A method for simulating characteristics of earth background infrared radiation in forest fire scene

Abstract

The application discloses a kind of earth background infrared radiation characteristic simulation method under forest fire scene, it is related to earth background infrared radiation simulation technical field, and can be used for the research of large-scale forest fire scene to the influence of earth background infrared radiation characteristic.Establish the distribution of fire point in detection area with MOD11A1 thermal anomaly fire product, obtain the physical parameters of fire field by MOD11A1 land temperature product, MCD19A2 land aerosol optical depth product and experimental data, and establish parameter grade classification model.Improve according to cellular automaton model, and establish fire field image model under large-scale space.Based on forest fire scene grade classification model, radiation calculation is carried out using MODTRAN, and radiation data model of forest fire scene is established, which is stored using sqlite database, so as to realize the rapid simulation of earth background infrared radiation regional image under forest fire scene, and support target infrared radiation identification.

Description

Technical Field

[0001] This invention relates to the field of Earth background infrared radiation simulation technology, specifically a method for simulating the characteristics of Earth background infrared radiation under forest fire scenarios. Background Technology

[0002] Target identification by early warning satellites is the first step in a space-based missile defense system. The radiation characteristics of different landforms in the Earth's background vary significantly, making the establishment of a comprehensive infrared radiation model of the Earth's background crucial for target identification by early warning satellites. Since the infrared radiation transmission of the Earth and its adjacent background is affected by surface type and weather, simulations of the infrared radiation characteristics of the Earth and its adjacent background must consider real-world environments, such as forest fire scenarios. Forest fire scenarios significantly impact the infrared radiation characteristics of the detection area, adding considerable difficulty to the simulation of infrared radiation against the target missile's background. To more accurately and efficiently identify the infrared radiation characteristics of targets, simulations of the infrared radiation characteristics of the Earth and its adjacent background under forest fire scenarios are necessary. Current simulations of the Earth's background infrared radiation characteristics are based on typical areas and do not consider the scenario of fires in the detection area. Image simulations of forest fires are primarily from the perspective of spread prediction; such small-scale simulations are not suitable for the large-scale spatial requirements of satellite fields of view. Summary of the Invention

[0003] To overcome the shortcomings of the prior art, this invention provides a method for simulating the infrared radiation characteristics of the Earth's background under forest fire scenarios. This method can take into account the effects of different fire scales and durations, and can simulate the infrared radiation characteristics of forest fire scenarios in any region of the Earth's background in any band from 2 to 14 μm.

[0004] The technical solution of this invention is as follows:

[0005] A method for simulating the infrared radiation characteristics of the Earth's background under forest fire scenarios includes the following steps:

[0006] Step 1: Establish the distribution of fire points in the detection area using the MOD11A1 thermal anomaly fire product, and obtain the physical parameters of the fire scene using the MODI11A1 land temperature product, the MCD19A2 land aerosol optical thickness product, and experimental data to establish a parameter level classification model.

[0007] Step 2: Improve the cellular automata model to establish a large-scale spatial fire scene image model.

[0008] Step 3: Based on the values ​​of the corresponding image grid points in the classification model and the fire scene image model, obtain the corresponding fire level, and use MODTRAN to perform radiation calculation to generate an image of the Earth's background infrared radiation region under the forest fire scene.

[0009] Preferably, in step 1, the ignition point of the detection area is detected using the MOD11A1 thermal anomaly fire product to identify the fire and obtain the fire area; the fire area is detected using the MOD11A1 land temperature product to obtain the numerical changes in surface temperature and surface emissivity; the change in aerosol thickness (AOD) in the fire area is obtained using the MCD19A2 land aerosol optical thickness product and converted into horizontal line-of-sight (VIS); the changes in water vapor content and CO2 concentration with fire intensity are obtained through experimental data; and a parameter level classification model is established based on the parameter changes of surface temperature, surface emissivity, horizontal line-of-sight (VIS), water vapor content, and CO2 under different fire intensities and after the fire has burned.

[0010] Preferably, in step 1, pixel products with emissivity data are obtained based on MOD11A1, and the pixel values ​​are converted into emissivity values:

[0011] ε=mD N +n

[0012] In the formula, D N ε is the pixel value, ε is the emissivity, and m and n are parameters specified in the MOD11A1 product specification.

[0013] Preferably, in step 1, an AOD distribution map is obtained based on MCD19A2 remote sensing data. To establish a model of the forest fire scene, the AOD is numerically derived and converted into VIS to meet the conditions for calculating MODTRAN radiation parameters.

[0014] MODTRAN uses weather line distance as an input parameter, and the weather line distance V is represented as:

[0015]

[0016] In the formula, c is the extinction coefficient of Rayleigh scattering at the Earth's surface;

[0017] In a horizontally uniform atmosphere, atmospheric aerosol optical thickness is introduced:

[0018]

[0019] k(z) is the extinction coefficient of the aerosol, including the absorption coefficient and the scattering coefficient; h1 is the altitude at the Earth's surface; h2 is the altitude at the top of the atmosphere; τ a It is the optical thickness of the aerosol.

[0020] Preferably, in step 2, the cell state includes unburned, burning, burned, and unable to ignite, and the burning state is divided into several burning levels; the initial state of the cell state is unburned and assigned a value of C=0. First, it is determined whether the cell state is unable to ignite based on the surface type. If it is, the value is assigned C=-2. If not, the ignition point is randomly determined and assigned a value of C=1. The ignition point is determined based on the regional fire spread based on the cell neighborhood. The range of the cell neighborhood is related to the scale of the forest fire. As the range expands, the probability of ignition decreases.

[0021] Preferably, in step 2, when the surface type corresponding to the cell point is seawater, snow, desert, or wetland, the cell state is "cannot ignite". (i,j) represents the coordinates of the center point of the cell region;

[0022] When a cell corresponds to other surface types, the ignition point is randomly determined, and the number of random ignition points is determined based on the initial scale of the input scene. In large-scale space, multiple areas in the detection area will be on fire, and the fire spreads within a certain range centered on different ignition points. The range of the cell's neighborhood is represented as follows: (i,j) represents the coordinates of the center point of the region, and (m,n) represents the coordinates of a point within the diffusion range. When D>1, the cell (m,n) is determined to be the ignition point.

[0023] Preferably, the cell rule is improved in three aspects: fire escalation, fire spread, and fire extinguishing. Specifically,

[0024] Determine the state of cell (i,j) at time t. Size,

[0025] when At that time, it is determined that there is a probability that the fire will escalate, and the probability L of each escalation level is further calculated.

[0026] The higher the level, the lower the probability of the fire escalating;

[0027] when When the fire is deemed to be spreading, the number of ignition points around the cell is further determined. Adjacent cells and next-nearest cells are counted separately. If only one of the adjacent cells around cell (i,j) has a fire level greater than 2, the fire count is directly increased by one, i.e., N+1. If only one of the four next-nearest cells around cell (i,j) has a fire level greater than 2, then... The probability increases by 1 fire count. When the number of fires around cell (i,j) is N > 2,

[0028] The number of ignition points at time t is represented as:

[0029]

[0030] when At that time, the condition for determining that the fire has been extinguished is The extent to which a wildfire is extinguished and its initial size are related to the duration of the fire.

[0031] As a preferred approach, an infrared radiation database for forest fire scenarios is established based on the infrared radiation calculation results of forest fire scenarios. The database is stored in an SQLite database, and the radiation data of forest fire scenarios is retrieved through join table queries.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] (1) Current simulations of the infrared radiation characteristics of the Earth's background do not consider the situation of fire in the detection area. This invention realizes the simulation of the infrared radiation characteristics of the Earth's background under forest fire scenarios based on satellite remote sensing data.

[0034] (2) By improving the cellular automata model, the fire simulation images are made suitable for large-scale detection needs under satellite field of view;

[0035] (3) Based on the fire level classification model and according to MODTRAN, the simulation of the Earth's background infrared radiation image under the forest fire scenario with arbitrary bands of 2 to 14 μm and accuracy of 0.1 μm was realized, and the data was stored in the database (general file io, which cannot quickly search for large amounts of data).

[0036] This invention lays the foundation for studying the radiation characteristics of detection areas under forest fire scenarios and provides support for target infrared radiation identification, which has important practical and technological significance. Attached Figure Description

[0037] Figure 1 This is a flowchart of forest fire scene parameter processing;

[0038] Figure 2 It is a map showing the distribution of fire points in the detection area;

[0039] Figure 3 It is a map showing the surface temperature distribution in the detection area;

[0040] Figure 4 This is a map showing the emission rate distribution in 2023;

[0041] Figure 5 This is a map showing the distribution of AOD in the detection area;

[0042] Figure 6 It is a Moor-type neighborhood graph;

[0043] Figure 7This is a forest fire scene diagram based on cellular automata;

[0044] Figure 8 This is a simulation diagram of the radiation characteristics of a small-scale forest fire scenario;

[0045] Figure 9 This is a simulation diagram of the radiation characteristics of a medium-sized forest fire scenario;

[0046] Figure 10 It is a simulation diagram of the radiation characteristics of a large-scale forest fire scene;

[0047] Figure 11 This is a simulation diagram of the radiation characteristics of a forest fire scene lasting 10 days;

[0048] Figure 12 This is a simulation diagram of the radiation characteristics of a forest fire scenario lasting 20 days. Detailed Implementation

[0049] To clearly illustrate the technical features of this patent, the following detailed description is provided through specific embodiments and in conjunction with the accompanying drawings.

[0050] This invention provides a method for simulating the Earth's background infrared radiation characteristics under forest fire scenarios. A forest fire severity model is established based on satellite data, filling the gap in existing technologies for simulating Earth's background radiation under forest fire scenarios. A forest fire radiation model is established for rapid calculation and storage in a database, which is stored in SQLite. Forest fire scenario radiation data can be retrieved through join queries. Based on an improved cellular automata model, this application establishes a large-scale spatial fire scene image model, which can randomly generate fire scene images based on the input fire scale and duration. The value of each point on the image grid corresponds to a different fire severity level. Then, the corresponding radiation data is found by searching a fire scene radiation database, and finally, a radiation simulation image of the fire scene in the detection area is generated. Specifically, it includes:

[0051] Step 1: Fire point detection was conducted in the fire area of ​​northern Canada in 2023 based on the MOD11A1 thermal anomaly fire product. The changes in surface temperature and surface emissivity of the detection area were obtained using the MOD11A1 land temperature product, and the changes in aerosol thickness in the detection area were obtained using the MCD19A2 land aerosol optical thickness product. Experimental data were used to obtain the changes in water vapor content and CO2 concentration with fire intensity. Finally, a fire field parameter classification model was established.

[0052] Step 2: Based on the cellular automata model, improvements were made to three aspects: cell state, cell neighborhood, and cell rules, making the image model suitable for large-scale detection fields of view from a satellite perspective.

[0053] Step 3: Based on the fire field parameter level classification model, radiation calculation is performed using MODTRAN, which can realize the simulation of infrared radiation characteristics of forest fire scenes in any region of the earth with different initial states and durations of forest fires in any band from 2 to 14 μm.

[0054] It should be noted that due to the large amount of data, this embodiment chooses an SQLite database for storage. A join query is used to retrieve the radiation of the forest fire scene to ensure the speed of calculating regional radiation.

[0055] Specifically:

[0056] This invention, based on satellite remote sensing data of the 2023 summer wildfires in northern Canada and combined with previous research on wildfires, acquires data from both surface and atmospheric parameters to analyze the typical characteristics of wildfire scenarios. Based on the analysis results, the fire severity is classified by changing physical parameters. The specific process is as follows: Figure 1 As shown.

[0057] To conduct radiation simulation of forest fire scenarios, it is first necessary to acquire data on these scenarios. Currently, satellite-borne sensors are used to map and assess the distribution of forest combustibles on a large spatial scale, offering high timeliness and convenient data acquisition, significantly saving manpower and resources. In the summer of 2023, a large-scale wildfire occurred in Fort Smith and Slade River region of northern Alberta, Canada, burning a total area of ​​170,000 square kilometers, breaking historical records. This data is of great value for studying forest fire parameters. Therefore, this invention uses satellite remote sensing data to acquire surface temperature and emissivity data of the fire-affected areas in northern Alberta, Canada.

[0058] This invention utilizes the MOD14A1 data product for fire scene identification to determine the fire area. MOD14A1 is a gridded, Level 3 product with a resolution of 1 km per day. The product information provides a macroscopic view of the fire situation, determining the location of the fire, its severity, and the ratio of smoldering fire to combustion.

[0059] This invention acquired data from the northern Canada exploration area from July 20, 2023 to August 10, 2023. Figure 2 This is the data processing result for July 29th. According to the surface type map of the area, it is mostly grassland and evergreen forest, with relatively few wetlands. The fire point map shows that multiple fires with high probability occurred in the area. In summary, this area can provide scene data support for obtaining physical parameters of forest fire scenarios in this invention.

[0060] When a fire occurs, surface temperature rises sharply, and surface emissivity changes. These two parameters dynamically change with the fire's intensity, significantly impacting radiative transfer. This invention uses MOD11A1 land temperature data combined with the fire identification described above to acquire surface parameters of a fire-affected area in northern Canada.

[0061] like Figure 3 The image shows the surface temperature distribution of the probe area on July 29, 2022 and 2023. A comparison clearly shows that the surface temperature in the probe area was higher in 2023, generally ranging from 300 to 310 K, with some areas reaching temperatures above 330 K. In contrast, the surface temperature in the probe area in 2022 was generally around 290 K.

[0062] For surface emissivity, since the product obtained from MOD11A1 is a pixel product, the emissivity data of bands 31 and 32 need to be processed. According to equation (1), the pixel values ​​are converted into emissivity values. Figure 4 This is a surface emissivity distribution map of the detection area on July 29, 2023. Comparing the two maps reveals that the emissivity of the detection area does not change significantly with different wavelengths; therefore, this invention treats the forest fire scene as a gray body model. Within the detection area, the emissivity does not vary much with location, fluctuating within a range of 0.02. Such differences in emissivity values ​​have a negligible impact on radiation. The following formula is used to convert the values ​​of all points in the detection area:

[0063] ε=0.002D N +0.49 (1)

[0064] In the formula, D N ε represents the pixel value, and ε represents the emissivity. Parameters 0.002 and 0.49 are both obtained from the official MODIS documentation.

[0065] Taking all factors into consideration, based on the calculated average value, the present invention sets the emissivity to 0.986 and the emissivity after combustion to 0.97.

[0066] Forest fires generate large amounts of smoke particles, affecting the optical thickness of aerosol particles and thus reducing atmospheric transmittance. The burning of biomass also releases significant amounts of CO2 and water vapor, increasing atmospheric CO2 concentration and humidity, which greatly impacts atmospheric radiative transport. This invention analyzes and obtains aerosol, CO2, and water vapor parameters from a fire area in northern Canada using MCD19A2 remote sensing data and previous experimental research. Figure 5The graphs show the AOD distribution for July 29, 2022 and 2023. The difference in AOD between the two years is clearly visible. In 2022, the AOD in most areas was 0.1–0.2, with an overall maximum not exceeding 0.3. In 2023, the AOD was generally in the range of 2.5–3, with some areas reaching 4.3. This invention uses field experimental data measured by Clements and Potter as the data source to obtain CO2 and H2O concentration data.

[0067] To build a model of a forest fire scenario, it is necessary to numerically derive AOD and convert it into VIS to meet the conditions for inputting MODTRAN radiation parameters for calculation.

[0068] Based on the above formula, MODTRAN uses the meteorological sight distance V (in km) as the input parameter and gives the following formula:

[0069]

[0070] In the formula: 0.01159 is the extinction coefficient of Rayleigh scattering at the Earth's surface (at 550 nm).

[0071] In a horizontally uniform atmosphere, atmospheric aerosol optical thickness is introduced:

[0072]

[0073] In the formula: k(z) is the extinction coefficient of the aerosol, including the absorption coefficient and the scattering coefficient; h1 is the height of the Earth's surface; h2 is the height of the top of the atmosphere; τ a It is the optical thickness of the aerosol.

[0074] Therefore, it can be seen that AOD is related to the profile distribution of aerosol extinction coefficient; while VIS is only related to the surface extinction coefficient and has no direct relationship with AOD. This invention uses empirical formulas for the mutual conversion of AOD and VIS summarized by previous research. The relationship between AOD and VIS was simulated using MODTRAN4 under three aerosol types (rural, urban, and marine), two seasons (summer and winter), and three water vapor contents (0 g / cm², 3 g / cm², and 6 g / cm²). Through data fitting, it was found that VIS and... The two have a strong linear relationship, and the following method is used:

[0075]

[0076] In the formula: according to the needs of the present invention, the values ​​of a and b are 0.12022071 and 0.29739269, respectively.

[0077] With the aim of establishing forest fire scenarios, a forest fire scenario level model was established based on the changes in parameters such as surface temperature, surface emissivity, horizontal line-of-sight (VIS), CO2 concentration, and H2O water vapor content under different fire intensities and after the fire has burned, as shown in Table 1 below:

[0078] Table 1 Classification of Forest Fire Scene Parameters

[0079]

[0080] Infrared detection models of forest fire scenarios are a crucial component in constructing models of Earth and its surrounding background radiation under complex scenarios. Based on different fire severity levels, considering the characteristics of forest fire spread and the large spatial scale of satellite observation, this invention establishes a dynamic image model of forest fire scenarios over time by improving the cellular automatic basis function model. Radiation calculations were performed using the MODTRAN4 atmospheric transport model, leveraging the characteristics of different parameters in forest fire scenario classification. The calculation results were then mapped to the input parameters, and a radiation database of forest fire scenarios was established using SQLite, categorized by fire severity level.

[0081] This model has four cell states: not on fire, on fire, previously on fire, and unable to ignite. The on fire state is further divided into five fire levels. In the initial cell state, whether a cell can ignite is determined by the surface type of the probe area. When the surface type corresponding to the cell is seawater, snow, desert, or wetland, the cell state is "unable to ignite." When a cell point corresponds to other surface types, the ignition point is randomly determined, and the number of random ignition points is determined based on the initial scale of the input scene. In large-scale spaces, multiple areas within the detection area may ignite. To address this, this invention uses different ignition points as centers to diffuse the fire in small-scale areas. This range is also related to the input forest fire scale, and the probability of ignition decreases as the range expands. As shown in Equation 5, (i,j) represents the coordinates of the center point of the area, and (m,n) represents the coordinates of a point within the diffusion range. When D < 1, cell (m,n) is determined to be the ignition point.

[0082]

[0083] It needs to be explained that regarding the cell states and state amplitudes at different levels, refer to Table 2 below. The initial state of each grid point is unburned (C=0). During initial assignment, based on the surface type, the grid points corresponding to seawater, snow, desert, and wetland types are in an unburnable state (C=-2); for other surface types, a portion of the area is randomly determined to be in a burning state (C=1). Specifically, several points are randomly selected within the area, and the burning area is defined within a certain range centered on these points. D represents the determination of a range.

[0084] Table 2 Cell states at different levels

[0085]

[0086] The neighborhood of this model's mesh uses a Moore-type neighborhood. A neighborhood consists of the four adjacent cells (top, bottom, left, and right) plus the four next-nearest cells diagonally, as shown below. Figure 6 As shown. The number of grids and the scale corresponding to the grid side length are determined according to the number of pixels and resolution of the satellite detection area. This invention uses a grid of 512×512, and the grid side length corresponds to an accuracy of 1km.

[0087] From a fire scene perspective, a fire involves several stages: escalation, spread, and extinguishing. It starts small, gradually increases in size, and eventually extinguishes, sometimes spreading to surrounding areas. Therefore, the cell rules in this model are designed to address these three aspects: escalation, spread, and extinguishing. They assess the surrounding fire situation and, based on the number of fires in the surrounding area, determine whether the current grid point will be affected by the fire. Specifically, the model is divided into adjacent cells and next-nearest cells. If an adjacent cell is on fire, the number of fires in the surrounding area is directly incremented by 1. If a next-nearest cell is on fire, there is a probability that the number of fires in the surrounding area will be incremented by 1 (this probability is relatively common considering the greater distance between next-nearest cells). Finally, the number of fires in the surrounding area is assessed. When the number of fires is greater than 2, the cell's state in the next time step is considered to be on fire.

[0088] The primary condition for a fire to escalate is Next, according to Equation 6, calculate the probability L of upgrading to each level. The higher the level, the lower the probability.

[0089]

[0090] The scale of fire spread is related to the initial size of the wildfire; the primary condition for fire spread is... and These represent the states of cell (i,j) at time t and t+1, respectively. If cell (i,j) is in the burning state, i.e. To determine the number of fire points around a cell, considering the distance factor, it is necessary to count adjacent cells and second-nearest cells separately. N represents the number of fire points around cell (i,j). If exactly one of the four adjacent cells around cell (i,j) has a fire level greater than 2, then the fire count is directly increased by one. If exactly one of the four second-nearest cells around cell (i,j) has a fire level greater than 2, then... The probability of fire increases by 1. When the number of fires around cell (i,j) is N > 2,

[0091]

[0092] The conditions for extinguishing a fire are: The extent to which a forest fire is extinguished is related to its initial size and burning time. The larger the initial size and the longer the burning time, the greater the area of ​​the forest fire that will be extinguished.

[0093] After the burning time iteration was completed, considering the actual fire scene conditions, a perturbation of 0.2 was set for the ignition point. Figure 7 This is a forest fire scene image model based on cellular automata, combining the image with real-world land surface types. The simulation area is the Fort Smith region in northern Canada. The initial model scale is large-scale, with a burn time of 8 days. Simulation results show that there are no fire points in areas with water, desert, and wetland land surfaces, and the fire spread is relatively consistent with real-world conditions.

[0094] The radiance of a forest fire scene model with a resolution of 0.1 μm was calculated using the MODTRAN radiance calculation software in the 2–14 μm band. Parameters were input according to the classification of surface and atmospheric parameters summarized in step 1. Parameter descriptions for the radiance calculation model: 1) Atmospheric model: A mid-latitude summer atmospheric model was selected based on the high surface temperature characteristic of forest fire scenes. 2) Aerosol type: Due to the large amount of smoke and dust produced by forest fires, 20% of urban aerosols are smoke-related aerosols formed by combustion products. Compared with rural and marine aerosols, urban aerosols are more suitable. 3) Surface temperature: Inputted according to different classifications of forest fire scenes, ranging from 310 K to 350 K, with a post-combustion surface temperature of 305 K. 4) Surface emissivity: The forest fire scene model is treated as a gray body model. Since the emissivity difference between different levels is not significant, it is uniformly set to a value of 0.986. The surface emissivity after combustion is 0.97. 5) VIS: Input according to different forest fire scene levels, ranging from 0.3 to 3.07 km. The VIS after combustion is 3.92 km. 6) CO2 concentration: Input according to different forest fire scene levels, ranging from 450 to 2560 ppmv. The CO2 concentration after combustion is 410 ppmv. 7) Water vapor concentration: Input according to different forest fire scene levels, ranging from 2.96 to 3.84 g / cm². The water vapor concentration after combustion is 3.05 g / cm². 8) Zenith angle: The zenith angle ranges from 90 to 180°, grouped in 10° increments. 9) Solar zenith angle: The solar zenith angle at the detector location, ranging from 0 to 90°, grouped in 10° increments. Other parameters should be entered using MODTRAN's default values.

[0095] To quickly calculate the Earth's background infrared radiation under forest fire scenarios, this invention establishes an infrared radiation database for forest fire scenarios based on the infrared radiation calculation results. The database parameters are described in the table below:

[0096] Table 2. Parameter Description of Infrared Radiation Database for Forest Fire Scenes

[0097]

[0098] Note: The parameters corresponding to the scene level are consistent with the parameter classification of the forest fire scene level model in Section 4.1.4.

[0099] The exploration area was located in California, USA (122.6°W, 38.9°N). Figure 8 The forest fire scenario is set to a small scale, taking into account the influence of solar radiation, and the simulation results are shown in two bands: 2-3 μm and 8-14 μm. Figure 9 and Figure 10 These are simulations of forest fire scenarios set as medium-sized and large-scale detection areas, respectively. Figure 11 and Figure 12 The irradiance distribution of the detection area was obtained by simulating forest fire scenarios by changing the duration of the fire.

[0100] There are many specific ways to implement this invention. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements can be made without departing from the principle of this invention, and these improvements should also be considered within the scope of protection of this invention.

Claims

1. A method for simulating the infrared radiation characteristics of the Earth's background under a forest fire scenario, characterized in that, Includes the following steps: Step 1: Establish the distribution of fire points in the detection area using the MOD11A1 thermal anomaly fire product, and obtain the physical parameters of the fire scene using the MOD11A1 land temperature product, the MCD19A2 land aerosol optical thickness product, and experimental data to establish a parameter level classification model. Step 2: Based on the cellular automata model, the cell states, cell neighborhoods, and cell rules were improved to establish a large-scale fire scene image model; among which, The cell states include unburned, burning, burned, and unable to ignite, and the burning state is divided into several burning levels. The initial state of the cell is unburned and assigned a value of C=0. First, it is determined whether the cell state is unable to ignite based on the surface type. If it is, the value is assigned C=-2. If not, the ignition point is randomly determined and assigned a value of C=1. The ignition point is determined based on the regional fire spread based on the cell neighborhood. The range of the cell neighborhood is related to the scale of the forest fire. As the range increases, the probability of ignition decreases. When the surface type corresponding to the cell point is seawater, snow, desert, or wetland, the cell state is "cannot ignite". , The coordinates of the center point of the cell region; When a cell corresponds to other surface types, the ignition point is randomly determined, and the number of random ignition points is determined based on the initial scale of the input scene. In large-scale space, multiple areas in the detection area will be on fire, and the fire spreads within a certain range centered on different ignition points. The range of the cell's neighborhood is represented as follows: , Represents the coordinates of the center point of the region. Represents the coordinates of a point within the diffusion range, when When, then determine the cell. For the ignition point, ; The cell rule is improved in three aspects: fire escalation, fire spread, and fire extinguishing. Specifically, Determine the cell exist state of time Size, when If the fire is deemed to have a probability of escalation, the probability of escalation at each level is further calculated. , The higher the level, the lower the probability of the fire escalating; when When the fire is deemed to have spread, the number of ignition points around the cell is further determined, and adjacent and next-near cells are counted separately. If the cell If exactly one of the surrounding adjacent cells has a fire level greater than 2, then the fire count is directly increased by one. If cells If there is exactly one cell among the four next-neighbor cells with a fire level greater than 2, then... The probability of ignition increases by 1 unit when the cell... Number of fires around hour, , The number of ignition points at time t is represented as: ， when At that time, the condition for determining that the fire has been extinguished is The extent to which a forest fire is extinguished and its initial size are related to the duration of the fire. Step 3: Based on the values ​​of the corresponding image grid points in the classification model and the fire scene image model, obtain the corresponding fire level, and use MODTRAN to perform radiation calculation to generate an image of the Earth's background infrared radiation region under the forest fire scene.

2. The method for simulating the infrared radiation characteristics of the Earth's background under a forest fire scenario according to claim 1, characterized in that, In step 1, the MOD11A1 thermal anomaly fire product is used to detect fire points in the detection area, identify the fire, and obtain the fire area. The MOD11A1 land temperature product is used to detect the fire area and obtain the numerical changes in surface temperature and surface emissivity. The MCD19A2 land aerosol optical thickness product is used to obtain the changes in aerosol thickness (AOD) in the fire area and convert it into horizontal line-of-sight (VIS). Experimental data are used to obtain the changes in water vapor content and CO2 concentration with fire intensity. A parameter level classification model is established based on the parameter changes of surface temperature, surface emissivity, horizontal line-of-sight (VIS), water vapor content, and CO2 under different fire intensities and after the fire has burned.

3. A method for simulating the infrared radiation characteristics of the Earth's background under a forest fire scenario according to claim 1 or 2, characterized in that, In step 1, pixel products with emissivity data are obtained based on MOD11A1, and the pixel values ​​are converted into emissivity values: ， In the formula, For pixel values, For emissivity, m and n are parameters specified in the MOD11A1 product specification.

4. The method for simulating the infrared radiation characteristics of the Earth's background under a forest fire scenario according to claim 2, characterized in that, In step 1, an AOD distribution map is obtained based on MCD19A2 remote sensing data. To establish a model of the forest fire scene, the AOD is numerically derived and converted into a VIS to meet the conditions for calculating MODTRAN radiometric parameters. MODTRAN uses meteorological sight distance as an input parameter. Represented as: ， In the formula, c is the extinction coefficient of Rayleigh scattering at the Earth's surface; In a horizontally uniform atmosphere, atmospheric aerosol optical thickness is introduced: ， This is the extinction coefficient of the aerosol, which includes the absorption coefficient and the scattering coefficient. h1 is the altitude at the Earth's surface, and h2 is the altitude at the top of the atmosphere. It is the optical thickness of the aerosol.

5. The method for simulating the infrared radiation characteristics of the Earth's background under a forest fire scenario according to claim 1, characterized in that, Based on the infrared radiation calculation results of forest fire scenarios, an infrared radiation database for forest fire scenarios is established and stored in an SQLite database. The radiation data of forest fire scenarios can be retrieved through join table queries.

Images