Forest fire early warning system and method based on meteorological and vegetation data analysis

Abstract

The invention discloses a forest fire danger early warning system and method based on meteorological and vegetation data analysis, and relates to the technical field of fire risk prediction and control. According to the forest fire danger early warning system and method based on meteorological and vegetation data analysis, fine grid division is carried out on a target forest area through a high-resolution DEM and a multispectral or hyperspectral image, and vegetation type identification is carried out on each monitoring subunit; calculating a wet-dry cycle index WDI, a monsoon effect index MFI and a dry thunderstorm risk factor DRF by combining hourly wind speed, wind direction, relative humidity, precipitation and vapor pressure difference meteorological data; through extracting canopy height, canopy density, under-forest shrub and grass layer thickness and high VOC tree species coverage rate, fuel continuity coefficient FCI and volatile matter flammability index VEI are calculated, and accurate quantification of continuity and flammability of vertical ladder fuel and horizontal canopy is realized. And a comprehensive fire danger index FFI is constructed and subjected to hierarchical management after judgment, so that the dynamism, the hierarchical property and the pertinence of fire danger early warning are realized, and the forest fire danger prevention and control precision and the patrol resource utilization efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of fire risk prediction and control technology, specifically a forest fire risk early warning system and method based on meteorological and vegetation data analysis. Background Technology

[0002] In current forest fire risk management practices, most methods rely primarily on meteorological factors or single vegetation indices for fire risk assessment, focusing on emergency response and firefighting after a fire has occurred. For the pre-fire stage, namely the prevention and control phase centered on patrols and preventative management, existing technologies remain insufficient. Particularly in monsoon climate regions, sudden changes in wind speed and direction during monsoon intrusion or retreat can lead to rapid drying of local dry fuel and an increase in potential fire risk. However, traditional methods fail to quantify the impact of such meteorological disturbances on fire risk potential and cannot guide the scientific allocation of patrol resources.

[0003] Furthermore, the vertical fuel structure of forests has a significant impact on the potential fire spread path. The thickness of the understory shrub and grass layer, the distribution of canopy height, and the canopy closure jointly determine the likelihood of surface fire spreading upwards to the canopy (the ladder fuel effect) and the horizontal fire spread capacity between canopies. Existing methods are insufficient in quantifying vertical fuel continuity and horizontal connectivity, making it difficult to accurately identify high-risk forest stands that have not yet experienced fires for focused patrols and preventative management.

[0004] Meanwhile, traditional fire risk monitoring systems suffer from low spatial resolution and lack a priority patrol mechanism for high ignition potential areas, resulting in low efficiency in patrol resource allocation and unsatisfactory preventative control effects. Therefore, there is an urgent need for a system that can identify high-risk areas, prioritize patrols, and formulate control strategies through comprehensive meteorological and vegetation analysis before a fire occurs, thereby improving the foresight and precision of forest fire risk prevention and control. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a forest fire risk early warning system and method based on meteorological and vegetation data analysis, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a forest fire risk early warning system based on meteorological and vegetation data analysis, comprising:

[0007] The monitoring area division module is used to divide the target forest area into several monitoring sub-units according to a set spatial resolution by regular grid division based on high-resolution digital elevation model (DEM) and remote sensing imagery, and to identify and extract the vegetation type in each monitoring sub-unit and mark it.

[0008] The meteorological monsoon characteristic analysis module is used to determine the monitoring date n and the latitude and longitude of the monitoring area based on GPS and astronomical epoch calculations. It divides the entire day into K time steps Δt and collects the hourly wind speed at the index of the kth time step. ,wind direction relative humidity Hourly precipitation With vapor pressure difference Establish a meteorological dataset; to construct:

[0009] The wet-dry cycle index (WDI) is used to characterize the rapid re-drying effect of fine fuels after heavy precipitation.

[0010] The Monsoon Effect Index (MFI) is used to characterize the exacerbating effect of sudden wind field changes during monsoon intrusion or retreat.

[0011] Furthermore, dry thunderstorm characteristics were identified, and a dry thunderstorm risk factor DRF was constructed, based on hourly lightning frequency. >5 times / hour, and hourly precipitation When the thickness is less than 1mm, it is determined to be a high ignition probability zone, and the first warning command is triggered.

[0012] The vegetation fuel characteristic analysis module is used to extract and calculate vegetation data within each monitoring sub-unit, including:

[0013] Canopy height distribution, canopy closure, and understory shrub and grass layer thickness are used to calculate the fuel continuity coefficient (FCI).

[0014] The proportion of combustible volatiles was extracted, including the coverage of pine, eucalyptus, rhododendron, and Lauraceae tree species, and a volatile flammability index (VEI) was constructed.

[0015] The fire risk comprehensive assessment module is used to construct the comprehensive fire risk index (FFI) for each monitoring sub-unit based on the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI). It then classifies the comprehensive fire risk index (FFI) of each monitoring sub-unit into risk levels, obtains the corresponding risk level and corresponding control strategies, and generates a first priority control group based on the first early warning instruction to prioritize the implementation of the corresponding control strategies.

[0016] Preferably, the monitoring area division module includes a data acquisition and preprocessing unit, a division unit, and a vegetation type identification unit;

[0017] The data acquisition and preprocessing unit is used to set up external data interfaces to receive or download multispectral or hyperspectral remote sensing image data from satellite / airborne / UAV platforms, as well as LiDAR point cloud and echo data from airborne or ground platforms. It performs radiometric calibration, atmospheric correction, geometric correction, denoising, and cloud / snow masking. The unit also performs ground / non-ground classification, outlier removal, and intensity normalization on the LiDAR point clouds. Furthermore, it unifies the coordinate reference system and projection of the data, performs spatiotemporal registration and co-location correction with the high-resolution digital elevation model (DEM), and resamples and crops the data according to the initial spatial resolution R to generate a "vegetation dataset" for subsequent analysis.

[0018] The subdivision unit is used to divide the target forest area into grids according to the initial spatial resolution R in the DEM coordinate system, obtain the set of monitoring sub-units, and complete the spatial labeling and indexing on the DEM;

[0019] Monitoring Subunit Set , to This represents the first to the Zth monitoring sub-units;

[0020] The vegetation type identification unit is used to adopt a supervised classification method. It is trained on the "vegetation dataset" of multispectral or hyperspectral images output by the data acquisition and preprocessing unit, and then classified by support vector machine to output the vegetation category of each monitoring sub-unit. The vegetation categories include coniferous forest, broad-leaved forest, mixed forest, shrubland and grassland.

[0021] Preferably, the vegetation type identification unit includes a feature extraction unit and a classification unit;

[0022] The feature extraction unit is used to extract spectral indices from the "vegetation dataset" of multispectral or hyperspectral images output by the data acquisition and preprocessing unit. These indices include: Normalized Difference Vegetation Index (NDVI), Red Edge Vegetation Index (REVI), and Shortwave Infrared Moisture Index (MSI), expressed as follows:

[0023]

[0024]

[0025]

[0026] Wherein, NIR is the near-infrared band reflectance, RED is the red band reflectance, RE is the red edge band reflectance, and SWIR is the short-wave infrared band reflectance;

[0027] The gray-level co-occurrence matrix P(i,j) is constructed through the following steps: Spatial texture features are extracted, specifically by quantizing the target band or principal component image to L=32 gray levels. The joint occurrence frequency of gray-level values ​​j in neighboring pixels with a window size w=7×7 pixels, a distance d=1, and an orientation θ∈{0°,45°,90°,135°} is used. The statistical results are then normalized to obtain the gray-level co-occurrence matrix P(i,j).

[0028]

[0029] Where N(i,j) is the number of times gray level i and gray level j co-occur; i represents the gray value of the reference pixel, j represents the gray value of the neighboring pixels of the reference pixel at a specified direction and distance, and P(i,j) represents the joint probability of the pixel with gray level i and the neighboring pixel with gray level j appearing in the window.

[0030] The texture index is obtained by calculating and averaging the directions, including:

[0031] Contrast CON: ;

[0032] Homogeneous HOM: ;

[0033] Energy / Angle Second Moment (ASM): ;

[0034] Entropy ENT: ; where P(i,j)=0, this term is ignored;

[0035] Correlation COR: in, Indicates the average gray value in the row direction. This indicates the average gray level along the column direction. Indicates the standard deviation in the row direction. Indicates the standard deviation in the column direction.

[0036] Classification units, used for classification based on spectral and textural indicators, include:

[0037] When the NDVI value of the monitoring subunit is between 0.65 and 0.85, the REVI value is between 0.10 and 0.20, the MSI value is between 0.45 and 0.55, and the texture index contrast (CON) is between 10 and 20, the homogeneity (HOM) is between 0.70 and 0.85, the energy (ASM) is between 0.55 and 0.70, the entropy (ENT) is between 3 and 4, and the correlation (COR) is between 0.85 and 0.95, the monitoring subunit is determined to be a coniferous forest.

[0038] When the NDVI value of the monitoring subunit is between 0.60 and 0.80, the REVI value is between 0.15 and 0.25, the MSI value is between 0.50 and 0.60, and the texture index CON is between 15 and 25, HOM is between 0.60 and 0.75, ASM is between 0.50 and 0.65, ENT is between 3.5 and 4.5, and COR is between 0.80 and 0.90, it is identified as a broad-leaved forest.

[0039] When the NDVI value of the monitoring subunit is between 0.55 and 0.75, the REVI value is between 0.12 and 0.22, the MSI value is between 0.50 and 0.65, and the texture index CON is between 18 and 28, HOM is between 0.55 and 0.70, ASM is between 0.45 and 0.60, ENT is between 4 and 5, and COR is between 0.75 and 0.85, it is determined to be a mixed forest.

[0040] When the NDVI value of the monitoring subunit is between 0.35 and 0.55, the REVI value is between 0.08 and 0.15, the MSI value is between 0.60 and 0.75, and the texture index CON is between 22 and 35, HOM is between 0.50 and 0.65, ASM is between 0.40 and 0.55, ENT is between 4.5 and 5.5, and COR is between 0.70 and 0.80, it is determined to be a shrub.

[0041] When the NDVI value of the monitoring subunit is between 0.20 and 0.40, the REVI value is between 0.05 and 0.10, the MSI value is between 0.65 and 0.80, and the texture index CON is between 12 and 20, HOM is between 0.45 and 0.60, ASM is between 0.35 and 0.50, ENT is between 4 and 5, and COR is between 0.65 and 0.75, it is determined to be grassland.

[0042] Preferably, the meteorological monsoon characteristic analysis module includes a meteorological data acquisition unit, a wet-dry cycle index calculation unit, and a monsoon effect index calculation unit;

[0043] The meteorological data acquisition unit is used to obtain hourly wind speeds within a monitoring date n, using a time step Δt to form time indices k=1,…,K. ,wind direction relative humidity Hourly precipitation With vapor pressure difference Establish a meteorological dataset;

[0044] And decompose scalar wind speed and wind direction into two-dimensional vectors: , ; This represents the wind speed component in the east-west direction. This indicates the wind speed component in the north-south direction;

[0045] The wet-dry cycle index calculation unit is used to characterize the effect of "rapid wetting of fine fuel by heavy precipitation, but subsequent sunny weather and high vapor pressure differential leading to rapid re-drying within 1-2 days". The calculation uses the most recent W-hour window, with W=48h, to calculate the humidity. and drying tensile strength coefficient DI;

[0046]

[0047] ;

[0048] Where W represents the set of the most recent W hours. The scaling parameter is set to 10mm. ∈[0,1] represents the near-end wettability, which is closer to 1 the greater the rainfall; This represents the number of valid time points within the window. When the research object is evapotranspiration, The effective evapotranspiration period is defined as the time when the atmosphere has reached a certain water deficit and the surface of plant leaves begins to experience evapotranspiration tension. This determines which moments within the statistical window qualify as valid samples. This indicates that the air is extremely dry, exceeding the vegetation's regulatory capacity, and there is a risk of causing the vegetation's stomata to close or even causing physiological stress. This indicates the reference vapor pressure differential risk identification threshold, set to 2 kPa, used to determine "whether it has entered an evaporation stress state";

[0049] Combined with humidity After dimensionless processing of the drying tensile strength coefficient DI, the wet-dry cycle index WDI is calculated:

[0050] .

[0051] Preferably, the monsoon effect index calculation unit is used to characterize the abrupt changes in wind field and the exacerbating effect of low-level jet streams during monsoon intrusion / retreat. When rapid changes in wind speed / direction favor fire spread and uphill movement, the wind vector difference is calculated to obtain the monsoon effect index MFI.

[0052]

[0053] in, This represents the wind speed component in the east-west direction. This indicates the wind speed component in the north-south direction; and This represents the change in the wind vector component between two adjacent time intervals.

[0054] Preferably, the meteorological monsoon characteristic analysis module also includes a dry thunderstorm risk identification unit, which is used to identify hourly lightning frequency. and hourly precipitation And preset lightning threshold And dry thunderstorm precipitation threshold , Set to 5 times / hour; Set to 1mm / h;

[0055] Define the Thunderstorm Risk Factor (DRF):

[0056] ;

[0057] If the thunderstorm risk factor DRF=1 at any time within the monitoring date n, the monitoring sub-unit corresponding to that date is marked as a 'high ignition probability zone' and the first warning command is triggered, indicating that there is a risk of lightning igniting dry vegetation or surface combustibles in the monitoring sub-unit. The dry thunderstorm risk layer is then output through raster aggregation.

[0058] Preferably, the vegetation fuel characteristic analysis module includes a canopy height distribution extraction unit, a canopy closure calculation unit, and an understory shrub and grass layer thickness measurement unit;

[0059] The canopy height distribution extraction unit is used to extract the canopy height distribution curve within the monitoring subunit from a forest 3D point cloud model constructed using LiDAR point cloud data or stereo image pairs, and to calculate the average canopy height Hmean, maximum canopy height Hmax, and the standard deviation σh of the height distribution.

[0060] ;

[0061] Where N is the total number of point clouds within the monitoring sub-unit. Let p be the height of the p-th point;

[0062] The canopy closure calculation unit is used to calculate the canopy closure C of the monitoring sub-unit based on remote sensing images in the "vegetation dataset" using point cloud porosity analysis technology. The range is 0-1, representing the proportion of the ground surface covered by the canopy. The formula for calculating canopy closure C is:

[0063] ;

[0064] Wherein, GapArea is the area of ​​visible voids on the ground surface, and TotalArea is the total area of ​​the monitoring sub-units;

[0065] The understory shrub and grass layer thickness measurement unit is used to calculate the average thickness (Hunder) of shrubs and herbaceous vegetation within the monitoring subunit based on the low-level point cloud echoes in the "vegetation dataset" from hyperspectral imagery. The calculation formula is as follows:

[0066] ;

[0067] Where M represents the number of low-level pixels within the monitoring sub-unit. The height of the qth low-level pixel is defined as the height between 0 and 2m above the ground.

[0068] Preferably, the vegetation fuel characteristic analysis module also includes a fuel continuity analysis unit and a combustible species identification unit;

[0069] The fuel continuity analysis unit is used to normalize and process the average canopy height Hmean, maximum canopy height Hmax, standard deviation of height distribution σh, canopy closure C, and average thickness of shrubs and herbaceous vegetation Hunder using a clipping function to obtain the relative canopy uniformity m, canopy vertical stratification intensity s, relative amount of understory "ladder fuel" thickness u, and horizontal canopy closure c.

[0070]

[0071]

[0072]

[0073]

[0074] Construct the horizontal continuity component and obtain the horizontal continuity index Hhor:

[0075]

[0076] Construct the vertical continuity component and obtain the vertical continuity index Hver:

[0077]

[0078] in, and As weight, , The thickness of the understory is slightly greater than the degree of stratification;

[0079] The fuel continuity index (FCI) is calculated by combining the horizontal continuity index (Hhor) and the vertical continuity index (Hver).

[0080]

[0081] in, and As weight, , ;

[0082] The combustible species identification unit is used to extract the coverage of high-VOC tree species in each monitoring subunit, including the coverage of pine, eucalyptus, rhododendron, and camphor family tree species, and to construct the volatile flammability index (VEI) for each monitoring subunit.

[0083] ;

[0084] in, Indicates the number of high-VOC tree species. The coefficients represent the volatile matter weighting coefficients for tree species in category a, where pine is 0.8, eucalyptus is 0.95, rhododendron is 0.7, and Lauraceae is 0.6. This represents the coverage rate of tree species of type a in the r-th monitoring sub-unit.

[0085] Preferably, the fire risk comprehensive assessment module includes a comprehensive fitting unit and a judgment unit;

[0086] The comprehensive fitting unit is used to extract the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI) for each monitoring subunit. After normalization, the comprehensive fire hazard index (FFI) for each monitoring subunit is constructed.

[0087]

[0088]

[0089] in, , , , , As weight;

[0090] The determination unit is used to classify the risk level of the Fire Risk Index (FFI) for each monitoring sub-unit, specifically as follows:

[0091] When FFI > 0.7, the label is set to the first risk level, and the first control strategy is generated, including: patrolling once a day, prohibiting all open flames, including picnicking, burning waste and fireworks; setting up obvious warning signs along the forest line and management roads, with at least one "No Open Flames" sign every 500 meters; and cleaning up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 10-meter wide strip along the patrol route and within 50 meters of the forest edge at least once a week.

[0092] When 0.4≤FFI≤0.7, the label is set as the second risk level, and a second control strategy is generated, including: patrolling once every 2 days, setting up at least one "No Open Flame" sign every 800 meters; cleaning up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 10-meter wide strip along the patrol route and within 30 meters of forest roads and forest edges once every 2 weeks;

[0093] When FFI < 0.4, the label is the third risk level, and the third control strategy is generated, including: patrol once a week, set up a "No Open Flame" sign every 1000 meters, and clean up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 3-5 meter wide area along the patrol route and within 10 meters of forest roads and forest edges once a month.

[0094] When a monitoring subunit is identified as a 'high ignition probability zone' and triggers the first warning command, the monitoring subunits that issued the first warning command are counted to form a first priority control group, and the corresponding control strategies are implemented first.

[0095] Preferably, it includes the following steps:

[0096] S1. Acquire high-resolution digital elevation model (DEM) and multispectral or hyperspectral remote sensing images. Divide the target forest area into several monitoring sub-units according to the set spatial resolution using regular grids. Identify and mark the vegetation type in each monitoring sub-unit.

[0097] S2. Based on the Global Positioning System (GPS) and astronomical epoch calculations, determine the monitoring date n and the latitude and longitude of the monitoring area. Divide the entire day into K time steps according to the time step Δt, and collect the hourly wind speed at the index of the kth time step. ,wind direction relative humidity Hourly precipitation With vapor pressure difference A meteorological dataset was established to construct the wet-dry cycle index (WDI), the monsoon effect index (MFI), and the dry thunderstorm risk factor (DRF). If the thunderstorm risk factor DRF = 1 at any time, the monitoring sub-unit corresponding to that date is marked as a 'high ignition probability zone' and the first warning instruction is triggered, indicating that there is a risk of lightning igniting dry vegetation or surface combustibles in that monitoring sub-unit. The dry thunderstorm risk layer is output through raster aggregation.

[0098] S3. Extract vegetation data from each monitoring sub-unit and calculate the fuel continuity coefficient (FCI) and volatile flammability index (VEI).

[0099] S4. Based on the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI), construct the comprehensive fire risk index (FFI) for each monitoring sub-unit, classify the risk level of the comprehensive fire risk index (FFI) of each monitoring sub-unit, obtain the corresponding risk level and corresponding control strategy, and combine with the first early warning instruction to generate a first priority control group for priority implementation of the corresponding control strategy.

[0100] This invention provides a forest fire risk early warning system and method based on meteorological and vegetation data analysis. It has the following beneficial effects:

[0101] (1) This invention achieves a refined and quantitative assessment of fire risk potential by comprehensively considering monsoon climate conditions and forest vertical fuel structure, effectively solving the problem that traditional fire risk monitoring methods rely solely on meteorological factors or single vegetation indices and are difficult to reflect the risk of rapid fire spread. This invention establishes the wet-dry cycle index (WDI) and the monsoon effect index (MFI), which can quantify the re-drying rate of fine fuel after heavy rainfall and the aggravating effect of sudden changes in wind speed and direction during monsoon intrusion or retreat on fire spread, thereby accurately identifying the potential risk of rapid local fire spread. At the same time, by extracting and analyzing the canopy height distribution, canopy closure, and understory shrub and grass layer thickness of each monitoring subunit, the fuel continuity coefficient (FCI) is calculated, and the volatile flammability index (VEI) is constructed in combination with the proportion of combustible volatiles, realizing the quantitative characterization of the influence of vertical fuel gradient and horizontal connectivity. It can accurately assess the potential path of surface fire spreading upward to the canopy and the lateral spread of fire between canopies, fully reflecting the ladder fuel effect and canopy connectivity effect. Furthermore, this invention constructs a dry thunderstorm risk factor (DRF) and combines it with a first early warning command to include high ignition probability areas in the first priority control group, enabling priority patrol and control of high-risk areas, significantly improving the efficiency of patrol resource utilization and fire risk prevention and control. Simultaneously, this invention utilizes high-resolution DEMs and remote sensing imagery to divide the monitoring area into regular grids, accurately extracting vegetation types, and achieving spatially differentiated and hierarchical management of forest stand heterogeneity. This allows fire risk warnings to implement differentiated control strategies for different sub-units, effectively reducing the suddenness and uncontrollability of fires, and enhancing the scientific, precise, and operational nature of forest fire risk prevention and control.

[0102] (2) By using hourly meteorological data to calculate the wet-dry cycle index and the monsoon effect index, the impact of precipitation re-drying and sudden changes in wind field on fire risk potential can be dynamically reflected, thereby realizing meteorological-driven dynamic fire risk early warning and improving the timeliness of prevention and control.

[0103] (3) By quantifying canopy connectivity, vertical ladder fuel thickness and distribution of high VOC tree species through fuel continuity coefficient and volatile flammability index, the horizontal spread and vertical ignition potential of fire can be scientifically assessed, and targeted control strategies can be formulated.

[0104] (4) The comprehensive fire risk index (FFI) is combined with the priority identification of high ignition probability areas to achieve hierarchical management and optimized allocation of key patrol resources, effectively reducing the fire incidence rate and the risk of rapid spread, and improving the efficiency of forest fire prevention and control. Attached Figure Description

[0105] Figure 1 This is a schematic diagram of the overall system flow of the present invention;

[0106] Figure 2 This is a flowchart illustrating the monitoring area division module of the present invention;

[0107] Figure 3 This is a flowchart illustrating the meteorological monsoon characteristic analysis module of the present invention.

[0108] Figure 4 This is a flowchart illustrating the vegetation fuel characteristic analysis module of the present invention.

[0109] Figure 5 This is a schematic diagram of the fire risk comprehensive assessment module of the present invention;

[0110] Figure 6 This is a schematic diagram of the method steps of the present invention. Detailed Implementation

[0111] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0112] Example 1

[0113] Please see Figures 1 to 5 This invention provides a forest fire risk early warning system based on meteorological and vegetation data analysis, comprising:

[0114] The monitoring area division module is used to divide the target forest area into several monitoring sub-units according to a set spatial resolution by regular grid division based on high-resolution digital elevation model (DEM) and remote sensing imagery, and to identify and extract the vegetation type in each monitoring sub-unit and mark it.

[0115] The meteorological monsoon characteristic analysis module is used to determine the monitoring date n and the latitude and longitude of the monitoring area based on GPS and astronomical epoch calculations. It divides the entire day into K time steps Δt and collects the hourly wind speed at the index of the kth time step. ,wind direction relative humidity Hourly precipitation With vapor pressure difference Establish a meteorological dataset; to construct:

[0116] The wet-dry cycle index (WDI) is used to characterize the rapid re-drying effect of fine fuels after heavy precipitation.

[0117] The Monsoon Effect Index (MFI) is used to characterize the exacerbating effect of sudden wind field changes during monsoon intrusion or retreat.

[0118] Furthermore, dry thunderstorm characteristics were identified, and a dry thunderstorm risk factor DRF was constructed, based on hourly lightning frequency. >5 times / hour, and hourly precipitation When the thickness is less than 1mm, it is determined to be a high ignition probability zone, and the first warning command is triggered.

[0119] The vegetation fuel characteristic analysis module is used to extract and calculate vegetation data within each monitoring sub-unit, including:

[0120] Canopy height distribution, canopy closure, and understory shrub and grass layer thickness are used to calculate the fuel continuity coefficient (FCI).

[0121] The proportion of combustible volatiles was extracted, including the coverage of pine, eucalyptus, rhododendron, and Lauraceae tree species, and a volatile flammability index (VEI) was constructed.

[0122] The fire risk comprehensive assessment module is used to construct the comprehensive fire risk index (FFI) for each monitoring sub-unit based on the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI). It then classifies the comprehensive fire risk index (FFI) of each monitoring sub-unit into risk levels, obtains the corresponding risk level and corresponding control strategies, and generates a first priority control group based on the first early warning instruction to prioritize the implementation of the corresponding control strategies.

[0123] In this embodiment, the present invention achieves a refined and quantitative assessment of fire risk potential by comprehensively considering monsoon climate conditions and forest vertical fuel structure. This effectively solves the problem that traditional fire risk monitoring methods rely solely on meteorological factors or single vegetation indices, making it difficult to reflect the risk of rapid fire spread. The present invention establishes the Wet-Dry Cycle Index (WDI) and the Monsoon Effect Index (MFI), which can quantify the re-drying rate of fine fuels after heavy rainfall and the exacerbating effect of sudden changes in wind speed and direction during monsoon intrusion or retreat on fire spread, thereby accurately identifying the potential risk of rapid local fire spread. Simultaneously, by extracting and analyzing the canopy height distribution, canopy closure, and understory shrub and grass thickness of each monitoring subunit, the fuel continuity coefficient (FCI) is calculated, and a volatile flammability index (VEI) is constructed by combining the proportion of combustible volatiles. This achieves a quantitative characterization of the influence of vertical fuel gradient and horizontal connectivity, accurately assessing the potential paths for surface fire to spread upwards to the canopy and for lateral fire spread between canopies, fully reflecting the ladder fuel effect and canopy connectivity effect. Furthermore, this invention constructs a dry thunderstorm risk factor (DRF) and combines it with a first early warning command to include high ignition probability areas in the first priority control group, enabling priority patrol and control of high-risk areas, significantly improving the efficiency of patrol resource utilization and fire risk prevention and control. Simultaneously, this invention utilizes high-resolution DEMs and remote sensing imagery to divide the monitoring area into regular grids, accurately extracting vegetation types, and achieving spatially differentiated and hierarchical management of forest stand heterogeneity. This allows fire risk warnings to implement differentiated control strategies for different sub-units, effectively reducing the suddenness and uncontrollability of fires, and enhancing the scientific, precise, and operational nature of forest fire risk prevention and control.

[0124] Example 2

[0125] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figures 1 to 5 Specifically, the monitoring area division module includes a data acquisition and preprocessing unit, a division unit, and a vegetation type identification unit;

[0126] The data acquisition and preprocessing unit is used to set up external data interfaces to receive or download multispectral or hyperspectral remote sensing image data from satellite / airborne / UAV platforms, as well as LiDAR point cloud and echo data from airborne or ground platforms. It performs radiometric calibration, atmospheric correction, geometric correction, denoising, and cloud / snow masking. The unit also performs ground / non-ground classification, outlier removal, and intensity normalization on the LiDAR point clouds. Furthermore, it unifies the coordinate reference system and projection of the data, performs spatiotemporal registration and co-location correction with the high-resolution digital elevation model (DEM), and resamples and crops the data according to the initial spatial resolution R to generate a "vegetation dataset" for subsequent analysis.

[0127] The division unit is used to divide the target forest area into grids according to the initial spatial resolution R, preferably 10m×10m, in the DEM coordinate system, to obtain a set of monitoring sub-units and complete spatial labeling and indexing on the DEM;

[0128] Monitoring Subunit Set , to This represents the first to the Zth monitoring sub-units;

[0129] The vegetation type identification unit is used to adopt a supervised classification method. It is trained on the "vegetation dataset" of multispectral or hyperspectral images output by the data acquisition and preprocessing unit, and then classified by support vector machine to output the vegetation category of each monitoring sub-unit. The vegetation categories include coniferous forest, broad-leaved forest, mixed forest, shrubland and grassland.

[0130] In this embodiment, the monitoring area division module is further subdivided into a data acquisition and preprocessing unit, a division unit, and a vegetation type identification unit, achieving high-precision and systematic management of forest areas, thereby significantly improving the accuracy and scientific rigor of fire risk warnings. In the data acquisition and preprocessing unit, multispectral / hyperspectral remote sensing images from satellite, airborne, and UAV platforms, as well as LiDAR point cloud and echo data, are integrated. Radiometric calibration, atmospheric correction, geometric correction, denoising, cloud and snow masking processing, ground / non-ground classification, outlier removal, and intensity normalization are performed to ensure the high quality and comparability of the original data. Simultaneously, coordinate system one, projection registration, and co-location correction with the high-resolution DEM enable precise spatiotemporal matching of multi-source data, providing a reliable foundation for subsequent analysis. The division unit divides the target forest area into grids according to a preferred spatial resolution (e.g., 10m × 10m), forming a clear set of monitoring sub-units. Spatial labeling and indexing are completed on the DEM, allowing for a refined expression of forest heterogeneity. The vegetation type identification unit utilizes supervised classification methods and support vector machines to train and classify the "vegetation dataset," achieving accurate identification of different vegetation types such as coniferous forests, broad-leaved forests, mixed forests, shrublands, and grasslands. This ensures that each monitoring sub-unit obtains clear vegetation category information. This method can fully reflect the impact of different forest stand structures and vertical fuel characteristics on fire risk, providing a scientific basis for comprehensive fire risk assessment and differentiated management. It significantly improves the accuracy and efficiency of forest fire risk early warning, optimizes patrol resource allocation, and reduces the risk of sudden fires and their spread.

[0131] Example 3

[0132] This embodiment is an explanation based on Embodiment 2. Please refer to it. Figures 1 to 5 Specifically, the vegetation type identification unit includes a feature extraction unit and a classification unit;

[0133] The feature extraction unit is used to extract spectral indices from the "vegetation dataset" of multispectral or hyperspectral images output by the data acquisition and preprocessing unit. These indices include: Normalized Difference Vegetation Index (NDVI), Red Edge Vegetation Index (REVI), and Shortwave Infrared Moisture Index (MSI), expressed as follows:

[0134]

[0135]

[0136]

[0137] Wherein, NIR is the near-infrared band reflectance, RED is the red band reflectance, RE is the red edge band reflectance, and SWIR is the short-wave infrared band reflectance;

[0138] The gray-level co-occurrence matrix P(i,j) is constructed through the following steps: Spatial texture features are extracted, specifically by quantizing the target band or principal component image to L=32 gray levels. The joint occurrence frequency of gray-level values ​​j in neighboring pixels with a window size w=7×7 pixels, a distance d=1, and an orientation θ∈{0°,45°,90°,135°} is used. The statistical results are then normalized to obtain the gray-level co-occurrence matrix P(i,j).

[0139]

[0140] Where N(i,j) is the number of times gray level i and gray level j co-occur; i represents the gray value of the reference pixel, j represents the gray value of the neighboring pixels of the reference pixel at a specified direction and distance, and P(i,j) represents the joint probability of the pixel with gray level i and the neighboring pixel with gray level j appearing in the window.

[0141] The texture index is obtained by calculating and averaging the directions, including:

[0142] Contrast CON: This indicates the degree of significance of pixel grayscale differences; a larger value indicates a coarser texture.

[0143] Homogeneous HOM: This indicates the degree of concentration of diagonal neighboring elements in a pixel; a larger value indicates a more uniform texture.

[0144] Energy / Angle Second Moment (ASM): This indicates the uniformity of grayscale distribution; a larger value indicates a more regular image texture.

[0145] Entropy ENT: ; where P(i,j)=0, this term is ignored. It represents the texture complexity; a larger value indicates a more irregular texture.

[0146] Correlation COR: in, Indicates the average gray value in the row direction. This indicates the average gray level along the column direction. Indicates the standard deviation in the row direction. The column-directed standard deviation is represented by COR, which indicates the degree of linear dependence between pixels. A value close to 1 indicates that the texture grayscale is highly correlated.

[0147] Classification units, used for classification based on spectral and textural indicators, include:

[0148] When the NDVI value of the monitoring subunit is between 0.65 and 0.85, the REVI value is between 0.10 and 0.20, the MSI value is between 0.45 and 0.55, and the texture index contrast (CON) is between 10 and 20, the homogeneity (HOM) is between 0.70 and 0.85, the energy (ASM) is between 0.55 and 0.70, the entropy (ENT) is between 3 and 4, and the correlation (COR) is between 0.85 and 0.95, the monitoring subunit is determined to be a coniferous forest.

[0149] When the NDVI value of the monitoring subunit is between 0.60 and 0.80, the REVI value is between 0.15 and 0.25, the MSI value is between 0.50 and 0.60, and the texture index CON is between 15 and 25, HOM is between 0.60 and 0.75, ASM is between 0.50 and 0.65, ENT is between 3.5 and 4.5, and COR is between 0.80 and 0.90, it is identified as a broad-leaved forest.

[0150] When the NDVI value of the monitoring subunit is between 0.55 and 0.75, the REVI value is between 0.12 and 0.22, the MSI value is between 0.50 and 0.65, and the texture index CON is between 18 and 28, HOM is between 0.55 and 0.70, ASM is between 0.45 and 0.60, ENT is between 4 and 5, and COR is between 0.75 and 0.85, it is determined to be a mixed forest.

[0151] When the NDVI value of the monitoring subunit is between 0.35 and 0.55, the REVI value is between 0.08 and 0.15, the MSI value is between 0.60 and 0.75, and the texture index CON is between 22 and 35, HOM is between 0.50 and 0.65, ASM is between 0.40 and 0.55, ENT is between 4.5 and 5.5, and COR is between 0.70 and 0.80, it is determined to be a shrub.

[0152] When the NDVI value of the monitoring subunit is between 0.20 and 0.40, the REVI value is between 0.05 and 0.10, the MSI value is between 0.65 and 0.80, and the texture index CON is between 12 and 20, HOM is between 0.45 and 0.60, ASM is between 0.35 and 0.50, ENT is between 4 and 5, and COR is between 0.65 and 0.75, it is classified as grassland. The vegetation type classification table is shown in Table 1 below.

[0153] Table 1: Classification of Vegetation Types

[0154]

[0155] In this embodiment, a feature extraction unit and a classification unit are introduced to achieve high-precision and automated identification of vegetation types within the monitoring sub-units, thereby significantly improving the scientific rigor and relevance of forest fire risk early warning. In the feature extraction unit, spectral indicators (such as NDVI, REVI, and MSI) from multispectral or hyperspectral imagery are comprehensively utilized with gray-level co-occurrence matrix texture features (including contrast CON, homogeneity HOM, energy ASM, entropy ENT, and correlation COR). This not only characterizes the vegetation's growth status, chlorophyll cover, and water content, but also reflects the forest's vertical structure and spatial heterogeneity, providing multidimensional information for finely distinguishing different vegetation types such as coniferous forests, broad-leaved forests, mixed forests, shrublands, and grasslands. In the classification unit, by setting multidimensional thresholds for spectral and texture indicators, rule-based accurate classification is achieved, ensuring clear labeling of vegetation types within the monitoring sub-units. This method not only reflects the impact of vertical and horizontal differences in forest stand structure on fire risk but also provides a reliable basis for fuel continuity assessment and fire risk index calculation, effectively improving the accuracy of comprehensive fire risk assessment and the relevance of patrol strategies, thus optimizing resource allocation.

[0156] Example 4

[0157] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figures 1 to 5 Specifically, the meteorological monsoon characteristic analysis module includes a meteorological data acquisition unit, a wet-dry cycle index calculation unit, and a monsoon effect index calculation unit.

[0158] The meteorological data acquisition unit is used to obtain hourly wind speeds within a monitoring date n, using a time step Δt to form time indices k=1,…,K. (m / s), wind direction (°), relative humidity (%), hourly precipitation (mm) and vapor pressure difference (kPa), establish a meteorological dataset;

[0159] And decompose scalar wind speed and wind direction into two-dimensional vectors: , ; This represents the wind speed component in the east-west direction. This indicates the wind speed component in the north-south direction;

[0160] The wet-dry cycle index calculation unit is used to characterize the effect of "rapid wetting of fine fuel by heavy precipitation, but subsequent sunny weather and high vapor pressure differential leading to rapid re-drying within 1-2 days". The calculation uses the most recent W-hour window, with W=48h, to calculate the humidity. and drying tensile strength coefficient DI;

[0161] ;

[0162] ;

[0163] Where W represents the set of the most recent W hours. The scaling parameter is set to 10mm. ∈[0,1] represents the near-end wettability, which is closer to 1 the greater the rainfall; This represents the number of valid time points within the window. When the research object is evapotranspiration, The effective evapotranspiration period is defined as the time when the atmosphere has reached a certain water deficit and the surface of plant leaves begins to experience evapotranspiration tension. This determines which moments within the statistical window qualify as valid samples. This indicates that the air is extremely dry and the atmospheric evaporation pull is strong, often exceeding the vegetation's regulatory capacity, which may cause the vegetation's stomata to close or even cause physiological stress. The reference vapor pressure differential risk identification threshold is set to 2 kPa and is used to determine whether "evaporation stress has been entered"; DI reflects the strength of the proximal drying driving force.

[0164] Combined with humidity After dimensionless processing of the drying tensile strength coefficient DI, the wet-dry cycle index WDI is calculated:

[0165] .

[0166] The formula means that when "there is indeed significant precipitation in the vicinity" ( When the DI is relatively high and the VPD is high within the following 48 hours (higher DI), it indicates that there is a phenomenon of "more flammable after wet-dry cycle";

[0167] Example 5

[0168] This embodiment is an explanation based on Embodiment 4. Please refer to it. Figures 1 to 5Specifically, the Monsoon Effect Index (MFI) calculation unit is used to characterize the abrupt changes in wind field and the exacerbating effect of low-level jet streams during monsoon intrusion / retreat. When rapid changes in wind speed / direction favor fire spread and uphill movement, the wind vector difference is calculated to obtain the Monsoon Effect Index (MFI).

[0169] ;

[0170] in, This represents the wind speed component in the east-west direction. This indicates the wind speed component in the north-south direction; and MFI indicates the change in wind vector components between two adjacent time points. A larger MFI indicates drastic changes in wind direction / speed between adjacent time points (or levels), resulting in an unstable or highly fluctuating wind field. A smaller MFI indicates smoother wind direction / speed between adjacent time points, resulting in a more stable wind field.

[0171] The meteorological monsoon characteristic analysis module acquires hourly meteorological parameters such as wind speed, wind direction, relative humidity, precipitation, and vapor pressure difference, and decomposes wind speed and direction into two-dimensional vectors, providing accurate dynamic meteorological data for subsequent fire hazard analysis. The wet-dry cycle index calculation unit can quantitatively characterize the effect of rapid fuel wetting by heavy precipitation followed by rapid fuel re-drying due to high vapor pressure difference and sunny weather. By calculating humidity and drying pull coefficient (DI), it identifies the effective evapotranspiration period and determines the evapotranspiration stress state, achieving sensitive capture of the impact of short-term wet-dry cycles on fire hazard. When recent precipitation is significant and atmospheric evapotranspiration pull increases within the following 48 hours, the WDI value increases, effectively reflecting the potential risk of "fuel becoming more flammable after the wet-dry cycle," thus providing a quantitative judgment of fire hazard potential. This method overcomes the limitations of traditional fire hazard early warning systems that rely solely on precipitation or temperature and humidity as single factors, enabling dynamic and precise assessment of local fire hazard risks in monsoon climate regions, providing a scientific basis for fire hazard classification, patrol scheduling, and priority control.

[0172] The meteorological monsoon characteristic analysis module also includes a dry thunderstorm risk identification unit, which is used to identify hourly lightning frequency. and hourly precipitation And preset lightning threshold And dry thunderstorm precipitation threshold , Set to 5 times / hour; Set to 1mm / h;

[0173] Define the Thunderstorm Risk Factor (DRF):

[0174] ;

[0175] The physical meaning is that when there is a lot of lightning but very little precipitation, it means that the atmospheric convection is strong enough to trigger lightning, but the raindrops are evaporated during the falling process due to the high temperature and low humidity environment (i.e., the virga phenomenon), and ultimately cannot effectively increase the surface humidity → the ignition potential increases.

[0176] If the thunderstorm risk factor DRF=1 at any time within the monitoring date n, the monitoring sub-unit corresponding to that date is marked as a 'high ignition probability zone' and the first warning command is triggered, indicating that there is a risk of lightning igniting dry vegetation or surface combustibles in the monitoring sub-unit. The dry thunderstorm risk layer is then output through raster aggregation.

[0177] In this embodiment, the dry thunderstorm risk identification unit monitors hourly lightning frequency and hourly precipitation, setting a lightning threshold of 5 times / hour and a dry thunderstorm precipitation threshold of 1 mm / h to identify dry thunderstorm events. A thunderstorm risk factor (DRF) is defined: when DRF=1, it indicates frequent lightning but weak precipitation. Physically, this reflects sufficiently strong atmospheric convection to trigger lightning, but raindrops evaporate during their fall due to the high temperature and low humidity environment (virga phenomenon), failing to effectively increase surface humidity, thus significantly increasing the ignition potential of surface combustibles and dry vegetation. Within the monitoring date n, if DRF=1 at any moment, the corresponding monitoring subunit is marked as a "high ignition probability zone," triggering the first warning command. A dry thunderstorm risk layer is generated through grid aggregation, enabling visualization and spatial distribution assessment of high ignition potential areas. This method can identify high-risk fire areas caused by dry thunderstorms in a timely and accurate manner, making up for the shortcomings of traditional fire risk warnings in identifying local severe convective weather. It provides a scientific basis for the implementation of priority patrol and control strategies, and effectively improves the pertinence and efficiency of forest fire risk prevention and control.

[0178] Example 6

[0179] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figures 1 to 5 Specifically, the vegetation fuel characteristic analysis module includes a canopy height distribution extraction unit, a canopy closure calculation unit, and a forest understory shrub and grass layer thickness measurement unit.

[0180] The canopy height distribution extraction unit is used to extract the canopy height distribution curve within the monitoring subunit from a forest 3D point cloud model constructed using LiDAR point cloud data or stereo image pairs, and to calculate the average canopy height Hmean, maximum canopy height Hmax, and the standard deviation σh of the height distribution.

[0181] ;

[0182] Where N is the total number of point clouds within the monitoring sub-unit. Let p be the height of the p-th point;

[0183] The canopy closure calculation unit is used to calculate the canopy closure C of the monitoring sub-unit based on remote sensing imagery and point cloud porosity analysis technology. The range is 0-1, representing the proportion of the ground surface covered by the canopy. The formula for calculating canopy closure C is:

[0184] ;

[0185] Wherein, GapArea is the area of ​​visible voids on the ground surface, and TotalArea is the total area of ​​the monitoring sub-units;

[0186] The understory shrub and grass layer thickness measurement unit is used to calculate the average thickness (Hunder) of shrubs and herbaceous vegetation within the monitoring subunit based on low-level point cloud echoes or high-resolution image texture features. The calculation formula is as follows:

[0187] ;

[0188] Where M represents the number of low-level pixels within the monitoring sub-unit. The height of the qth low-level pixel; the low level includes the height between 0 and 2m above the ground.

[0189] In this embodiment, the refined extraction and analysis of the vertical structure of vegetation within the monitoring sub-unit enables a quantitative assessment of fire risk potential. The canopy height distribution extraction unit utilizes LiDAR point clouds or stereo imagery to construct a three-dimensional forest point cloud model, extracting the canopy height distribution curve within the monitoring sub-unit and calculating the average canopy height Hmean, maximum canopy height Hmax, and standard deviation σh of the height distribution. This accurately characterizes the potential path of surface fire spreading upwards to the canopy and the possibility of vertical fire propagation between canopies. The canopy closure calculation unit, based on remote sensing imagery and point cloud porosity analysis technology, calculates the canopy closure C, reflecting the proportion of the ground surface covered by the canopy, providing a basis for assessing the ability of fire to spread horizontally. The understory shrub and grass layer thickness measurement unit calculates the average thickness Hunder of shrubs and herbaceous vegetation within a height range of 0–2 meters using low-level point cloud echoes or high-resolution image texture features. This helps quantify the "ladder fuel" effect, i.e., the risk of surface fire spreading upwards along low-level vegetation to the canopy. This module can comprehensively and precisely characterize the vertical and horizontal fuel structure of forests, providing scientific data support for comprehensive fire risk assessment. It makes up for the lack of quantitative analysis of stand heterogeneity and vertical fuel continuity in traditional methods, and improves the accuracy and pertinence of fire risk early warning.

[0190] Example 7

[0191] This embodiment is an explanation based on Embodiment 6. Please refer to it. Figures 1 to 5 Specifically, the vegetation fuel characteristic analysis module includes a fuel continuity analysis unit and a combustible species identification unit;

[0192] The fuel continuity analysis unit is used to normalize and process the average canopy height Hmean, maximum canopy height Hmax, standard deviation of height distribution σh, canopy closure C, and average thickness of shrubs and herbaceous vegetation Hunder using a clipping function to obtain the relative canopy uniformity m, canopy vertical stratification intensity s, relative amount of understory "ladder fuel" thickness u, and horizontal canopy closure c.

[0193] ;

[0194] ;

[0195] ;

[0196] ;

[0197] The clipping function lip(x,0,1)=min(max(x,0),1) can uniformly set m,s,um,s,um,s,u to 0 and handle the shrub / grassland cases separately to avoid division by the minimum value, which may cause anomalies.

[0198] Physical meaning: m: relative canopy uniformity (the ratio of average canopy height to the highest canopy height). The closer to 1, the more uniform the canopy and the more even the canopy top, which is conducive to horizontal canopy-to-canopy spread. s: vertical canopy stratification intensity (a relative measure of height dispersion). The larger the value, the more layers there are and the more obvious the tiers, which is conducive to the vertical connection of the ground surface, understory, and canopy ("ladder fuel").

[0199] u: Relative thickness of the understory "ladder fuel". The larger the value, the easier it is for combustibles to "bridge" between the ground and the canopy, which is beneficial for vertical ignition. c: Horizontal connectivity / canopy closure. The larger the value, the more continuous the canopy is horizontally, which is beneficial for the lateral spread of crown fire. If Hmax < 1 m (shrub / grass / bare ground scenario), we can set m = s = u = 0;

[0200] Construct the horizontal continuity component and obtain the horizontal continuity index Hhor:

[0201] ;

[0202] Among them, high canopy coverage (larger c) and more uniform canopy height (larger m) are more conducive to horizontal propagation of the crowns; any deficiency will reduce the probability of the crown fire continuing to cross (multiplicative suppression of the "weakest link effect").

[0203] Construct the vertical continuity component and obtain the vertical continuity index Hver:

[0204] ;

[0205] Among them, u: the thicker the understory, the easier it is to ignite surface fires in the tree canopy; s: the greater the canopy height dispersion and the more layers there are, the easier it is to form a continuous vertical combustible zone. and As weight, , The thickness of the understory is slightly greater than the degree of stratification;

[0206] The fuel continuity index (FCI) is calculated by combining the horizontal continuity index (Hhor) and the vertical continuity index (Hver).

[0207] ;

[0208] in, and As weight, , In most forest fire behaviors, the continuity of the horizontal canopy has a more direct impact on the large-scale spread rate, therefore ;FCI∈[0,1], the larger the value, the more continuous the space fuel and the higher the fire propagation potential.

[0209] Hmean represents the crown height level of the "dominant layer"; after being normalized with Hmax, m is obtained, which is used to measure the crown uniformity and affects the lateral continuity of the crown.

[0210] Hmax: Used as a reference elevation for scale normalization to make different forest stands comparable; it also reflects the potential vertical corridor length.

[0211] σh: Canopy height difference (layering degree); the larger the value, the more layers and levels the fuel continuity analysis unit has, which is conducive to the upward penetration of the fire.

[0212] C: Canopy density, which measures whether the horizontal combustible zone is continuous; the higher the density, the easier it is for coronavirus to spread.

[0213] Hunder: The thickness of the understory shrub and grass layer directly corresponds to the abundance of "ladder fuel"; the larger the layer, the more conducive it is to the growth of the canopy on the ground.

[0214] The combustible species identification unit is used to extract the coverage of high-VOC tree species in each monitoring subunit, including the coverage of pine, eucalyptus, rhododendron, and camphor family tree species, and to construct the volatile flammability index (VEI) for each monitoring subunit.

[0215] ;

[0216] Where A represents the number of high-VOC tree species, The coefficients represent the volatile matter weighting coefficients for tree species in category a, where pine is 0.8, eucalyptus is 0.95, rhododendron is 0.7, and Lauraceae is 0.6. This represents the coverage rate of tree species of type a in the r-th monitoring sub-unit;

[0217] A higher VEI indicates a higher proportion of high-VOC tree species coverage in the unit, higher susceptibility to combustion, and a higher potential fire hazard and rapid fire spread.

[0218] In this embodiment, the vegetation fuel characteristic analysis module achieves a quantitative and refined assessment of forest fire risk potential through fuel continuity analysis and combustible species identification. The fuel continuity analysis unit, based on the average canopy height Hmean, maximum canopy height Hmax, height distribution standard deviation σh, canopy closure C, and understory shrub and grass thickness Hunder of the monitoring subunit, calculates the relative canopy uniformity m, canopy vertical stratification intensity s, relative amount of understory "ladder fuel" thickness u, and horizontal connectivity c after trimming and normalization. This helps quantify the canopy's lateral propagation potential, vertical fire chaining ability, and the fire uplift channel from the ground surface to the understory to the canopy. The horizontal continuity index Hhor reflects the lateral continuity of the canopy, while the vertical continuity index Hver characterizes the potential for surface fire to be led to the canopy via understory combustibles. The combination of these two indices generates the fuel continuity coefficient FCI. A higher value indicates more continuous spatial fuel and a higher fire propagation potential, scientifically reflecting the contribution of the ladder fuel effect, horizontal canopy connectivity, and vertical stratification to fire spread. The combustible species identification unit further extracts the coverage rate of high-VOC tree species (such as pine, eucalyptus, rhododendron, and Lauraceae) within each monitoring sub-unit, and constructs a Volatile Flammability Index (VEI) by combining the tree species volatile matter weighting coefficient. A higher VEI value indicates a larger proportion of high-VOC tree species, higher susceptibility to combustion, and a greater potential fire hazard and faster fire spread. This module comprehensively and quantitatively characterizes the vertical and horizontal fuel structure features and combustible material distribution of forests, providing a scientific basis for comprehensive fire risk assessment and refined management strategies, improving the accuracy and targeting of fire risk warnings, and compensating for the neglect of vertical fuel continuity and high-flammable tree species coverage by traditional methods.

[0219] Example 8

[0220] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figures 1 to 5 Specifically, the fire risk comprehensive assessment module includes a comprehensive fitting unit and a judgment unit;

[0221] The comprehensive fitting unit is used to extract the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI) for each monitoring subunit. After normalization, the comprehensive fire hazard index (FFI) for each monitoring subunit is constructed.

[0222] ;

[0223] ;

[0224] in, , , , , As weight;

[0225] The determination unit is used to classify the risk level of the Fire Risk Index (FFI) for each monitoring sub-unit, specifically as follows:

[0226] When FFI > 0.7, the label indicates the first risk level, meaning that the monitoring subunit has continuous fuel supply, abundant combustibles, and unfavorable weather conditions for fire control, resulting in a high risk of rapid fire spread. The first control strategy is generated, including: daily patrols, prohibiting all open flames, including picnics, burning waste, and fireworks; setting up clear warning signs along forest lines and management roads, with at least one "No Open Flames" sign every 500 meters; and cleaning up dead branches, leaves, haystacks, waste wood, and other flammable waste in a 10-meter wide strip along the patrol route and within 50 meters of forest edges at least once a week.

[0227] When 0.4≤FFI≤0.7, the label is Level 2, indicating that the fuel conditions of the monitoring subunit are moderate, the meteorological conditions present a certain fire risk potential, and the fire spread rate is moderate. A second control strategy is generated, including: patrolling once every 2 days, setting up at least one "No Open Flame" sign every 800 meters; cleaning up dead branches, leaves, haystacks, waste wood and other flammable waste in a 10-meter wide strip along the patrol route and within 30 meters of forest roads and forest edges once every 2 weeks.

[0228] When FFI < 0.4, the label is Level 3, indicating that the fuel in the monitoring sub-unit is sparse or wet, the meteorological conditions are favorable for suppressing the spread of fire, and the fire risk is low. A third control strategy is generated, including: weekly patrols, setting up one "No Open Flame" sign every 1000 meters, and cleaning up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 3-5 meter wide strip along the patrol route and within 10 meters of forest roads and forest edges once a month.

[0229] When a monitoring subunit is identified as a 'high ignition probability zone' and triggers the first warning command, the monitoring subunits that issued the first warning command are counted to form a first priority control group, and the corresponding control strategies are implemented first.

[0230] In this embodiment, the integrated fitting unit and risk level determination enable quantitative assessment and refined management of the fire hazard potential of each monitoring sub-unit. The integrated fitting unit normalizes and weights the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI) to generate a comprehensive fire hazard index (FFI), which scientifically reflects the combined impact of meteorological conditions, fuel structure, and combustible material distribution on fire spread. The assessment unit classifies monitoring sub-units into three risk levels based on their Fire Fissure Index (FFI) values. The first risk level, with FFI > 0.7, indicates continuous fuel supply, abundant combustibles, unfavorable weather conditions for fire control, and a high potential for rapid fire spread. This triggers strict control strategies, such as high-frequency patrols, a complete ban on fire, and clearing dead branches, leaves, and flammable materials along patrol routes and forest edges. The second risk level, with FFI ≤ 0.4 and FFI ≤ 0.7, indicates moderate fuel conditions, some fire potential under favorable weather conditions, and a moderate fire spread rate. This triggers moderate-intensity control strategies, such as reduced patrol frequency, increased spacing between fire prohibition signs, and moderate cleanup frequency. The third risk level, with FFI < 0.4, indicates sparse or moist fuel, favorable weather conditions for suppressing fire spread, and a low fire risk. This triggers low-intensity control strategies, such as extended patrol cycles, increased sign spacing, and reduced cleanup frequency. Simultaneously, when a monitoring sub-unit is identified as a 'high ignition probability zone' and triggers the first warning command, a first-priority control group will be formed, prioritizing the implementation of corresponding control strategies to achieve rapid response and optimized resource allocation in high-risk areas. This module enables dynamic, tiered fire risk early warning and targeted management based on meteorological and fuel conditions, significantly improving the accuracy and effectiveness of forest fire risk prevention and control.

[0231] Example 9

[0232] A forest fire risk early warning method based on meteorological and vegetation data analysis is described in the following reference: Figure 6 This includes the following steps:

[0233] S1. Acquire high-resolution digital elevation model (DEM) and multispectral or hyperspectral remote sensing images. Divide the target forest area into several monitoring sub-units according to the set spatial resolution using regular grids. Identify and mark the vegetation type in each monitoring sub-unit.

[0234] S2. Based on the Global Positioning System (GPS) and astronomical epoch calculations, determine the monitoring date n and the latitude and longitude of the monitoring area. Divide the entire day into K time steps according to the time step Δt, and collect the hourly wind speed at the index of the kth time step. ,wind direction relative humidity Hourly precipitation With vapor pressure difference A meteorological dataset was established to construct the wet-dry cycle index (WDI), the monsoon effect index (MFI), and the dry thunderstorm risk factor (DRF). If the thunderstorm risk factor DRF = 1 at any time, the monitoring sub-unit corresponding to that date is marked as a 'high ignition probability zone' and the first warning instruction is triggered, indicating that there is a risk of lightning igniting dry vegetation or surface combustibles in that monitoring sub-unit. The dry thunderstorm risk layer is output through raster aggregation.

[0235] S3. Extract vegetation data from each monitoring sub-unit and calculate the fuel continuity coefficient (FCI) and volatile flammability index (VEI).

[0236] S4. Based on the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI), construct the comprehensive fire risk index (FFI) for each monitoring sub-unit, classify the risk level of the comprehensive fire risk index (FFI) of each monitoring sub-unit, obtain the corresponding risk level and corresponding control strategy, and combine with the first early warning instruction to generate a first priority control group for priority implementation of the corresponding control strategy.

[0237] Figure 6The clearly visible grid represents the monitoring sub-units for refined management of the target forest area, corresponding to step S1 of this method: "dividing the target forest area into regular grids according to a set spatial resolution." The layered structure depicted in the view, including understory shrubs, tree trunks, and canopies, along with symbolic satellite, weather station, and lightning symbols, represent the sources of vegetation and meteorological data acquisition, respectively. This corresponds to step S1: "acquiring high-resolution digital elevation model (DEM) and multispectral or hyperspectral remote sensing imagery," and step S2: establishing the meteorological dataset. The technical roadmap on the right clearly reveals the logical processing flow of this method. The first box, "Data Acquisition and Grid Division," summarizes step S1; the second box, "Meteorological Risk Analysis," corresponds to the process in step S2 of calculating WDI, MFI, and DRF based on meteorological data; the third box, "Quantification of Combustible Material Characteristics," corresponds to the process in step S3 of calculating FCI and VEI based on vegetation data; and the last box, "Comprehensive Assessment and Graded Early Warning," corresponds to step S4, which is the final step of constructing the comprehensive fire risk index FFI based on all the aforementioned indices, and performing risk classification and generating control strategies. In this embodiment, the target forest area is finely gridded using high-resolution DEM and multispectral or hyperspectral imagery, and vegetation type is identified for each monitoring sub-unit to achieve spatially precise acquisition of fuel distribution information; combined with hourly meteorological data such as wind speed, wind direction, relative humidity, precipitation, and vapor pressure difference, the wet-dry cycle index WDI, monsoon effect index MFI, and dry thunderstorm risk factor DRF are calculated, which can dynamically reflect the impact of extreme weather such as sudden wind field changes, heavy precipitation re-drying, and dry thunderstorms on fire risk potential during monsoon intrusion or retreat; by extracting canopy height... By considering factors such as canopy density, understory shrub and grass thickness, and coverage of high-VOC tree species, the fuel continuity coefficient (FCI) and volatile flammability index (VEI) are calculated to accurately quantify the fuel continuity and flammability of vertical ladders and horizontal canopies. A comprehensive fire risk index (FFI) is constructed by integrating these meteorological and fuel indicators. Based on the numerical values, monitoring sub-units are managed in a hierarchical manner. High-ignition-probability areas are prioritized to form a first-priority control group, achieving dynamic, hierarchical, and targeted fire risk warnings. This effectively improves the accuracy of forest fire prevention and control and the efficiency of patrol resource utilization, significantly reducing the risk of fire occurrence and rapid spread.

[0238] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0239] The above formulas are all derived from software simulation using a large amount of data and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the protection scope of the present invention.

Claims

1. A forest fire risk early warning system based on meteorological and vegetation data analysis, characterized in that, include: The monitoring area division module is used to divide the target forest area into several monitoring sub-units according to a set spatial resolution by regular grid division based on high-resolution digital elevation model (DEM) and remote sensing imagery, and to identify and extract the vegetation type in each monitoring sub-unit and mark it. The meteorological monsoon characteristic analysis module is used to determine the monitoring date n and the latitude and longitude of the monitoring area based on GPS and astronomical epoch calculations. It divides the entire day into K time steps Δt and collects the hourly wind speed at the index of the kth time step. ,wind direction relative humidity Hourly precipitation With vapor pressure difference Establish a meteorological dataset to construct: The wet-dry cycle index (WDI) is used to characterize the rapid re-drying effect of fine fuels after heavy precipitation. The Monsoon Effect Index (MFI) is used to characterize the exacerbating effect of sudden wind field changes during monsoon intrusion or retreat. Furthermore, dry thunderstorm characteristics were identified, and a dry thunderstorm risk factor DRF was constructed, based on hourly lightning frequency. >5 times / hour, and hourly precipitation When the thickness is less than 1mm, it is determined to be a high ignition probability zone, and the first warning command is triggered. The vegetation fuel characteristic analysis module is used to extract and calculate vegetation data within each monitoring sub-unit, including: Canopy height distribution, canopy closure, and understory shrub and grass layer thickness are used to calculate the fuel continuity coefficient (FCI). The proportion of combustible volatiles was extracted, including the coverage of pine, eucalyptus, rhododendron, and Lauraceae tree species, and a volatile flammability index (VEI) was constructed. The fire risk comprehensive assessment module is used to construct the comprehensive fire risk index (FFI) for each monitoring sub-unit based on the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI). It then classifies the comprehensive fire risk index (FFI) of each monitoring sub-unit into risk levels, obtains the corresponding risk level and corresponding control strategies, and generates a first priority control group based on the first early warning instruction to prioritize the implementation of the corresponding control strategies.

2. The forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 1, characterized in that, The monitoring area division module includes a data acquisition and preprocessing unit, a division unit, and a vegetation type identification unit; The data acquisition and preprocessing unit is used to set up an external data interface to receive or download multispectral or hyperspectral remote sensing image data from satellite / airborne / UAV platforms, as well as LiDAR point cloud and echo data from airborne or ground platforms, and to perform radiometric calibration, atmospheric correction, geometric correction, denoising, and cloud / snow masking processing; to perform ground / non-ground classification, outlier removal, and intensity normalization on the LiDAR point cloud; to unify the coordinate reference system and projection of the above data, to perform spatiotemporal registration and co-location correction with the high-resolution digital elevation model (DEM), and to resample and crop according to the initial spatial resolution R to generate a "vegetation dataset" for subsequent analysis; The division unit is used to divide the target forest area into grids according to the initial spatial resolution R in the DEM coordinate system, obtain a set of monitoring sub-units, and complete spatial labeling and indexing on the DEM. Monitoring Subunit Set , to This represents the first to the Zth monitoring sub-units; The vegetation type identification unit is used to employ a supervised classification method, train on the "vegetation dataset" of multispectral or hyperspectral images output by the data acquisition and preprocessing unit, classify it using a support vector machine, and output the vegetation category of each monitoring subunit. The vegetation categories include coniferous forest, broad-leaved forest, mixed forest, shrubland, and grassland.

3. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 2, characterized in that, The vegetation type identification unit includes a feature extraction unit and a classification unit; The feature extraction unit is used to extract spectral indices based on the "vegetation dataset" of multispectral or hyperspectral images output by the data acquisition and preprocessing unit. These spectral indices include: Normalized Difference Vegetation Index (NDVI), Red Edge Vegetation Index (REVI), and Shortwave Infrared Moisture Index (MSI), expressed as follows: Wherein, NIR is the near-infrared band reflectance, RED is the red band reflectance, RE is the red edge band reflectance, and SWIR is the short-wave infrared band reflectance; The gray-level co-occurrence matrix P(i,j) is constructed through the following steps: Spatial texture features are extracted, specifically by quantizing the target band or principal component image to L=32 gray levels. The joint occurrence frequency of gray-level values ​​j in neighboring pixels with a window size w=7×7 pixels, a distance d=1, and an orientation θ∈{0°,45°,90°,135°} is used. The statistical results are then normalized to obtain the gray-level co-occurrence matrix P(i,j). Where N(i,j) is the number of times gray level i and gray level j co-occur; i represents the gray value of the reference pixel, j represents the gray value of the neighboring pixels of the reference pixel at a specified direction and distance, and P(i,j) represents the joint probability of the pixel with gray level i and the neighboring pixel with gray level j appearing in the window. The texture index is obtained by calculating and averaging the directions, including: Contrast CON: ; Homogeneous HOM: ; Energy / Angle Second Moment (ASM): ; Entropy ENT: ; where P(i,j)=0, this term is ignored; Correlation COR: in, Indicates the average gray value in the row direction. This indicates the average gray level along the column direction. Indicates the standard deviation in the row direction. Indicates the standard deviation in the column direction. The classification unit, used for classification based on spectral and texture indices, includes: When the NDVI value of the monitoring subunit is between 0.65 and 0.85, the REVI value is between 0.10 and 0.20, the MSI value is between 0.45 and 0.55, and the texture index contrast (CON) is between 10 and 20, the homogeneity (HOM) is between 0.70 and 0.85, the energy (ASM) is between 0.55 and 0.70, the entropy (ENT) is between 3 and 4, and the correlation (COR) is between 0.85 and 0.95, the monitoring subunit is determined to be a coniferous forest. When the NDVI value of the monitoring subunit is between 0.60 and 0.80, the REVI value is between 0.15 and 0.25, the MSI value is between 0.50 and 0.60, and the texture index CON is between 15 and 25, HOM is between 0.60 and 0.75, ASM is between 0.50 and 0.65, ENT is between 3.5 and 4.5, and COR is between 0.80 and 0.90, it is identified as a broad-leaved forest. When the NDVI value of the monitoring subunit is between 0.55 and 0.75, the REVI value is between 0.12 and 0.22, the MSI value is between 0.50 and 0.65, and the texture index CON is between 18 and 28, HOM is between 0.55 and 0.70, ASM is between 0.45 and 0.60, ENT is between 4 and 5, and COR is between 0.75 and 0.85, it is determined to be a mixed forest. When the NDVI value of the monitoring subunit is between 0.35 and 0.55, the REVI value is between 0.08 and 0.15, the MSI value is between 0.60 and 0.75, and the texture index CON is between 22 and 35, HOM is between 0.50 and 0.65, ASM is between 0.40 and 0.55, ENT is between 4.5 and 5.5, and COR is between 0.70 and 0.80, it is determined to be a shrub. When the NDVI value of the monitoring subunit is between 0.20 and 0.40, the REVI value is between 0.05 and 0.10, the MSI value is between 0.65 and 0.80, and the texture index CON is between 12 and 20, HOM is between 0.45 and 0.60, ASM is between 0.35 and 0.50, ENT is between 4 and 5, and COR is between 0.65 and 0.75, it is determined to be grassland.

4. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 1, characterized in that, The meteorological monsoon characteristic analysis module includes a meteorological data acquisition unit, a wet-dry cycle index calculation unit, and a monsoon effect index calculation unit. The meteorological data acquisition unit is used to acquire hourly wind speed within the monitoring date n, using a time step Δt to form a time index k=1,…,K. ,wind direction relative humidity Hourly precipitation With vapor pressure difference Establish a meteorological dataset; And decompose scalar wind speed and wind direction into two-dimensional vectors: , ; This represents the wind speed component in the east-west direction. This indicates the wind speed component in the north-south direction; The wet-dry cycle index calculation unit is used to characterize the effect of "rapid wetting of fine fuel by heavy precipitation, but subsequent sunny weather and high vapor pressure difference leading to rapid re-drying within 1-2 days". It uses the most recent W-hour window for calculation, setting W=48h, to calculate the humidity. and drying tensile strength coefficient DI; Where W represents the set of the most recent W hours. The scaling parameter is set to 10mm. This represents the number of valid time points within the window. When the research object is evapotranspiration, The effective evapotranspiration period is defined as the time when the atmosphere has reached a certain level of water deficit, and the surface of plant leaves begins to experience transpiration pull. This determines which moments within the statistical window qualify as valid samples. This indicates that the air is extremely dry, exceeding the vegetation's regulatory capacity, and there is a risk of causing the vegetation's stomata to close or even causing physiological stress. This indicates the reference vapor pressure differential risk identification threshold, set to 2 kPa, used to determine "whether it has entered an evaporation stress state"; Combined with humidity After dimensionless processing of the drying tensile strength coefficient DI, the wet-dry cycle index WDI is calculated: 。 5. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 4, characterized in that, The monsoon effect index calculation unit is used to characterize the abrupt changes in wind field and the exacerbating effect of low-level jet streams during monsoon intrusion / retreat. When rapid changes in wind speed / direction favor fire spread and uphill movement, the wind vector difference is calculated to obtain the monsoon effect index MFI. in, This represents the wind speed component in the east-west direction. This indicates the wind speed component in the north-south direction; and This represents the change in the wind vector component between two adjacent time intervals.

6. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 5, characterized in that, The meteorological monsoon characteristic analysis module also includes a dry thunderstorm risk identification unit, which is used to identify hourly lightning frequency. and hourly precipitation And preset lightning threshold And dry thunderstorm precipitation threshold , Set to 5 times / hour; Set to 1mm / h; Define the Thunderstorm Risk Factor (DRF): If the thunderstorm risk factor DRF=1 at any time within the monitoring date n, the monitoring sub-unit corresponding to that date is marked as a 'high ignition probability zone' and the first warning command is triggered, indicating that there is a risk of lightning igniting dry vegetation or surface combustibles in the monitoring sub-unit. The dry thunderstorm risk layer is then output through raster aggregation.

7. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 1, characterized in that, The vegetation fuel characteristic analysis module includes a canopy height distribution extraction unit, a canopy closure calculation unit, and a forest understory shrub and grass layer thickness measurement unit. The canopy height distribution extraction unit is used to extract the canopy height distribution curve within the monitoring subunit using a forest 3D point cloud model constructed from LiDAR point cloud data or stereo image pairs, and to calculate the average canopy height Hmean, the maximum canopy height Hmax, and the standard deviation σh of the height distribution. Where N is the total number of point clouds within the monitoring sub-unit. Let p be the height of the p-th point; The canopy closure calculation unit is used to calculate the canopy closure C of the monitoring subunit based on remote sensing images in the "vegetation dataset" and through point cloud porosity analysis technology. The range is 0~1, representing the proportion of the ground surface covered by the canopy. The formula for calculating the canopy closure C is: Wherein, GapArea is the area of ​​visible voids on the ground surface, and TotalArea is the total area of ​​the monitoring sub-unit; The understory shrub and grass layer thickness measurement unit is used to calculate the average thickness Hunder of shrubs and herbaceous vegetation within the monitoring subunit based on the low-level point cloud echoes in the "vegetation dataset" of hyperspectral imagery. The calculation formula is as follows: Where M represents the number of low-level pixels within the monitoring sub-unit. The height of the qth low-level pixel is defined as the height between 0 and 2m above the ground.

8. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 7, characterized in that, The vegetation fuel characteristic analysis module also includes a fuel continuity analysis unit and a combustible species identification unit; The fuel continuity analysis unit is used to normalize and process the average canopy height Hmean, maximum canopy height Hmax, standard deviation of height distribution σh, canopy closure C, and average thickness Hunder of shrubs and herbaceous vegetation using a clipping function to obtain the relative canopy uniformity m, canopy vertical stratification intensity s, relative amount of understory "ladder fuel" thickness u, and horizontal canopy closure c. Construct the horizontal continuity component and obtain the horizontal continuity index Hhor: Construct the vertical continuity component and obtain the vertical continuity index Hver: in, and As weight, , The thickness of the understory is slightly greater than the degree of stratification; The fuel continuity index (FCI) is calculated by combining the horizontal continuity index (Hhor) and the vertical continuity index (Hver). in, and As weight, , ; The combustible species identification unit is used to extract the coverage rate of high VOC tree species in each monitoring subunit, including the coverage rate of pine, eucalyptus, rhododendron, and camphor family tree species, and to construct the volatile flammability index (VEI) for each monitoring subunit. in, Indicates the number of high-VOC tree species. The coefficients represent the volatile matter weighting coefficients for tree species in category a, where pine is 0.8, eucalyptus is 0.95, rhododendron is 0.7, and Lauraceae is 0.

6. This represents the coverage rate of tree species of type a in the r-th monitoring sub-unit.

9. A forest fire risk early warning system based on meteorological and vegetation data analysis according to claim 1, characterized in that, The fire risk comprehensive assessment module includes a comprehensive fitting unit and a judgment unit; The comprehensive fitting unit is used to extract the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI) for each monitoring subunit. After normalization, the comprehensive fire hazard index (FFI) for each monitoring subunit is constructed. in, , , , , As weight; The determination unit is used to classify the risk level of the comprehensive fire risk index (FFI) of each monitoring sub-unit, specifically as follows: When FFI > 0.7, the label is set to the first risk level, and the first control strategy is generated, including: patrolling once a day, prohibiting all open flames, including picnicking, burning waste and fireworks; setting up obvious warning signs along the forest line and management roads, with at least one "No Open Flames" sign every 500 meters; and cleaning up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 10-meter wide strip along the patrol route and within 50 meters of the forest edge at least once a week. When 0.4≤FFI≤0.7, the label is set as the second risk level, and a second control strategy is generated, including: patrolling once every 2 days, setting up at least one "No Open Flame" sign every 800 meters; cleaning up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 10-meter wide strip along the patrol route and within 30 meters of forest roads and forest edges once every 2 weeks; When FFI < 0.4, the label is the third risk level, and the third control strategy is generated, including: patrol once a week, set up a "No Open Flame" sign every 1000 meters, and clean up dead branches, leaves, haystacks, waste wood and other flammable garbage in a 3-5 meter wide area along the patrol route and within 10 meters of forest roads and forest edges once a month. When a monitoring subunit is identified as a 'high ignition probability zone' and triggers the first warning command, the monitoring subunits that issued the first warning command are counted to form a first priority control group, and the corresponding control strategies are implemented first.

10. A forest fire risk early warning method based on meteorological and vegetation data analysis, applied to the forest fire risk early warning system based on meteorological and vegetation data analysis as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. Acquire high-resolution digital elevation model (DEM) and multispectral or hyperspectral remote sensing images. Divide the target forest area into several monitoring sub-units according to the set spatial resolution using regular grids. Identify and mark the vegetation type in each monitoring sub-unit. S2. Based on the Global Positioning System (GPS) and astronomical epoch calculations, determine the monitoring date n and the latitude and longitude of the monitoring area. Divide the entire day into K time steps according to the time step Δt, and collect the hourly wind speed at the index of the kth time step. ,wind direction relative humidity Hourly precipitation With vapor pressure difference Establish a meteorological dataset; The wet-dry cycle index (WDI), the monsoon effect index (MFI), and the dry thunderstorm risk factor (DRF) are constructed. If the thunderstorm risk factor DRF = 1 at any time, the monitoring sub-unit corresponding to that date is marked as a 'high ignition probability zone' and the first warning command is triggered, indicating that there is a risk of lightning igniting dry vegetation or surface combustibles in the monitoring sub-unit. The dry thunderstorm risk layer is output through raster aggregation. S3. Extract vegetation data within each monitoring sub-unit and calculate the fuel continuity coefficient (FCI) and volatile flammability index (VEI). S4. Based on the wet-dry cycle index (WDI), monsoon effect index (MFI), dry thunderstorm risk factor (DRF), fuel continuity coefficient (FCI), and volatile flammability index (VEI), construct the comprehensive fire risk index (FFI) for each monitoring sub-unit, classify the risk level of the comprehensive fire risk index (FFI) of each monitoring sub-unit, obtain the corresponding risk level and corresponding control strategy, and combine with the first early warning instruction to generate a first priority control group for priority implementation of the corresponding control strategy.

Images