A forest fire risk prediction method based on a weighted superposition model

Abstract

This application discloses a forest fire risk prediction method based on a weighted superposition model, comprising: S1, dividing the target area into several grids; S2, setting up a first quadrat in each grid or at each grid intersection, taking soil samples in each first quadrat and calculating the flammability index of the soil layer; S3, selecting representative grids, setting up second quadrats in each representative grid, collecting litter in each second quadrat and calculating the flammability index of the litter layer; S4, calculating the fire resistance index of the canopy based on the stand characteristics and the burning characteristics of tree species branches and leaves in the target area; S5, after the post-fire growing season, selecting representative grids in the forest fire-affected area, and calculating the fire resistance index of different tree species based on the diameter of all charred trunks and suckers and the total number of trunks and suckers in the representative grids; S6, standardizing the index data and dividing them into several levels to obtain the fire risk level.

Description

Technical Field

[0001] This application relates to the field of natural disaster prevention and control technology, specifically to a forest fire risk prediction method based on a weighted superposition model. Background Technology

[0002] Existing forest fire prediction technologies mainly fall into two categories: First, remote sensing imagery-based forest fire prediction technology. This relies on collecting multi-source monitoring data on forest areas, including temperature, precipitation, wind speed, wind direction, relative humidity, vegetation type, accumulation of combustible materials on the ground, and terrain orientation. It then uses machine learning, artificial intelligence, neural networks, and other methods combined with dynamic risk assessment models to predict fire risk. Second, experimental measurement of the flammability of tree species. This technology collects data on tree species, specifically gathering flammability indicators from litter, leaves, branches, core material, and bark. It then proposes prediction formulas to obtain fire risk predictions for the forest canopy.

[0003] While the above technologies can predict forest fires to some extent, their application has significant limitations. Forest fire prediction technologies based on remote sensing imagery suffer from shortcomings such as insufficient detail (insufficient resolution, inability to obtain real-time data such as the three-dimensional distribution of combustibles), lack of clarity (cloud and canopy obstruction), inaccurate calculations (model errors and false alarms), and difficulty in application. This is because models trained in a specific region (such as northern forests) may have poor generalization ability when transferred to a different ecosystem (such as Mediterranean shrublands or tropical rainforests) due to completely different combustible types, climates, and topography. Forest fire prediction technologies based on experimental measurements of tree species flammability have a core weakness: they typically only consider the forest canopy or the flammability of the surface litter layer, lacking assessment of soil flammability and the regenerative capacity of trees after a fire, resulting in significant discrepancies between predicted and actual results. In addition, although flammability tests can accurately measure parameters such as ignition point and calorific value of a single tree species, they cannot reproduce the influence of key dynamic factors such as wind force, terrain, and sudden changes in humidity in the field on the initiation or control of fire.

[0004] Meanwhile, the spatial heterogeneity of forest stand structure (such as mixed distribution of tree species and continuity of combustible load) and sudden human factors (such as lightning strikes and fire sources) cannot be effectively simulated in controlled experiments, which may lead to significant deviations in the prediction results at the macro scale and in actual emergency response, and thus cannot fully support accurate spatiotemporal early warning.

[0005] Application content

[0006] The purpose of this application is to provide a forest fire risk prediction method based on a weighted superposition model, and the specific technical solution is as follows:

[0007] A forest fire risk prediction method based on a weighted superposition model includes: S1, dividing the target area into several grids; S2, setting up first quadrats in each grid or at each grid intersection, collecting soil samples in each first quadrat and calculating the soil flammability index; S3, selecting representative grids, setting up second quadrats in each representative grid, collecting litter in each second quadrat and calculating the litter layer flammability index; S4, calculating the canopy flame retardancy index based on forest stand characteristics and tree species leaf combustion characteristics within the target area; S5, analyzing the forest... After the post-fire growing season, representative grids are selected in the burned area. The fire resistance index of different tree species is calculated based on the diameter of all charred trunks and suckers in the representative grids and the total number of trunks and suckers. S6. The flammability index of the soil layer, the flammability index of the litter layer, the flame retardancy index of the canopy, and the fire resistance index of different tree species calculated in S5 are numerically standardized. The standardized data are equally divided into several levels, and the fire risk level is calculated according to the weight of each index.

[0008] The calculation of the soil flammability index in S2 includes: S2.1, using the quincunx sampling method and the classic ring sampler method to collect soil samples in each first sample plot; S2.2, measuring the organic matter content, texture coefficient, soil layer thickness, soil moisture content, soil bulk density, and acidity / alkalinity of the collected soil layers; S2.3, normalizing or logarithmizing the measured indicators; S2.4, calculating the soil flammability index using the processed data.

[0009] The calculation of the flammability index of the litter layer in S3 includes: S3.1, periodically collecting litter in each second quadrat; S3.2, drying the collected litter and weighing the fallen leaves, branches, and bark; S3.3, calculating the steady-state litter accumulation in the target area using the Olsen model based on the collected litter data; S3.4, calculating the flammability index of the litter layer based on the obtained steady-state litter accumulation.

[0010] The calculation of the canopy flame retardancy index in S4 includes: S4.1, obtaining forest stand characteristics through publicly available data of the target area; S4.2, obtaining the burning characteristics of tree species branches and leaves through publicly available results; S4.3, calculating the canopy flame retardancy index after standardizing the data obtained in S4.1 and S4.2.

[0011] S5 calculates the fire resistance index for different tree species, including: S5.1, measuring the diameter of all charred trunks and suckers in each representative grid and counting the total number of charred trunks and suckers; S5.2, calculating the composite trunk area for each sampled plant based on the diameter of the main burning trunk and the number of burning trunks of the same plant, and then calculating the composite trunk diameter; S5.3, calculating the composite sucker area for each sampled plant based on the diameter of the main burning sucker and the number of burning suckers of the same plant, and then calculating the composite sucker diameter; S5.4, calculating the fire resistance index for different tree species based on the number of tree species, the composite trunk diameter, and the composite sucker diameter.

[0012] The expression for calculating the soil flammability index in S2 is:

[0013] FFI=(OMC)×(D)×(1 / W)×(Ts)×(1 / BD)×(Ap),

[0014] Among them, OMC is the organic matter content, measured by the loss on ignition method; the higher the value, the higher the flammability. D is the soil layer thickness; the higher the value, the more fuel is in the deeper layers. W is the soil layer moisture content, measured by the gravimetric moisture content conversion method to obtain the instantaneous volumetric moisture content, and is expressed in reciprocal form to reflect the degree of dryness. Ts is the texture coefficient, assigned according to the soil texture type: sandy soil = 1.2, loam = 1.0, clay = 0.8, with sandy soil having a higher coefficient. BD is the soil layer bulk density, expressed in reciprocal form to reflect the promoting effect of low bulk density soil. Ap is the acidity / alkalinity index, assigned based on pH value: 1.2 for pH < 5.5, 1.0 for pH = 5.5–7.0, and 0.8 for pH > 7.0; acidic soil has a higher coefficient, which is conducive to the accumulation of combustibles.

[0015] The expression for the steady-state litter accumulation in S3 is:

[0016] L = I / K,

[0017] Where L is the steady-state litter accumulation, I is the annual litter input, and K is the decomposition rate constant. The value of k is measured in the field through the "litter bag experiment": fresh litter of representative tree species in the target forest sub-compartment is dried at 65℃ to obtain an accurate initial dry weight, and then packaged into nylon mesh bags with a mesh size of about 1-2 mm. Subsequently, the bags are randomly placed in the natural environment of the forest patch to ensure full contact with the litter layer. After 1, 3, 6, and 12 months, the litter bags are collected, dried again at 65℃, and weighed to obtain the remaining dry weight at time t. Based on these data, the remaining mass ratio at each time point can be calculated. The logarithmically transformed remaining mass ratio is linearly regressed with time t, and the slope of the resulting regression line is the negative value of the decomposition rate constant k.

[0018] The expression for the flammability index of the litter layer in S3 is:

[0019] FRI = L × 2e (-0.45+0.987×ln(DF)+0.0338×T-0.0345×H+0.0234×W) ,

[0020] Where DF is the drought factor, T is the air temperature, H is the relative humidity, and W is the wind speed; the expression for DF is:

[0021] DF = 10·e -0.03·P ,

[0022] When performing DF calculation, historical meteorological data is reviewed to find the end point of a continuous precipitation event as the starting date for DF value calculation. The data is then superimposed, and a sufficient total precipitation is selected. The starting date for DF value calculation is the total precipitation in the past 5-7 days that exceeds 100 mm. P is the precipitation since the end of the continuous rainfall event.

[0023] The expression for the composite trunk area of ​​each sampled plant in S5 is as follows:

[0024] ECTA = π × (MTD / 2) 2 ×TN,

[0025] Among them, MTD refers to the diameter of the most prominent and thickest trunk of a plant that has been burned, measured using the standard of diameter at breast height (DBH), which is the diameter of the trunk measured at a height of 1.3 meters above the ground; TN refers to the number of burned trunks produced on the same plant.

[0026] The expression for the area of ​​the compound suckers of each sampled plant in S5 is as follows:

[0027] ECRA = π × (MRD / 2) 2 ×RN,

[0028] Wherein, MRD is the diameter of the thickest new shoot that grows from the base of the plant after a fire, and RN is the total number of new shoots that sprout from the same plant after a fire.

[0029] The expression for the diameter of the composite trunk in S5 is:

[0030]

[0031] The expression for the diameter of the compound sucker in S5 is:

[0032]

[0033] The expressions for the fire resistance index of different tree species in S5 are as follows:

[0034] RCI = Number of tree species × (ECRD / ECTD).

[0035] The expression for the fire risk level index in S6 is:

[0036] ZHFX = W FFI ×FFI+W FRI ×FRI-W FCI ×FCI-W RCI ×RCI,

[0037] Among them, FFI is the soil flammability index, FRI is the litter layer flammability index, FCI is the canopy flame retardancy index, RCI is the fire resistance index of different tree species, and W FFI W FRI W FCI and W RCI These are the weights of the corresponding indices.

[0038] The beneficial effects of this application lie in its comprehensive consideration of the flammability indices of the soil layer, surface layer, and canopy layer. It also incorporates the fire resistance index (post-fire regeneration capacity – recovery capacity) of forest sub-compartments into the fire risk classification of forest sub-compartments, resulting in a more comprehensive range of factors and taking into account multiple possible spatial levels of fire occurrence. This application employs a weighted overlay model, scientifically calculating and standardizing each fire influencing factor before representing it in a hierarchical manner. The logic is clear, the process transparent, and it is easy to implement, enabling a comprehensive and quantitative consideration of multiple influencing factors. This technology represents a classic paradigm combining field investigation (experimentation) with GIS spatial analysis. The factor values ​​obtained from field investigations (such as surface litter load, soil properties, canopy flammability characteristics, and post-disaster regeneration rate) are first-hand, real data with a much higher accuracy than secondary data obtained through remote sensing inversion or model extrapolation. This directly overcomes the shortcomings of optical remote sensing, such as susceptibility to cloud and fog obstruction, difficulty in penetrating the canopy, and insufficient resolution, ensuring the reliability of the model input data and reducing errors at the source. This application can also output data in GIS as a heat map, a crucial step in transforming complex scientific data into intuitive decision-making information. It can not only clearly show the spatial distribution of risk levels in different areas, but also accurately locate high-risk areas, thus providing direct and visual scientific basis for the precise allocation of fire prevention resources (such as the location of watchtowers, patrol route planning, and configuration of fire-fighting facilities) and the formulation of emergency response plans, greatly enhancing the practical value of the results.

[0039] Instruction manual illustrations

[0040] Figure 1 This is a schematic diagram showing the spatial location of the forest compartments in this application;

[0041] Figure 2 This is a schematic diagram of principal component analysis of the burning characteristics of tree branches and leaves in this application;

[0042] Figure 3 This is a schematic diagram in the form of a heat map showing the comprehensive fire risk index of each forest compartment in this application. Specific Implementation

[0043] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to specific embodiments and accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of this application. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0044] A forest fire risk prediction method based on a weighted superposition model includes:

[0045] S1. Divide the target area into several grids.

[0046] Specifically, in practical applications, forest survey plot data is imported into ArcGIS software and divided into several grids of 40m×40m on the map, with each grid containing as few as possible a single species.

[0047] S2. Set up the first sample plot in each grid or at each grid intersection, take soil samples in each first sample plot and calculate the flammability index of the soil layer.

[0048] Specifically, the calculation of the soil flammability index includes: S2.1, using the quincunx sampling method and the classic ring sampler method to collect soil samples in each first sample plot; S2.2, measuring the organic matter content, texture coefficient, soil layer thickness, soil moisture content, soil bulk density, and pH index of the collected soil layer; S2.3, normalizing or logarithmizing the measured indicators; S2.4, calculating the soil flammability index using the processed data. The expression is:

[0049] FFI=(OMC)×(D)×(1 / W)×(Ts)×(1 / BD)×(Ap),

[0050] Among them, OMC is the organic matter content, measured by the loss on ignition method; the higher the value, the higher the flammability. D is the soil layer thickness; the higher the value, the more fuel is in the deeper layers. W is the soil layer moisture content, measured by the gravimetric moisture content conversion method to obtain the instantaneous volumetric moisture content, and is expressed in reciprocal form to reflect the degree of dryness. Ts is the texture coefficient, assigned according to the soil texture type: sandy soil = 1.2, loam = 1.0, clay = 0.8, with sandy soil having a higher coefficient. BD is the soil layer bulk density, expressed in reciprocal form to reflect the promoting effect of low bulk density soil. Ap is the acidity / alkalinity index, assigned based on pH value: 1.2 for pH < 5.5, 1.0 for pH = 5.5–7.0, and 0.8 for pH > 7.0; acidic soil has a higher coefficient, which is conducive to the accumulation of combustibles.

[0051] When performing normalization or logarithmic processing, OMC, D, W, etc. can be divided by the reference maximum value (e.g., OMC_max = 80%, D_max = 100cm, W_max = 50%).

[0052] S3. Select representative grids, and set up second quadrats in each representative grid. Collect litter in each second quadrat and calculate the flammability index of the litter layer.

[0053] Specifically, the calculation of the litter layer flammability index includes: S3.1, regularly collecting litter in each second quadrat. In practical applications, based on the main tree species and growth status of the forest subcommittee, the 40m×40m grid type to be surveyed is selected. In the representative grid, 1m×1m quadrats are arranged along three parallel transects, and litter is collected regularly (monthly); the fallen leaves, branches, bark, etc., are dried and weighed, continuing for at least one year to obtain a reliable annual average litter value; S3.2, the collected litter is dried and the fallen leaves, branches, and bark are weighed; S3.3, based on the collected litter data, the steady-state litter accumulation in the target area is calculated using the Olsen model; S3.4, the flammability index of the litter layer is calculated based on the obtained steady-state litter accumulation. The expression for steady-state litter accumulation is:

[0054] L = I / K,

[0055] Where L is the steady-state litter accumulation, I is the annual litter input, and K is the decomposition rate constant. The value of K is measured in the field through the "litter bag experiment": fresh litter of representative tree species in the target forest sub-compartment is dried at 65℃ to obtain an accurate initial dry weight, and then packaged into nylon mesh bags with a mesh size of about 1-2 mm. Subsequently, the bags are randomly placed in the natural environment of the forest patch to ensure full contact with the litter layer. After 1, 3, 6, and 12 months, the litter bags are collected, dried again at 65℃, and weighed to obtain the remaining dry weight at time t. Based on these data, the remaining mass ratio at each time point can be calculated. The logarithmically transformed remaining mass ratio is linearly regressed with time t, and the slope of the regression line is the negative value of the decomposition rate constant k.

[0056] The expression for the flammability index of litter layer is:

[0057] FRI = L × 2e (-0.45+0.987×ln(DF)+0.0338×T-0.0345×H+0.0234×W) ,

[0058] Where DF is the drought factor, T is the air temperature, H is the relative humidity, and W is the wind speed; the expression for DF is:

[0059] DF = 10 × e (-0.03×P) ,

[0060] When performing DF calculation, historical meteorological data is reviewed to find the end point of a continuous precipitation event as the starting date for DF value calculation. The data is then superimposed, and a sufficient total precipitation is selected. The starting date for DF value calculation is the total precipitation in the past 5-7 days that exceeds 100 mm. P is the precipitation since the end of the continuous rainfall event.

[0061] S4. Calculate the flame retardancy index of the canopy based on the forest stand characteristics and the burning characteristics of tree branches and leaves within the target area.

[0062] Specifically, the calculation of the canopy flame retardancy index includes: S4.1, obtaining forest stand characteristics through publicly available data of the target area; S4.2, obtaining the burning characteristics of tree species branches and leaves through publicly available results; S4.3, calculating the canopy flame retardancy index after standardizing the data obtained in S4.1 and S4.2.

[0063] S5. After the post-fire growing season, select representative grids for the forest fire-affected areas. Calculate the fire resistance index of different tree species based on the diameter of all charred trunks and suckers within the representative grids, as well as the total number of trunks and suckers.

[0064] Specifically, the calculation of the fire resistance index for different tree species includes: S5.1, measuring the diameter of all charred trunks and suckers in each representative grid and counting the total number of charred trunks and suckers; S5.2, calculating the composite trunk area for each sampled plant based on the diameter of the main burning trunk and the number of burning trunks of the same plant, and then calculating the composite trunk diameter accordingly; S5.3, calculating the composite sucker area for each sampled plant based on the diameter of the main burning sucker and the number of burning suckers of the same plant, and then calculating the composite sucker diameter accordingly; S5.4, calculating the fire resistance index for different tree species based on the number of tree species, the composite trunk diameter, and the composite sucker diameter. The expression for the composite trunk area of ​​each sampled plant is:

[0065] ECTA = π × (MTD / 2) 2 ×TN,

[0066] Among them, MTD refers to the diameter of the most prominent and thickest trunk of a plant that has been burned, measured using the standard of diameter at breast height (DBH), which is the diameter of the trunk measured at a height of 1.3 meters above the ground; TN refers to the number of burned trunks produced on the same plant.

[0067] The expression for the area of ​​compound suckers for each sampled plant is as follows:

[0068] ECRA = π × (MRD / 2) 2 ×RN,

[0069] Wherein, MRD is the diameter of the thickest new shoot that grows from the base of the plant after a fire, and RN is the total number of new shoots that sprout from the same plant after a fire.

[0070] The expression for the diameter of a composite tree trunk is:

[0071]

[0072] The expression for the diameter of compound suckers is:

[0073]

[0074] The expressions for the fire resistance index of different tree species are as follows:

[0075] RCI = Number of tree species × (ECRD / ECTD).

[0076] S6. The flammability index of the soil layer, the flammability index of the litter layer, the flame retardancy index of the canopy, and the fire resistance index of different tree species calculated in S5 are numerically standardized. The standardized data are equally divided into several levels, and the fire risk level is calculated according to the weight of each index.

[0077] Specifically, forestry fire prevention experts were invited to use the Delphi method to compare the importance of four influencing factors in forest fire risk estimation (soil flammability index, litter flammability index, canopy fire retardancy index, and forest fire resistance index). The first step was to verify expert consensus; therefore, Kendall's W test was used to check for consistency. If W > 0.5, strong consistency was considered. Under the premise of consistency, SPSS software was used to calculate the weight of each factor. After determining the weights, the comprehensive risk value of each forest fire risk level index unit was calculated based on the fire risk level index expression. This step can also be done using the "raster calculator" or "weighted overlay" tool in GIS to obtain a comprehensive risk raster. The fire risk level index expression is:

[0078] ZHFX = W FFI ×FFI+W FRI ×FRI-W FCI ×FCI-W RCI ×RCI,

[0079] Among them, FFI is the soil flammability index, FRI is the litter layer flammability index, FCI is the canopy flame retardancy index, RCI is the fire resistance index of different tree species, and W FFI W FRI W FCI and W RCI These are the weights for the corresponding index levels.

[0080] To make this application easier to understand, the following explanation is based on specific experiments.

[0081] Based on the data from the Third National Forestry Survey, 20 forest compartments were selected in the forests of Yajiang County, Ganzi Prefecture, Sichuan Province. Ten of these compartments are located in areas where fires have occurred in the past 20 years, while the other ten have never experienced a fire. Each compartment measures 40m × 40m (containing a single species). The specific locations are shown in Table 1, and a location map is provided below. Figure 1 As shown.

[0082] Table 1

[0083]

[0084] Calculation of the flammability index (FFI) of the soil layer:

[0085] The organic matter content, soil porosity, and proportion of each particle size in the soil layers of the small plots were measured. Indicators such as soil layer thickness, annual soil moisture content (drought level), soil bulk density, and pH were also analyzed. The following data (Table 2) were obtained through field measurements and calculations in the sample plots and then normalized.

[0086] Table 2

[0087]

[0088] The FFI value is calculated using FFI = (OMC) × (D) × (1 / W) × (Ts) × (1 / BD) × (Ap) (Table 3). Finally, based on the theoretical minimum and maximum values ​​calculated using the FFI formula, the FFI values ​​are standardized and equally divided into 5 levels.

[0089] Table 3

[0090]

[0091]

[0092] Flammability Risk Index (FRI) assessment of litter layer:

[0093] In a representative grid, 1m×1m quadrats were arranged along three parallel transects, and litter was collected periodically (monthly). The fallen leaves, branches, bark, etc., were dried and weighed, and the annual average litter value was obtained after one year, as shown in the table below. The litter data for 12 months were dried at 65℃ and weighed to obtain the remaining dry weight (Mt) at time t, and the remaining mass ratio (Mt / M0) at each time point was calculated. A linear regression was performed on the logarithmically transformed remaining mass ratio (ln(Mt / M0)) with time t. The slope of the resulting regression line is the negative value of the decomposition rate constant k (k = -slope), as shown in Table 4 below.

[0094] Table 4

[0095]

[0096]

[0097] We then calculated the spring drought factor (DF) for this region. Based on the meteorological data measured in 2025, the area where the forest compartment is located experienced continuous rainfall from April 2nd to April 5th, with a total precipitation exceeding 100 mm. Therefore, we used April 6th as the starting date for the calculation, with P representing the total precipitation of 20 mm from April 6th to 11th. Thus, the DF for April 11th was calculated as follows:

[0098] DF = 10 × e (-0.03×P) =5.49

[0099] Based on the FRI calculation formula, the FRI values ​​of 20 forest patches are shown in Table 5 below. They are then classified into 5 levels in SPSS.

[0100] Table 5

[0101]

[0102] Flame Retardant Index (FCI) Assessment of Forest Canopy

[0103] Based on the tree species involved in this study, and by consulting published literature, such as Zhong Jin, Zhu Yuyi, Qin Chao, et al. Evaluation of the combustibility of branches and leaves of 70 common tree species in Sichuan Province. Anhui Agricultural Sciences, 2024, 52(19):103-109+117, the combustibility characteristics of branches and leaves of all tree species (including moisture content, crude fat content, ignition point, calorific value and ash content of branches and leaves) were obtained, and the following Table 6 was improved by combining the third national forest survey data:

[0104] Table 6

[0105]

[0106] After standardizing the data in Table 6, principal component analysis was performed to obtain... Figure 2 In the PCA analysis, the loading coefficients or contribution values ​​of ignition point, average diameter at breast height (DBH), and planting density were low and therefore excluded, resulting in the data in Table 7.

[0107] Table 7

[0108]

[0109]

[0110] The formula for calculating the FCI value is:

[0111] FCI = 0.432 × moisture content - 0.279 × crude fat - 0.478 × calorific value + 0.044 × ash content + 0.021 × canopy closure / coverage.

[0112] Based on the above formula, we calculated the fire resistance FCI values ​​of the 20 forest compartments and their groupings as shown in Table 8 below:

[0113] Table 8

[0114]

[0115] Fire resistance index assessment of forest sub-compartments

[0116] In the local fire sites, during the first rainy season after the fire, forest compartments containing relevant tree species were identified, and representative 40m×40m grids were divided as quadrats. Subsequently, one year after the end of the post-fire growing season, detailed measurements were taken of the diameter of all charred trunks and suckers (Table 9), as well as the total number of trunks and suckers.

[0117] Table 9

[0118] Spruce 2 18 0.7 55 fir 22 26 1.3 24 Sichuan-Yunnan alpine oak (shrubland) 3 190 0.5 400 mountain pine 11 24 2.2 20 Rhododendron simsii 1 156 0.4 400

[0119] Next, using only the main burning trunk diameter (MTD) and the number of burning trunks (TN), we calculated the estimated composite trunk area (ECTA) for each sampled plant, using the following formula:

[0120] ECTA = π × (MTD / 2) 2 ×TN,

[0121] After a fire, a regenerating plant may have a prominent main trunk, surrounded by several thinner secondary trunks or sprouts. MTD specifically refers to the most important main trunk. Measurement method: The standard measurement is typically DBH (Diameter at Breast Height), which measures the trunk diameter at a height of 1.3 meters above the ground. For trees after a fire, even if the bark is burned, the diameter of the remaining trunk is measured at the original standard location. TN refers to the number of burned trunks that originate from the same plant individual (the same rootstock). These trunks are "multi-trunk" (multi-trunk per plant). They share the same root system but branch out into multiple trunks above ground. TN counts how many such main trunks a plant has. After a fire, knowing the number of trunks is crucial for calculating the overall damaged area or potential recovery potential.

[0122] Similarly, using the diameter of the main shoot (MRD) (i.e., the thickest shoot) and the number of burning shoots (RN), we calculated the estimated area of ​​the compound shoot (ECRA) for each sampled plant:

[0123] ECRA = π × (MRD / 2) 2 ×RN,

[0124] Meanwhile, the estimated composite trunk diameter (ECTD) and estimated composite sprout diameter (ECRD) of each sampled plant are also calculated using the following formula, converting the estimated composite area (ECTA / ECRA) into a derived value of the equivalent circular diameter, which facilitates comparison with the diameter of a single trunk / sprout.

[0125] After calculation, we obtained the data in Table 10 below:

[0126] Table 10

[0127] Spruce 56.52 21.15575 8.485281374 5.191338941 fir 9878.44 31.8396 112.1784293 6.368673331 Sichuan-Yunnan alpine oak (shrubland) 1342.35 78.5 41.35214626 10 mountain pine 2279.64 75.988 53.88877434 9.838699101 Rhododendron simsii 122.46 50.24 12.489996 8

[0128] We substitute ECRD and ECRD into the Fire Resistance Index (RCI) formula as follows, and then divide them equally into 5 levels.

[0129] RCI = Number of tree species × (ECRD / ECTD)

[0130] The final result is Table 11 below:

[0131] Table 11

[0132]

[0133] Calculation, classification, and visualization of the Comprehensive Fire Risk Index (ZHFX) for forest sub-compartments

[0134] Nineteen forestry fire prevention experts were invited to compare the importance of four influencing factors (soil flammability index, litter flammability index, canopy flammability index, and fire resistance index) in forest stand fire risk estimation using the Delphi method. First, Kendall's W test was used to check their consistency; the W value was 0.805, > 0.5, indicating consensus among the experts. Then, their weights were determined in SPSS, resulting in a suitable ZHFX calculation formula for the local area, as follows:

[0135] ZHFX=0.17×FFI+0.29×FRI-0.41×FCI-0.13×RCI,

[0136] Then, based on this formula and the risk grouping results of each indicator, the comprehensive risk value of each small class spatial unit is calculated, as shown in Table 12 below:

[0137] Table 12

[0138]

[0139] Finally, the GIS uses heat maps to display the comprehensive fire risk index (ZHFX) of each forest compartment, such as... Figure 3 As shown.

Claims

1. A forest fire risk prediction method based on a weighted superposition model, characterized in that, include: S1. Divide the target area into several grids; S2. Set up the first sample plot in each grid or at each grid intersection, take soil samples in each first sample plot and calculate the flammability index of the soil layer. S3. Select representative grids and deploy second quadrats in each representative grid. Collect litter in each second quadrat and calculate the flammability index of the litter layer. Specifically: S3.

1. Collect litter periodically in each second quadrat; S3.

2. Dry the collected litter and weigh the fallen leaves, branches, and bark; S3.

3. Calculate the steady-state litter accumulation in the target area using the Olsen model based on the collected litter data; S3.

4. Calculate the flammability index of the litter layer based on the obtained steady-state litter accumulation. S4. Calculate the flame retardancy index of the canopy based on the forest stand characteristics and the burning characteristics of tree branches and leaves within the target area; S5. After the post-fire growing season, representative grids are selected for the forest fire-affected areas. The fire resistance index of different tree species is calculated based on the diameters of all charred trunks and suckers within each representative grid, as well as the total number of trunks and suckers. Specifically: S5.

1. In each representative grid, the diameters of all charred trunks and suckers are measured, and the total number of charred trunks and suckers is counted. S5.

2. The composite trunk area of ​​each sampled plant is calculated using the diameter of the main burning trunk and the number of burning trunks of the same plant, and the composite trunk diameter is calculated accordingly. S5.

3. The composite sucker area of ​​each sampled plant is calculated using the diameter of the main burning sucker and the number of burning suckers of the same plant, and the composite sucker diameter is calculated accordingly. S5.

4. The fire resistance index of different tree species is calculated based on the number of tree species, the composite trunk diameter, and the composite sucker diameter. S6. The flammability index of the soil layer calculated in S2, the flammability index of the litter layer calculated in S3, the flame retardancy index of the canopy calculated in S4, and the fire resistance index of different tree species calculated in S5 are numerically standardized. Each standardized data is equally divided into several levels, and the fire risk level is calculated according to the weight of each index level.

2. The forest fire risk prediction method based on a weighted superposition model as described in claim 1, characterized in that, The calculation of the soil flammability index in S2 includes: S2.1 In each first plot, the quincunx method is used, and the classic ring cutter method is used to extract soil. S2.2 Measure the organic matter content, texture coefficient, soil layer thickness, soil layer water content, soil layer bulk density and acid-base index in the extracted soil layer; S2.3 Normalize or logarithmize the measured indicators; S2.4 Calculate the soil flammability index using the processed data.

3. The forest fire risk prediction method based on a weighted superposition model as described in claim 1, characterized in that, The calculation of the flame retardancy index of the canopy in S4 includes: S4.1 Obtain forest stand characteristics through publicly available data of the target area; S4.2 Obtain the combustion characteristics of tree branches and leaves through publicly available results; S4.3 Calculate the flame retardancy index of the canopy after standardizing the data obtained in S4.1 and S4.

2.

4. The forest fire risk prediction method based on the weighted superposition model as described in claim 2, characterized in that, The expression for calculating the soil flammability index in S2 is as follows: FFI =( OMC )×( D )×(1 / W )×( Ts )×(1 / BD )×( Ap )， in, OMC The organic matter content of the soil sample plot is determined by the loss on ignition method; the higher the value, the higher the flammability. D This represents the thickness of the soil layer; the higher the value, the more fuel is present in deeper layers. W The soil moisture content is measured using the gravimetric moisture content conversion method to obtain the instantaneous volumetric moisture content, and the reciprocal form is used to reflect the degree of dryness. Ts The texture coefficient is assigned according to the soil texture type: sandy soil = 1.2, loam = 1.0, clay = 0.8, with sandy soil having a higher coefficient. BD The reciprocal of the soil bulk density is used to reflect the promoting effect of low bulk density soils; Ap It is an acidity / alkalinity index, assigned based on pH value. It is 1.2 when pH < 5.5, 1.0 when pH = 5.5~7.0, and 0.8 when pH > 7.

0. A higher acidity index is conducive to the accumulation of combustibles.

5. The forest fire risk prediction method based on a weighted superposition model as described in claim 1, characterized in that, The expression for the steady-state litter accumulation in S3 is: L=I / K, Where L is the steady-state litter accumulation, I is the annual litter input, and K is the decomposition rate constant. The value of K is measured in the field through the "litter bag experiment": fresh litter of representative tree species in the target forest sub-compartment is dried at 65°C to obtain an accurate initial dry weight, and then packaged into nylon mesh bags with a mesh size of about 1-2 mm. Subsequently, the bags are randomly placed in the natural environment of the forest patch to ensure full contact with the litter layer. After 1, 3, 6, and 12 months, the litter bags are collected, dried again at 65°C, and weighed to obtain the remaining dry weight at time t. Based on these data, the remaining mass ratio at each time point can be calculated. The logarithmically transformed remaining mass ratio is linearly regressed with time t, and the slope of the resulting regression line is the negative value of the decomposition rate constant k. The expression for the flammability index of the litter layer in S3 is: FRI=L×2e (−0.45+0.987×ln(DF)+0.0338×T−0.0345×H+0.0234×W) , Where DF is the drought factor, T is the air temperature, H is the relative humidity, and W is the wind speed; the expression for DF is: DF=10 × e −0.03×P , When performing DF calculation, historical meteorological data is reviewed to find the end point of a continuous precipitation event as the starting date for DF value calculation. The data is then superimposed, and a sufficient total precipitation is selected. The starting date for DF value calculation is the total precipitation in the past 5-7 days that exceeds 100 mm. P is the precipitation since the end of the continuous rainfall event.

6. The forest fire risk prediction method based on a weighted superposition model as described in claim 1, characterized in that, The expression for the composite trunk area of ​​each sampled plant in S5 is as follows: ECTA= π× (MTD / 2) 2 × TN, MTD refers to the diameter of the largest and thickest trunk of a plant that has been burned. The measurement method is based on the standard of diameter at breast height (DBH), which is the diameter of the trunk measured at a height of 1.3 meters above the ground. TN refers to the number of burned trunks produced on the same plant. The expression for the area of ​​the composite sucker of each sampled plant in S5 is as follows: ECRA= π× (MRD / 2) 2 × RN, Where MRD is the diameter of the thickest new shoot that grows from the base of the plant after the fire, and RN is the total number of new shoots that sprout from the same plant after the fire. The expression for the diameter of the composite trunk in S5 is: , The expression for the diameter of the compound sucker in S5 is: ; The fire resistance index expressions for different tree species in S5 are as follows: RCI = Number of tree species × (ECRD / ECTD).

7. The forest fire risk prediction method based on a weighted superposition model as described in claim 1, characterized in that, The expression for the fire risk level index in S6 is as follows: ZHFX = W FFI ×FFI+W FRI ×FRI-W FCI ×FCI-W RCI ×RCI, Among them, FFI is the soil flammability index, FRI is the litter layer flammability index, FCI is the canopy flame retardancy index, RCI is the fire resistance index of different tree species, and W FFI W FRI W FCI and W RCI These are the weights of the corresponding indices.

Images