first Distribution density of the type of exposure vector, For the first Sensitivity coefficient of the type of exposure vector, For the first The average distance between the exposed carrier and the disaster source. This is the distance attenuation coefficient. The model represents the number of exposed carrier types. It enables the quantitative calculation of exposure to disaster-prone environments, with higher exposure values ​​indicating greater impact of disasters on the carriers.

[0083] Furthermore, the disaster-bearing body vulnerability assessment model is constructed from three dimensions: disaster resistance capacity, resilience capacity, and vulnerability, comprehensively reflecting the vulnerability of disaster-bearing bodies to floods and droughts. The disaster resistance capacity dimension selects indicators such as infrastructure completeness, emergency response capacity, and irrigation guarantee rate; the resilience capacity dimension selects indicators such as economic development level, medical resource coverage rate, and ecological resilience index; and the vulnerability dimension selects indicators such as population density, proportion of elderly population, and agricultural dependence, forming a core assessment indicator system encompassing population density, economic development level, infrastructure completeness, emergency response capacity, and ecological resilience index. The weights of each indicator are determined using fuzzy hierarchical analysis. First, a fuzzy judgment matrix is ​​constructed, and the relative importance between indicators is represented by triangular fuzzy numbers. Then, the fuzzy judgment matrix is ​​converted into a clear weight vector, and the weights of each indicator are obtained after normalization. The indicator data is fuzzified using membership functions. An ascending semi-trapezoidal membership function is used for positive indicators (such as infrastructure completeness), and a descending semi-trapezoidal membership function is used for negative indicators (such as population density), thus constructing a fuzzy evaluation matrix. ,in For the first The first indicator corresponds to the first Membership degree of each vulnerability level, To assess the number of indicators, Define the number of vulnerability levels (set to 4); construct a comprehensive vulnerability assessment model, with the following formula: ,in, As a vulnerability index for disaster-bearing bodies, For the first The weight of each indicator, For the first The standardized values ​​of each indicator are calculated; the correlation degree of each vulnerability level is calculated, and the vulnerability level of the disaster-bearing body is determined by the principle of maximum correlation degree. The vulnerability index and the level classification results are combined to complete the vulnerability assessment of the disaster-bearing body.

[0084] Furthermore, such as Figure 3 As shown, the multi-model coupling operation calculates the dynamic risk assessment results of floods and droughts by constructing a risk coupling matrix and designing a multivariate nonlinear coupling algorithm. First, a risk coupling matrix is ​​constructed with a dimension of 3×3. The rows and columns correspond to the flood and drought hazard (H), the exposure to the disaster-prone environment (E), and the vulnerability of the disaster-bearing body (V), respectively. Matrix elements... This represents the coupling degree between the i-th evaluation result and the j-th evaluation result, calculated using the coupling coordination degree model. The formula is: ,in , These are the standardized values ​​of the two evaluation results, and the coupling degree is... The value ranges from [0,1], with larger values ​​indicating stronger interactions. Based on the coupling matrix, the interaction relationship among the three is clarified. A multivariate nonlinear coupling algorithm is used to establish a risk assessment model, introducing a coupling coordination coefficient to adjust the influence weights of the three. The algorithm formula is as follows: ,in, This represents the dynamic risk value for floods and droughts. The standardized value of the flood and drought disaster risk assessment results ( (or the standardized value of the drought risk level). This represents the standardized value of the environmental exposure assessment results for pregnancy-related disasters. The standardized value of the vulnerability assessment result of the disaster-bearing body. , , The coefficients for the interaction terms (calibrated using historical disaster data) are used. Before the coupling operation, the original assessment results of flood and drought disaster risk, disaster-prone environment exposure, and disaster-bearing body vulnerability need to be transformed to the [0,1] interval using the extreme value standardization method to ensure data dimension consistency. After the operation is completed, the dynamic risk value of flood and drought disaster is obtained.

[0085] Furthermore, the dynamic risk update mechanism includes three core components: incremental data updates, risk change detection, and iterative optimization of model parameters. In the incremental data update phase, a baseline update cycle is set. (Based on regional disaster characteristics, the interval is set to 1 month, 3 months, or 6 months). Based on remote sensing image time-series data, land use change data, and real-time monitoring data (such as real-time precipitation and soil moisture monitoring data), an incremental update algorithm is used to periodically update the fused dataset. Spatial change detection identifies areas where land information data has changed, and only data in these changed areas is reprocessed to avoid redundant calculations on the entire dataset. The updated data must maintain spatial reference and data format consistency with the original dataset. In the risk change detection phase, a risk change detection model is constructed. A spatiotemporal difference analysis algorithm is used to quantify the spatial migration characteristics and temporal evolution patterns of risk. The formula for calculating risk change is: ,in, Spatial coordinates The change in risk at time t compared to time t-1. Let be the risk value at time t. The risk value at time t-1 The time interval between the two assessments This is the risk attenuation coefficient; simultaneously, the change detection index is calculated. ,in It is the L2 norm. The relative risk change rate (CDI) is used to determine the significance of risk changes (CDI ≥ 10% indicates a significant change). During the model parameter iterative optimization phase, based on the risk change characteristics and the CDI threshold determination results, an incremental learning algorithm is used to iteratively optimize the parameters of the dynamic risk assessment model for floods and droughts. For areas of significant change, the corresponding indicator weights, model correction coefficients, and other parameters are updated without retraining the entire model, achieving rapid dynamic updates of risk assessment results and forming a closed-loop mechanism of assessment, update, and reassessment.

[0086] Furthermore, the visualization assessment results are presented in multiple formats and at multiple levels to ensure that the assessment results are intuitive, easy to understand, and easy to apply. For spatial distribution visualization, Geographic Information System (GIS) technology is used to generate a spatial distribution map of flood and drought disaster risk levels. Combined with a 3D terrain model (based on a DEM), a three-dimensional visualization of risk distribution is achieved. A hierarchical color scheme (using red, orange, yellow, and green to correspond to extremely high, high, medium, and low risk levels respectively) visually presents the spatial distribution differences of risks. The map must include standard elements such as a scale bar, legend, north arrow, and coordinate system, and support interactive operations such as map zooming, panning, and querying. For temporal evolution visualization, a time-series risk change curve is constructed. A sliding window algorithm (with a window size set to 3 update cycles) is introduced to smooth data fluctuations and clearly show the dynamic evolution trend of flood and drought disaster risks at different times. The curve must be labeled with key information such as risk value range, time nodes, and significant change points, and support comparative display of risk change curves across multiple regions. In terms of report output, the system generates a risk assessment report that integrates text and graphics. The report includes risk level statistics (area proportion and affected areas for each risk level), high-risk area identification (clarifying the spatial location, scope, and main risk drivers of high-risk areas), contribution rate analysis of risk drivers (calculating the contribution rate of each indicator to the risk value through sensitivity analysis), and targeted disaster prevention and mitigation recommendations (proposing engineering and non-engineering disaster prevention and mitigation measures based on risk characteristics). The report supports export in PDF, Word, and other formats to meet the needs of different application scenarios.

[0087] This implementation details how a systematic heterogeneous land information adaptation and fusion mechanism provides high-quality data support for assessment. By leveraging a targeted multi-dimensional assessment model and a multi-dimensional nonlinear coupling algorithm, it achieves accurate quantification of flood and drought disaster risks. The innovative dynamic update mechanism can quickly respond to changes in the underlying surface, forming a closed-loop optimization. Finally, the assessment results are presented through various forms of visualization.

[0088] Based on Example 1, the Ganjiang River Basin was selected as the study area, mainly consisting of cultivated land, forest land, grassland, and urban built-up areas. It is a typical flood- and drought-prone basin in southern my country, exhibiting both typicality and representativeness in terms of floods and droughts. Multi-dimensional land information source data was collected for the study area. Remote sensing inversion data was obtained using multispectral remote sensing images and synthetic aperture radar images, covering key parameters such as vegetation cover, surface temperature, and precipitation distribution. The spatial resolution of the remote sensing images was 30m. Land use status data adopted the national land use classification standard, clearly distinguishing 12 secondary land categories, including paddy fields, dry land, and forested land. Soil physicochemical properties data were obtained through testing at 1,200 soil sampling points deployed within the watershed, covering indicators such as soil texture and soil organic matter content. Among them, loam accounted for 42.5%, clay for 28.3%, and sandy soil for 29.2%. Topographic and geomorphological feature data were extracted based on a digital elevation model. The average slope within the watershed is 15.3°, and the river network density is 0.32 km / km². Socioeconomic statistics include population density and spatial distribution of GDP. The average population density within the watershed is 235 people / km², and the irrigation guarantee rate can reach over 85% in the plain area, but only about 42% in the mountainous area.

[0089] Multi-dimensional land information source data were collected and processed for heterogeneous data adaptation. A georeferencing model was used to spatially correct the remote sensing inversion data. A seven-parameter coordinate transformation algorithm was used to unify land information data from different projection coordinate systems to the WGS-84 coordinate system. The planar position error of the corrected data was controlled within 0.5 pixels. An outlier detection algorithm was constructed based on the Laida criterion to identify and remove outlier samples from soil physicochemical property data and socioeconomic statistics. A total of 127 outlier samples were removed. To address the issue of missing data, Kriging interpolation was used for continuous data and discrete data... The data was filled with 356 missing values ​​using nearest neighbor interpolation to ensure dataset integrity. A plane fitting method was used to calculate topographic factors such as slope, aspect, and roughness based on the digital elevation model, and geometric registration was performed with the geomorphological data, achieving a registration accuracy of over 98%. A multi-source data weight allocation model was constructed, combining the information contribution and reliability indicators of each data dimension to determine the fusion weights for each dimension. Specific data are shown in the table below. A feature-level fusion algorithm was used to generate a fused dataset with a unified spatial benchmark, unified data format, and unified resolution, providing high-quality data support for subsequent model construction.

[0090] Table 1. Weighting and Information Contribution of Land Information Data Integration in the Ganjiang River Basin

[0091] Data types Fusion weights Information contribution Reliability indicators RS 0.28 0.89 0.92 LULC 0.22 0.85 0.95 SP 0.18 0.78 0.88 DEM 0.15 0.75 0.90 SES 0.17 0.82 0.85

[0092] Note: RS represents remote sensing inversion data; LULC represents land use status data; SP represents soil physicochemical properties data; DEM represents topographic and geomorphological features data; SES represents socioeconomic statistics.

[0093] Based on a fusion dataset, three models were constructed to assess the risk of floods and droughts: a model for assessing exposure to disaster-prone environments and a model for assessing the vulnerability of disaster-bearing bodies. The flood and drought risk assessment model comprises two sub-models: flood and drought. The flood risk assessment sub-model selects extreme rainfall, topographic slope, river network density, soil saturated moisture content, and land use type as assessment indicators. The comprehensive weights of each indicator are determined using the analytic hierarchy process (AHP) and the entropy weight method, with extreme rainfall weighted at 0.23, river network density at 0.21, topographic slope at 0.18, soil saturated moisture content at 0.15, land use type at 0.13, and topographic aspect at 0.10. The flood risk index is calculated through indicator standardization and weighted summation, combined with a logistic regression model. The flood risk in the Ganjiang River basin is classified into four levels: low, medium, high, and extremely high, with the extremely high-risk area accounting for 7.8% of the total area. The drought risk assessment sub-model selects vegetation water supply and demand index, soil moisture anomaly, potential evapotranspiration to precipitation ratio, water availability, and irrigation guarantee rate as assessment indicators. Grey relational analysis is used to screen core indicators, with a correlation coefficient of 0.82 for vegetation water supply and demand index, 0.79 for soil moisture anomaly, and 0.76 for potential evapotranspiration to precipitation ratio. A classification model is constructed based on the support vector machine algorithm to output the drought risk level, with the extremely high-risk area accounting for 9.2% of the total area. To visually compare the correlation between the indicator weights and risk indices of the two sub-models, a three-dimensional surface plot of the correlation between the flood and drought risk indicator weights and indices in the Ganjiang River basin is generated, as shown below. Figure 4 As shown, the X-axis represents the combined classification of 6 flood indicators and 5 drought indicators, with clearly labeled abbreviations of indicator names. The Y-axis represents the comprehensive weight of the indicators, ranging from 0 to 0.3, and the Z-axis represents the hazard index, ranging from 0 to 1. The figure uses a smooth blue surface to represent the weight-index relationship of the flood sub-model and a smooth red surface to represent the weight-index relationship of the drought sub-model. The slope changes of the two surfaces visually reflect the impact of increasing indicator weights on the hazard index. Each grid node on the surface is labeled with its corresponding specific value. For example, in the flood sub-model, the extreme rainfall weight of 0.23 corresponds to an index node value of 0.89, and in the drought sub-model, the vegetation water supply and demand index weight of 0.27 corresponds to an index node value of 0.91. There are no topographical elements present; the data correlation characteristics are only presented through the surface morphology.

[0094] The disaster-prone environment exposure assessment model categorizes exposure carriers into four types: natural ecosystems, agricultural production land, urban built-up areas, and infrastructure. Exposure area, spatial distribution density, and distance from the disaster source are selected as assessment factors. A quantitative model is constructed, calculating the average disaster-prone environment exposure in the basin to be 0.42, with an average exposure of 0.76 for urban built-up areas and 0.28 for natural ecosystems. The disaster-bearing body vulnerability assessment model selects indicators from three dimensions: disaster resistance, resilience, and vulnerability. Fuzzy analytic hierarchy process (AHP) is used to determine the weights: population density (0.25), economic development level (0.22), infrastructure completeness (0.20), emergency response capacity (0.18), and ecological resilience index (0.15). The calculated average watershed vulnerability index is 45.3. To illustrate the numerical distribution patterns of exposure and vulnerability indices for the four types of exposure carriers, a two-dimensional scatter curve of disaster-prone environment exposure versus disaster-bearing body vulnerability in the Ganjiang River basin is obtained, as shown below. Figure 5 As shown, the X-axis represents the exposure to the disaster-prone environment, ranging from 0 to 1, and the Y-axis represents the vulnerability index of the disaster-bearing body, ranging from 0 to 100. The figure uses four types of scatter points—circles, triangles, squares, and hexagons—to represent four types of carriers: natural ecosystems, agricultural production land, urban built-up areas, and infrastructure, respectively. Each type of scatter point is the same color. Curve fitting was performed on the scatter points of each of the four types of carriers, resulting in four smooth fitting curves. The slope of the curves intuitively reflects the trend of the impact of increased exposure on the vulnerability index of different carriers. For example, the fitting curve for urban built-up areas has the steepest slope, indicating that for every 0.1 increase in exposure, the vulnerability index increases by an average of 12.3, clearly presenting the risk correlation characteristics of different carriers.

[0095] The three models were coupled using a risk coupling matrix and a multivariate nonlinear coupling algorithm. The constructed 3×3 risk coupling matrix clearly defines the interaction between flood and drought hazard (H), exposure to the disaster-prone environment (E), and vulnerability of the disaster-bearing body (V). The coupling degree between H and E is 0.85, H and V is 0.79, and E and V is 0.82. Based on the multivariate nonlinear coupling algorithm, the dynamic risk value of flood and drought in the basin was calculated to range from 0.05 to 0.92, with a mean of 0.38, and the area of ​​extremely high risk zone accounted for 6.7%. To present the quantitative correlation between the coupling degree of the three factors and the dynamic risk value, a three-dimensional histogram of the coupling degree of the three factors and the dynamic risk value in the Ganjiang River basin was generated, as shown below. Figure 6As shown, the X-axis represents the three coupling degree types: HE, HV, and EV; the Y-axis represents the coupling degree values, ranging from 0 to 1; and the Z-axis represents the dynamic risk value, also ranging from 0 to 1. The graph sets four sets of bar curves for each coupling degree type, corresponding to low, medium, high, and extremely high risk levels. These four sets of curves are distinguished by green, yellow, orange, and red colors, respectively. The height of the bar curve represents the average risk value at the corresponding risk level. For example, the extremely high risk level bar curve height corresponding to HE coupling degree 0.85 is 0.89, and the extremely high risk level bar curve height corresponding to HV coupling degree 0.79 is 0.83. The difference in curve height intuitively reflects the degree of influence of different coupling degrees on the risk value. The overall chart structure is clear, distinguishing risk levels only by the height and color of the bar curves, without any complex terrain elements.

[0096] A dynamic risk update mechanism was established. Based on time-series remote sensing imagery, land use change data, and real-time monitoring data within the basin, an incremental update algorithm was used to periodically update the fused dataset, achieving a change area identification accuracy of over 95%. A risk change detection model was constructed, with the calculated risk change detection index ranging from 0.02 to 0.35. Areas with significant changes (index ≥ 0.10) accounted for 12.3% of the total area. An incremental learning algorithm was used to iteratively optimize the model parameters, improving the model update efficiency by over 60%. To compare the efficiency differences between incremental updates and traditional full updates under different data scales, a risk update efficiency-data scale change curve for the Ganjiang River basin was obtained, as shown below. Figure 7 As shown, the X-axis represents the data update scale, ranging from 0-50%, and the Y-axis represents the update time, ranging from 0-120 minutes. The graph uses a solid blue line to represent the time consumption curve of the incremental update algorithm and a dashed red line to represent the time consumption curve of the full update algorithm. Both curves show a smooth downward trend. Key nodes are marked on the curves; for example, when the update scale is 10%, the incremental update takes 12 minutes and the full update takes 48 minutes; when the update scale is 30%, the incremental update takes 28 minutes and the full update takes 95 minutes. The spacing between the curves visually reflects the efficiency advantage of incremental updates at different data scales. The overall chart is concise and clear, presenting the technical effect solely through the comparison of the two curves.

[0097] Based on the risk assessment results, the area proportion, distribution area and core driving factors of each risk level are statistically analyzed. The specific data is shown in the table below. The key influencing factors and prevention and control priorities of different risk levels are identified by combining the data in the table.

[0098] Table 2. Statistics on the Risk Levels of Floods and Droughts in the Ganjiang River Basin

[0099] Risk level Risk value range Area percentage Main distribution area Core driving factors Low risk 0.05-0.20 42.3% Upstream mountainous areas and areas with high vegetation cover Low risk, low exposure, and high ecological resilience Medium risk 0.21-0.40 35.7% Midstream hilly area, transition zone between forest and cultivated land Moderate risk, moderate exposure High risk 0.41-0.60 15.3% Middle and lower reaches plain agricultural area High exposure, moderate vulnerability Extremely high risk 0.61-0.92 6.7% Urban built-up areas and densely river-networked areas in the middle and lower reaches High risk, high exposure, high vulnerability

[0100] To quantify the weighting of core driving factors at each risk level, a stacked curve of risk level-driving factor weighting in the Ganjiang River Basin was obtained, as shown below. Figure 8 As shown, the X-axis represents four risk levels: low, medium, high, and extremely high, while the Y-axis represents the weight percentage of driving factors, ranging from 0% to 100%. For each risk level, three stacked curve segments are set, representing the three driving factors: hazard, exposure, and vulnerability. These three curve segments are distinguished by blue, green, and red, respectively. The total height of the stacked curves is 100%, and the height of each curve segment represents the weight percentage of the corresponding driving factor. For example, in the extremely high risk level, the hazard curve segment accounts for 45% of the height, the exposure curve segment accounts for 35%, and the vulnerability curve segment accounts for 20%. In the low risk level, the hazard curve segment accounts for 15%, the exposure curve segment accounts for 25%, and the vulnerability curve segment accounts for 60%. The difference in curve segment heights intuitively reflects the dominant driving factors for different risk levels. The overall chart contains no terrain-related content; it only presents data characteristics through changes in the height of the stacked curves.

[0101] The system outputs visualized assessment results, constructs time-series risk change curves, and introduces a sliding window algorithm to smooth data fluctuations, clearly demonstrating the dynamic evolution trend of basin risk. It generates an integrated text and graphic risk assessment report, including risk level statistics, high-risk area identification, risk driver contribution rate analysis, and targeted disaster prevention and mitigation recommendations. To present the temporal evolution pattern of risk values ​​across the entire basin, a three-dimensional curve of the time-series change of dynamic risk values ​​for floods and droughts in the Ganjiang River basin is obtained, as shown below. Figure 9 As shown, this figure is a three-dimensional coordinate system surface curve. The X-axis represents the assessment time nodes, labeled sequentially according to the update cycle. The Y-axis represents the assessment unit number within the watershed, with a value range of 0-1000. The Z-axis represents the dynamic risk value, with a value range of 0-1. The figure uses a smooth three-dimensional surface curve to represent the temporal change trend of the risk value of the entire watershed. The fluctuations of the curve reflect the risk value fluctuations of each assessment unit at different time nodes. At the same time, the evolution trajectory of extremely high risk values ​​is marked with red highlighted lines. The direction of the highlighted lines intuitively presents the temporal change characteristics of high-risk areas. For example, at the 5th time node, the number of assessment units with extremely high risk values ​​increased from 32 to 58, and the peak value of the curve increased significantly. The overall chart only presents the risk evolution pattern through the shape of the surface curve and the highlighted lines, without any topographic elements.

[0102] This implementation and verification, through a large amount of measured data and precise model calculations, validated the effectiveness and applicability of this method in the dynamic risk assessment of floods and droughts in the Ganjiang River Basin. The assessment results are highly consistent with the actual disaster occurrence in the basin, and the accuracy rate of identifying extremely high-risk areas reaches over 92%, which can provide scientific support for the prevention and control of floods and droughts in the Ganjiang River Basin.

[0103] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any changes, modifications, substitutions, integrations, and parameter changes made to these embodiments within the spirit and principles of the present invention, without departing from the principles and spirit of the present invention, through conventional substitutions or to achieve the same function, fall within the scope of protection of the present invention.

Claims

1. A method for dynamic risk assessment of floods and droughts that integrates multi-source land information, characterized in that, include: Collect multi-dimensional land information source data, which includes remote sensing inversion data, land use status survey data, soil physicochemical properties data, topographic and geomorphological features data, and socio-economic statistics data; Heterogeneous data adaptation processing is performed on land information source data from various dimensions to construct a fusion dataset with unified spatial benchmarks and data standards. Based on the fused dataset, a flood and drought disaster risk assessment model, a disaster-prone environment exposure assessment model, and a disaster-bearing body vulnerability assessment model were constructed respectively. The dynamic risk assessment results of floods and droughts are obtained by coupling the aforementioned flood and drought disaster risk assessment model, disaster-prone environment exposure assessment model, and disaster-bearing body vulnerability assessment model. Establish a dynamic risk update mechanism, iteratively optimize the dynamic risk assessment results of floods and droughts based on time-updated land information data, and output visualized assessment results.

2. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The heterogeneous data adaptation processing of land information source data in various dimensions includes: A georeferencing model is used to spatially correct remote sensing inversion data, and a coordinate transformation algorithm is used to unify land information data from different projection coordinate systems to the target coordinate system. Anomaly detection algorithms are used to identify and remove outlier samples from soil physicochemical properties data and socioeconomic statistics data, and data interpolation methods are used to fill in missing values. The planar fitting method is used to calculate topographic factors such as slope, aspect, and roughness based on the digital elevation model. After geometric registration with the geomorphological data, topographic and geomorphological feature data are formed. A multi-source data weight allocation model is constructed, and the fusion weight of each data dimension is determined by combining the information contribution and reliability indicators of each data dimension. A fused dataset is generated through a feature-level fusion algorithm.

3. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The flood and drought hazard assessment model includes a flood hazard assessment sub-model and a drought hazard assessment sub-model. The construction process of the flood hazard assessment sub-model is as follows: Extreme rainfall, topographic slope, topographic aspect, river network density, soil saturated moisture content, and land use type were selected as indicators for flood risk assessment. The subjective weights of each evaluation indicator are determined based on the analytic hierarchy process (AHP), the objective weights of each indicator are calculated using the entropy weight method, and the comprehensive weights are determined by the coupled weight formula. A flood hazard index calculation model is constructed. Through index standardization, weighted summation, and index hierarchical mapping, the flood hazard assessment results are obtained. The formula for the flood hazard index calculation model is as follows: ,in, The flood risk index, For the first The combined weight of each indicator For the first The standardized value of each indicator, These are the model correction coefficients. To assess the number of indicators.

4. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 3, characterized in that, The construction process of the drought risk assessment sub-model is as follows: The vegetation water supply and demand index, soil moisture anomaly, potential evapotranspiration to precipitation ratio, water availability and irrigation guarantee rate were selected as indicators for drought risk assessment. A correlation matrix of indicators was constructed, and the grey relational analysis method was used to quantify the correlation between each indicator and drought disaster, and to screen the core influencing indicators. A drought hazard classification model is constructed based on the support vector machine algorithm. A high-dimensional feature space is mapped using a radial basis function kernel. The drought hazard classification model satisfies the following: ,in, The output results are the drought risk level. For the Lagrange multipliers of support vector machines, The class labels for the training samples. For radial basis kernel functions, , For kernel function parameters, Let be the feature vector of the sample to be evaluated. For the first Each training sample feature vector For the number of support vectors, For model bias terms; The model parameters are optimized using a training sample set, and standardized assessment index data is input to output drought risk level assessment results.

5. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The construction process of the environmental exposure assessment model for pregnancy-related disasters is as follows: The types of exposure carriers are classified, including natural ecosystems, agricultural production land, urban built-up areas, and infrastructure; For different types of exposure carriers, exposure area, spatial distribution density, and distance from the disaster source are selected as exposure assessment factors; An exposure quantification model is constructed, and the exposure intensity of each exposure carrier is calculated through spatial overlay analysis. Combined with the exposure sensitivity coefficient, the exposure assessment results of the disaster-prone environment are obtained. The formula for the exposure quantification model is as follows: ,in, For environmental exposure during pregnancy, For the first The area percentage of the surface area exposed to the carrier. For the first Distribution density of the type of exposure vector, For the first Sensitivity coefficient of the type of exposure vector, For the first The average distance between the exposed carrier and the disaster source. This is the distance attenuation coefficient. Number of exposed carrier types.

6. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The process of constructing the vulnerability assessment model for disaster-bearing bodies is as follows: Vulnerability assessment indicators were selected from three dimensions: disaster resistance, resilience, and vulnerability, including population density, economic development level, infrastructure completeness, emergency response capability, and ecological resilience index. The weights of each evaluation index are determined by fuzzy hierarchical analysis, and the index data are fuzzified by membership function to construct a fuzzy evaluation matrix. A comprehensive vulnerability assessment model is constructed to calculate the correlation degree and determine the vulnerability level. The formula for the comprehensive vulnerability assessment model is as follows: ,in, As a vulnerability index for disaster-bearing bodies, For the first The weight of each indicator, For the first The standardized value of each indicator, To assess the number of indicators, we obtain the vulnerability assessment results of the disaster-bearing body.

7. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The dynamic risk assessment results for floods and droughts obtained through the coupled computation of three models include: Construct a risk coupling matrix to clarify the interaction among the three factors: the risk of water and drought disasters, the exposure of disaster-prone environments, and the vulnerability of disaster-bearing bodies, and quantify the degree of coupling between each pair. A risk assessment model is established using a multivariate nonlinear coupling algorithm, introducing a coupling coordination coefficient to adjust the influence weights of the three factors. The formula for the multivariate nonlinear coupling algorithm is as follows: ,in, This represents the dynamic risk value for floods and droughts. The results of the flood and drought disaster risk assessment The results of the environmental exposure assessment for pregnancy-related disasters. The results of the vulnerability assessment of disaster-bearing bodies, , , These are the coefficients of the interaction term.

8. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The process of establishing the aforementioned dynamic risk update mechanism is as follows: A data update cycle is set, and based on remote sensing image time series data, land use change data and real-time monitoring data, an incremental update algorithm is used to periodically update the fused dataset, and only the data of the changed areas are reprocessed. A risk change detection model was constructed, and the spatial migration characteristics and temporal evolution patterns of flood and drought disaster risks were identified through a spatiotemporal difference analysis algorithm. Based on the characteristics of risk changes, an incremental learning algorithm is used to iteratively optimize the parameters of the dynamic risk assessment model for floods and droughts, without retraining the entire model, thus enabling rapid dynamic updates of risk assessment results.

9. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 8, characterized in that, The formula for setting the data update cycle is: ,in, This refers to the actual update cycle. As the baseline update cycle, For the first The rate of change of land-related information data For the first Weights of class data For the first Sensitivity of data type to risk assessment For the number of data types, For environmental stability coefficient, This represents the intensity coefficient of regional disaster activity.

10. The method for dynamic risk assessment of floods and droughts that integrates multi-source land information according to claim 1, characterized in that, The output of the visualization evaluation results includes: Geographic Information System (GIS) technology is used to generate a spatial distribution map of flood and drought disaster risk levels. Combined with a three-dimensional terrain model, the risk distribution is visualized in three dimensions. The spatial distribution differences of risks are presented intuitively through a hierarchical color scheme. A time-series risk change curve is constructed, and a sliding window algorithm is introduced to smooth data fluctuations, clearly showing the dynamic evolution trend of flood and drought disaster risks in different periods; Generate risk assessment reports, including risk level statistics, high-risk area identification, risk driver contribution rate analysis, and targeted disaster prevention and mitigation recommendations, and support integrated text and graphic output.

Images