Geographic spatio-temporal characteristic-based wildfire disaster risk assessment method

Abstract

The invention discloses a wildfire disaster risk assessment method based on geographic spatial-temporal characteristics. The method comprises the following steps: S1, determination and acquisition of assessment elements, wherein the assessment elements comprise spatial-temporal characteristics, combustible material information, meteorological elements and human activity factors; s2, data preprocessing: cleaning, integrating and standardizing the collected evaluation elements; s3, feature extraction: extracting key features from the preprocessed spatial and temporal features, combustible information, meteorological elements and human activity factors; s4, evaluation model construction: constructing and optimizing a wildfire disaster risk evaluation model in combination with GNN and ST-GCN; s5, risk assessment: inputting the preprocessed data into an assessment model, performing wildfire risk assessment, and generating a risk assessment report according to an assessment result; and S6, early warning and emergency response: according to a risk assessment result, early warning information is sent out in time, and an emergency response plan is made.

Description

Technical Field

[0001] The present invention relates to the technical field of wildfire disaster risk assessment, and in particular to a wildfire disaster risk assessment method based on geographic temporal and spatial characteristics. Background Art

[0002] As a destructive natural disaster, wildfires have a significant impact on the global ecology and climate. Wildfire disasters not only threaten human life and property safety, but also cause great damage to the natural environment and ecosystem. For example, wildfires can lead to a reduction in forest area, increased soil erosion, and loss of biodiversity. At the same time, the large amount of carbon dioxide and other greenhouse gases released during the burning of wildfires will also aggravate the trend of global warming. Wildfire disaster risk assessment is an important means to prevent and mitigate the impact of wildfire disasters. By assessing the risk of wildfire disasters, high-risk areas and potential sources of danger can be identified, providing a scientific basis for the formulation of effective prevention and response measures. At the same time, wildfire disaster risk assessment can also help government departments and all sectors of society understand the severity and potential impact of wildfire disasters, and enhance the public's awareness and ability of disaster prevention and mitigation.

[0003] In the existing technology, traditional methods mainly rely on information from a single source, such as historical fire data and meteorological data. They lack diversified data support and are difficult to deeply explore the complex relationships and laws between data, making it difficult to accurately predict the occurrence and spread of wildfire disasters. Therefore, a wildfire disaster risk assessment method based on geographic and spatiotemporal characteristics is proposed. Summary of the invention

[0004] The purpose of this invention is to solve the shortcomings of traditional methods that mainly rely on information from a single source such as historical fire data and meteorological data, lack of diversified data support, and difficulty in deeply exploring the complex relationships and laws between data, and thus difficulty in accurately predicting the occurrence and spread trends of wildfire disasters. A wildfire disaster risk assessment method based on geographic and spatiotemporal characteristics is proposed to address the shortcomings of traditional methods that mainly rely on information from a single source such as historical fire data and meteorological data, lack of diversified data support, and difficulty in deeply exploring the complex relationships and laws between data, and thus difficulty in accurately predicting the occurrence and spread trends of wildfire disasters.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A wildfire disaster risk assessment method based on geographic spatiotemporal characteristics includes the following steps: S1: Determination and collection of assessment factors: Assessment factors include spatiotemporal characteristics, combustible information, meteorological factors and human activity factors. The spatiotemporal characteristics are obtained by collecting and analyzing historical data on wildfire occurrence, including time (month, season, year) and space (geographic location, terrain, vegetation distribution) characteristics. The combustible information is obtained by remote sensing technology to obtain the type, density and moisture content of surface vegetation and assess the type and quantity of combustibles. The meteorological factors are obtained by meteorological stations, satellite remote sensing or numerical weather forecast models to obtain meteorological data related to wildfire occurrence. The human activity factors are the impact of human activities on wildfire occurrence, such as agricultural activities, forestry operations, camping and picnics, etc. S2: Data preprocessing: cleaning, integrating and standardizing the collected evaluation factors; S3: Feature extraction: Extract key features from the preprocessed spatiotemporal features, combustible information, meteorological elements and human activity factors as inputs to the assessment model; S4: Assessment model construction: Combine GNN and ST-GCN to construct and optimize the wildfire disaster risk assessment model. GNN represents geographic spatial features (such as terrain, vegetation distribution) and human activity factors (such as population distribution, road network) as a graph structure, where nodes represent geographic locations or activity points, and edges represent spatial relationships between geographic locations or associations between activity points. ST-GCN captures both spatial and temporal dependencies. ST-GCN uses GNN to capture spatial relationships and temporal dependencies in time series data, thereby predicting the occurrence and spread of wildfires. S5: Risk assessment: input the preprocessed data into the assessment model to conduct wildfire risk assessment and generate a risk assessment report based on the assessment results; S6: Early warning and emergency response: Based on the risk assessment results, timely issue early warning information to remind relevant departments and the public to pay attention to wildfire risks, and formulate emergency response plans, including personnel evacuation, material deployment and fire-fighting measures.

[0006] The above further includes: Furthermore, in S3, we use geographic information system technology to visualize wildfire data and analyze the distribution patterns of wildfires in time and space. Based on the analysis results, we extract key spatiotemporal features, such as the peak season for wildfires and high-risk areas. The specific steps are as follows: Import data into GIS: Import the pre-processed wildfire data into GIS, ensure that the data format is compatible with GIS, and import the wildfire data into the map in the form of points, lines or surfaces. Each data point represents a wildfire event, lines are used to represent the spread of wildfires, and surfaces are used to represent areas affected by wildfires, high-risk areas or areas that have been extinguished. Data visualization: Use the visualization function of the geographic information system to display wildfire data in the form of maps, charts or reports, using different colors, points or symbols to indicate the severity, frequency or type of wildfires. At the same time, you can use bar charts or line charts to show the frequency of wildfires in different seasons or years;

[0007] Spatiotemporal analysis: Use spatial cluster analysis to analyze the distribution of wildfires in time and space, identify areas and seasons with high incidence of wildfires, and use distance matrix to calculate the distance between points. The distance matrix in spatial cluster analysis is ,in, is the distance matrix, is the distance between point i and point j, and n is the number of calculation points; Extract key spatiotemporal features: Based on the spatiotemporal analysis results, extract key spatiotemporal features, such as wildfire peak seasons and high-risk areas.

[0008] Further, in S3, the combustible information is sorted and analyzed, and the influence of vegetation type, density and moisture content on the flammability of combustibles is analyzed using a linear regression model. The linear regression model formula is: ,in, The flammability of the combustibles. Factors that affect the flammability of combustibles: is the regression coefficient, Error term; based on the data analysis results, extract key combustible material characteristics, such as flammable vegetation types, high-density combustible material areas, etc.

[0009] Furthermore, in S3, correlation analysis is used to evaluate the association between human activities and wildfire risk, calculate the contribution of human activities to wildfire risk, and identify high-risk human activity types and high-frequency activity areas. Based on the correlation analysis results, key human activity features are extracted, such as high-risk human activity types (such as camping, burning garbage, etc.) and high-frequency activity areas (such as forest edges, grassland areas, etc.). The correlation analysis calculation formula is: ; in, is the correlation coefficient, and are the observed values ​​of human activity data and wildfire risk data, respectively. and are the means of the two sets of data, and the correlation coefficient The value range is between -1 and 1. The larger the absolute value, the stronger the correlation between the two sets of data.

[0010] Furthermore, in S3, feature extraction is performed on meteorological elements, including temperature feature extraction, humidity feature extraction, and wind speed and wind direction feature extraction.

[0011] Furthermore, in S4, the specific steps of building the evaluation model are as follows: GNN model construction: Graph structure definition: Based on geographic spatial features and human activity factors, a graph structure is constructed, where nodes represent geographic locations or activity points (such as villages, cities, forests, etc.), and edges represent spatial relationships between geographic locations or associations between activity points (such as road connections, terrain adjacency, etc.); Feature extraction: Extract feature vectors for each node and edge, including terrain features, vegetation features, population density, road density, etc.; GNN layer design: Select graph convolution layer to capture complex relationships in the graph structure; The graph convolution formula is ,in, represents the node feature matrix of the lth layer, represents the adjacency matrix of the graph, represents the weight matrix of the lth layer, represents the activation function; ST-GCN model construction: spatiotemporal graph structure definition: based on the GNN graph structure, time series data is introduced to construct the spatiotemporal graph structure; spatiotemporal feature extraction: spatiotemporal feature vectors are extracted for each node and edge, including temporal dependencies (such as seasonal changes in wildfires) and spatial dependencies (such as the impact of terrain on the spread of wildfires) in time series data; ST-GCN layer design: combining the advantages of graph convolutional networks (GCN) and recurrent neural networks (RNN), the ST-GCN layer is designed to capture spatial and temporal dependencies at the same time; the spatiotemporal convolution formula is ,in, represents the spatial convolution operation, Represents the temporal convolution operation. ST-GCN captures spatial relationships by first performing spatial convolution, and then performing temporal convolution to capture temporal dependencies. ST-GCN layers are used to capture both spatial and temporal dependencies. First, the spatial convolution layer is used to extract spatial features from nodes in the graph structure. Then, the temporal convolution layer is used to capture the temporal dependencies of the extracted spatial features.

[0012] Loss function definition: Select the loss function (such as cross entropy loss, mean square error loss, etc.) according to the task requirements of wildfire risk assessment; Optimizer selection: Select Adam for updating model parameters; Model training: The preprocessed data is input into the model for training. During the training process, the gradient of the loss function is calculated through the back propagation algorithm, and the model parameters are updated using the optimizer;

[0013] Model evaluation: Use validation set data to evaluate the model, calculate accuracy, recall, F1 score and other evaluation indicators, and evaluate the performance of the model; Model optimization: Optimize the model based on the evaluation results, such as adjusting the model structure, increasing the number of training rounds, using data enhancement technology, etc.

[0014] Furthermore, the specific steps of designing the ST-GCN layer are: Spatial convolution layer: Use the spatial convolution layer to extract spatial features of nodes in the graph structure. This layer captures the spatial relationship between nodes and updates the feature vectors of the nodes. The operation of the spatial convolution layer is based on the adjacency matrix of the graph and the feature matrix of the nodes;

[0015] Temporal convolution layer: Use the temporal convolution layer to capture the temporal dependency of the extracted spatial features. This layer can be regarded as an RNN or CNN layer that performs convolution operations along the time dimension. It can capture the temporal dependency in time series data and update the feature vector of the node to include temporal information. Activation function: After each convolution operation, ReLU is applied to increase the nonlinearity of the model; Batch Normalization: To speed up the training process and improve the stability of the model, a batch normalization layer is added after each convolutional layer; Stacking multiple ST-GCN layers: In order to capture more complex spatiotemporal dependencies, multiple ST-GCN layers are stacked, and each layer performs further feature extraction and update based on the output of the previous layer.

[0016] Further, in S5, the risk assessment report includes a probability distribution map of wildfire occurrence, a schematic diagram of potential spread paths, and an estimate of possible losses.

[0017] Furthermore, in S6, during the emergency response process, the dynamic changes of wildfires are continuously monitored, and the emergency response plan is adjusted according to the actual situation. At the same time, after the wildfire is extinguished, the effectiveness of the emergency response is evaluated and summarized. in, Represents the emergency response effect evaluation model.

[0018] The present invention has the following beneficial effects: In the present invention, GNN captures the complex relationship between geographic space information and human activity information. Through GNN, factors such as geographical location, terrain, vegetation distribution and human activities are represented as graph structures, so as to have a deeper understanding of the interactions and influences between them. ST-GCN combines the advantages of GCN and RNN, while capturing spatial and temporal dependencies. This means that it uses historical data and real-time data to analyze the spatiotemporal correlation and dynamics of wildfire disasters, so as to more accurately predict the occurrence and spread of wildfires. This assessment method based on geographic spatiotemporal features not only takes into account the traditional factors of wildfire occurrence, but also deeply integrates time series and spatial discrete features. It can more accurately capture the spatiotemporal dynamics of wildfires, thereby improving the accuracy and reliability of risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a step diagram of a wildfire disaster risk assessment method based on geographic spatiotemporal characteristics proposed by the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] See also Figure 1 As shown, the present invention is a wildfire disaster risk assessment method based on geographic spatiotemporal characteristics, comprising the following steps: S1: Determination and collection of assessment factors: Assessment factors include spatiotemporal characteristics, combustible information, meteorological factors and human activity factors. The spatiotemporal characteristics are obtained by collecting and analyzing historical data on wildfire occurrence, including time (month, season, year) and space (geographic location, terrain, vegetation distribution) characteristics. The combustible information is obtained by remote sensing technology to obtain the type, density and moisture content of surface vegetation and assess the type and quantity of combustibles. The meteorological factors are obtained by meteorological stations, satellite remote sensing or numerical weather forecast models to obtain meteorological data related to wildfire occurrence. The human activity factors are the impact of human activities on wildfire occurrence, such as agricultural activities, forestry operations, camping and picnics, etc. S2: Data preprocessing: cleaning, integrating and standardizing the collected evaluation factors; S3: Feature extraction: Extract key features from the preprocessed spatiotemporal features, combustible information, meteorological elements and human activity factors as inputs to the assessment model; S4: Assessment model construction: Combine GNN and ST-GCN to construct and optimize the wildfire disaster risk assessment model. GNN represents geographic spatial features (such as terrain, vegetation distribution) and human activity factors (such as population distribution, road network) as a graph structure, where nodes represent geographic locations or activity points, and edges represent spatial relationships between geographic locations or associations between activity points. ST-GCN captures both spatial and temporal dependencies. ST-GCN uses GNN to capture spatial relationships and temporal dependencies in time series data, thereby predicting the occurrence and spread of wildfires. S5: Risk assessment: input the preprocessed data into the assessment model to conduct wildfire risk assessment and generate a risk assessment report based on the assessment results; S6: Early warning and emergency response: Based on the risk assessment results, timely issue early warning information to remind relevant departments and the public to pay attention to wildfire risks, and formulate emergency response plans, including personnel evacuation, material deployment and fire-fighting measures.

[0022] In one embodiment, for the above S3, in S3, geographic information system technology is used to visualize wildfire data and analyze the distribution patterns of wildfires in time and space. According to the analysis results, key spatiotemporal features are extracted, such as wildfire high-incidence seasons, high-risk areas, etc. The specific steps are as follows: Import data into GIS: Import the pre-processed wildfire data into GIS, ensure that the data format is compatible with GIS, and import the wildfire data into the map in the form of points, lines or surfaces. Each data point represents a wildfire event, lines are used to represent the spread of wildfires, and surfaces are used to represent areas affected by wildfires, high-risk areas or areas that have been extinguished. Data visualization: Use the visualization function of the geographic information system to display wildfire data in the form of maps, charts or reports, using different colors, points or symbols to indicate the severity, frequency or type of wildfires. At the same time, you can use bar charts or line charts to show the frequency of wildfires in different seasons or years;

[0023] Spatiotemporal analysis: Use spatial cluster analysis to analyze the distribution of wildfires in time and space, identify areas and seasons with high incidence of wildfires, and use distance matrix to calculate the distance between points. The distance matrix in spatial cluster analysis is ,in, is the distance matrix, is the distance between point i and point j, and n is the number of calculation points; for example, through spatial cluster analysis, it is found that the forest area in a certain region is a high-incidence area of ​​wildfires in summer because the temperature and humidity conditions in the area in summer are conducive to the occurrence of wildfires; Extract key spatiotemporal features: Based on the spatiotemporal analysis results, extract key spatiotemporal features, such as wildfire seasons, high-risk areas, etc. These features will serve as important inputs for subsequent risk assessment and early warning models.

[0024] Suppose there is a dataset containing records of wildfires in the past ten years, and we analyze it according to the above steps: Data collection and preprocessing: Obtain data sets from fire departments, including information on the time, location, and size of each wildfire. Clean and organize the data to ensure data quality.

[0025] Data import into GIS system: Use ArcGIS software to import the dataset into the map in the form of points.

[0026] Data visualization: Use different colored dots on the map to represent wildfire events of different severity. Also, use a bar chart to show the frequency of wildfires in different seasons.

[0027] Spatiotemporal analysis: Use ArcGIS's spatial clustering analysis function to identify areas and seasons with high incidence of wildfires.

[0028] Extract key spatiotemporal features: Based on the analysis results, summer is extracted as the peak season for wildfires, and a certain forest area is extracted as a high-risk area.

[0029] In one embodiment, for the above S3, in S3, combustible information is sorted and analyzed, and the influence of vegetation type, density and moisture content on the flammability of combustibles is analyzed using a linear regression model. The linear regression model formula is: ,in, The flammability of the combustibles. Factors that affect the flammability of combustibles: is the regression coefficient, Error term: Based on the data analysis results, extract key fuel characteristics, such as flammable vegetation types, high-density fuel areas, etc. For example, through model prediction, it is found that the mixed area of ​​pine trees and shrubs in a certain area has high flammability, and the vegetation density in this area is high, so it is identified as a key fuel area.

[0030] In one embodiment, for the above S3, in S3, correlation analysis is used to evaluate the association between human activities and wildfire risk, calculate the contribution of human activities to wildfire risk, and identify high-risk human activity types and high-frequency activity areas. According to the results of the correlation analysis, key human activity features are extracted, such as high-risk human activity types (such as camping, burning garbage, etc.) and high-frequency activity areas (such as forest edges, grassland areas, etc.). The correlation analysis calculation formula is: ; in, is the correlation coefficient, and are the observed values ​​of human activity data and wildfire risk data, respectively. and are the means of the two sets of data, and the correlation coefficient The value range is between -1 and 1. The larger the absolute value, the stronger the correlation between the two sets of data.

[0031] The correlation coefficients between camping activities, agricultural activities, and burning of garbage and wildfire risk were calculated using correlation analysis. The results showed that the correlation coefficient between camping activities and wildfire risk was the highest (r=0.75), indicating that camping activities contribute the most to wildfire risk.

[0032] Extract key human activity characteristics: Based on the results of association analysis and risk assessment, camping is extracted as a high-risk activity type, and the activity has a higher incidence in summer and forest edge areas. Therefore, it is recommended to strengthen patrols and monitoring of forest edge areas in summer to reduce the risk of wildfires.

[0033] In one embodiment, for the above S3, in S3, feature extraction is performed on meteorological elements, including temperature feature extraction, humidity feature extraction, and wind speed and wind direction feature extraction.

[0034] In one embodiment, for the above S4, in S4, the evaluation model is constructed in the following specific steps: GNN model construction: Graph structure definition: Based on geographic spatial features and human activity factors, a graph structure is constructed, where nodes represent geographic locations or activity points (such as villages, cities, forests, etc.), and edges represent spatial relationships between geographic locations or associations between activity points (such as road connections, terrain adjacency, etc.); Feature extraction: Extract feature vectors for each node and edge, including terrain features, vegetation features, population density, road density, etc.; GNN layer design: Select graph convolution layer to capture complex relationships in the graph structure; The graph convolution formula is ,in, represents the node feature matrix of the lth layer, represents the adjacency matrix of the graph, represents the weight matrix of the lth layer, represents the activation function; uses the graph convolution layer to update the feature vector of the node; ST-GCN model construction: spatiotemporal graph structure definition: based on the GNN graph structure, time series data is introduced to construct the spatiotemporal graph structure; spatiotemporal feature extraction: spatiotemporal feature vectors are extracted for each node and edge, including temporal dependencies (such as seasonal changes in wildfires) and spatial dependencies (such as the impact of terrain on the spread of wildfires) in time series data; ST-GCN layer design: combining the advantages of graph convolutional networks (GCN) and recurrent neural networks (RNN), the ST-GCN layer is designed to capture spatial and temporal dependencies at the same time; the spatiotemporal convolution formula is ,in, represents the spatial convolution operation, Represents the temporal convolution operation. ST-GCN captures spatial relationships by first performing spatial convolution, and then performing temporal convolution to capture temporal dependencies. ST-GCN layers are used to capture both spatial and temporal dependencies. First, the spatial convolution layer is used to extract spatial features from nodes in the graph structure. Then, the temporal convolution layer is used to capture the temporal dependencies of the extracted spatial features.

[0035] Loss function definition: Select the loss function (such as cross entropy loss, mean square error loss, etc.) according to the task requirements of wildfire risk assessment; Optimizer selection: Select Adam for updating model parameters; Model training: input the preprocessed data into the model for training. During the training process, the gradient of the loss function is calculated through the back propagation algorithm, and the model parameters are updated using the optimizer;

[0036] Model evaluation: Use validation set data to evaluate the model, calculate accuracy, recall, F1 score and other evaluation indicators, and evaluate the performance of the model; Model optimization: Optimize the model based on the evaluation results, such as adjusting the model structure, increasing the number of training rounds, using data enhancement technology, etc.

[0037] In one embodiment, for the above-mentioned design of the ST-GCN layer, the specific steps of designing the ST-GCN layer are: Spatial convolution layer: Use the spatial convolution layer to extract spatial features of nodes in the graph structure. This layer captures the spatial relationship between nodes and updates the feature vectors of the nodes. The operation of the spatial convolution layer is based on the adjacency matrix of the graph and the feature matrix of the nodes;

[0038] Temporal convolution layer: Use the temporal convolution layer to capture the temporal dependency of the extracted spatial features. This layer can be regarded as an RNN or CNN layer that performs convolution operations along the time dimension. It can capture the temporal dependency in time series data and update the feature vector of the node to include temporal information. Activation function: After each convolution operation, ReLU is applied to increase the nonlinearity of the model; Batch Normalization: To speed up the training process and improve the stability of the model, a batch normalization layer is added after each convolutional layer; Stacking multiple ST-GCN layers: In order to capture more complex spatiotemporal dependencies, multiple ST-GCN layers are stacked, and each layer performs further feature extraction and update based on the output of the previous layer.

[0039] In one embodiment, for the above S5, in S5, the risk assessment report includes a probability distribution map of wildfire occurrence, a schematic diagram of potential spreading paths, and an estimate of possible losses.

[0040] In one embodiment, for the above S6, in S6, during the emergency response process, the dynamic changes of the wildfire are continuously monitored, and the emergency response plan is adjusted according to the actual situation. At the same time, after the wildfire is extinguished, the effect of the emergency response is evaluated and summarized. in, Represents the emergency response effect evaluation model.

[0041] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A wildfire disaster risk assessment method based on geographic spatiotemporal characteristics, characterized in that: The following steps are involved: S1: Determination and collection of assessment factors: Assessment factors include spatiotemporal characteristics, combustible information, meteorological factors and human activity factors. The spatiotemporal characteristics are obtained by collecting and analyzing historical data on wildfire occurrence, including time and space characteristics. The combustible information is obtained by remote sensing technology to obtain the type, density and moisture content of surface vegetation and assess the type and quantity of combustibles. The meteorological factors are obtained by meteorological stations, satellite remote sensing or numerical weather forecast models to obtain meteorological data related to wildfire occurrence. The human activity factors are the impact of human activities on wildfire occurrence. S2: Data preprocessing: cleaning, integrating and standardizing the collected evaluation factors; S3: Feature extraction: Extract key features from the preprocessed spatiotemporal features, combustible information, meteorological elements and human activity factors as inputs to the assessment model; S4: Assessment model construction: Combine GNN and ST-GCN to construct and optimize the wildfire disaster risk assessment model. GNN represents geographic spatial characteristics and human activity factors as a graph structure, where nodes represent geographic locations or activity points, edges represent spatial relationships between geographic locations or associations between activity points, and ST-GCN captures both spatial and temporal dependencies. ST-GCN uses GNN to capture spatial relationships and temporal dependencies in time series data, thereby predicting the occurrence and spread of wildfires. S5: Risk assessment: input the preprocessed data into the assessment model to conduct wildfire risk assessment and generate a risk assessment report based on the assessment results; S6: Early warning and emergency response: Based on the risk assessment results, timely issue early warning information and formulate emergency response plans.

2. A wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1, characterized in that: In S3, we use geographic information system technology to visualize wildfire data and analyze the distribution of wildfires in time and space. Based on the analysis results, we extract key spatiotemporal features. The specific steps are as follows: Data import into GIS: Import the pre-processed wildfire data into GIS; Data visualization: Use the visualization function of geographic information systems to display wildfire data, using different colors, points or symbols to indicate the severity, frequency or type of wildfires; Spatiotemporal analysis: Use spatial cluster analysis to analyze the distribution of wildfires in time and space. Spatial cluster analysis uses a distance matrix to calculate the distance between points. The distance matrix in spatial cluster analysis is ,in, is the distance matrix, is the distance between point i and point j, and n is the number of calculation points; Extract key spatiotemporal features: Extract key spatiotemporal features based on the spatiotemporal analysis results.

3. The wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1, characterized in that: In S3, combustible information is collated and analyzed, and the influence of vegetation type, density and moisture content on the flammability of combustibles is analyzed using a linear regression model. The linear regression model formula is: ,in, The flammability of the combustibles. Factors that affect the flammability of combustibles: is the regression coefficient, Error term; extract key combustible material characteristics based on data analysis results.

4. The wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1 is characterized in that: In S3, correlation analysis is used to evaluate the association between human activities and wildfire risk, calculate the contribution of human activities to wildfire risk, identify high-risk human activity types and high-frequency activity areas based on the results of correlation analysis, and extract key human activity features. The correlation analysis calculation formula is: ； in, is the correlation coefficient, and are the observed values ​​of human activity data and wildfire risk data, respectively. and are the means of the two sets of data, and the correlation coefficient The value range is between -1 and 1. The larger the absolute value, the stronger the correlation between the two sets of data.

5. The wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1, characterized in that: In S3, feature extraction is performed on meteorological elements, including temperature feature extraction, humidity feature extraction, and wind speed and wind direction feature extraction.

6. The wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1, characterized in that: In S4, the specific steps of building the evaluation model are as follows: GNN model construction: Graph structure definition: Based on geographic spatial features and human activity factors, a graph structure is constructed, where nodes represent geographic locations or activity points, and edges represent spatial relationships between geographic locations or associations between activity points; Feature extraction: Extract feature vectors for each node and edge; GNN layer design: Select graph convolution layer to capture complex relationships in the graph structure; The graph convolution formula is ,in, represents the node feature matrix of the lth layer, represents the adjacency matrix of the graph, represents the weight matrix of the lth layer, represents the activation function; ST-GCN model construction: spatiotemporal graph structure definition: based on the GNN graph structure, time series data is introduced to build a spatiotemporal graph structure; spatiotemporal feature extraction: spatiotemporal feature vectors are extracted for each node and edge, including temporal dependency and spatial dependency in time series data; ST-GCN layer design: ST-GCN layer is designed to capture spatial and temporal dependencies at the same time; the spatiotemporal convolution formula is ,in, represents the spatial convolution operation, Represents the temporal convolution operation. ST-GCN captures spatial relationships by first performing spatial convolution and then performing temporal convolution to capture temporal dependencies. Loss function definition: Select the loss function based on the task requirements of wildfire disaster risk assessment; Optimizer selection: Select Adam for updating model parameters; Model training: The preprocessed data is input into the model for training. During the training process, the gradient of the loss function is calculated through the back propagation algorithm, and the model parameters are updated using the optimizer; Model evaluation: Use validation set data to evaluate the model, calculate evaluation indicators, and evaluate the performance of the model; Model optimization: Optimize the model based on the evaluation results.

7. A wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 6, characterized in that: The specific steps of designing the ST-GCN layer are: Spatial convolution layer: Use the spatial convolution layer to extract spatial features of nodes in the graph structure; Temporal convolution layer: Use the temporal convolution layer to capture the temporal dependency of the extracted spatial features; Activation function: After each convolution operation, ReLU is applied to increase the nonlinearity of the model; Batch Normalization: Add a batch normalization layer after each convolutional layer; Stacking multiple ST-GCN layers: Stacking multiple ST-GCN layers, each layer will perform further feature extraction and update based on the output of the previous layer.

8. The wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1, characterized in that: In S5, the risk assessment report includes a probability distribution map of wildfire occurrence, a schematic diagram of potential spread paths, and an estimate of possible losses.

9. The wildfire disaster risk assessment method based on geographic spatiotemporal characteristics according to claim 1, characterized in that: In S6, during the emergency response process, the dynamic changes of wildfires are continuously monitored and the emergency response plan is adjusted according to the actual situation. At the same time, after the wildfire is extinguished, the effectiveness of the emergency response is evaluated and summarized. in, Represents the emergency response effect evaluation model.

Images