Regional landslide disaster meteorological risk early warning method and device and storage medium

Abstract

The invention discloses a regional landslide disaster meteorological risk early warning method and device and a storage medium, and belongs to the technical field of geological disaster early warning. The regional landslide disaster meteorological risk early warning method comprises the following steps: acquiring a disaster pregnancy sensitivity grade of each evaluation unit in a target region, and comprises the following steps: S1, according to the target region, selecting a reference region, the target region being located in the reference region, or the spatial similarity between the target region and the reference region being higher than a first threshold value, and if the spatial similarity between the target region and the reference region is higher than the first threshold value, determining that the disaster pregnancy sensitivity grade is lower than the first threshold value; the landslide sample of the reference area is greater than a second threshold, and the rainfall record of the reference area is greater than a third threshold; s2, constructing a disaster-pregnancy sensitivity evaluation model of the target area based on the disaster-pregnancy sensitivity evaluation model of the reference area; and S3, performing disaster-pregnancy sensitivity grade division on each evaluation unit in the target area by using the disaster-pregnancy sensitivity evaluation model of the target area. According to the invention, under the condition that the region lacks historical landslide disaster event records, refined regional landslide disaster meteorological risk early warning and forecasting can still be realized.

Description

Technical Field

[0001] This invention relates to the field of geological disaster early warning technology, and in particular to a method, equipment and storage medium for early warning of meteorological risks of regional landslide disasters. Background Technology

[0002] Besides being related to specific topography and geological conditions, rainfall is a significant factor in triggering landslide geological disasters. Regional landslide geological disaster meteorological risk early warning is a crucial technical means to shift from passive disaster relief to proactive prevention. The core of this approach involves constructing a refined regional landslide geological disaster meteorological risk early warning model based on the regional geological environment and disaster-prone characteristics.

[0003] Previously, landslide meteorological risk early warning models were built for larger regions such as provinces and cities. However, with the increasing demands for disaster prevention and mitigation, there is a need to build more refined early warning and forecasting models for smaller scales such as districts, counties, and townships. As the regional scale becomes more refined, factors such as the lack of large-scale field geological surveys, a small number of historical landslide samples, and unclear historical landslide occurrence times in a particular area can lead to a scarcity of disaster event samples suitable for model building. Consequently, it becomes impossible to construct a refined regional landslide meteorological risk early warning model, let alone wait until landslide records have accumulated to a certain extent before carrying out relevant landslide disaster early warning and forecasting work. Summary of the Invention

[0004] The purpose of this invention is to provide a method, device and storage medium for early warning of meteorological risks of regional landslide disasters, in view of the existing technical status, so as to achieve refined early warning and forecasting of meteorological risks of regional landslide disasters even in the absence of historical records of landslide disaster events in the region.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for early warning of meteorological risks of regional landslide disasters includes the following steps: Construct a three-dimensional criterion matrix in a three-dimensional coordinate system, which includes three dimensions: disaster sensitivity level, rainfall-induced disaster warning level, and short-term predicted rainfall. Obtain the disaster-prone sensitivity level, rainfall-induced disaster warning level, and short-term predicted rainfall for each evaluation unit in the target area. Based on the constructed three-dimensional criterion matrix, derive the landslide disaster meteorological risk warning level for each evaluation unit. The landslide disaster meteorological risk warning levels of each evaluation unit in the target area are merged to obtain the landslide disaster meteorological risk warning level of the target area; The steps to obtain the pregnancy susceptibility level for each evaluation unit in the target area are as follows: S1. Select a reference area based on the target area, and the target area is located within the reference area, or the spatial similarity between the target area and the reference area is higher than the first threshold, the landslide samples in the reference area are greater than the second threshold, and the rainfall records in the reference area are greater than the third threshold. S2. Construct a pregnancy sensitivity assessment model for the target area based on the pregnancy sensitivity assessment model of the reference area. The steps are as follows: S2.1. Taking each evaluation unit in the reference area as the evaluation object, select historical disaster points and non-historical disaster points as positive and negative samples respectively, and select several basic environmental factors. Use a deep neural network to construct an initial disaster-prone sensitivity evaluation model. Train the initial disaster-prone sensitivity evaluation model through basic environmental factors and positive and negative sample sets, save the network model structure and network parameters after training, and obtain the disaster-prone sensitivity evaluation model of the reference area. S2.2 Obtain the disaster-prone sensitivity evaluation model of the reference area. Add a geological environment regional feature adjuster before the output layer. One of the input data of the geological environment regional feature adjuster is the spatial similarity between the target area and the reference area. Take each evaluation unit in the target area as the evaluation object. Select historical disaster points and non-historical disaster points as positive samples and negative samples, respectively. Select the same basic environmental factors as in S2.1. Retrain the disaster-prone sensitivity evaluation model of the reference area. During training, freeze the network parameters saved in S2.1 and train only the geological environment regional feature adjuster to obtain the disaster-prone sensitivity evaluation model of the target area. S3. Use the pregnancy and fertility sensitivity assessment model of the target area to classify the pregnancy and fertility sensitivity level of each assessment unit in the target area.

[0006] Furthermore, the steps to obtain the rainfall-induced disaster warning level for each evaluation unit in the target area are as follows: Collect existing rainfall events and landslide disaster data for the target area, filter rainfall events related to landslide disasters (i.e., disaster-causing rainfall events), and the remaining events are rainfall events unrelated to landslides (i.e., non-disaster-causing rainfall events); The cumulative rainfall and duration of each non-disastrous rainfall event were statistically analyzed and plotted on a logarithmic coordinate system. Based on the distribution of different non-disastrous rainfall events, linear warning functions for different warning levels were fitted. The fitting formulas for the warning functions are as follows: E=αD β +C Where E is the cumulative rainfall, D is the rainfall duration, and α, β, and C are constants; By inputting the cumulative rainfall and rainfall history of each evaluation unit in the target area into a logarithmic coordinate system, and based on the linear early warning function of different early warning levels, the rainfall-induced disaster early warning level of each evaluation unit in the target area is obtained.

[0007] Furthermore, the evaluation unit is a grid unit or a landslide unit.

[0008] Furthermore, the spatial similarity between the target region and the reference region is calculated as follows: Within the target region and the reference region, cluster analysis is performed to divide each region into multiple clusters, corresponding to different spatial types; Obtain the centroid vector of each cluster and construct a matrix representing the geological environment characteristics of different regions; The cosine similarity between the geological environment feature matrix of the target area and the geological environment feature matrix of the reference area is calculated as the spatial similarity between the two.

[0009] Furthermore, when selecting non-historical disaster points as negative samples, random selection or semi-supervised selection methods are used.

[0010] Furthermore, the basic environmental factors include elevation, slope, aspect, plane curvature, profile curvature, topographic relief, soil and rock type, normalized vegetation index, normalized building index, total gully density, topographic humidity index, modified normalized differential water index, and road density.

[0011] Furthermore, before training the landslide disaster-prone sensitivity assessment model using basic environmental factors and positive and negative sample sets, one of the two basic environmental factors with a correlation greater than the fourth threshold is removed.

[0012] Furthermore, before training the landslide hazard sensitivity assessment model using basic environmental factors and positive and negative sample sets, the basic environmental factors are quantified using frequency ratio or weight of evidence methods.

[0013] A regional landslide disaster meteorological risk early warning device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aforementioned regional landslide disaster meteorological risk early warning method.

[0014] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for early warning of meteorological risks of regional landslide disasters.

[0015] The beneficial effects of this invention are as follows: This invention discloses a method, device, and storage medium for early warning of meteorological risks of regional landslide disasters. Even in the absence of historical landslide disaster records in a region, it can still achieve refined early warning and forecasting of meteorological risks of regional landslide disasters by constructing a disaster-prone sensitivity evaluation model of the target region based on a reference region through intelligent algorithms. This safeguards the safety of people's lives and property and provides effective support for geological disaster prevention and mitigation. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the process of obtaining the disaster-prone sensitivity level of each evaluation unit in a regional landslide disaster meteorological risk early warning method according to the present invention. Figure 2 This is a schematic diagram of a landslide disaster meteorological risk early warning method of the present invention, which constructs a disaster-prone sensitivity evaluation model for the target area based on a reference area; Figure 3 This is a flowchart illustrating a method for early warning of meteorological risks of regional landslide disasters according to the present invention. Figure 4 This is a schematic diagram of the three-dimensional criterion matrix in the meteorological risk early warning method for regional landslide disasters of the present invention; Figure 5 This is a schematic diagram of the process for obtaining the rainfall-induced disaster warning level of each evaluation unit in a regional landslide disaster meteorological risk early warning method of the present invention; Figure 6 This is a logarithmic coordinate system used to obtain the rainfall-induced disaster warning level of each evaluation unit in the target area in the regional landslide disaster meteorological risk early warning method of the present invention (without filtering rainfall events related to landslide disasters). Figure 7 This is a logarithmic coordinate system used in the method for early warning of meteorological risks of regional landslide disasters in this invention to obtain the rainfall-induced disaster warning level of each evaluation unit in the target area (rainfall events related to landslide disasters have been filtered out). Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific examples described herein are merely illustrative and not intended to limit the scope of the invention.

[0018] Please see Figures 3-4 As shown, a method for early warning of meteorological risks of regional landslide disasters includes the following steps: Construct a three-dimensional criterion matrix in a three-dimensional coordinate system, which includes three dimensions: disaster sensitivity level, rainfall-induced disaster warning level, and short-term predicted rainfall. Obtain the disaster-prone sensitivity level, rainfall-induced disaster warning level, and short-term predicted rainfall for each evaluation unit in the target area. Based on the constructed three-dimensional criterion matrix, derive the landslide disaster meteorological risk warning level for each evaluation unit. The landslide disaster meteorological risk warning levels of each evaluation unit in the target area are merged to obtain the landslide disaster meteorological risk warning level of the target area.

[0019] In this embodiment, for example, the evaluation unit is a grid unit or a landslide unit.

[0020] Please see Figures 1-2 As shown in the above technical solution, specifically, the steps for obtaining the pregnancy sensitivity level of each evaluation unit in the target area are as follows: S1. Select a reference area based on the target area, and the target area is located within the reference area, or the spatial similarity between the target area and the reference area is higher than the first threshold, the landslide samples in the reference area are greater than the second threshold, and the rainfall records in the reference area are greater than the third threshold, that is, satisfy one of the above two conditions. It should be noted that the reference area can be a single area or a set of multiple similar areas, and the first threshold, second threshold and third threshold are set based on experience; S2. Construct a pregnancy sensitivity assessment model for the target area based on the pregnancy sensitivity assessment model of the reference area. The steps are as follows: S2.1. Taking each evaluation unit in the reference area as the evaluation object, select historical disaster points and non-historical disaster points as positive and negative samples respectively, and select several basic environmental factors. Use a deep neural network to construct an initial disaster-prone sensitivity evaluation model. The network structure can be RNN, CNN, MLP, etc., or attention mechanism can be added according to the sample size. The number of network layers should not be less than 3. Train the initial disaster-prone sensitivity evaluation model with basic environmental factors and positive and negative sample sets, save the network model structure and network parameters after training, and obtain the disaster-prone sensitivity evaluation model of the reference area. S2.2. Obtain the disaster-prone sensitivity evaluation model of the reference area. Add a geological environment regional feature adjuster before the output layer. One of the input data of the geological environment regional feature adjuster is the spatial similarity between the target area and the reference area. Take each evaluation unit in the target area as the evaluation object. Select historical disaster points and non-historical disaster points as positive samples and negative samples, respectively. Select the same basic environmental factors as in S2.1. Retrain the disaster-prone sensitivity evaluation model of the reference area. During training, freeze the network parameters saved in S2.1 and train only the geological environment regional feature adjuster to obtain the disaster-prone sensitivity evaluation model of the target area. Figure 2In the diagram, h1, h2, h3, h4, and h5 are the input data of the geological environment regional feature adjuster, s1, s2, and s3 are the intermediate neurons of the geological environment regional feature adjuster, and u1, u2, u3, and u4 are the output data of the geological environment regional feature adjuster. S3. Use the pregnancy and fertility sensitivity assessment model of the target area to classify the pregnancy and fertility sensitivity level of each assessment unit in the target area.

[0021] The spatial similarity between the target region and the reference region is calculated as follows: Within the target region and the reference region, cluster analysis is performed to divide each region into multiple clusters, corresponding to different spatial types. For example, the first region is divided into three spatial types, and the second region is divided into four spatial types. Obtain the centroid vector of each cluster and construct a geological environment feature matrix representing different regions. For example, if six basic environmental factors are selected, the target region has three clusters, and the reference region has four clusters, then the feature matrix of the target region is a 3.6-dimensional feature matrix A, and the feature matrix of the reference region is a 4.6-dimensional feature matrix B. Calculate the cosine similarity between the geological environment feature matrix A of the target area and the geological environment feature matrix B of the reference area, and use it as the spatial similarity between the two.

[0022] The basic environmental factors include elevation, slope, aspect, plane curvature, profile curvature, topographic relief, soil and rock type, normalized vegetation index, normalized building index, total gully density, topographic humidity index, modified normalized differential water index, and road density.

[0023] More specifically, before training the landslide disaster-prone sensitivity assessment model using basic environmental factors and positive and negative sample sets, one of the two basic environmental factors with a correlation greater than the fourth threshold is removed.

[0024] In other words, when performing correlation analysis on the selected basic environmental factors, those with high correlation should be eliminated to avoid mutual interference and duplication of information among the basic environmental factors, which could affect the prediction accuracy of the disaster sensitivity model. For example, if the correlation between two basic environmental factors is greater than 0.5, only one should be retained, generally retaining those with low correlation to other factors.

[0025] More specifically, before training the landslide hazard sensitivity assessment model using basic environmental factors and positive and negative sample sets, the basic environmental factors are quantified using frequency ratio or weight of evidence methods.

[0026] In this embodiment, for example, when selecting non-historical disaster points as negative samples, random selection or semi-supervised selection methods are used.

[0027] Please see Figures 5-7 As shown, in the above technical solution, specifically, the steps for obtaining the rainfall-induced disaster warning level of each evaluation unit in the target area are as follows: Collect existing rainfall events and landslide disaster data for the target area, filter rainfall events related to landslide disasters (i.e., disaster-causing rainfall events), and the remaining events are rainfall events unrelated to landslides (i.e., non-disaster-causing rainfall events); The cumulative rainfall and duration of each non-disastrous rainfall event were statistically analyzed and plotted on a logarithmic coordinate system. Based on the distribution of different non-disastrous rainfall events, linear warning functions for different warning levels were fitted. The fitting formulas for the warning functions are as follows: E=αD β +C Where E is the cumulative rainfall, D is the rainfall duration, and α, β, and C are constants; By inputting the cumulative rainfall and rainfall history of each evaluation unit in the target area into a logarithmic coordinate system, and based on the linear early warning function of different early warning levels, the rainfall-induced disaster early warning level of each evaluation unit in the target area is obtained.

[0028] It should be noted that currently, statistical methods are generally used to determine the rainfall threshold that induces landslides. This involves collecting historical landslide disaster data and current rainfall data, and statistically analyzing the rainfall thresholds that lead to landslides. Such methods require sufficient landslide disaster records as support. However, in areas with poor landslide disaster record accumulation or short periods of systematic rainfall monitoring, the quality and quantity of historical data used limit the time it takes to accumulate sufficient records before conducting relevant landslide disaster early warning and forecasting work. Therefore, this application also improves the method for obtaining the rainfall-induced disaster early warning level for each evaluation unit in the target area.

[0029] As an application, a regional landslide disaster meteorological risk early warning device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aforementioned regional landslide disaster meteorological risk early warning method.

[0030] As an application, a computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the above-described method for early warning of meteorological risks of regional landslide disasters.

[0031] In summary, even in areas lacking historical records of landslide disasters, this invention, through intelligent algorithms and based on a reference area, constructs a disaster-prone sensitivity assessment model for the target area. This still enables refined regional landslide disaster meteorological risk early warning and forecasting, safeguarding people's lives and property, and providing effective support for geological disaster prevention and mitigation efforts.

[0032] This invention is not limited to the specific embodiments described above. Those skilled in the art can implement this invention using various other specific embodiments based on the content disclosed herein. Therefore, any design that adopts the design structure and concept of this invention and makes some simple changes or modifications falls within the scope of protection of this invention.

Claims

1. A method for early warning of meteorological risks of regional landslide disasters, characterized in that: Includes the following steps: Construct a three-dimensional criterion matrix in a three-dimensional coordinate system, which includes three dimensions: disaster sensitivity level, rainfall-induced disaster warning level, and short-term predicted rainfall. Obtain the disaster-prone sensitivity level, rainfall-induced disaster warning level, and short-term predicted rainfall for each evaluation unit in the target area. Based on the constructed three-dimensional criterion matrix, derive the landslide disaster meteorological risk warning level for each evaluation unit. The landslide disaster meteorological risk warning levels of each evaluation unit in the target area are merged to obtain the landslide disaster meteorological risk warning level of the target area; The steps to obtain the pregnancy susceptibility level for each evaluation unit in the target area are as follows: S1. Select a reference area based on the target area, and the target area is located within the reference area, or the spatial similarity between the target area and the reference area is higher than the first threshold, the landslide samples in the reference area are greater than the second threshold, and the rainfall records in the reference area are greater than the third threshold. S2. Construct a pregnancy sensitivity assessment model for the target area based on the pregnancy sensitivity assessment model of the reference area. The steps are as follows: S2.

1. Taking each evaluation unit in the reference area as the evaluation object, select historical disaster points and non-historical disaster points as positive and negative samples respectively, and select several basic environmental factors. Use a deep neural network to construct an initial disaster-prone sensitivity evaluation model. Train the initial disaster-prone sensitivity evaluation model through basic environmental factors and positive and negative sample sets, save the network model structure and network parameters after training, and obtain the disaster-prone sensitivity evaluation model of the reference area. S2.2 Obtain the disaster-prone sensitivity evaluation model of the reference area. Add a geological environment regional feature adjuster before the output layer. One of the input data of the geological environment regional feature adjuster is the spatial similarity between the target area and the reference area. Take each evaluation unit in the target area as the evaluation object. Select historical disaster points and non-historical disaster points as positive samples and negative samples, respectively. Select the same basic environmental factors as in S2.

1. Retrain the disaster-prone sensitivity evaluation model of the reference area. During training, freeze the network parameters saved in S2.1 and train only the geological environment regional feature adjuster to obtain the disaster-prone sensitivity evaluation model of the target area. S3. Use the pregnancy and fertility sensitivity assessment model of the target area to classify the pregnancy and fertility sensitivity level of each assessment unit in the target area.

2. The method for early warning of meteorological risks of regional landslide disasters according to claim 1, characterized in that: The steps to obtain the rainfall-induced disaster warning level for each evaluation unit in the target area are as follows: Collect data on existing rainfall events and landslide disasters in the target area, filter out disaster-causing rainfall events, and the remaining events are non-disaster-causing rainfall events; The cumulative rainfall and duration of each non-disastrous rainfall event were statistically analyzed and plotted on a logarithmic coordinate system. Based on the distribution of different non-disastrous rainfall events, linear warning functions for different warning levels were fitted. The fitting formulas for the warning functions are as follows: E=αD β +C Where E is the cumulative rainfall, D is the rainfall duration, and α, β, and C are constants; By inputting the cumulative rainfall and rainfall history of each evaluation unit in the target area into a logarithmic coordinate system, and based on the linear early warning function of different early warning levels, the rainfall-induced disaster early warning level of each evaluation unit in the target area is obtained.

3. The method for early warning of meteorological risks of regional landslide disasters according to claim 1, characterized in that: The evaluation unit is a grid unit or a landslide unit.

4. The method for early warning of meteorological risks of regional landslide disasters according to claim 1, characterized in that: The spatial similarity between the target region and the reference region is calculated as follows: Within the target region and the reference region, cluster analysis is performed to divide each region into multiple clusters, corresponding to different spatial types; Obtain the centroid vector of each cluster and construct a matrix representing the geological environment characteristics of different regions; The cosine similarity between the geological environment feature matrix of the target area and the geological environment feature matrix of the reference area is calculated as the spatial similarity between the two.

5. The method for early warning of meteorological risks of regional landslide disasters according to claim 1, characterized in that: When selecting non-historical disaster points as negative samples, random selection or semi-supervised selection methods are used.

6. The method for early warning of meteorological risks of regional landslide disasters according to claim 1, characterized in that: The basic environmental factors include elevation, slope, aspect, plane curvature, profile curvature, topographic relief, soil and rock type, normalized vegetation index, normalized building index, total gully density, topographic humidity index, modified normalized differential water index, and road density.

7. A method for early warning of meteorological risks of regional landslide disasters according to claim 1 or 6, characterized in that: Before training the landslide disaster-prone sensitivity assessment model using basic environmental factors and positive and negative sample sets, one of the two basic environmental factors with a correlation greater than the fourth threshold is removed.

8. A method for early warning of meteorological risks of regional landslide disasters according to claim 1 or 6, characterized in that: Before training the landslide hazard sensitivity assessment model using basic environmental factors and positive and negative sample sets, the basic environmental factors are quantified using frequency ratio or weight of evidence methods.

9. A regional landslide disaster meteorological risk early warning device, characterized in that: The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method for early warning of meteorological risks of regional landslide disasters as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method for early warning of meteorological risks of regional landslide disasters as described in any one of claims 1 to 8.

Images