Monitoring method and system for high-frequency dynamic remote sensing of grassland fire by geostationary meteorological satellite

Abstract

The application discloses a kind of static meteorological satellite prairie fire high-frequency dynamic remote sensing monitoring method and system, based on static meteorological satellite, wherein static meteorological satellite prairie fire high-frequency dynamic remote sensing monitoring method includes: obtaining static meteorological satellite data, and data preprocessing is generated monitoring area grid data.Based on monitoring area grid data, target pixel in monitoring area grid data is determined.Based on target pixel, fire point pixel is generated.Based on fire point pixel, open fire area is generated.Based on open fire area, output fire monitoring result.By this, the static meteorological satellite prairie fire high-frequency dynamic remote sensing monitoring method of the application can discover and monitor the state of prairie fire with high timeliness and high accuracy rate, and output fire monitoring result.

Description

Technical Field

[0001] This invention relates to the field of grassland fire monitoring technology, and in particular to a method and system for high-frequency dynamic remote sensing monitoring of grassland fires using geostationary meteorological satellites. Background Technology

[0002] In recent years, grassland fires have occurred frequently. They are a type of natural disaster characterized by their suddenness, destructiveness, and difficulty in handling and rescue, causing irreversible damage to the natural ecosystem and human life and property. Using satellite remote sensing to monitor grassland fires is currently an effective technical means with a wide monitoring range, short monitoring cycle, and fewer restrictions.

[0003] In existing technologies, spectral reflectance correction needs to consider the variation of the solar zenith angle from 0° to 90°. Reflectance information acquired at different times needs to be corrected to obtain more accurate reflectance. The current solar zenith angle correction method is shown in the following formula:

[0004] R′=R / cos(θ sz )

[0005] In the formula, R is the original reflectivity of the satellite, R′ is the reflectivity corrected for the solar zenith angle, and θ sz This is the solar zenith angle.

[0006] When θ sz When θ is small, reflectivity correction is not affected by the solar zenith angle; when θ is small... sz When approaching 90°, cos(θ) sz When the value is infinitely close to 0, it causes an anomaly in R′, which in turn leads to the failure of spectral reflectance correction.

[0007] Furthermore, existing fire monitoring methods cannot achieve high-timeliness and high-accuracy monitoring, nor can they achieve continuous tracking and monitoring of panoramic and full-process situational changes.

[0008] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0009] The purpose of this invention is to provide a method and system for high-frequency dynamic remote sensing monitoring of grassland fires using geostationary meteorological satellites. This method can detect and monitor the status of grassland fires with high timeliness and accuracy, and output the fire monitoring results.

[0010] To achieve the above objectives, in a first aspect, the present invention provides a high-frequency dynamic remote sensing monitoring method for grassland fires using geostationary meteorological satellites. Based on geostationary meteorological satellites, the method includes: acquiring geostationary meteorological satellite data and performing data preprocessing to generate grid data for a monitoring area; determining target pixels in the grid data based on the monitoring area data; generating fire point pixels based on the target pixels; generating the open flame area based on the fire point pixels; and outputting the fire monitoring results based on the open flame area.

[0011] In one embodiment of the present invention, acquiring geostationary meteorological satellite data and performing data preprocessing to generate monitoring area grid data specifically involves: acquiring geostationary meteorological satellite Earth observation data and performing data preprocessing to generate monitoring area grid data. The preprocessing of the geostationary meteorological satellite Earth observation data includes geolocation information correction, data projection, and reflectance correction.

[0012] In one embodiment of the present invention, geolocation information correction of geostationary meteorological satellite Earth observation data is specifically performed using the following formula:

[0013] Lon delt =Lon FY4B -Lon shp

[0014] Lat delt =Lat FY4B -Lat shp

[0015] In the formula, Lon delt and Lat delt The offsets for longitude and latitude, respectively. FY4B and Lat FY4B The longitude and latitude of FY4B are respectively used for positioning, Lon shp and Lat shp These are the longitude and latitude coordinates for the coastline and lake vectors, respectively.

[0016] In one embodiment of the present invention, reflectance correction of geostationary meteorological satellite Earth observation data is specifically performed using the following formula:

[0017] R′=R / cos(θ sz ×(1.0-1.3×sin(0.05×θ sz )))

[0018] In the formula, R is the original reflectivity of the satellite, R′ is the reflectivity corrected for the solar zenith angle, and θ sz This is the solar zenith angle.

[0019] In one embodiment of the present invention, determining the target pixel in the monitoring area grid data based on the monitoring area grid data specifically involves: extracting cloud pixel information based on the monitoring area grid data; identifying the cloud pixel information using a first identification strategy to obtain cloud pixels; removing cloud pixels to determine the target pixel in the monitoring area grid data.

[0020] In one embodiment of the present invention, generating a fire point pixel based on a target pixel specifically involves: identifying suspected high-temperature fire point pixels in the vicinity of the target pixel, and calculating the mean and standard deviation of the background brightness temperature in the vicinity of the target pixel to generate thermal anomaly pixels. A second identification strategy is used to identify the thermal anomaly pixels, generating a first undetermined fire point pixel. A third identification strategy is used to identify the first undetermined fire point pixel, generating a second undetermined fire point pixel. A cloud area fire point supplementary identification strategy is used to supplement the identification of cloud pixels, generating a third undetermined fire point pixel. The first undetermined fire point pixel, the second undetermined fire point pixel, and the third undetermined fire point pixel are combined to generate a fire point pixel.

[0021] In one embodiment of the present invention, the third identification strategy specifically includes: cloud pollution identification strategy, cloud edge or desert area edge strategy and fixed artificial heat source identification strategy.

[0022] Specifically, the cloud contamination identification strategy is as follows: if the first undetermined fire point pixel meets the following conditions, it will not be considered a fire point pixel.

[0023] R vis ≥R visbg +RC_th&T 13 ≤T 13_ -TC_th

[0024] In the formula, R vis Here, RC_th is the visible light channel reflectance value, TC_th is the visible light cloud contamination identification threshold, and T_th is the far-infrared cloud interference identification threshold. 13 For far-infrared channels, T represents the background reflectance value. 13_g This represents the background temperature value of the far-infrared channel.

[0025] Specifically, the cloud edge or desert area edge strategy is as follows: if the first undetermined fire point pixel meets the following conditions, it will not be considered a fire point pixel.

[0026] T7≤T 7_ +C7×δT 7_ &T 7_ ≤T 7__ +C 7_ ×δT 7___g

[0027] In the formula, T7 is the mid-infrared channel, C7 is the cloud edge impact detection threshold, and C7_3 T is the threshold for identifying the impact of desert edges. 7_g δT represents the background temperature value for the mid-infrared channel. 7_ T represents the standard deviation of the effective background brightness temperature in the mid-infrared region. 7_3 T represents the pixel brightness temperature difference between mid-infrared and far-infrared regions. 7_13_ δT represents the effective background brightness temperature difference between the mid-infrared and far-infrared regions. 7_3__bg The standard deviation of the pixel brightness temperature difference between the effective background of mid-infrared and far-infrared regions;

[0028] The specific strategy for identifying fixed artificial heat sources is as follows: manually eliminating non-hot abnormal points.

[0029] Secondly, this invention provides a geostationary meteorological satellite high-frequency dynamic remote sensing monitoring system for grassland fires. Based on geostationary meteorological satellites, the system includes: an acquisition and generation module, a determination module, a fire point pixel generation module, an open fire area generation module, and an output module. The acquisition and generation module acquires geostationary meteorological satellite data and performs data preprocessing to generate grid data for the monitoring area. The determination module determines target pixels in the grid data based on the monitoring area grid data. The fire point pixel generation module generates fire point pixels based on the target pixels. The open fire area generation module generates the open fire area based on the fire point pixels. The output module outputs the fire monitoring results based on the open fire area. The acquisition and generation module is communicatively connected to the determination module, the determination module is communicatively connected to the fire point pixel generation module, the fire point pixel generation module is communicatively connected to the open fire area generation module, and the open fire area generation module is communicatively connected to the output module.

[0030] Thirdly, the present invention provides a geostationary meteorological satellite high-frequency dynamic remote sensing monitoring computing device for grassland fires, comprising: one or more processors; and a storage device for storing one or more programs, wherein when one or more programs are executed by one or more processors, the one or more processors implement the geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described above.

[0031] Fourthly, the present invention provides a computer-readable storage medium storing a program that, when executed by a processor, implements the above-described method for high-frequency dynamic remote sensing monitoring of grassland fires using geostationary meteorological satellites.

[0032] Compared with existing technologies, the geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method and system for grassland fires according to the present invention can detect and monitor the status of grassland fires with high timeliness and high accuracy, and output fire monitoring results. Attached Figure Description

[0033] Figure 1This is a flowchart illustrating a method for high-frequency dynamic remote sensing monitoring of grassland fires using a geostationary meteorological satellite according to an embodiment of the present invention.

[0034] Figure 2 This is a detailed flowchart illustrating a geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires according to an embodiment of the present invention.

[0035] Figure 3 This is a schematic diagram of the process for determining a target pixel according to an embodiment of the present invention.

[0036] Figure 4 This is a schematic diagram of the process for generating fire pixels according to an embodiment of the present invention.

[0037] Figure 5 This is a schematic diagram of a geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring system according to an embodiment of the present invention.

[0038] Figure 6 This is a schematic diagram of the spatiotemporal distribution of a fire scene according to an embodiment of the present invention.

[0039] Figure 7 This is a schematic diagram of the spatiotemporal distribution of a fire scene according to an embodiment of the present invention.

[0040] Figure 8 This is a schematic diagram of the spatiotemporal distribution of a fire scene according to an embodiment of the present invention.

[0041] Figure 9 This is a schematic diagram of the spatiotemporal distribution of a fire scene according to an embodiment of the present invention.

[0042] Figure 10 This is a schematic diagram of the estimated open flame area according to an embodiment of the present invention.

[0043] Figure 11 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0044] Figure 12 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0045] Figure 13 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0046] Figure 14 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0047] Figure 15 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0048] Figure 16 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0049] Figure 17 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0050] Figure 18 This is a schematic diagram of fire point monitoring according to an embodiment of the present invention.

[0051] Figure 19 This is a schematic diagram of the structure of a geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device according to an embodiment of the present invention. Detailed Implementation

[0052] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the embodiments of the present invention, and not all structures.

[0053] To facilitate understanding, the main implementation concepts of the various embodiments of the present invention will be briefly described first.

[0054] The key to fire point pixel identification lies in the comparison with the background temperature. The calculation of background brightness temperature is particularly important. The initial background window area is 5×5 pixels. The background pixel brightness temperature is the average temperature of the background pixels in the window area, i.e., T3.9bg = mean(T3.9). However, it is necessary to remove cloud areas, water bodies, and high-temperature suspicious fire point pixels. The identification condition for high-temperature suspicious pixels is T3.9 > Tth or T3.9 > T'3.9bg + △T3.9bg.

[0055] Where T3.9 represents the brightness temperature value of the 3.9μm channel; Tth represents the brightness temperature threshold of the 3.9μm channel, which by default can be the sum of the average value of all pixels within the window and the corresponding twice the standard deviation; T'3.9bg represents the average brightness temperature of the 3.9μm channel within the window area for the same land use type; △T3.9bg represents the difference in brightness temperature between the suspected fire point pixel and the background, which can be expressed as 2.5 times the standard deviation of pixels of the same land use type. If less than 20% of the clear-sky pixels in the 5×5 window area meet the above conditions, the window area is expanded to 7×7, 9×9, ..., 51×51. If the requirements are still not met, the pixel is abandoned in the calculation and marked as a non-fire point pixel.

[0056] Fire point pixel identification: If a pixel meets the following two conditions, it can be preliminarily identified as a fire point pixel:

[0057] 1. T3.9>T3.9bg+n1×δT3.9bg;

[0058] 2. △T3.9_11>△T3.9_11bg+n2×δT3.9_11bg;

[0059] Where: δT3.9bg represents the standard deviation of the 3.9μm brightness temperature in the background window; △T3.9_11 = T3.9 - T11 (T11 represents the brightness temperature value of the 11μm channel); △T3.9_11bg(T3.9bg - T11bg) represents the average value of △T3.9_11 in the background window, and δT3.9_11bg represents the standard deviation of △T3.9_11 in the background window. The main purpose of this condition setting is to distinguish the differences in the inherent brightness temperature values ​​of pixels of different underlying surface types within the window. When the pixel types within the window area are relatively consistent, the values ​​of δT3.9_11bg and δT3.9_11bg are smaller. During the fire point identification process, when δT3.9_11bg and δT3.9_11bg are less than 2K, 2K is used as the substitute; when they are greater than 4K, 4K is used as the substitute. n1 and n2 are background coefficients, which vary with different monitoring areas, times, and angles. For the northern grassland region, n1 can be set to 3, and n2 can be set to 3.5. The existing technology described above does not achieve adaptive dynamic adjustment of the threshold.

[0060] The inventors, having discovered the technical defects described in the background art, hope to find a way to detect and monitor the status of grassland fires with high timeliness and accuracy, and output the fire monitoring results.

[0061] Example 1

[0062] like Figures 1 to 4 As shown in Example 1, a high-frequency dynamic remote sensing method for monitoring grassland fires using geostationary meteorological satellites is provided. Based on geostationary meteorological satellites, the method includes:

[0063] Step S1: Acquire geostationary meteorological satellite data and perform data preprocessing to generate grid data for the monitoring area.

[0064] Specifically, geostationary meteorological satellite Earth observation data is acquired and preprocessed to generate grid data for the monitoring area. The preprocessing of geostationary meteorological satellite Earth observation data includes geolocation information correction, data projection (equal latitude and longitude projection), and reflectance correction.

[0065] Specifically, in practice, based on FY-4B satellite data from January to June 2022 and the corresponding elevation (DEM) data, relatively fixed vector data with obvious geometric characteristics, such as coastlines and lakes, were manually processed. The elevation (DEM) of the entire country from 0m to 4000m was divided into eight intervals at 500m levels. In each elevation interval, 3-5 sampling points were selected from typical coastlines and lakes without cloud cover, totaling 267 points. Using the coastline and lake vectors as a reference, the FY-4B positioning data offset was determined. The longitude and latitude offsets of the sampling points in each of the eight intervals were statistically calculated. The geostationary meteorological satellite Earth observation data was corrected for geographic positioning information using the following formula:

[0066] Lon delt =Lon FY4B -Lon shp

[0067] Lat delt =Lat FY4B -Lat shp

[0068] In the formula, Lon delt and Lat delt The offsets for longitude and latitude, respectively. FY4B and Lat FY4B The longitude and latitude of FY4B are respectively used for positioning, Lon shp and Lat shp The invention locates the longitude and latitude of the coastline and lake vectors, respectively. Existing monitoring methods do not have a step to correct for geographic location information. It is precisely because of this step that the invention is highly timely and accurate.

[0069] Specifically, regarding spectral emissivity correction, during FY-4B Earth observations, the solar zenith angle varied from 0 to 90°. Reflectivity information acquired at different times required correction processing to obtain more accurate reflectivity. The traditional solar zenith angle correction method is shown in the formula:

[0070] R′=R / cos(θ sz )

[0071] When θ sz When θ is small, reflectivity correction is not affected by the solar zenith angle; when θ is small... sz When approaching 90°, cos(θ) sz The value of R′ approaches 0, causing an anomaly. Therefore, for large-scale solar zenith angle observations, the original correction method needs to be improved, as shown in the following formula:

[0072] R′=R / cos(θ sz×(1.0-1.3×sin(0.05×θ sz )))

[0073] In the formula, R is the original reflectivity of the satellite, R′ is the reflectivity corrected for the solar zenith angle, and θ sz This is the solar zenith angle.

[0074] Step S2: Based on the grid data of the monitoring area, determine the target pixels in the grid data of the monitoring area. Specifically, Step S2 consists of: Step S201: Extracting cloud pixel information based on the grid data of the monitoring area; Step S202: Identifying the cloud pixel information using a first identification strategy to obtain the cloud pixels; Step S203: Removing the cloud pixels to determine the target pixels in the grid data of the monitoring area.

[0075] Specifically, the following operations are performed based on the raster data of the monitoring area:

[0076] 1. Cloud Pixel Identification: Utilizing five identification strategies (the first strategy), a union logic is employed. A pixel is identified as a cloud pixel upon triggering a specific identification condition, and is then marked for removal. Specifically: Cloud cover is complex and variable, and has the greatest impact on the algorithm. On one hand, the presence of clouds can easily obscure ground fire information, causing energy attenuation of fire points entering the sensor and leading to missed detections. On the other hand, specular reflection from clouds can cause misidentification of fire points. Therefore, cloud pixel identification is a crucial factor in fire point identification. Based on the MODIS cloud identification method, cloud pixel information is extracted according to the different light radiation characteristics of cloud, water, and land pixels. The identification methods are shown in Table 1. A pixel is identified as a cloud pixel when any one of the criteria is met.

[0077] Table 1

[0078]

[0079] In Table 1, T7, T 13 and T 14 These are the brightness temperatures of the mid-infrared channel, far-infrared channel, and far-infrared split-window channel, respectively. vis θsz represents the reflectance value of the visible light channel, and θsz represents the solar zenith angle.

[0080] By removing cloud pixels from the grid data of the monitoring area, the target pixels in the grid data of the monitoring area can be determined.

[0081] Step S3: Based on the target pixel, generate fire point pixels. Specifically, Step S3 includes: Step S301: Identify suspected high-temperature fire point pixels in the vicinity of the target pixel, and calculate the mean and standard deviation of the background brightness temperature in the vicinity of the target pixel to generate thermal anomaly pixels. Step S302: Identify the thermal anomaly pixels using a second identification strategy to generate a first undetermined fire point pixel. Step S303: Identify the first undetermined fire point pixel using a third identification strategy to generate a second undetermined fire point pixel. Step S304: Supplement the identification of cloud pixels using a cloud area fire point supplementation identification strategy to generate a third undetermined fire point pixel. Step S305: Combine the first, second, and third undetermined fire point pixels to generate a fire point pixel. The third identification strategy specifically includes: a cloud pollution identification strategy, a cloud edge or desert area edge strategy, and a fixed artificial heat source identification strategy.

[0082] Specifically, the cloud contamination identification strategy is as follows: if the first undetermined fire point pixel meets the following conditions, it will not be considered a fire point pixel.

[0083]

[0084] In the formula, R vis Here, RC_th is the visible light channel reflectance value, TC_th is the visible light cloud contamination identification threshold, and T_th is the far-infrared cloud interference identification threshold. 13 For far-infrared channels, T represents the background reflectance value. 13_g This represents the background temperature value of the far-infrared channel.

[0085] Specifically, the cloud edge or desert area edge strategy is as follows: if the first undetermined fire point pixel meets the following conditions, it will not be considered a fire point pixel.

[0086] T7≤T 7_ +C7×δT 7_ &T 7_ ≤T 7__ +C 7_ ×δT 7___g

[0087] In the formula, T7 is the mid-infrared channel, C7 is the cloud edge impact detection threshold, and C 7_3 T is the threshold for identifying the impact of desert edges. 7_g δT represents the background temperature value for the mid-infrared channel. 7_ T represents the standard deviation of the effective background brightness temperature in the mid-infrared region. 7_3 T represents the pixel brightness temperature difference between mid-infrared and far-infrared regions. 7_13_ δT represents the effective background brightness temperature difference between the mid-infrared and far-infrared regions. 7_3__bg The standard deviation of the pixel brightness temperature difference between the effective background of mid-infrared and far-infrared regions;

[0088] The specific strategy for identifying fixed artificial heat sources is as follows: manually eliminating non-hot abnormal points.

[0089] Specifically, the following operations are performed based on the target pixel:

[0090] 2. Background brightness temperature calculation: Identify suspected high-temperature fire points in the vicinity of the target pixel; calculate the mean and standard deviation of the background brightness temperature in the vicinity of the target pixel. Specifically: For the identification of suspected fire points in the background, select pixels within the top 20% of the target pixels that exhibit high temperatures as thermal anomaly pixels. These high-temperature pixels are then individually identified as potential fire points, and suspected fire point pixels are excluded. The method for identifying thermal anomaly pixels is as follows:

[0091] T7≥T 13 +100×R vis +CB

[0092] In the formula, CB is the background threshold, which is usually taken as 20K.

[0093] Background brightness temperature mean and standard deviation calculation: This involves calculating the background temperature information of the target pixel, i.e., the average value of neighboring pixels, including the average mid-infrared brightness temperature, the average far-infrared brightness temperature, the average difference between mid-infrared and far-infrared, and the average visible light reflectance, etc., used as the basis for fire point confirmation. The calculation method is as follows:

[0094] Select the monitorable pixels from the 7×7 neighboring pixels of the target pixel and exclude pixels that are immediately adjacent to the detected pixel. If the number of monitorable pixels is less than 20% of the total number of neighboring pixels, expand to 9×9, 11×11, and so on up to 19×19. If the condition is still not met, abandon the monitoring of the pixel.

[0095] Calculate the average and standard deviation of the mid-infrared brightness temperature, the average and standard deviation of the far-infrared brightness temperature, the average and standard deviation of the difference between the mid-infrared and far-infrared brightness temperatures, and the average and standard deviation of the visible light channel for the monitorable pixels in the neighborhood.

[0096]

[0097]

[0098] In the above formula, c represents the sequence number, which in the text refers to the mid-infrared channel, the far-infrared channel, and the difference between the mid-infrared and far-infrared channels, respectively; n represents the total number of measurable pixels in the neighborhood; T ci The values ​​of the identified pixel are: mid-infrared brightness temperature, far-infrared brightness temperature, mid-infrared and far-infrared brightness temperature difference, and visible light reflectance; i represents the detectable pixels in the surrounding neighborhood of the target pixel; T represents the... c_bg These represent the average brightness temperatures of surrounding background pixels in the mid-infrared, far-infrared, and mid-infrared-far-infrared regions, respectively, δT. c_bgThese are the standard deviations of brightness temperature for mid-infrared, far-infrared, and the difference between mid-infrared and far-infrared, respectively. When the standard deviation is less than 2K, the standard deviation is set to 2K; if the standard deviation is greater than 3.5K, the standard deviation is set to 3.5K.

[0099] 3. Heat source extraction

[0100] Satellite heat source extraction mainly relies on the characteristics of mid-infrared thermal anomalies, which can be divided into two categories: absolute conditions and relative conditions. When a pixel meets condition 1) or 2), it can be identified as a fire point pixel (the first undetermined fire point pixel). Since the FY-4B geostationary satellite observes the sun from all angles, the solar zenith angle, cloud cover, and non-vegetation type have a significant impact on the calculation of background pixel brightness temperature. When the solar zenith angle exceeds 60°, reflectivity correction is prone to distortion. To reduce misjudgment, the identification threshold needs to be increased. Therefore, the influence of solar zenith angle, cloud cover, and non-vegetation information needs to be considered.

[0101] 1) T7 > 360K&R vis <0.7&θ sz >3°

[0102] 2) T7≥T 7_bg +a×δT 7_bg &T 7_13 ≥T 7_13_bg +a×δT 7_13_bg

[0103]

[0104] In the formula, 'a' is a coefficient function related to the proportion of non-vegetation pixels, the proportion of cloud pixels, and the solar zenith angle; 'Pv' is the proportion of non-vegetation pixels within the region (slow variable); 'Pc' is the proportion of cloud pixels within the region (fast variable); and 'T'... 7_bg The value is the background temperature value of the mid-infrared channel. Existing methods use a fixed threshold, while this invention changes the traditional fixed threshold method to an adaptive threshold, that is, the relevant threshold changes with the observation state, thereby making the monitoring timeliness and accuracy of this invention high.

[0105] 4. Reprocessing of heat source points: Three identification strategies (the third strategy) are employed, using union logic to trigger an identification condition, classifying the pixel as a non-hotspot and marking it for removal. The identification strategies include: cloud contamination identification strategy, cloud edge or desert area edge identification strategy, and fixed artificial heat source point strategy. Specifically:

[0106] Satellite monitoring acquires thermal anomaly signals from the mid-infrared channel. Besides actual fire sources, this includes information from non-fire sources, such as specular reflections from cloud edges and highly reflective underlying surfaces, as well as non-fire sources like bare ground, man-made structures, and factories. Therefore, the extracted heat source points need further processing to remove non-fire anomalies.

[0107] (1) Cloud interference removal

[0108] If a fire pixel (the first pending fire pixel) meets the following conditions, it is considered to be affected by cloud contamination and will not be considered a fire pixel:

[0109]

[0110] RC_th is the visible light cloud pollution detection threshold, with an initial value of 0.15; TC_th is the far-infrared cloud interference detection threshold, with an initial value of 5K. This represents the background reflectance value.

[0111] (2) Removal of cloud edge and desert area edge effects

[0112] If a fire point pixel is confirmed to meet the following conditions, it is considered to be affected by the edge of a cloud or desert area and will not be considered a fire point pixel:

[0113] T7≤T 7_bg +C7×δT 7_bg &T 7_13 ≤T 7_13_bg +C 7_13 ×δT 7_13__bg

[0114] In the formula, T7 is the mid-infrared channel; C7 is the cloud edge impact detection threshold, initially set to 8; C 7_13 The threshold for identifying the impact of desert edges is initially set to 8; T 7_bg δT represents the background temperature value for the mid-infrared channel. 7_bg T represents the standard deviation of the effective background brightness temperature in the mid-infrared region. 7_13 T represents the pixel brightness temperature difference between mid-infrared and far-infrared regions. 7_13_bg δT represents the effective background brightness temperature difference between the mid-infrared and far-infrared regions. 7_13__bg The standard deviation of the pixel brightness temperature difference between the effective background in the mid-infrared and far-infrared regions is given.

[0115] (3) Manually remove non-hot anomalies, mainly referring to fixed targets in locations such as factories, which are stored in a conventional heat source database (dataset).

[0116] Existing monitoring methods lack a thermal anomaly reprocessing procedure. This invention proposes that satellite monitoring acquires mid-infrared thermal anomaly signals, which, in addition to actual heat sources, also include non-heat source information, such as cloud edges, specular reflections from highly reflective underlying surfaces, and non-heat sources like bare land, man-made structures, and factories. Therefore, reprocessing the extracted heat source points to remove non-heat anomalies can significantly improve the accuracy of the monitoring method of this invention.

[0117] 5. Cloud area fire point supplementation, specifically including cloud area pixel background brightness temperature estimation and cloud area pixel fire point identification calculation. Cloud area fire point identification is mainly used for fire point identification under thin clouds. The steps are as follows:

[0118] (1) Calculate the average value of cloud pixels

[0119] Calculate the average value of a cloud area when a cloud pixel meets the following conditions.

[0120] T7≥TC7th&T 7_3 ≥T 7_3 th&R vis ≤RC vis th

[0121] TC7th is the mid-infrared brightness temperature threshold for cloud pixels, initially set to 325K; T 7_13 th is the threshold value for the difference in brightness temperature between the mid-infrared and far-infrared regions of a cloud pixel, initially set to 30K; RC vis th is the visible light reflectance threshold of cloud pixels, with an initial value of 0.6.

[0122] (2) Determine whether the pixels in the cloud area meet the conditions for fire detection.

[0123] If T7≥T 7_bg +T fire Based on the condition, determine that the pixel is a fire point in the cloud area (the third undetermined fire point pixel);

[0124] TC fire th is the threshold for identifying fire points in cloud pixels, with an initial value of 20K.

[0125] At this point, by combining the first undetermined fire point pixel, the second undetermined fire point pixel, and the third undetermined fire point pixel, the fire point pixel can be obtained.

[0126] Step S4, based on the fire point pixels, generates the open flame area. This is done by estimating the open flame area using a single-channel sub-pixel fire point area estimation method. Specifically:

[0127] 6. Method for estimating open flame areas

[0128] The sub-pixel fire point area, i.e. the open flame area, is estimated using a single-channel sub-pixel fire point area estimation method. The estimation formula is shown in equation (1). When the mid-infrared channel is not saturated, the mid-infrared channel is used for estimation, as shown in equation (2). When the mid-infrared channel is saturated, the far-infrared channel is used for estimation, as shown in equation (3).

[0129] S f =P×S (1)

[0130] P=(L MIR_mix (T MIR_mix )-L MIR_bg (TMIR_bg )) / (L MIR (T f )-L MIR_bg (T MIR_bg (2)

[0131] P=(L FIR_mix (T FIR_mix )-L FIR_bg (T FIR_bg )) / (L FIR (T f )-L FIR_bg (T FIR_bg (3)

[0132] In the formula: Sf is the sub-pixel fire point area, P is the sub-pixel fire point area ratio, and S is the pixel area. Tf is the sub-pixel fire point temperature, set to 750K. L MIR_mix L FIR_mix The emissivity of the mixed pixel for the mid-infrared and far-infrared channels, L MIR_bg L FIR_bg T represents the background pixel emissivity of the mid-infrared and far-infrared channels, respectively. MIR_mix T FIR_mix These are the brightness temperatures of the mixed mid-infrared and far-infrared pixels, T. MIR_bg T FIR_bg The background pixel brightness temperatures for the mid-infrared and far-infrared channels are respectively, L MIR For the mid-infrared channel, L FIR It is a far-infrared channel.

[0133] Step S5: Output the fire monitoring results based on the area of ​​the open flame zone.

[0134] In practical application, from April 18 to 20, 2022, a large grassland fire occurred in Mongolia, near the border of Xilin Gol League in Inner Mongolia Autonomous Region, my country, with an actual burned area exceeding 2,500 km². 2 This is the most threatening grassland fire to the border of Inner Mongolia Autonomous Region in recent years. Using the FY-4B geostationary satellite fire point identification algorithm and FY-4B / AGRI data proposed in this invention, we continuously tracked and monitored the entire process of this grassland fire, obtaining dynamic information on its development, including the area of ​​open flames at each observation time, the spatiotemporal dynamic range of the fire, and the spatial distribution of high-intensity fire zones within the fire area.

[0135] 1. Spatiotemporal dynamics analysis of fire spread

[0136] This invention extracted fire point information from FY-4B / AGRI for 35 hours from 15:00 on April 18, 2022 to 00:00 on April 20, 2022, and generated a fire point observation time series diagram, see [link / reference needed]. Figure 6 The blue to red areas in the image represent fire observations from the beginning to the end. As shown in the image, the fire initially appeared (at 15:00 on April 18th (Beijing time), the same below) near the border of Xilin Gol League in Inner Mongolia Autonomous Region (dark blue in the image). It then spread eastward, reaching the vicinity of the border of Hinggan League around 15:00 on the 19th. The fire area spanned approximately 93 kilometers east to west and was about 61 kilometers wide at its widest point north to south, forming a large burned area. Figure 6 The bottom center is the timeline, with colors in the order of blue-green-red. The darkest area on the far left of the dotted area at the top of the timeline is blue, the area to the right of the blue is green, and the darkest area above the green is red.

[0137] Monitoring and analysis of the spread of the grassland fire using the FY-4B satellite revealed that it unfolded in three phases. The first phase (see...) Figure 7 From 15:00 on April 18th to 06:00 on April 19th, during the initial stage of the fire, the fire spread eastward and northward with the wind direction; the second stage (see...) Figure 8 The third stage (see...) marks the rapid spread of the fire. From 06:00 to 17:00 on April 19th, the fire spread shifted from a northeast to a southeast direction, rapidly reaching the borders of Xilingol League and Hinggan League in my country, forming a large-scale burned area. Figure 9 During the stage of fire reversal and spread, from 20:00 on April 19th to 00:00 on April 24th, the fire spread from the northern part of the area in a north-northeast direction, approaching the border of Hinggan League and Hulunbuir League. Meanwhile, a small number of fire spots remained in the southeastern part of the fire area. Figure 6 Available all time; Figure 7 The period is from 15:00 on the 18th to 06:00 on the 19th. Figure 8 From 06:00 to 17:00 on the 19th; Figure 9 The period is from 17:00 on the 19th to 00:00 on the 20th. Among them, Figure 7 In the upper part of the timeline, the darkest area on the far left of the dotted region is blue, the area to the right of the blue is green, and the darkest area to the upper right of the green is red; among them, Figure 8 In the upper part of the timeline, the darkest area on the far left of the dotted region is blue, the area to the right of the blue is green, and the darkest area to the right of the green is red; among them, Figure 9 The darkest areas of the dotted regions at the top of the timeline are green, the slightly darker areas are blue, and the darkest areas are red.

[0138] 2. Dynamic Changes in the Extent of Open Flame Zones in the Fire Scene

[0139] Using the multi-time continuous monitoring results from FY-4B, the open flame area was extracted hourly. The changes in open flame area show that there were two peak open flame areas in this fire, at 09:00 and 21:00 on the 19th. Figure 10Analysis of the fire's spread timeline indicates that during the first stage of the fire, the area of ​​open flame gradually increased, but the rate of increase was relatively small. The largest open flame area occurred at 04:00 on the 19th, approximately 28 hectares. 2 The second phase began at 06:00 on the 19th, during which the fire intensified rapidly, and the area of ​​open flame increased quickly, reaching its maximum at 09:00 on the 19th, approximately 125 hectares. 2 The fire initially shrank, then gradually decreased; in the third phase, starting at 17:00 on the 19th, the fire intensified again, reaching a second peak at 21:00, with the open flame area covering approximately 87.1 hectares. 2 Analysis of the fire's development indicates that in the first stage, due to the fire's recent occurrence and its incubation period, the initial burned area was relatively small, resulting in a limited increase in burned area due to the spread of the fire. The second stage was characterized by rapid fire development. Figure 8 It is evident that the affected area exceeds 180,000 hectares. 2 The maximum instantaneous open flame area occurred during this period; in the third stage, apart from the spread in the north, the rest were all sporadic fire points within the original burned area, and the overall burned area was significantly reduced compared to the second stage.

[0140] 3. Analysis of the distribution of high-intensity fire points in the fire scene

[0141] To further analyze the quality of FY-4B / AGRI fire spot monitoring and obtain information on fire spots of different intensities (especially high-intensity fire spots) in the fire scene, this invention extracted original satellite observation images covering the entire fire scene at 15:00, 19:00, and 23:00 on the 18th of 2022 and at 06:00, 09:00, 13:00, 17:00, and 21:00 on the 19th of 2022, respectively. These images represent typical satellite images of the fire's initiation, development, and different stages of the fire.

[0142] In the infrared channel signal display for fire point monitoring, according to Wien's displacement law, temperature is inversely proportional to the center wavelength; different channels react as the fire intensity changes. This invention employs a combination of mid-infrared, far-infrared, and short-wave infrared signals, i.e., a three-channel composite image at 3.7µm, 11µm, and 2.23µm, which maintains consistency in daytime observations and also displays the distribution of different fire intensities within the fire field. Figures 11 to 18 As shown in the series of images, the fire scene is primarily displayed in three colors: red, yellow, and white. Red represents a single mid-infrared channel signal, yellow represents mid-infrared and far-infrared channel signals, and white represents a combined three-channel signal. When the fire is small, thermal anomalies are mainly observed in the mid-infrared channel. Figure 11 This moment marks the initial stage of the fire, when the fire is not strong and the energy is mainly concentrated in the 3.7µm channel, with the fire point appearing red. As the fire intensifies, thermal anomalies are observed in both far-infrared and near-infrared radiation. Figure 13The periphery is red in the mid-infrared channel, and the center is yellow, displayed by a combination of mid-infrared and far-infrared rays; when the fire intensifies to a certain extent, a thermal signal appears in the 2.22µm short-wave infrared, and the fire point appears white, as shown. Figure 16 As observed, the central area of ​​the fire tongue (the front of the fire spread) is white, surrounded by yellow pixels, while the perimeter of the fire shows a red outline. This indicates that the fire is strongest at the fire tongue and weakest at the fire edge. This demonstrates the sensitivity characteristics of different channels of the FY-4B / AGRI to different fire intensities, making it effective for disaster prevention and emergency support services in forest and grassland fire prevention. Figures 11 to 18 In the middle, the irregular outer frame is yellow, the bright spots within the irregular area are white, and the dark spots around the white bright spots are red.

[0143] Example 2

[0144] like Figure 5 As shown in Example 2, a high-frequency dynamic remote sensing monitoring system for grassland fires using a geostationary meteorological satellite is provided. Based on a geostationary meteorological satellite, the system includes: an acquisition and generation module 1, a determination module 2, a fire point pixel generation module 3, an open fire area generation module 4, and an output module 5. The acquisition and generation module 1 acquires geostationary meteorological satellite data and performs data preprocessing to generate grid data for the monitoring area. The determination module 2 determines the target pixels in the grid data based on the monitoring area grid data. The fire point pixel generation module 3 generates fire point pixels based on the target pixels. The open fire area generation module 4 generates the open fire area based on the fire point pixels. The output module 5 outputs the fire monitoring results based on the open fire area. The acquisition and generation module 1 is communicatively connected to the determination module 2, the determination module 2 is communicatively connected to the fire point pixel generation module 3, the fire point pixel generation module 3 is communicatively connected to the open fire area generation module 4, and the open fire area generation module 4 is communicatively connected to the output module 5.

[0145] Example 3

[0146] like Figure 19 As shown in Example 3, a geostationary meteorological satellite high-frequency dynamic remote sensing monitoring computing device for grassland fires is provided, comprising: one or more processors; and a storage device for storing one or more programs. When one or more programs are executed by one or more processors, the one or more processors implement the geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described above.

[0147] Figure 19 This is a schematic diagram of the structure of a geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device according to an embodiment of the present invention. Figure 19 The geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0148] like Figure 19 As shown, the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device is presented in the form of a general-purpose computing device. The components of the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device may include, but are not limited to: one or more processors or processing units, memory, and buses connecting different system components (including memory and processing units).

[0149] A bus refers to one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0150] Geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing equipment typically includes a variety of computer-readable media. These media can be any available media that can be accessed by the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing equipment, including volatile and non-volatile media, and portable and non-portable media.

[0151] The memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory. The geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system may be used to read and write non-removable, non-volatile magnetic media (…). Figure 19 Not shown; usually referred to as a "hard drive"). Although Figure 19 Not shown, disk drives for reading and writing to removable non-volatile disks (e.g., "floppy disks") and optical disc drives for reading and writing to removable non-volatile optical discs (e.g., CD-ROMs, DVD-ROMs, or other optical media) may be provided. In these cases, each drive may be connected to a bus via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0152] A program / utility having a set (at least one) of program modules can be stored, for example, in memory. Such program modules include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules typically perform the functions and / or methods described in the embodiments of this invention.

[0153] The geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device can also communicate with one or more external devices (e.g., keyboard, pointing device, monitor, etc.), and with one or more devices that enable users to interact with the device, and / or with any device that enables the device to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through an input / output (I / O) interface. Furthermore, the device can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via a network adapter. Figure 19 As shown, the network adapter communicates with other modules of the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device via a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring computing device, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0154] The processing unit executes various functional applications and data processing by running programs stored in the memory, such as implementing the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring method provided in any embodiment of the present invention.

[0155] Example 4

[0156] Example 4 provides a computer-readable storage medium storing a program that, when executed by a processor, implements the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring method as described in any embodiment of the present invention.

[0157] In summary, the geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method and system for grassland fires of the present invention can detect and monitor the status of grassland fires with high timeliness and high accuracy, and output fire monitoring results.

[0158] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A method for high-frequency dynamic remote sensing monitoring of grassland fires using geostationary meteorological satellites, characterized in that: The method includes: Acquire geostationary meteorological satellite data and perform data preprocessing to generate grid data for the monitoring area; Based on the grid data of the monitoring area, the target pixels in the grid data of the monitoring area are determined; Based on the target pixel, generate the fire point pixel; Based on the fire point pixels, the area of ​​the open flame zone is generated; Based on the area of ​​the open flame zone, output the fire monitoring results; Specifically, based on the grid data of the monitoring area, the target pixels in the grid data of the monitoring area are determined as follows: Based on the grid data of the monitored area, cloud pixel information is extracted; The cloud pixel information is identified using the first identification strategy to obtain cloud pixels; Remove the cloud pixels to determine the target pixels in the grid data of the monitoring area; Specifically, generating fire point pixels based on the target pixels involves: The target pixel is identified as a suspected high-temperature fire point pixel in the vicinity of the target pixel, and the mean and standard deviation of the background brightness temperature in the vicinity of the target pixel are calculated to generate thermal anomaly point pixels. The thermal anomaly point pixels are identified using the second identification strategy to generate the first undetermined fire point pixel; The first undetermined fire point pixel is identified using the third identification strategy to generate the second undetermined fire point pixel; The cloud pixels are supplemented and identified using a cloud fire point supplementation and identification strategy to generate a third undetermined fire point pixel. By combining the first undetermined fire point pixel, the second undetermined fire point pixel, and the third undetermined fire point pixel, a fire point pixel is generated. Specifically, the area of ​​the open flame zone generated based on the fire point pixels is as follows: Based on the fire point pixel, the fire point area estimation method of single-channel sub-pixel is used to determine the fire point area ratio according to the difference in emissivity between the fire point pixel and its background in the preset infrared channel, and the open flame area is generated by combining the pixel area.

2. The geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described in claim 1, characterized in that, The process of acquiring geostationary meteorological satellite data, performing data preprocessing, and generating grid data for the monitoring area is as follows: acquiring Earth observation data from geostationary meteorological satellites, performing data preprocessing, and generating grid data for the monitoring area. The preprocessing of the geostationary meteorological satellite Earth observation data includes geolocation information correction, data projection, and reflectance correction.

3. The geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described in claim 2, characterized in that, The geolocation information correction for the Earth observation data from the geostationary meteorological satellite is specifically performed using the following formula: ， ， In the formula, and These are the offsets for longitude and latitude, respectively. and These are the longitude and latitude coordinates for FY4B. and These are the longitude and latitude coordinates for the coastline and lake vectors, respectively.

4. The geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described in claim 2, characterized in that, The reflectance correction for the Earth observation data from the geostationary meteorological satellite is specifically performed using the following formula: ， In the formula, R is the original reflectivity of the satellite. The reflectance is the value corrected for the solar zenith angle. This is the solar zenith angle.

5. The geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described in claim 1, characterized in that, The third identification strategy specifically includes: cloud pollution identification strategy, cloud edge or desert area edge strategy, and fixed artificial heat source identification strategy; Specifically, the cloud contamination identification strategy is as follows: if the first undetermined fire point pixel is confirmed to meet the following conditions, it will not be considered a fire point pixel. & ， In the formula, R vis Here, RC_th is the visible light channel reflectance value, TC_th is the visible light cloud contamination identification threshold, and T_th is the far-infrared cloud interference identification threshold. 13 For far-infrared channels, This represents the background reflectance value. This represents the background temperature value of the far-infrared channel. Specifically, the cloud edge or desert area edge strategy is as follows: if the first undetermined fire point pixel is confirmed to meet the following conditions, it will not be considered a fire point pixel. & ， In the formula, For mid-infrared channel, Threshold for identifying the impact of cloud edges Threshold for identifying the impact of desert edges This represents the background temperature value for the mid-infrared channel. The standard deviation of the effective background brightness temperature in the mid-infrared region. The pixel brightness temperature difference between mid-infrared and far-infrared. The effective background brightness temperature difference between mid-infrared and far-infrared. The standard deviation of the pixel brightness temperature difference between the effective background of mid-infrared and far-infrared regions; Specifically, the fixed artificial heat source identification strategy is as follows: manually eliminating non-hot abnormal points.

6. A geostationary meteorological satellite high-frequency dynamic remote sensing monitoring system for grassland fires, based on the geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described in any one of claims 1-5, characterized in that, The system includes: The acquisition and generation module is used to acquire geostationary meteorological satellite data, perform data preprocessing, and generate grid data for the monitoring area. The determination module is used to determine the target pixels in the grid data of the monitoring area based on the grid data of the monitoring area; Fire point pixel generation module, used to generate fire point pixels based on the target pixel; The open flame area generation module is used to generate the open flame area based on the fire point pixels; and The output module is used to output fire monitoring results based on the area of ​​the open flame zone; The acquisition and generation module is communicatively connected to the determination module, the determination module is communicatively connected to the fire point pixel generation module, the fire point pixel generation module is communicatively connected to the open flame area generation module, and the open flame area generation module is communicatively connected to the output module.

7. A geostationary meteorological satellite high-frequency dynamic remote sensing monitoring and computing device for grassland fires, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the geostationary meteorological satellite grassland fire high-frequency dynamic remote sensing monitoring method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the geostationary meteorological satellite high-frequency dynamic remote sensing monitoring method for grassland fires as described in any one of claims 1-5.

Images