A Model-Based Workflow-Based Method and System for Early Warning and Risk Assessment of Landslides in Natural Disasters

Abstract

This invention proposes a natural disaster landslide early warning risk assessment method and system based on a model workflow. The method involves sorting indicator factors and algorithm models at different scales; constructing and publishing a tool set; and constructing a natural disaster landslide early warning risk assessment workflow by invoking the indicator factor tool set and the model tool set according to risk assessment requirements. The natural disaster landslide early warning risk assessment workflow is then invoked to perform a risk assessment, generating a model result dataset. Based on this model result dataset, the system is published as a WMS map service using different color schemes corresponding to different risk values. This invention allows for the flexible construction of different landslide hazard models, enabling better protection against natural disaster landslides.

Description

Technical Field

[0001] This invention belongs to the field of natural disaster risk assessment technology, and in particular relates to a natural disaster landslide early warning risk assessment method and system based on model workflow. Background Art

[0002] Landslides are a common geological hazard, characterized by their suddenness, destructive power, and wide-ranging impact. In recent years, landslides have occurred frequently, causing severe casualties and property losses. Current research on landslide disaster assessment models remains focused on causative factors and model algorithms. Early information systems were mostly based on client / server architecture and GIS component technology, which have the following drawbacks:

[0003] 1. File-based storage is primarily used for data storage;

[0004] 2. The model lacks flexibility in constructing different models for the hazard-causing factors;

[0005] 3. The systems are mostly built for evaluating models of a single type of natural disaster, and lack universality;

[0006] 4. The model evaluation relies heavily on manual intervention and lacks automation.

[0007] 5. The technology is relatively outdated, and the maintenance costs are high. Summary of the Invention

[0008] This invention proposes a model-based workflow-based method and system for risk assessment of landslide early warning in natural disasters. It can flexibly construct different landslide disaster models and better protect against landslides.

[0009] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0010] A model-based workflow-based method for early warning and risk assessment of natural disasters such as landslides includes:

[0011] S1. The index factors involved in the algorithms for assessing the hazard of disaster-causing factors, sensitivity of the disaster-prone environment, vulnerability of the landslide itself, and comprehensive risk of landslides are sorted out according to different scales; the index factor data is stored in PostGIS according to the rules based on the different types of data, and real-time data is accessed through API;

[0012] S2. Algorithm models for analyzing the hazard of disaster-causing factors, sensitivity of the disaster-prone environment, vulnerability of the landslide itself, and comprehensive risk of landslides according to different scales;

[0013] S3. Build and publish a toolset, which includes an indicator factor toolset and a model toolset. The indicator factor toolset is a data preprocessing toolset for the indicator factors, and the model toolset is a toolset for algorithmic models of disaster-causing factor hazard, disaster-prone environment sensitivity, landslide vulnerability, and comprehensive landslide risk.

[0014] S4. According to the risk assessment requirements, call the indicator factor tool set and model tool set, and construct the natural disaster landslide early warning risk assessment workflow by outputting the indicator factor tool set to three model tool sets, including the disaster-causing factor hazard model tool set, the disaster-causing environment sensitivity model tool set, and the landslide vulnerability model tool set, and outputting the three model tool sets to the landslide comprehensive risk model tool set.

[0015] S5. Call the natural disaster landslide early warning risk assessment workflow to conduct a risk assessment and obtain the model result dataset;

[0016] S6. Based on the model result dataset, and combined with different color schemes corresponding to different risk values, publish it as a WMS map service.

[0017] Furthermore, in step S1, the indicator factors include:

[0018] Disaster-causing factors: Indicators of the hazard of disaster-causing factors, including rainfall, earthquake intensity, geological conditions, slope, and aspect;

[0019] Disaster-prone factors: Indicators of environmental sensitivity to disasters, including geological structure, groundwater level changes, and vegetation cover.

[0020] Intrinsic vulnerability factors: Indicators of landslide vulnerability, including building density, population density, and infrastructure distribution;

[0021] Comprehensive landslide risk: The comprehensive landslide risk is obtained by combining the hazard of disaster-causing factors, the sensitivity of the disaster-prone environment, and the vulnerability of the landslide itself.

[0022] Furthermore, in step S2, the algorithm model includes:

[0023] Hazardous factor risk model: Based on the hazardous factors, the random forest or support vector machine algorithm is used for evaluation;

[0024] Disaster-prone environment sensitivity model: Based on disaster-prone factors, the model is evaluated using logistic regression or Bayesian network algorithms;

[0025] Landslide vulnerability model: Based on the vulnerability factors of building density and population density, the model is evaluated using multi-criteria decision analysis, comprehensive index method or fuzzy matter-element evaluation method;

[0026] Comprehensive landslide risk model: Combining the hazard model of disaster-causing factors, the sensitivity model of disaster-prone environment, and the vulnerability model of landslide itself, a comprehensive assessment is conducted using the weighted superposition method or the analytic hierarchy process.

[0027] Furthermore, the toolset construction method in step S3 includes:

[0028] A utility base class is set as the abstract base class, providing internationalization support through i18nFunction(), and a unified interface Execute() is defined, requiring utility implementation classes, as subclasses, to implement the core logic. Utility implementation classes inherit from the utility base class, overriding the Execute() method to operate on the DatasetVector class for computation. A factory base class dynamically binds the relationship between utility implementation classes and the base class through an SPI registration mechanism. The factory base class instantiates utility implementation classes based on the registration information and calls their Execute() method to execute tasks, while the utility base class provides the common logic.

[0029] In another aspect, this invention proposes a natural disaster landslide early warning risk assessment system based on model workflow, comprising:

[0030] Factor module: The module organizes the indicator factors involved in the algorithms for assessing the hazard of disaster-causing factors, sensitivity to disaster-prone environments, vulnerability of landslides, and comprehensive risk of landslides according to different scales; and stores the indicator factor data in PostGIS according to the rules based on the different types of data, with real-time data accessed via API.

[0031] Model module: Algorithm models for assessing the hazard of disaster-causing factors, sensitivity to the disaster-prone environment, vulnerability of the landslide itself, and comprehensive landslide risk at different scales;

[0032] Toolset Module: Constructs and publishes toolsets, which include an indicator factor toolset and a model toolset. The indicator factor toolset is a data preprocessing toolset for the indicator factors, and the model toolset is a toolset for algorithmic models of disaster-causing factor hazard, disaster-prone environment sensitivity, landslide vulnerability, and comprehensive landslide risk.

[0033] Workflow module: According to the risk assessment requirements, the indicator factor tool set and model tool set are called. The output of the indicator factor tool set is sent to three model tool sets, including the disaster-causing factor hazard model tool set, the disaster-causing environment sensitivity model tool set, and the landslide vulnerability model tool set. The output of the three model tool sets is sent to the landslide comprehensive risk model tool set in the following order to construct the natural disaster landslide early warning risk assessment workflow.

[0034] Calling module: Calls the natural disaster landslide early warning risk assessment workflow to perform risk assessment and obtain the model result dataset;

[0035] Service publishing module: Based on the model result dataset, and combined with different color schemes corresponding to different risk values, publish it as a WMS map service.

[0036] Furthermore, in the factor module, the indicator factors include:

[0037] Disaster-causing factors: Indicators of the hazard of disaster-causing factors, including rainfall, earthquake intensity, geological conditions, slope, and aspect;

[0038] Disaster-prone factors: Indicators of environmental sensitivity to disasters, including geological structure, groundwater level changes, and vegetation cover.

[0039] Intrinsic vulnerability factors: Indicators of landslide vulnerability, including building density, population density, and infrastructure distribution;

[0040] Comprehensive landslide risk: The comprehensive landslide risk is obtained by combining the hazard of disaster-causing factors, the sensitivity of the disaster-prone environment, and the vulnerability of the landslide itself.

[0041] Furthermore, in the model module, the algorithm model includes:

[0042] Hazardous factor risk model: Based on the hazardous factors, the random forest or support vector machine algorithm is used for evaluation;

[0043] Disaster-prone environment sensitivity model: Based on disaster-prone factors, the model is evaluated using logistic regression or Bayesian network algorithms;

[0044] Landslide vulnerability model: Based on the vulnerability factors of building density and population density, the model is evaluated using multi-criteria decision analysis, comprehensive index method or fuzzy matter-element evaluation method;

[0045] Comprehensive landslide risk model: Combining the hazard model of disaster-causing factors, the sensitivity model of disaster-prone environment, and the vulnerability model of landslide itself, a comprehensive assessment is conducted using the weighted superposition method or the analytic hierarchy process.

[0046] Furthermore, in the toolset module, building the toolset includes setting a tool base class as an abstract base class, providing internationalization support through i18nFunction(), and defining a unified interface Execute(), requiring tool implementation classes as subclasses to implement the core logic; setting tool implementation classes to inherit from the tool base class, overriding the Execute() method to operate on the DatasetVector vector dataset class to implement computation; setting a factory base class to dynamically bind the relationship between tool implementation classes and the base class through the SPI registration mechanism, the factory base class instantiates tool implementation classes according to the registration information, calls their Execute() method to execute tasks, while the tool base class provides common logic.

[0047] Compared with existing technologies, the landslide early warning risk assessment method and system based on model workflow proposed in this invention have the following beneficial effects:

[0048] 1. Improved assessment accuracy: By accessing meteorological and geological data in real time, the accuracy and effectiveness of landslide disaster risk assessment have been improved;

[0049] 2. Enhanced system flexibility: Users can customize the parameters involved in the calculation, flexibly construct different landslide disaster models, and adapt to different application scenarios;

[0050] 3. Reduced maintenance costs: Based on the SuperMap iDesktopX extended development API interface, the system architecture is simple and the maintenance cost is low;

[0051] 4. Enhanced early warning capabilities: The system automatically monitors data changes and updates risk assessment results in real time, providing strong support for landslide disaster early warning and effectively protecting people's lives and property. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the toolset data flow according to an embodiment of the present invention;

[0054] Figure 3 This is a UML diagram of the toolkit for constructing PGA indicator factors according to an embodiment of the present invention;

[0055] Figure 4 This is a schematic diagram of the model workflow construction process according to an embodiment of the present invention. Detailed Implementation

[0056] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0057] To make the purpose and features of this invention patent more apparent and understandable, the invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0058] This embodiment is based on the SuperMap GIS platform framework and uses iDesktopXJava as the API interface to build a workflow for landslide disaster risk assessment models covering different scales. SuperMap iserver publishes the model workflow as a service. The browser calls the published model service, automatically calculates the risk value, and automatically publishes the risk map service with different color symbols according to different risk values.

[0059] The method proposed in this embodiment is as follows: Figure 1 As shown, it includes:

[0060] Step 1: Organize the data and indicator factors involved in the algorithms for assessing the hazard of disaster-causing factors, the sensitivity of the disaster-prone environment, the vulnerability of the landslide itself, and the risk of the comprehensive landslide model according to different scales; the different scales mainly refer to different administrative spatial scales, such as the data and indicator factors of landslide locations in all spatial locations within a province, county, or township.

[0061] Depending on the type of data, PostGIS stores data according to certain naming rules. In one example, the naming rules are as follows: dataset abbreviation (e.g., the abbreviation for earthquake intensity, PGA) + year, month, day, hour, minute, second - administrative division code. The naming rules for indicators, factors, and models are the same.

[0062] Real-time data is accessed via API to ensure its timeliness and accuracy. For example, meteorological data comes from the API interface provided by the National Meteorological Science Data Center, and the meteorological data mainly involves rainfall, wind force, and temperature.

[0063] Disaster-causing factors: Indicators of the hazard of disaster-causing factors, including rainfall, earthquake intensity, geological conditions, slope, and aspect;

[0064] Disaster-prone factors: Indicators of environmental sensitivity to disasters, including geological structure, groundwater level changes, and vegetation cover.

[0065] Intrinsic vulnerability factors: Indicators of landslide vulnerability, including building density, population density, and infrastructure distribution;

[0066] Comprehensive landslide risk results: The comprehensive landslide risk is obtained by combining the results of the hazard model of hazard-causing factors, the results of the sensitivity of the disaster-causing environment, and the results of the vulnerability of the landslide itself.

[0067] Step 2: Analyze the algorithms for disaster-causing factor hazard, disaster-prone environment sensitivity, landslide vulnerability, and comprehensive landslide risk model according to different scales;

[0068] Hazard risk model for disaster-causing factors: Based on disaster-causing factors such as rainfall and earthquake intensity, machine learning algorithms such as random forest and support vector machine are used for evaluation.

[0069] Disaster-prone environment sensitivity model: Based on disaster-prone factors such as geological structure and groundwater level changes, the model is evaluated using algorithms such as logistic regression and Bayesian networks.

[0070] Landslide vulnerability model: Based on factors such as building density and population density, the model is assessed using methods such as multi-criteria decision analysis (MCDA).

[0071] Comprehensive landslide risk model: Combining disaster-causing model, disaster-prone model and ontological vulnerability model, a comprehensive assessment is conducted using methods such as weighted superposition method and analytic hierarchy process (AHP).

[0072] Step 3: Based on the desktop GIS software SuperMap iDesktopX extension development, build toolsets according to the different indicator factors, disaster-causing factor hazard models, disaster-prone environment sensitivity models, landslide vulnerability models, and landslide comprehensive risk models as sorted above, and set up a "toolbox". The completed toolsets are placed in the toolbox according to the directory for management.

[0073] Each indicator / factor / model toolset must meet the requirements for data input and data output, such as... Figure 2 The diagram illustrates the data flow of each toolset, with each toolset configured based on its intended functionality. The indicator factor toolset takes as input the raw data corresponding to that indicator factor, encapsulates the preprocessing algorithm logic for that raw data, and outputs the preprocessed data results based on the algorithm logic. Taking the PGA (Pulse Gauge Index) toolet as an example, the input is the raw data about PGA. This raw data is input into the PGA indicator factor toolset, where it undergoes preprocessing algorithm logic (which may include detrending and mean removal, filtering, resampling, Fourier transform, vector transformation, etc.). Through the application of these preprocessing algorithms, preprocessed PGA data is obtained, effectively improving the quality of the PGA data and converting unsuitable data for model analysis into vectors, thus improving the accuracy of subsequent model analysis.

[0074] The model toolset takes as input the output of multiple indicator factor toolsets, encapsulates the model algorithm logic for that toolset, and outputs the model analysis results based on that algorithm logic. Figure 4 Taking the disaster-prone environment sensitivity model toolset as an example, the input data includes: the results output by the geomorphic type index factor toolset, the results output by the vegetation coverage index factor, and the results output by other related index factor toolsets (such as groundwater level changes); after being processed by the disaster-prone environment sensitivity model algorithm logic (optional logistic regression, Bayesian network, etc.) to perform classification and probability prediction, the disaster-prone probability data is output.

[0075] The process of building toolsets is as follows Figure 3 As shown, Figure 3 Taking the PGA indicator factor toolkit as an example, the construction process includes:

[0076] A `BaseProcess` class (utility base class) is set up as the abstract base class, providing internationalization support through `i18nFunction()` and defining a unified interface (such as `Execute()`). Subclasses (utility implementation classes) are required to implement the core logic. A `PGAProcessFactor` class (utility implementation class) inherits from `BaseProcess`, overriding the `Execute()` method to operate on the `DatasetVector` vector dataset class, implementing the specific preprocessing calculation of PGA indicator factors. A `PGAFactory` class (factory base class) dynamically binds the relationship between the utility implementation class and the base class through the SPI registration mechanism. `AbstractAnnotatedProcessFactory` and its annotations (such as `AnnotatedProcessBean`) manage the metadata of the PGA indicator factor toolset, achieving object creation and decoupling. The specific process is as follows: the factory base class instantiates `PGAProcessFactor` based on the registration information and calls its `Execute()` method to execute the task, while the utility base class `BaseProcess` provides common logic (such as input validation). This design, through SPI extension and the factory pattern, achieves efficient registration, flexible expansion, and unified invocation of processing tools.

[0077] The process of building the model toolset is the same as that of building the indicator factor toolset above, and will not be repeated here.

[0078] After completing the construction of the indicator factor toolset and model toolset, publish all the toolsets as JAR files and place them in the lib path of SuperMap iDesktopX. Restart SuperMap iDesktopX and the developed toolsets will be displayed in the toolbox.

[0079] Step 4: Based on the indicator factor toolset and model toolset built in Step 3, in the processing automation module of the desktop GIS software SuperMap iDesktopX, drag the corresponding indicator factor toolset and model toolset into the current window according to the desired model results. Construct a workflow based on the order of outputting the indicator factor toolset to the three major model toolsets (hazard factor risk model toolset, disaster-prone environment sensitivity model toolset, and landslide vulnerability model toolset), and outputting the three major model toolsets to the landslide comprehensive risk model toolset. After completion, run and test the feasibility of the model and verify the accuracy of the model results in SuperMap iDesktopX. If there are no problems in the above steps, fill in the SuperMap iServer connection information on the SuperMap iDesktopX publishing page and click Publish to publish it as a RESTful model service. Compared with previous technologies, model calculations are usually performed manually and distributed, and the final result is calculated. There is no publishing as a service for invocation. This method of building a model workflow and publishing it as a callable service will greatly shorten the model running time, reduce manual intervention, and facilitate the model service to be called on other platforms. See the model workflow construction process. Figure 4 As can be seen, administrative divisions serve as an input to each set of indicator factors, implying that different administrative divisions represent different scales, requiring the selection of indicator factor data at the corresponding scale for input.

[0080] Step 5: Call the model service published in Step 4 from the web system frontend to generate the model dataset in the database. Previously, when calling the model service, the process from raw data input to output was typically completed in one step, focusing on the model result data while neglecting to store or record process data. In this model service call, process data, including indicator factors and model data, will also be generated synchronously in the database during model generation, allowing for later analysis and traceability of the model results.

[0081] Step 6: Combine different color schemes corresponding to different risk values ​​and publish as a WMS map service.

[0082] Based on the model result dataset generated in step 5, a column of attribute values ​​will represent the risk value of the current model within the spatial range of the current administrative division (province, district / county, township). Different risk values ​​(7-10 red, 4-7 yellow, 6-3 blue, 0-3 transparent) represent high risk, medium risk, low risk, and no risk, respectively. An SLD style file supporting the GeoServer WMS service is generated in QGIS. Add the GeoServer-Manager development dependency to the Java backend program, configure the workspace, database connection information, and the generated SLD, and automatically publish it as a WMS service. Compared to previous technologies, displaying risk values ​​in different styles more intuitively shows the risk level, while automatically publishing as a WMS service replaces the tedious manual service publishing, and can also be shared as a service.

[0083] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for risk assessment of landslide early warning in natural disasters based on model workflow, characterized in that, include: S1. The index factors involved in the algorithm for sorting out the hazard of disaster-causing factors, the sensitivity of disaster-prone environment, the vulnerability of landslide itself, and the comprehensive risk of landslide according to different scales; Based on the different types of indicator factor data, they are stored in PostGIS according to the naming rules, and real-time data is accessed via API; S2. Algorithm models for analyzing the hazard of disaster-causing factors, sensitivity of the disaster-prone environment, vulnerability of the landslide itself, and comprehensive risk of landslides according to different scales; S3. Construct and publish a toolset, which includes an indicator factor toolset and a model toolset. The indicator factor toolset is a data preprocessing toolset for the indicator factor. The input is the original acquired data corresponding to the indicator factor, the encapsulation is the preprocessing algorithm logic for the original acquired data of the indicator factor, and the output is the preprocessed data result based on the algorithm logic. The model toolset is a toolset for algorithmic models of disaster-causing factor hazard, disaster-prone environment sensitivity, landslide vulnerability, and comprehensive landslide risk. The input is the output results of multiple indicator factor toolsets, the encapsulation is the model algorithm logic for this model toolset, and the output is the model analysis result based on the model algorithm logic. S4. According to the risk assessment requirements, call the indicator factor tool set and model tool set, and construct the natural disaster landslide early warning risk assessment workflow by outputting the indicator factor tool set to three model tool sets, including the disaster-causing factor hazard model tool set, the disaster-causing environment sensitivity model tool set, and the landslide vulnerability model tool set, and outputting the three model tool sets to the landslide comprehensive risk model tool set. S5. Call the natural disaster landslide early warning risk assessment workflow to conduct a risk assessment and obtain the model result dataset; S6. Based on the model result dataset, and combined with different color schemes corresponding to different risk values, publish it as a WMS map service.

2. The landslide early warning risk assessment method based on model workflow according to claim 1, characterized in that, In step S1, the indicator factors include: Disaster-causing factors: Indicators of the hazard of disaster-causing factors, including rainfall, earthquake intensity, geological conditions, slope, and aspect; Disaster-prone factors: Indicators of environmental sensitivity to disasters, including geological structure, groundwater level changes, and vegetation cover. Intrinsic vulnerability factors: Indicators of landslide vulnerability, including building density, population density, and infrastructure distribution; Comprehensive landslide risk: The comprehensive landslide risk is obtained by combining the hazard of disaster-causing factors, the sensitivity of the disaster-prone environment, and the vulnerability of the landslide itself.

3. The landslide early warning risk assessment method based on model workflow according to claim 1, characterized in that, In step S2, the algorithm model includes: Hazardous factor risk model: Based on the hazardous factors, the random forest or support vector machine algorithm is used for evaluation; Disaster-prone environment sensitivity model: Based on disaster-prone factors, the model is evaluated using logistic regression or Bayesian network algorithms; Landslide vulnerability model: Based on the vulnerability factors of building density and population density, the model is evaluated using multi-criteria decision analysis, comprehensive index method or fuzzy matter-element evaluation method; Comprehensive landslide risk model: Combining the hazard model of disaster-causing factors, the sensitivity model of disaster-prone environment, and the vulnerability model of landslide itself, a comprehensive assessment is conducted using the weighted superposition method or the analytic hierarchy process.

4. The landslide early warning risk assessment method based on model workflow according to claim 1, characterized in that, The toolset construction method in step S3 includes: A utility base class is set as the abstract base class, providing internationalization support through i18nFunction(), and a unified interface Execute() is defined, requiring utility implementation classes, as subclasses, to implement the core logic. Utility implementation classes inherit from the utility base class, overriding the Execute() method to operate on the DatasetVector class for computation. A factory base class dynamically binds the relationship between utility implementation classes and the base class through an SPI registration mechanism. The factory base class instantiates utility implementation classes based on the registration information and calls their Execute() method to execute tasks, while the utility base class provides the common logic.

5. A natural disaster landslide early warning risk assessment system based on model workflow, characterized in that, include: Factor module: The module organizes the indicator factors involved in the algorithms for assessing the hazard of disaster-causing factors, sensitivity to disaster-prone environments, vulnerability of landslides, and comprehensive risk of landslides according to different scales; and stores the indicator factor data in PostGIS according to the rules based on the different types of data, with real-time data accessed via API. Model module: Algorithm models for assessing the hazard of disaster-causing factors, sensitivity to the disaster-prone environment, vulnerability of the landslide itself, and comprehensive landslide risk at different scales; Toolset Module: This module builds and publishes toolsets, including an indicator factor toolset and a model toolset. The indicator factor toolset is a data preprocessing toolset for the indicator factor. It takes the raw data corresponding to the indicator factor as input, encapsulates the preprocessing algorithm logic for the raw data of that indicator factor, and outputs the preprocessed data results based on the algorithm logic. The model toolset is a toolset for algorithmic models of disaster-causing factor hazard, disaster-prone environment sensitivity, landslide vulnerability, and comprehensive landslide risk. It takes the output results of multiple indicator factor toolsets as input, encapsulates the model algorithm logic for this model toolset, and outputs the model analysis results based on the model algorithm logic. Workflow module: According to the risk assessment requirements, the indicator factor tool set and model tool set are called. The output of the indicator factor tool set is sent to three model tool sets, including the disaster-causing factor hazard model tool set, the disaster-causing environment sensitivity model tool set, and the landslide vulnerability model tool set. The output of the three model tool sets is sent to the landslide comprehensive risk model tool set in the following order to construct the natural disaster landslide early warning risk assessment workflow. Calling module: Calls the natural disaster landslide early warning risk assessment workflow to perform risk assessment and obtain the model result dataset; Service publishing module: Based on the model result dataset, and combined with different color schemes corresponding to different risk values, publish it as a WMS map service.

6. The landslide early warning risk assessment system for natural disasters based on model workflow according to claim 5, characterized in that, In the factor module, the indicator factors include: Disaster-causing factors: Indicators of the hazard of disaster-causing factors, including rainfall, earthquake intensity, geological conditions, slope, and aspect; Disaster-prone factors: Indicators of environmental sensitivity to disasters, including geological structure, groundwater level changes, and vegetation cover. Intrinsic vulnerability factors: Indicators of landslide vulnerability, including building density, population density, and infrastructure distribution; Comprehensive landslide risk: The comprehensive landslide risk is obtained by combining the hazard of disaster-causing factors, the sensitivity of the disaster-prone environment, and the vulnerability of the landslide itself.

7. The landslide early warning risk assessment system for natural disasters based on model workflow according to claim 5, characterized in that, In the model module, the algorithm model includes: Hazardous factor risk model: Based on the hazardous factors, the random forest or support vector machine algorithm is used for evaluation; Disaster-prone environment sensitivity model: Based on disaster-prone factors, the model is evaluated using logistic regression or Bayesian network algorithms; Landslide vulnerability model: Based on the vulnerability factors of building density and population density, the model is evaluated using multi-criteria decision analysis, comprehensive index method or fuzzy matter-element evaluation method; Comprehensive landslide risk model: Combining the hazard model of disaster-causing factors, the sensitivity model of the disaster-prone environment, and the vulnerability model of the landslide itself, a comprehensive assessment is conducted using the weighted superposition method or the analytic hierarchy process (AHP). 。 8. The landslide early warning risk assessment system for natural disasters based on model workflow according to claim 5, characterized in that, In the toolset module, building the toolset involves setting a tool base class as an abstract base class, providing internationalization support through i18nFunction(), and defining a unified interface Execute(), requiring tool implementation classes, as subclasses, to implement the core logic; setting tool implementation classes to inherit from the tool base class, overriding the Execute() method to operate on the DatasetVector vector dataset class to perform calculations; and setting a factory base class to dynamically bind the relationship between tool implementation classes and the base class through the SPI registration mechanism. The factory base class instantiates tool implementation classes based on the registration information and calls their Execute() method to execute tasks, while the tool base class provides the common logic.

Images