Satellite remote sensing and trajectory simulation-based wild fire smoke plume attenuation correction method and system

Abstract

The present application relates to satellite remote sensing and atmospheric environment monitoring technical field, disclose a kind of based on satellite remote sensing and trajectory simulation's wild fire smoke plume attenuation correction method and system.The method obtains wild fire smoke plume boundary, target active gas satellite observation, active fire point and meteorological data, concentration is completed to invalid or missing measurement pixel in plume boundary, determines each satellite pixel observation quality;Backward trajectory is established in each pixel, matches candidate fire point and determines transportation time;According to the fire radiation power of candidate fire point and target active gas retention ratio determines contribution weight, distributes pixel observation quality and inverts initial emission mass, and the initial emission mass of wild fire smoke plume is obtained by summarizing.This application can consider the mass attenuation difference caused by different burning intensity and transmission duration of multiple fire points, improve the rationality and reliability of wild fire emission mass inversion, and can be used for wild fire emission estimation and emission inventory verification.

Description

Technical Field

[0001] This invention relates to the field of satellite remote sensing and atmospheric environment monitoring technology, specifically to a wildfire plume attenuation correction method and system based on satellite remote sensing and trajectory simulation. Background Technology

[0002] Wildfires release large amounts of gaseous pollutants such as nitrogen oxides, formaldehyde, and carbon monoxide into the atmosphere, significantly impacting regional air quality and the physicochemical properties of the atmosphere. Accurately estimating fire emissions is a prerequisite and foundation for effective atmospheric chemical simulation, air quality assessment, and wildfire environmental impact analysis. However, current emission estimation methods suffer from significant uncertainties. Traditional methods often rely on parameters such as burned area, combustible load, and emission factors for indirect estimation, failing to account for the spatiotemporal heterogeneity of factors like meteorological conditions, combustion stage, and combustible dryness / wetness on wildfire combustion and emissions. This leads to substantial discrepancies between different estimates, even on a global average scale, where differences can reach several times.

[0003] Satellite remote sensing technology has the advantages of wide coverage and the ability to directly acquire the spatial distribution characteristics of reactive gas concentrations such as nitrogen dioxide in wildfire plumes. However, reactive gases are continuously consumed during transport due to their participation in chemical reactions. The observation values ​​at the moment of satellite transit correspond to the plume state after a certain period of transport, rather than the initial emission state of the fire source. If the emission amount is directly inverted from the concentration observed by satellite, the mass loss during transport will be ignored, leading to a systematic underestimation. Although existing technologies can use air mass trajectory models to analyze the transport path of atmospheric pollutants and can use parameters such as the atmospheric lifetime of reactive gases to consider the concentration decay of point sources or relatively stable emission sources (such as industrial emissions), these methods are difficult to directly apply to the complex formation process of real wildfire plumes. This is mainly reflected in the significant spatiotemporal dynamic changes of wildfire sources, and the strong spatiotemporal heterogeneity of their emission status. The gas at a certain location within the plume may originate from emissions from multiple different fire points at different times, and be the result of mixing after experiencing different degrees of decay. At the same time, the emission sources, atmospheric transport, and decay degrees of different parts of the plume may also vary greatly. These complex scenarios pose challenges to accurately correcting the mass decay of the plume. Currently, there is a lack of corresponding technical means to achieve differentiated attenuation correction within the plume in such real wildfire scenarios.

[0004] Therefore, how to make full use of the spatial concentration information of the smoke plume observed by satellite, combined with the dynamics of the fire source and atmospheric transport information, to make fine corrections for the mass attenuation of active gases at different locations in the smoke plume during the transport process, so as to improve the accuracy of the initial emission mass estimation of wildfires, is a technical problem to be solved in this field. Summary of the Invention

[0005] This invention aims to address the technical problem that existing methods for retrieving the mass of reactive gases in wildfire plumes are insufficient to effectively distinguish and correct the actual mass decay at different spatial locations within the plume, as the concentration observed by satellite deviates from the initial emission state due to chemical consumption during atmospheric transport. Furthermore, the gas sources, transport paths, and attenuation levels at different locations within the actual wildfire plume vary.

[0006] To address the above problems, this invention provides a wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation, comprising the following steps:

[0007] S1. Acquire wildfire plume boundary data, target reactive gas satellite observation data, satellite active fire point data, and meteorological data. When there are satellite pixels with invalid or missing target reactive gas observation concentrations within the plume boundary, complete the target reactive gas concentrations of the corresponding satellite pixels. Determine the observation quality of each satellite pixel based on the target reactive gas concentrations of each satellite pixel within the plume boundary.

[0008] S2. Establish the backward trajectory of the air mass for each satellite pixel, and determine the candidate fire points that match each backward trajectory based on the spatiotemporal matching relationship between the trajectory points and the active fire points of the satellite; for the satellite pixels that match the candidate fire points, determine the transportation time from each candidate fire point to the corresponding satellite pixel.

[0009] S3. When there are multiple candidate fire points, the contribution weight of each candidate fire point to the observation quality of the corresponding satellite pixels is determined according to the fire radiation power of each candidate fire point and the retention ratio of the target active gas corresponding to the transportation time, and the observation quality of the satellite pixels is allocated according to the contribution weight.

[0010] S4. Based on the retention ratio of the target reactive gas corresponding to each candidate fire point, the allocated observation quality is inverted into the initial emission quality of each candidate fire point, and the sum is used to obtain the initial emission quality of the corresponding satellite pixel; when there is only one candidate fire point, the initial emission quality of the corresponding satellite pixel is directly inverted based on the retention ratio of the target reactive gas corresponding to that candidate fire point.

[0011] S5. Summarize the initial emission quality of each satellite pixel within the plume boundary that has obtained the initial emission quality, and determine the initial emission quality of the wildfire emission plume.

[0012] Optionally, the wildfire plume is a spatially isolated plume, distinct from the atmospheric background, and the corresponding fire source is located within the plume boundary. The satellite active fire point data includes at least the fire point location, fire point detection time, and fire radiation power. The extraction period for active fire point data is determined based on the plume boundary formation time and the atmospheric lifetime of the target reactive gas, and satellite observation concentration data of the target reactive gas corresponding to the plume boundary observation time is selected to ensure that the plume boundary, satellite observation concentration, and active fire point data have a temporal and spatial correspondence.

[0013] In one implementation, satellite pixels within the plume boundary are extracted from the target reactive gas satellite observation concentration data. Valid target reactive gas concentration satellite pixels and invalid or missing satellite pixels are identified based on satellite product quality indicators, and the proportion of valid observation pixels is calculated. When the proportion reaches a set threshold, a set of usable concentration satellite pixels is formed from the satellite pixels with valid concentration observations. Satellite pixels to be supplemented are selected sequentially according to the original storage order of the satellite pixels. Spatially adjacent satellite pixels are selected from the set of usable concentration satellite pixels. Each satellite pixel is used to estimate the concentration based on the average concentration of the target reactive gas. The completed satellite pixels are then added to the set of available concentration satellite pixels until all satellite pixels with invalid or missing concentration measurements are completed. The value can be 1 to 4. In subsequent quality calculations, the original observed concentration is used for satellite pixels with valid observed concentrations, and the estimated concentration is used for satellite pixels that have been completed.

[0014] Concentration of target reactive gas column from satellite pixels Subtracting atmospheric background concentration The concentration of wildfire emissions was obtained. :

[0015] ;

[0016] when The unit is mol / m 2 At that time, the observation quality of reactive gas of the target within the satellite pixel Determined according to the following formula:

[0017] ;

[0018] when The unit is molecules / cm 2 At that time, the observation quality of reactive gas of the target within the satellite pixel Determined according to the following formula:

[0019] ;

[0020] in, This refers to the mass remaining in a satellite pixel after the target reactive gas has been chemically consumed during atmospheric transport, i.e., the observed mass. For satellite pixels with effective observed concentrations, The target reactive gas column concentration provided for satellite products; for satellite pixels with concentration-complemented data, To complete the estimated column concentration; The area of ​​a satellite pixel is expressed in square meters (m²). 2 ; The molar mass of the target reactive gas is expressed in g / mol. It is Avogadro's constant; To reduce the satellite pixel area from m 2 Convert to cm 2 The conversion factor.

[0021] Optionally, for each satellite pixel within the plume boundary, the position and observation time of the satellite pixel are used as the starting conditions for the backtracking trajectory of the air mass, and the corresponding backtracking trajectory data is obtained through an air mass backtracking trajectory model. The air mass backtracking trajectory model can be the HYSPLIT model. The duration of the air mass backtracking trajectory can be determined according to the atmospheric lifetime of the target reactive gas. The value is determined by a factor of 1.5. For trajectory points in the backward trajectory, the active fire points of the satellite are matched with the trajectory points in time and space to determine candidate fire points. Candidate fire points satisfy the following time and space matching conditions:

[0022] ;

[0023] and:

[0024] ;

[0025] in, For the detection time of active fire points of the satellite, This represents the time corresponding to the trajectory point in the backward trajectory. The maximum allowable time difference; The location of the satellite's active fire point. For the position of the corresponding trajectory point, The maximum permissible spatial distance. Satellite active fire points that meet the above time and spatial conditions are identified as candidate fire points corresponding to the backward trajectory of the corresponding satellite pixel. When the same satellite active fire point and multiple trajectory points in the same backward trajectory all meet the matching conditions, each matching relationship is retained, each matching relationship participates in subsequent calculations, and each is counted separately in the number of candidate fire points. The transportation time is determined according to the corresponding trajectory point.

[0026] For multiple candidate fire points matching the backward trajectory of the same satellite pixel, the contribution weight of each candidate fire point is determined jointly based on the fire radiation power of each candidate fire point and the retention ratio of the target reactive gas after the corresponding transportation time. The retention ratio of target reactive gas corresponding to each candidate ignition point for

[0027] ;

[0028] in, The candidate fire point number to be calculated. For the first The transportation time corresponding to each candidate fire point. The atmospheric lifetime of the target reactive gas. The transport time of the target reactive gas The percentage of emissions retained relative to the initial emissions.

[0029] No. The contribution weight of each candidate fire point to the observation quality of the corresponding satellite pixel. for:

[0030] ;

[0031] in, For the first Fire radiation power of each candidate ignition point; The summation index is used to iterate through all candidate fire points corresponding to the current satellite pixel; For the first Fire radiation power of each candidate ignition point; For the first The retention ratio of the target reactive gas corresponding to each candidate ignition point; This represents the total number of candidate fire points corresponding to the current satellite pixels, and , =1,2,……, .

[0032] Based on the aforementioned contribution weights, the total mass of target reactive gas observed by the current satellite pixels is allocated to different candidate ignition points. Remaining mass corresponding to each candidate fire point for:

[0033] ;

[0034] in, For the first The mass of the target reactive gas emitted by each candidate fire point, after transmission and chemical consumption, remains in the corresponding satellite pixel.

[0035] Furthermore, according to the first The retention ratio of the target reactive gas corresponding to each candidate ignition point is used to invert the remaining mass, and the... Initial emission mass corresponding to each candidate ignition point ,for:

[0036] ;

[0037] The initial emission masses of all candidate fire points corresponding to the current satellite pixel are summed to obtain the initial emission mass corresponding to that satellite pixel. :

[0038] ;

[0039] in, This represents the initial emission mass of the target reactive gas corresponding to the current satellite pixel.

[0040] In one implementation, when the overall height of the wildfire plume is known, the average height of the plume is used as the starting height of the backward trajectory of each satellite pixel within the plume boundary. When the overall height of the wildfire plume is unknown, multiple different starting heights can be set within the possible height range of the plume, and backward trajectories for each satellite pixel within the plume boundary are established for each starting height. For each starting height, the proportion of satellite pixels whose backward trajectories match candidate fire points is calculated relative to the total number of satellite pixels involved in the processing. The starting height where the proportion reaches a set threshold is determined as the effective height layer, and the initial emission mass of the satellite pixels is calculated based on the backward trajectory corresponding to each effective height layer.

[0041] When multiple effective altitude layers exist, in one implementation, the average value of the initial emission mass obtained for the same satellite pixel at multiple effective altitude layers is taken as the final initial emission mass of the satellite pixel.

[0042] For satellite pixels whose backward trajectory does not match a candidate fire point, a spatially adjacent reference pixel can be selected from the satellite pixels within the plume boundary whose backward trajectory has already matched a candidate fire point. For the first... A reference pixel can be used to determine the quality attenuation correction factor based on the initial emission quality and the observation quality of that reference pixel. For satellite pixels whose backward trajectory does not match the candidate fire point, the mass attenuation correction coefficient of the satellite pixel is determined according to the mass attenuation correction coefficients corresponding to the selected multiple reference pixels, and the determined mass attenuation correction coefficient is used to correct its observation quality, so that satellite pixels within the plume boundary that do not directly trace to the candidate fire point can also obtain the corresponding initial emission quality.

[0043] This invention also provides a wildfire plume attenuation correction system based on satellite remote sensing and trajectory simulation. The system includes a data acquisition component, a memory, and a processor. The data acquisition component acquires wildfire plume boundary data, target reactive gas satellite observation data, satellite active fire point data, and meteorological data, and stores them in the memory. When there are satellite pixels with invalid or missing target reactive gas observation concentrations, the processor completes the corresponding target reactive gas concentrations, obtains the observation quality of each satellite pixel within the plume boundary based on the target reactive gas satellite observation data, establishes a backward trajectory of the gas mass for each pixel, and obtains candidate fire points and their transport times based on the spatiotemporal correspondence between the backward trajectory and satellite active fire points. When there are multiple candidate fire points for a corresponding satellite pixel, the contribution weight is determined based on the fire radiation power and target reactive gas retention ratio of each candidate fire point, and the observation quality of the satellite pixel is allocated accordingly. The initial emission quality of each candidate fire point is inverted based on the target reactive gas retention ratio corresponding to each candidate fire point and summed to obtain the initial emission quality of the corresponding satellite pixel.

[0044] For satellite pixels whose backward trajectory does not match a candidate fire point, the processor can correct the observation quality of the corresponding satellite pixels based on the quality attenuation correction coefficient of the reference pixels that are spatially adjacent within the plume boundary and have matched a candidate fire point.

[0045] When there is only one candidate fire point in the corresponding satellite pixel, the processor directly inverts the initial emission mass of the corresponding satellite pixel based on the retention ratio of the target reactive gas corresponding to the candidate fire point.

[0046] The processor aggregates the initial emission quality of each satellite pixel within the plume boundary that has obtained its initial emission quality, and determines the initial emission quality of the wildfire plume.

[0047] This invention is based on the residual mass of reactive gas actually observed by satellite pixels. It establishes a spatiotemporal correspondence between satellite pixels and candidate fire points through the backward trajectory of the gas cloud. Combining the fire radiation power of the candidate fire points and the proportion of reactive gas retention determined by the transport time, it distinguishes the contribution of different candidate fire points to the observed quality of satellite pixels. Then, it performs mass attenuation inversion on the corresponding observed quality, thus reasonably characterizing the mass change of reactive gas under the combined effects of fire points with different combustion intensities and transport durations. Simultaneously, for pixels with invalid or missing local concentrations in satellite observations, concentration can be supplemented using neighboring satellite pixels with available concentrations to improve the integrity of the plume observation quality field. When the plume height is unknown, effective height layers can be selected through multiple trajectory starting heights and fire point matching. For satellite pixels that do not match candidate fire points, regardless of whether the plume height is known or unknown, compensation can be made using the mass attenuation correction coefficient of neighboring reference pixels, thereby improving the integrity and stability of the quality correction results under complex observation conditions. The initial emission quality of wildfire plumes obtained in this way can more closely approximate the actual emissions of wildfires, providing a more reliable data foundation for subsequent wildfire emission assessment and comparative analysis of related emission data. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation.

[0049] Figure 2 This image shows the implementation effect of NO2 mass atmospheric attenuation correction for an isolated wildfire plume in a certain area on May 1, 2018.

[0050] Figure 2 (a) in the figure is a spatial distribution map of NO2 tropospheric column concentration and backward trajectory of some satellite pixels in the plume region, as observed by satellite.

[0051] Figure 2 (b) in the figure shows the spatial distribution of NO2 observation quality of each satellite pixel in the plume region before correction;

[0052] Figure 2 (c) in the figure is the spatial distribution map of the initial NO2 emission mass of each satellite pixel in the plume region after atmospheric mass attenuation correction;

[0053] Figure 2 (d) in the figure represents the spatial distribution map of the atmospheric chemical mass attenuation correction factor (MECF) for each satellite pixel in the plume region.

[0054] Figure 2 (e) in the figure shows the correlation between the NO2 observation quality of each satellite pixel and the corresponding aerosol quality before correction;

[0055] Figure 2(f) in the figure shows the correlation between the initial NO2 emission mass of each satellite pixel after correction and the corresponding aerosol mass;

[0056] Figure 3 This image shows the implementation effect of batch correction for NO2 mass atmospheric attenuation of 86 spatially isolated wildfire plumes in a certain region from April to May 2018.

[0057] Figure 3 (a) in the figure shows the correlation between the total NO2 mass of each wildfire plume before correction and the corresponding total aerosol mass;

[0058] Figure 3 (b) in the figure shows the correlation between the initial total mass of NO2 emissions from each wildfire plume and the corresponding total mass of aerosols after correction.

[0059] Figure 4 In response to Figure 3 The figure shows a comparison of the implementation effects of mass decay correction for 86 spatially isolated wildfire plumes using different NO2 atmospheric lifetime parameters.

[0060] Figure 4 (a) in the figure shows the correlation between the initial total mass of NO2 emissions after correction and the corresponding total mass of aerosols when the NO2 atmospheric lifetime is set to 2 hours.

[0061] Figure 4 (b) in the figure shows the correlation between the initial total mass of NO2 emissions after correction and the corresponding total mass of aerosols for each wildfire plume when the NO2 atmospheric lifetime is set to 6 hours. Detailed Implementation

[0062] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0063] The following examples are used to illustrate the implementation process and parameter selection method of the present invention, and are not intended to limit the scope of protection of the present invention. Without departing from the processing relationships, the target reactive gas type, satellite data product, meteorological data product, air mass backward trajectory model, and parameters adapted to the data resolution can all be selected according to the actual application conditions.

[0064] Example 1

[0065] This embodiment provides a wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation. The method uses target reactive gas satellite pixels within the wildfire plume boundary as the basic processing object. It utilizes satellite observation data to determine the reactive gas observation quality of each satellite pixel and traces the candidate fire points and their transport processes corresponding to each satellite pixel through the backward trajectory of the gas cloud. When a satellite pixel corresponds to multiple candidate fire points, the contribution of different candidate fire points to the satellite pixel observation quality is determined based on the fire radiation power of each candidate fire point and the retention ratio of the target reactive gas within the corresponding transport time. Based on this, the observation quality allocation and initial emission quality recovery are completed, and the initial emission quality of the entire wildfire plume is further obtained. For satellite pixels within the plume boundary with invalid or missing local observation concentrations, this embodiment performs concentration supplementation before calculating the observation quality. When the plume height is unknown, trajectory results applicable to the current plume are screened through multiple initial height layers. For satellite pixels whose backward trajectory does not match candidate fire points, compensation is performed using the quality attenuation correction coefficient of neighboring reference pixels. This compensation method is applicable to both known and unknown plume heights.

[0066] This embodiment uses nitrogen dioxide (NO2) as the target reactive gas to illustrate the specific implementation process. A spatially isolated wildfire plume within the range of 126.6°E to 131.7°E and 50.7°N to 55.4°N in a certain region on May 1, 2018, is selected as the single plume target. Furthermore, 86 spatially isolated wildfire plumes within the range of 80°E to 130°E and 10°N to 60°N in a certain region from April to May 2018 are selected for batch processing. The spatially isolated wildfire plumes refer to those spatially distinguishable from the surrounding atmospheric background, mainly formed by recent wildfire emissions, and whose corresponding fire sources are located within the plume boundary. This reduces the impact of large-scale, mixed plumes formed by different wildfire events on fire source tracing and quality correction. In other embodiments, the target reactive gas can also be selected based on satellite observation capabilities and atmospheric lifetime characteristics; for example, formaldehyde (HCHO) or carbon monoxide (CO) can be selected.

[0067] The following combination Figure 1 The process of satellite observation data processing, observation quality determination, backtracking trajectory establishment of air masses, candidate fire point matching, mass allocation, and initial emission mass recovery is explained separately.

[0068] In this embodiment, plume boundary data can be obtained using existing wildfire plume boundary identification methods, and must include at least the plume boundary formation time and the latitude and longitude information of each boundary point. When the target reactive gas is NO2, satellite observation data uses the S5P TROPOMI orbital-level tropospheric column concentration product; satellite active fire point data uses the Himawari-8 AHI geostationary satellite fire product, which provides fire point detection results at 10-minute intervals, and each fire point data includes at least latitude and longitude, UTC detection time, and fire radiation power (FRP). Meteorological data used for backward trajectory simulation can use daily meteorological data such as gfs0p25, which must include at least the wind field and related meteorological elements required for air mass trajectory calculation. If the acquired meteorological data cannot be directly read by the HYSPLIT model, it can be converted to ARL format using existing data format conversion methods. When applied to other regions or other target reactive gases, similar data products that can provide corresponding plume boundaries, gas column concentrations, active fire points, and meteorological information can also be used, without being limited to the data sources listed in this embodiment.

[0069] The temporal and spatial resolutions of different data sources are often not entirely consistent, therefore a unified data benchmark needs to be established before trajectory matching. Specifically, the plume boundary formation time, satellite pixel observation time, fire detection time, and backward trajectory output time should be unified to UTC time; the plume boundary points, satellite pixel centers, fire locations, and trajectory point locations should be unified to latitude and longitude in the same geographic coordinate system. Satellite pixels can use the pixel area directly given in the product file (if applicable), or calculate the actual area using the pixel boundary directly given in the product file, or calculate the area based on the four corner coordinates of the pixel using existing spherical or projected area algorithms.

[0070] The formation time and latitude / longitude of the plume boundary points are obtained from the plume boundary data. This embodiment uses the NO2 atmospheric lifetime... A 4-hour period is used. The atmospheric lifetime here describes the timescale for effective chemical removal of NO2, typically defined as the time it takes for the concentration to decay exponentially to 1 / e of its initial value. The representative lifetime of tropospheric NO2 can range from 2 to 6 hours; this embodiment uses the midpoint of this range, 4 hours. In practical applications, Typical lifetimes for the target region and season can be obtained from existing research literature, or, when conditions permit, can be directly measured or estimated based on OH radicals, solar radiation, or observational data for the target region. When changing the target reactive gas, simply select the corresponding atmospheric lifetime for that gas.

[0071] To construct the set of fire points to be matched, this embodiment uses the time of plume boundary formation as a benchmark and extracts the fire points before and after that time. Data on active fire points located within the plume boundary during the specified time period. When a 4-hour period is used, the location, detection time, and fire radiation power (FRP) of fire points are extracted within 8 hours before and after the plume boundary formation time. This relatively wide fire point extraction period is primarily used to reduce the possibility of missing potentially relevant fire points due to time differences between fire point detection time and plume formation. This extraction period is only used to establish a set of fire points to be matched and does not mean that all fire points within the set will participate in subsequent mass inversion. Subsequently, a backward trajectory of the air mass is established starting from the satellite pixel observation time, and further screening is performed based on the temporal and spatial relationships between the trajectory points and the fire points to be matched. Only fire points that meet the spatiotemporal matching conditions are identified as candidate fire points and used for subsequent transport time determination and mass decay calculation.

[0072] After obtaining the NO2 satellite orbit data at the plume boundary and corresponding time, the system first identifies pixels with valid concentration observations and pixels with invalid or missing concentrations within the plume boundary based on satellite product quality indicators, and calculates the proportion of valid concentration pixels to the total number of satellite pixels. In this embodiment, the threshold for the proportion of valid observation pixels is set to 0.75. When the proportion is below 0.75, the corresponding plume is determined to not meet the data integrity requirements of this embodiment and is not included in subsequent quality correction. When the proportion reaches or exceeds 0.75, pixels with invalid or missing concentrations are supplemented. The proportion threshold can be adjusted according to the satellite product quality and inversion accuracy requirements; for example, a threshold not lower than 0.70 can be selected. Specifically, during supplementation, a set of usable concentration satellite pixels is first constructed from the valid concentration pixels. Invalid or missing concentration pixels are then selected sequentially according to the original storage order of the pixel records in the satellite product. For the pixel to be supplemented, spatially adjacent pixels are selected from the set of usable concentration satellite pixels. Each pixel is used to calculate the average concentration of the target reactive gas, and this average is used as the estimated concentration of the current pixel. After completion, the current pixel is added to the set of available concentration satellite pixels, enabling it to participate in the concentration estimation of subsequent invalid or missing pixels. This process is iterated until all invalid or missing pixels within the plume boundary have obtained estimated concentrations. In this embodiment, the following is taken: =3; In other embodiments, The appropriate value can be selected from 1 to 4 based on the number of pixels in the plume and spatial heterogeneity. After completing the above concentration supplementation, calculate the actual NO2 mass observed by each satellite pixel within the plume at the time of satellite transit. Satellite column concentration characterizes the content of target reactive gases within the atmospheric column above a unit ground area. Therefore, first subtract the atmospheric background concentration that is not contributed by this wildfire, and then combine the pixel area and molecular molar mass to obtain the target reactive gas mass contributed by the wildfire within the current pixel.

[0073] For the current satellite pixels, the NO2 tropospheric column concentration given in the satellite product... Subtracting atmospheric background column concentration The column concentration of wildfire emissions contribution was obtained. This embodiment uses the lowest NO2 column concentration observed by satellite pixels within the plume boundary as the background concentration. Optionally, when a more stable upwind background region or background statistics exist in a certain application scenario, existing background estimation techniques can also be used, such as the upwind region average, the low-concentration pixel average, or the low quantile value. This background value is sufficient as long as it characterizes the baseline concentration not formed by emissions from this wildfire. All three conditions must be met:

[0074]

[0075] in, The target reactive gas column concentration used for mass calculation of current satellite pixels, for the original effective observed pixels, The observation column concentration provided for satellite products; for satellite pixels with density-complemented data, To complete the estimated column concentration; This represents the atmospheric background column concentration. This represents the column concentration resulting from wildfire emissions after background subtraction. When the value is less than or equal to zero, the pixel can be marked as having no effective wildfire enhancement according to the existing anomaly or background pixel processing method, and will not be included in subsequent positive quality inversion; when If the value is greater than zero, continue calculating the observation quality.

[0076] The NO2 orbital stage column concentration used in this embodiment is in mol / m². Multiply by the current pixel area By calculating the molar mass (MM) of NO2 molecules, we can obtain the observed mass of NO2 in that pixel at the moment of satellite transit that was contributed by wildfires and still remains in the atmosphere. :

[0077]

[0078] in, The remaining mass of the target reactive gas observed by the satellite pixel, in grams; The area of ​​a satellite pixel is expressed in square meters (m²). 2 ; The value represents the molecular molar mass of the target reactive gas, expressed in g / mol. The mass of the target reactive gas, after atmospheric transport and chemical removal, that remains in the current pixel during satellite transit is characterized. Subsequently, the mass of this observation is allocated and attenuated based on the candidate fire source and its transport time.

[0079] After calculating the satellite pixel observation quality, a backward trajectory of the air mass is established for each satellite pixel within the plume boundary. This backward trajectory uses the center position of the satellite pixel and the observation time as starting conditions, and calculates the transport path of the corresponding air mass before the observation time based on the concurrent meteorological wind field, thereby obtaining the position of the air mass at different historical moments and the corresponding transport time. Through this backward trajectory, a spatiotemporal correspondence can be established between the satellite pixel and any potentially active fire points it may pass through upstream, providing a foundation for subsequent candidate fire point identification and determination of the transport time from each candidate fire point to the satellite pixel.

[0080] This embodiment uses HYSPLIT as the air mass backward trajectory model. HYSPLIT itself is an existing air mass trajectory calculation tool, and its conventional installation, meteorological file configuration, and trajectory operation can all be achieved using existing technologies. This embodiment utilizes the time and spatial location of the trajectory points output by HYSPLIT to provide data for subsequent fire point matching and transport time calculation. In principle, other models that can output equivalent backward trajectory information can also be used, such as Lagrange transport models like FLEXPART; as long as the time-space path of the air mass in satellite pixels over a past period can be obtained, it can be used for the same subsequent processing.

[0081] For each NO2 satellite pixel within the plume boundary, the latitude and longitude of the pixel center are used as the starting position of the backward trajectory, the satellite observation time of that pixel is used as the starting time of the backward trajectory, and the trajectory starting height is determined based on the plume height information. When plume height information corresponding to the satellite observation time is available, the average plume height can be used as the trajectory starting height for each satellite pixel; when the plume height is unknown, multiple starting heights are set within the possible height range of the plume, and backward trajectory simulations are carried out separately. In this embodiment, since synchronous plume height information is not available, based on the typical height range of wildfire plumes in Asia, four starting heights of 500m, 1000m, 1500m, and 2000m are set for each satellite pixel, thereby obtaining backward trajectory data corresponding to four height layers. For each starting height layer, the proportion of satellite pixels whose backward trajectory can match candidate fire points is counted to the total number of satellite pixels involved in the processing, and the height layers whose proportion reaches a set threshold are determined as valid height layers; height layers that do not reach the threshold are not included in subsequent quality correction. In this embodiment, the ratio threshold is set to 0.75 to filter out effective height layers that meet the fire point matching ratio requirements.

[0082] The total time of backward simulation (TRT) in HYSPLIT is based on the atmospheric lifetime of the target reactive gas. The value is determined to be 1.5 times, i.e., TRT = 1.5. In this embodiment The time interval is set to 4 hours, therefore the TRT is 6 hours; the upper limit of the trajectory altitude is set to 10000m, and the vertical motion of atmospheric particles is based on the vertical velocity field provided by the gfs0p25 meteorological data. If there is no vertical velocity data for the corresponding location, it can be treated as 0. The trajectory output time interval is set to 10 minutes. For a certain satellite pixel, for example, if its satellite observation time is UTC06:00, in In the case of a 4-hour timeframe, the model traces backwards for 6 hours from 06:00 to UTC00:00, outputting the time, latitude, longitude, altitude, and air pressure of a trajectory point every 10 minutes. After multiple pixels run independently, a pixel-level backward trajectory set covering the entire plume boundary is formed.

[0083] This embodiment establishes a correlation between the backward trajectory duration and the atmospheric lifetime of the target reactive gas, instead of setting it to a fixed 6 hours. The TRT is set to 1.5. At that time, the theoretical retention rate corresponding to the end of the backward trajectory is approximately 22.3%, thus enabling it to perform well in different... The relative attenuation tracing scale remains basically consistent across different values. For NO2 used in this embodiment, Taking 4 hours corresponds to a TRT of 6 hours; this 6 hours is only... =Specific results under 4-hour conditions. If If the duration is adjusted to 2 hours or 6 hours, the backward trajectory duration will be adjusted accordingly to 3 hours or 9 hours. The aforementioned extraction of the two preceding and following times is based on the plume boundary formation time. Fire point data is used to establish a broad set of fire points to be matched, while the actual time range for fire point source tracing and matching along the backward trajectory is 1.5 days prior to the satellite observation time. This approach can cover recent emissions that still have a certain remaining proportion at the time of satellite observation, while avoiding the expansion of trajectory uncertainty and candidate fire point search range due to the use of a fixed tracing duration unrelated to the lifetime of the target reactive gas.

[0084] After obtaining the backward trajectory, spatiotemporal matching of fire points is performed for each initial altitude layer. For each trajectory point on the trajectory, its UTC time and spatial location are read, and then fire points that satisfy the maximum time difference and maximum spatial distance conditions for detection time and spatial location are searched in the set of active fire points to be matched within the plume boundary. The detection time in the fire point product represents the sampling time of the satellite for the continuously burning fire source, and is not equivalent to the ignition time of the fire point. Therefore, this embodiment does not set an additional restriction that the fire point detection time must be earlier than the satellite pixel observation time, but directly matches based on the difference between the fire point detection time and the trajectory point time. Fire points that meet both conditions are considered as candidate fire points that may deliver NO2 to the current satellite pixel along the trajectory. The matching relationship can be written as:

[0085]

[0086] in, UTC detection time for active fire points The current trajectory point's UTC time. The maximum allowable time difference; Location of the fire. For the location of the trajectory point, This represents the maximum permissible spatial distance. This distance can be calculated using existing geographic distance algorithms, such as great circle distance or projected planar distance based on latitude and longitude. In actual processing, if a fire point and a trajectory point simultaneously meet the above temporal and spatial matching conditions, the matching relationship between the fire point and the trajectory point is retained. When the same fire point and multiple trajectory points in the same backward trajectory all meet the matching conditions, these matching relationships are not uniquely filtered but are retained separately. Each matching relationship participates in subsequent calculations and is counted separately in the candidate fire point count. The transportation time is determined based on the time of the corresponding trajectory point.

[0087] This embodiment will The time window is set to 30 minutes. Himawari-8 AHI generates fire detection results every 10 minutes, so a 30-minute time difference window can cover continuous fire detection results within approximately one hour, helping to reduce the impact of smoke plume obstruction and the low sensitivity of geostationary satellites to small fires causing missed fire detections. On the other hand, the fire plume (FRP) reflects the instantaneous combustion intensity of the fire at the moment of detection, while NO2 at the trajectory point may be formed by continuous emissions from the fire over a period of time. If the time window is set too narrow, fluctuations in a single instantaneous FRP will have a significant impact on the matching results. Optionally, The time window can be adjusted within the range of 10 to 40 minutes based on the time resolution of the fire point data. The larger the time window, the higher the matching probability is usually, but if it is too large, it may also introduce fire points that are irrelevant to the current trajectory time.

[0088] This embodiment will The range is set to 20km. One consideration is that the trajectories are output discretely over 10 minutes, allowing for continuous air mass movement between trajectory points. For example, at an atmospheric altitude of 2km and a wind speed of 60m / s, an air mass moves approximately 36km in 10 minutes, meaning the distance between adjacent trajectory points is 36km. If a fire point is located between two adjacent trajectory points, its distance to at least one trajectory point can be controlled within 20km. In the trajectory statistics of the 86 plumes in this embodiment, the distance between adjacent trajectory points is less than 40km, therefore, 20km can reduce the omission of fire points along the route caused by discrete trajectory sampling. Optionally, It can be adjusted within the range of 10 to 30 km by combining the trajectory output interval, regional wind speed, and fire point product spatial resolution.

[0089] When a backward trajectory matches at least one candidate fire point within a set simulation period, the satellite pixel corresponding to that trajectory is identified as a fire-matched pixel; when the backward trajectory does not match any candidate fire points within the entire simulation period, the corresponding satellite pixel is identified as a fire-free pixel. For the same starting height layer, the number of fire-matched pixels within the plume boundary is counted, and their proportion to the total number of satellite pixels processed is calculated. This proportion is used to evaluate the overall matching degree of the backward trajectory corresponding to the current starting height layer to the actual active fire points, and is not included in the calculation of the contribution weight of individual candidate fire points. When the proportion of fire-matched pixels corresponding to a certain starting height layer is low, it indicates that most backward trajectories established at that starting height cannot effectively trace back to the actual active fire points within the plume boundary. Therefore, the trajectory results corresponding to that height layer should not be directly used for subsequent plume quality correction; when the proportion reaches a set threshold, that starting height is identified as an effective height layer, and the backward trajectory results corresponding to that height layer are used to determine the subsequent candidate fire point transport time and initial emission quality inversion.

[0090] In this embodiment, the threshold for the proportion of pixels with fire points matching is set to 0.75, meaning that at least 75% of the pixels involved in the processing need to be able to find candidate fire points through the backward trajectory of the corresponding height layer before that height layer is retained for quality inversion. Under a 4-hour timeframe, all 86 plumes used for batch analysis must have at least one height layer that meets the threshold. Plumes that do not meet the condition are removed before batch result statistics. If a plume fails to reach the threshold in terms of the proportion of matching fire points at all preset starting heights, the plume is determined to have failed trajectory matching quality control and will not be included in subsequent quality inversion or batch statistics. Optionally, this threshold can be set between 0.5 and 1.0: the threshold can be increased when there are more samples and greater concern is given to inversion reliability; the threshold can be appropriately decreased when there are fewer samples and more plumes need to be retained.

[0091] For the effective altitude layers obtained through the above screening, the observation quality is further allocated and the initial emission quality is restored based on the number of candidate fire points corresponding to each satellite pixel. When the backward trajectory of a satellite pixel matches only one candidate fire point, the observation quality of that satellite pixel is determined as the remaining quality corresponding to that candidate fire point, and the retention ratio of the target reactive gas is calculated based on the transport time corresponding to that candidate fire point. Then, the initial emission quality corresponding to that candidate fire point is restored using the retention ratio. When the backward trajectory of a satellite pixel matches multiple candidate fire points, it indicates that the target reactive gas observed by the current satellite pixel may be contributed by one or more fire points at different matching time periods of the backward trajectory. Therefore, the contribution weight is determined based on the fire radiation power of each candidate fire point and the retention ratio of the target reactive gas under the corresponding transport time. The observation quality of the satellite pixel is allocated according to the contribution weight, and the initial emission quality corresponding to each candidate fire point is restored.

[0092] Suppose that the backward trajectory of a certain satellite cell is matched to a total of The first candidate ignition point. For the first... For each candidate ignition point, the retention ratios of FRP and NO2 after the corresponding transportation time are first used. Determine its contribution weight to the current satellite pixel observation quality. Contribution weighting is based on:

[0093]

[0094] in, For the first Fire radiation power of each candidate ignition point; For the first The NO2 retention rate corresponding to each candidate ignition point; This represents the number of candidate fire points involved in the calculation of the backward trajectory of this pixel. When the same fire point corresponds to multiple retained trajectory point matching relationships, they are counted separately. For summing the denominators, use the traversal indices. and They represent the first The fire radiation power and retention rate of each candidate fire point.

[0095] No. The retention rate of each candidate fire point is determined by its transportation time. Atmospheric lifetime of target reactive gas Sure:

[0096]

[0097] in, For the first The transportation time corresponding to each candidate fire point is determined by the difference between the time of the trajectory point matched by the candidate fire point and the satellite pixel observation time; The atmospheric lifetime of the target reactive gas. The closer it is to 1, the shorter the transmission time and the less chemical loss during the transmission process corresponding to the candidate ignition point; The smaller the value, the longer the transmission time and the more mass has been chemically removed.

[0098] For example, the backward trajectory of a satellite pixel matches two candidate fire points. The first candidate fire point has an FRP of 100, corresponding to a transportation time of 1 hour, while the second candidate fire point has an FRP of 50, corresponding to a transportation time of 4 hours. (NO2 atmospheric lifetime) A 4-hour timeframe was used. Based on the transportation time corresponding to each candidate fire point, the NO2 retention ratios for the first and second candidate fire points were calculated to be approximately 0.779 and 0.368, respectively. Further multiplying the FRP (Fuel Radiation Power) of each candidate fire point by its corresponding retention ratio, and determining the contribution weight based on the proportion of each product in the sum of all candidate fire point products, the contribution weights for the first and second candidate fire points were approximately 0.809 and 0.191, respectively. These calculations show that when determining the contribution of multiple candidate fire points to the observation quality of the same satellite pixel, fire radiation power is used to characterize the relative combustion intensity of the candidate fire points, and the retention ratio is used to characterize the degree to which NO2 emitted by the corresponding fire point can reach the satellite pixel location after different transportation times. For candidate fire points with lower combustion intensity and longer transportation times, their corresponding observation contribution weights are correspondingly reduced, thus allowing the allocation of satellite pixel observation quality to simultaneously reflect the differences in combustion intensity and transportation attenuation among different candidate fire points.

[0099] Based on the contribution weight, the total remaining mass actually observed by the current pixel is... Assigned to different candidate fire points. The remaining observation mass corresponding to each candidate fire point for:

[0100]

[0101] in, This indicates the current satellite pixel observation quality allocated to the first... The target reactive gas mass of a candidate fire point refers to the mass of the target reactive gas emitted by that candidate fire point that remains in the current satellite pixel at the time of satellite observation after a corresponding transportation time. Taking the aforementioned two candidate fire points as examples, when the observed mass of the current satellite pixel... When the value is 100g, based on the contribution weights of the first and second candidate ignition points being approximately 0.809 and 0.191 respectively, the corresponding value of the first candidate ignition point can be obtained. Approximately 80.9g, corresponding to the second candidate ignition point Approximately 19.1g. Therefore, the target reactive gas mass actually observed by the current satellite pixels is allocated according to the relative contribution of each candidate fire point at the satellite observation time, obtaining the remaining mass after transmission corresponding to each candidate fire point, which provides a basis for subsequently restoring its initial emission mass based on the retention ratio corresponding to each candidate fire point.

[0102] The initial emission masses corresponding to each candidate ignition point were then restored. For the first... For each candidate ignition point, the observed residual mass is divided by the corresponding retention ratio to obtain the initial emission mass corresponding to that candidate ignition point. :

[0103]

[0104] in, Indicates the first The initial emission mass of the target reactive gas corresponding to each candidate fire point. Taking the aforementioned satellite pixel observation mass of 100g as an example, the observation mass allocated to the first candidate fire point is approximately 80.9g, with a corresponding NO2 retention ratio of approximately 0.779. Therefore, the initial emission mass of the first candidate fire point is approximately 103.9g. The observation mass allocated to the second candidate fire point is approximately 19.1g, with a corresponding NO2 retention ratio of approximately 0.368. Therefore, the initial emission mass of the second candidate fire point is approximately 51.9g. By restoring the allocated observation mass using the retention ratio corresponding to each candidate fire point, the mass decay of the target reactive gas caused by different transport times can be compensated, thereby obtaining the initial emission mass corresponding to each candidate fire point before entering the atmospheric transport process. Therefore, in the case of multiple candidate fire points, the mass processing is performed in the order of observation mass allocation and initial emission mass restoration. The differences in transport time of each candidate fire point are reflected in the corresponding retention ratio and mass restoration results.

[0105] All corresponding to the current satellite pixels The initial emission masses of each candidate fire point are summed to obtain the total initial emission mass corresponding to that satellite pixel. :

[0106]

[0107] To facilitate subsequent processing of fire-free matching pixels and observation of the attenuation degree of different pixels, the atmospheric chemical mass attenuation correction factor (MECF) for this satellite pixel can be further defined:

[0108]

[0109] MECF represents the multiple of the initial emitted mass of a pixel after recovery relative to the actual remaining mass observed by the satellite. Under normal circumstances where chemical depletion occurs, MECF is typically not less than 1; all other things being equal, the longer the transport time and the greater the chemical loss that needs to be recovered, the larger the MECF will generally be.

[0110] After performing atmospheric mass attenuation correction on all pixels whose backward trajectories match candidate fire points, correction is then applied to pixels that do not match candidate fire points. For pixels whose backward trajectories do not match candidate fire points, this embodiment selects reference pixels from satellite pixels that are spatially adjacent within the plume boundary and have already completed fire point tracing. The mass attenuation correction coefficient of the reference pixels is then used to compensate these pixels to obtain their corresponding initial emission mass. This process is independent of whether the plume height is known: when the plume height is known, it can be performed on the trajectory results corresponding to the adopted average plume height; when the plume height is unknown, it can be performed separately on each of the selected effective height layers. Specifically, the pixel closest to the current target pixel is selected from the pixels within the plume boundary that have fire point matches. A reference pixel is taken in this embodiment. =3. Spatial distance can be determined using existing latitude and longitude distance algorithms. The value can be selected between 1 and 4, which takes into account the number of available pixels for small plumes while avoiding the introduction of reference pixels that are too far away or have significant differences in source and transmission path.

[0111] Let the first The quality attenuation correction factor for each reference pixel is: Then, for non-fire point matching pixels, the following is adopted: The arithmetic mean of the correction coefficients of each reference pixel is used as :

[0112]

[0113] If there are obvious abnormal reference pixels in certain application scenarios, existing outlier identification methods can be used to perform quality control first, and then the average value of the effective reference pixels can be calculated. The basic processing principle of this embodiment is still to use the attenuation ratio of neighboring pixels that have been directly traced to the source to determine the attenuation degree of pixels that have not been directly traced to the source.

[0114] Matching no-firepoint pixels and Multiply by each other to obtain the corresponding initial emission mass. :

[0115]

[0116] With this compensation, both fire-matched and non-fire-matched pixels within the current trajectory height layer can obtain initial emission mass, thus forming a spatially continuous plume mass field.

[0117] For cases where the plume height is unknown, trajectory matching and pixel-by-pixel quality correction are performed at the four initial height layers of 500m, 1000m, 1500m, and 2000m, respectively. If the proportion of matched fire points at a certain height layer is less than 0.75, the result of that height layer is not used; height layers with a proportion of 0.75 are determined as effective height layers. If there is only one effective height layer, the initial emission quality per pixel of that layer is directly used; if there are multiple effective height layers, the initial emission quality of the same satellite pixel at each effective height layer is arithmetically averaged to obtain the final initial emission quality of that pixel. Subsequently, the final initial emission quality of each pixel is compared with its... In comparison, the final MECF can be obtained, and the final initial emission mass of all pixels within the plume boundary can be summed to obtain the initial NO2 emission mass of the entire isolated wildfire plume in space.

[0118] If the overall height of the plume is known, the multi-height layer screening described above can be simplified. The average height of the plume obtained from synchronous observations can be directly used as the starting height of the backward trajectory of each satellite pixel within the plume boundary, and fire point matching and quality inversion are performed only on the backward trajectory of this starting height layer. For example, when lidar, stereo satellite, or other synchronous observations have indicated that the main body of the plume is located at approximately 1500m, this average plume height can be used for the trajectory calculation of each pixel, without the need to set multiple candidate heights from 500m to 2000m. This method shares the same candidate fire point matching, contribution weights, and quality recovery relationships as the processing when the height is unknown.

[0119] Figure 2 This example illustrates the specific results of NO2 mass atmospheric attenuation correction for a single wildfire plume in a certain area on May 1, 2018. (By...) Figure 2 As shown in (a), the high NO2 value areas observed by satellite correspond to the spatial distribution of wildfire plumes. The backward trajectories established by each satellite pixel extend upstream along the plume transport direction and can form a spatiotemporal correspondence with active fire points within the plume boundary. Based on this, the candidate fire points and corresponding transport times that different satellite pixels may correspond to can be determined. After completing the completion of invalid or missing concentration pixels within the plume boundary, the NO2 observation quality of each pixel during satellite transit is calculated based on the concentration of the satellite observation column. Its spatial distribution is as follows: Figure 2As shown in (b) above, this observation quality reflects the mass of NO2 remaining in each pixel at the time of satellite observation after emission from the fire point, atmospheric transport, and chemical consumption. Further, using the candidate fire points and transport times determined by the backward trajectory, the observation quality is allocated to pixels with multiple candidate fire points, and their initial emission quality is restored according to the NO2 retention ratio corresponding to each candidate fire point. For pixels where the backward trajectory does not match a candidate fire point, compensation is performed using the MECF of neighboring reference pixels, ultimately obtaining... Figure 2 Image (c) shows the initial NO2 emission mass distribution per pixel. Figure 2 (b) and Figure 2 As can be seen from the spatial distribution changes in (c), different pixels exhibit varying degrees of quality recovery due to differences in their corresponding fire source locations and transportation times. This difference is further reflected in... Figure 2 In the spatial distribution of MECF shown in (d), the overall MECF is not less than 1, and it tends to increase as the pixel moves away from the upwind fire source. This indicates that pixels with longer transport distances and longer transport times experienced more significant NO2 mass decay during satellite observation, thus requiring a corresponding increase in the mass to be recovered. Therefore, Figure 2 The data in (a) to (d) form a continuous spatial correspondence from the satellite concentration field and the source of the fire to the recovery of the initial emission quality.

[0120] After completing the pixel-by-pixel quality restoration described above, the correction results are further compared using the mass of aerosols emitted synchronously with the wildfire. In this embodiment, the aerosols are not subjected to mass decay restoration processing tailored to the NO2 atmospheric chemical lifetime; instead, they are treated as co-emission components of the wildfire to compare the correspondence between NO2 mass correction and emissions from the same source. For the 108 satellite pixels contained in this plume, such as... Figure 2 As shown in (e), the Pearson correlation coefficient R between the NO2 observation mass and the corresponding aerosol mass before correction is 0.52, and the linear fitting slope is 0.01 kg / kg. After completing the candidate fire point tracing, observation mass allocation, and mass decay recovery using the aforementioned method, as shown in (e), the results are as follows: Figure 2 As shown in (f), the Pearson correlation coefficient R between the initial NO2 emission mass and the corresponding aerosol mass increased to 0.62, and the linear fitting slope increased to 0.02 kg / kg. Both aerosols and NO2 originate from wildfire combustion. This embodiment focuses on mass recovery of NO2 during atmospheric transport due to chemical decay. After correction, the Pearson correlation coefficient between the initial NO2 emission mass and the aerosol mass increased from 0.52 to 0.62, and the linear fitting slope increased from 0.01 kg / kg to 0.02 kg / kg. This indicates that after correction for transport time and chemical decay, the correspondence between the initial NO2 emission mass and the components emitted concurrently with wildfires was improved.

[0121] After completing the pixel-by-pixel quality correction for individual wildfire plumes, the same processing procedure was used to batch process 86 spatially isolated wildfire plumes in a certain region from April to May 2018. For each plume, the following steps were performed: filling in missing and invalid satellite pixels within the plume boundary, calculating satellite pixel observation quality, establishing backward trajectories, matching candidate fire points, allocating observation quality, and restoring initial emission quality. The quality of each satellite pixel within the plume boundary was then summarized to obtain the total NO2 observation quality and the total initial emission quality at the plume scale. Figure 3 Figure (a) shows the relationship between the total observed NO2 mass and the corresponding total aerosol mass of each plume before correction, with a Pearson correlation coefficient R of 0.76 and a linear fitting slope of 0.028 kg / kg. After mass decay correction using the aforementioned method, the Pearson correlation coefficient R between the initial total NO2 emission mass and the corresponding total aerosol mass increased to 0.83, and the linear fitting slope increased to 0.048 kg / kg. Figure 3 (as shown in (b)). Compared to Figure 2 Pixel-by-pixel analysis of a single plume Figure 3 Extending the processing scale further to multiple independent plumes demonstrates that the initial emission masses of different wildfire plumes can be obtained separately by following the same fire source tracing, mass allocation, and decay recovery process, and can be used for batch emission analysis at the plume scale.

[0122] Based on the aforementioned batch processing, further alterations can be made to the atmospheric lifetime of NO2. The parameters were set and reprocessed according to the aforementioned relationships to analyze the impact of atmospheric lifetime values ​​on mass recovery results and applicable plume quantities.

[0123] when When the duration is 2 hours, the duration of the backward trajectory simulation output is correspondingly shortened to 3 hours. For example... Figure 4 As shown in (a), the Pearson correlation coefficient R between the initial total NO2 emission mass and the aerosol mass is 0.86, and the linear fitting slope is 0.041 kg / kg. Since the backward trajectory period that can participate in fire point matching is... =When the 6-hour period of 4 hours is shortened to 3 hours, the number of candidate ignition points and ignition point matching pixels that can be matched for some plumes is reduced accordingly. Among them, 9 plumes no longer meet the threshold for the proportion of ignition point matching pixels, and the final number of plumes participating in the statistics is 77; when When the duration is 6 hours, the duration of the backward trajectory is correspondingly extended to 9 hours. At this time, if... Figure 4 As shown in (b), the Pearson correlation coefficient R is 0.81, and the linear fit slope is 0.039 kg / kg. As a control, See the calibration results at 4 hours. Figure 3 In (b), all 86 plumes were included in the statistics.

[0124] Different atmospheric lifetime values ​​will change the NO2 retention ratio corresponding to each candidate fire point, and will also be calculated based on the backward trajectory simulation total duration TRT=1.5. Changing the actual source-tracing period of the backward trajectory affects the allocation of observational quality and the initial emission quality recovery results, and may also affect the number of plumes that can ultimately be included in the statistics. Therefore, in practical applications, the atmospheric lifetime of the target reactive gas under the corresponding regional and seasonal conditions can be determined. When measured or estimated results exist, the corresponding results can be used. When direct measurement data is lacking, selection can also be made based on the corresponding region, season, and atmospheric lifetime range of the target gas from existing studies. This embodiment uses... The specific implementation parameter is 4 hours.

[0125] For other target reactive gases, such as HCHO or CO, if reliable satellite column concentrations are available and the corresponding atmospheric lifetimes are known, the same logic can be applied: first, convert the corresponding satellite column concentrations into pixel observation mass; then, use the gas mass backward trajectory to establish the source relationship between pixels and candidate fire points; calculate the retention ratio after transportation based on the corresponding gas lifetime and transportation time; and complete the observation mass allocation and initial mass recovery. After the target gas is replaced, the molecular molar mass... Atmospheric lifespan And according to 1.5 The total trajectory duration and the range of fire point data matching that corresponds to the gas lifetime need to be adjusted accordingly, while other data streams do not need to be changed.

[0126] When the target reactive gas satellite product used is not in mol / m 2 Instead, it is measured in molecules / cm 2 When given column concentration, the aforementioned backward trajectory, candidate fire point matching, and mass decay inversion processes do not need to be changed; only the mass conversion relationship corresponding to the number of molecules is used in the observation mass conversion stage.

[0127]

[0128] in, The unit is molecules / cm². The unit is m². Used to change the cell area from m 2 Convert to cm 2 , The target active gas molecule's molar mass, The constant of Avogadro is 6.022 × 10⁻⁶. 23 molecules / mol.

[0129] This embodiment correlates satellite pixel observation quality, backflush trajectory, fire radiation power of candidate fire points, transport time, and target reactive gas retention ratio. This allows the observation quality of each satellite pixel to be allocated according to different candidate fire points, and quality attenuation recovery is implemented separately based on the transport process corresponding to each candidate fire point. This method can simultaneously consider the contribution differences caused by the combustion intensity and transport duration of different fire points, and is particularly suitable for situations where multiple fire points jointly affect the same satellite pixel, thereby reducing deviations caused by ignoring transport differences in the allocation of fire source contributions and the recovery of initial emission quality. Simultaneously, by supplementing locally invalid or missing concentration pixels with neighboring concentrations, gaps in the plume mass field caused by local missing measurements in satellite products can be reduced. When the plume height is unknown, multi-starting-height trajectory screening can improve the adaptability of trajectory layer selection. For satellite pixels that do not match candidate fire points, regardless of whether the plume height is known or unknown, compensation can be performed using the nearest reference pixel (MECF) to improve the completeness of the quality correction results within the plume boundary. The aforementioned single-plume and multi-plume processing results further demonstrate that, after mass attenuation correction, the correlation between the initial NO2 emission mass and the mass of wildfire-emitted aerosols is improved. This indicates that the results can more reasonably reflect the emission status in the early stages of wildfire plume formation, providing a more reliable data foundation for wildfire emission mass estimation and related emission data analysis.

[0130] Example 2

[0131] This embodiment provides a wildfire plume attenuation correction system based on satellite remote sensing and trajectory simulation, used to execute the wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation described in Embodiment 1. The system includes a data acquisition component, a memory, and a processor. The data acquisition component receives wildfire plume boundary data, target reactive gas satellite observation data, satellite active fire point data, and concurrent meteorological data. The memory stores the above input data, as well as satellite pixels, air mass backward trajectories, candidate fire points, and mass calculation results generated during processing. The processor determines the satellite pixel observation quality, traces the fire source, and restores the initial emission quality based on the correspondence between different data. The data acquisition component, memory, and processor can exchange data via a data bus, network interface, shared storage, or inter-process communication. Specific communication and interface methods can be implemented using existing computer data processing technologies.

[0132] The data acquisition component first establishes a correspondence between the plume, satellite pixels, and observation time. For a wildfire plume to be processed, a corresponding plume data record is created, storing the plume boundary, the corresponding time, and the set of satellite pixels within the plume boundary. Each satellite pixel is associated with its spatial location, satellite observation time, target reactive gas column concentration, and pixel area. Active fire points are associated with their locations, detection times, and fire radiation power (FRP). Meteorological data is saved or retrieved according to the time and spatial range required for trajectory simulation. Through this data organization method, satellite pixels, backward trajectories, and candidate fire points within the same plume maintain a clear attribution relationship during subsequent processing.

[0133] In one implementation, the data acquisition component can read pre-downloaded data via a local file interface. For example, when the target reactive gas is NO2, it can read S5P TROPOMI satellite observation files, along with Himawari-8 AHI active fire point data and gfs0p25 meteorological data for backward trajectory simulation. For applications requiring continuous processing of multiple wildfire events, data within a specified time and spatial range can be automatically obtained via network data interfaces, data service interfaces, or database interfaces. Downloading, decoding, format conversion, and coordinate transformation of different data products can all be achieved using existing technologies, provided the input data provides the spatial location, time, target reactive gas observation values, fire point FRPs, and meteorological parameters required for trajectory calculation as specified in Embodiment 1.

[0134] The processor determines the satellite pixels involved in the current plume processing from satellite observation data based on the plume boundary. For satellite pixels with invalid or missing target reactive gas concentrations, concentration completion is performed according to the method described in Example 1. The target reactive gas observation quality of each satellite pixel is obtained based on the background concentration and pixel area. The obtained observation quality is associated with and saved with the satellite pixel identifier, so that the subsequent trajectory results, candidate fire points, and quality recovery results can all be mapped to specific satellite pixels. When the plume height is known, the processor establishes the gas mass backward trajectory for each satellite pixel based on the determined plume height; when the plume height is unknown, backward trajectories are established based on multiple preset starting heights, and the trajectory data corresponding to different starting heights are saved separately to avoid mixing trajectory results from different height layers.

[0135] After the backward trajectory of the air mass is established, the processor matches the trajectory points with active satellite fire points of the same period based on their temporal and spatial locations. The duration of the backward trajectory is determined as 1.5 times the atmospheric lifetime τ of the target active gas. The processor retains the matching relationships between each fire point and trajectory point that meet the spatiotemporal matching conditions. The transport time of candidate fire points is determined based on the corresponding trajectory point time, and the fire point identifier, fire radiation power (FRP), and transport time corresponding to the candidate fire points are associated with the current satellite pixels. When multiple trajectory starting altitudes are set, the processor also calculates the proportion of satellite pixels that can match candidate fire points at each starting altitude to the total number of satellite pixels involved in the processing, and filters the effective altitude layer according to a set threshold, ensuring that subsequent quality recovery uses trajectory results that meet the fire point matching requirements.

[0136] For a satellite pixel with one candidate fire point, the processor determines the retention ratio based on the transport time and atmospheric lifetime of the target reactive gas corresponding to the candidate fire point, and uses this retention ratio to restore the satellite pixel's observation quality to the corresponding initial emission quality. For a satellite pixel with multiple candidate fire points, the processor reads the FRP and transport time of each candidate fire point, determines the corresponding retention ratio based on the transport time, and uses the FRP and retention ratio to jointly determine the contribution weight of each candidate fire point; then, the current satellite pixel's observation quality is allocated according to the contribution weight, and the corresponding initial emission quality is restored based on the retention ratio corresponding to each candidate fire point. Finally, the initial emission qualities corresponding to each candidate fire point are summed to form the initial emission quality of the satellite pixel.

[0137] When the backward trajectory of some satellite pixels does not match a candidate fire point, the processor can select reference pixels from spatially adjacent satellite pixels that can match the candidate fire point, as described in Embodiment 1. Based on the observation quality and initial emission quality of the reference pixels, the processor determines the corresponding MECF (Mean Earth Flow Compensation Function), and then uses the MECF to perform quality compensation on the satellite pixels that do not match a candidate fire point. This compensation process can be used for single trajectory starting heights when the plume height is known, and also for effective height layers obtained by screening when the plume height is unknown. After processing each satellite pixel within the plume boundary, the processor summarizes the final initial emission quality of each satellite pixel to obtain the initial emission quality of the current wildfire plume.

[0138] The system output data may include the initial total emission mass of the plume, the initial emission mass per satellite pixel, the MECF of each satellite pixel, the effective height layer, the correspondence between satellite pixels and candidate fire points, and the transport time of candidate fire points. These results can be saved as CSV, NetCDF, or other files suitable for storing raster and attribute data, or written to a database for later retrieval. Figure 2 The satellite-by-satellite pixel spatial distribution and Figure 3 and Figure 4The statistical results at the plume scale shown can all be further generated based on the corresponding data output by the system. File saving, database writing, and graphic rendering can be achieved using existing data processing and visualization technologies.

Claims

1. A wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation, characterized in that, include: S1. Acquire wildfire plume boundary data, target reactive gas satellite observation data, satellite active fire point data and meteorological data. When there are satellite pixels with invalid or missing target reactive gas observation concentrations within the plume boundary, complete the target reactive gas concentrations of the corresponding satellite pixels. Determine the observation quality of each satellite pixel based on the target reactive gas concentrations of each satellite pixel within the plume boundary. S2. Establish the backward trajectory of the air mass for each satellite pixel, and determine the candidate fire points that match each backward trajectory based on the spatiotemporal matching relationship between the trajectory points and the active fire points of the satellite. For satellite pixels that match candidate fire points, determine the transportation time from each candidate fire point to the corresponding satellite pixel; S3. When there are multiple candidate fire points, determine the contribution weight of each candidate fire point to the observation quality of the corresponding satellite pixels based on the fire radiation power of each candidate fire point and the retention ratio of the target active gas corresponding to the corresponding transportation time, and allocate the observation quality of satellite pixels according to the contribution weight. S4. Based on the target reactive gas retention ratio corresponding to each candidate fire point, the allocated observation quality is inverted into the initial emission quality of each candidate fire point, and the sum is used to obtain the initial emission quality of the corresponding satellite pixel; When there is only one candidate fire point, the initial emission mass of the corresponding satellite pixel can be directly inverted based on the retention ratio of the target reactive gas corresponding to the candidate fire point. S5. Summarize the initial emission quality of each satellite pixel within the plume boundary that has obtained the initial emission quality, and determine the initial emission quality of the wildfire emission plume.

2. The wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation according to claim 1, characterized in that: The wildfire plume is a spatially isolated plume, distinct from the atmospheric background, and the candidate fire points are located within the plume boundary; The satellite active fire data includes at least the fire location, detection time, and fire radiation power. Based on the plume boundary formation time and the atmospheric lifetime of the target reactive gas Determine the extraction period for active fire point data and select the target reactive gas satellite observation concentration data corresponding to the observation time of the plume boundary.

3. The wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation according to claim 1, characterized in that: Satellite pixels within the plume boundary are extracted from the satellite observation concentration data of the target reactive gas. Valid satellite pixels and invalid or missing satellite pixels are identified based on satellite product quality marks. The proportion of valid satellite pixels to the total number of satellite pixels is calculated. When the ratio reaches the set threshold, the set of satellite pixels with effective observed concentration is formed by the satellite pixels with usable concentration. Satellite pixels with invalid or missing observation concentrations are completed sequentially according to their original storage order. Specifically, for each satellite pixel to be completed, spatially adjacent pixels are selected from the set of available concentration satellite pixels. Each satellite pixel, with The average concentration of the target reactive gas in each satellite pixel is used as the estimated concentration for the current satellite pixel to be supplemented. The satellite pixels that have been supplemented are added to the set of available concentration satellite pixels for subsequent satellite pixel concentration supplementation, until all satellite pixels with invalid or missing concentrations have obtained estimated concentrations. The range is 1 to 4; In subsequent quality calculations, the original observed concentration is used for satellite pixels with valid observed concentrations, while the estimated concentration is used for satellite pixels that have been completed. Concentration of target reactive gas column from satellite pixels Subtracting atmospheric background concentration The concentration of wildfire emissions was obtained. : ； when The unit is mol / m 2 At that time, satellite pixel observation quality for: ； when The unit is molecules / cm 2 At that time, satellite pixel observation quality for: ； in, For satellite pixel area, The molecular molar mass of the target reactive gas. is Avogadro's constant.

4. The wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation according to claim 1, characterized in that: Candidate fire points satisfy the following spatiotemporal matching conditions: ； and: ； in, and These are the fire detection time and the trajectory point time, respectively. and These are the fire point location and the trajectory point location, respectively. and These are the maximum allowed time difference and the maximum allowed spatial distance, respectively. When the same satellite active fire point and multiple trajectory points in the same backward trajectory meet the spatiotemporal matching conditions, the matching relationship between the fire point and each trajectory point is retained, each matching relationship is used in subsequent calculations and counted in the number of candidate fire points, and the transportation time is determined according to the time of the corresponding trajectory point. No. The retention ratio of target reactive gas corresponding to each candidate ignition point and contribution weight They are respectively: ； ； in, For the first Fire radiation power of each candidate ignition point For the first The transportation time from each candidate fire point to its corresponding satellite pixel. The atmospheric lifetime of the target reactive gas. The number of candidate fire points participating in the current satellite pixel calculation. and The first The fire radiation power and the retention ratio of target reactive gas at each candidate ignition point, and , =1,2,……, ; according to: ； and: ； Obtain the Remaining mass corresponding to each candidate fire point and initial emission quality and in accordance with: ； The initial emission mass of satellite pixels is obtained.

5. The wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation according to claim 1, characterized in that: The backward trajectory of the air mass was obtained using the HYSPLIT model; The duration of the gas mass's backward trajectory is the atmospheric lifetime of the target reactive gas. 1.5 times; When the overall height of the wildfire plume is known, the average height of the plume is used as the starting height of the backward trajectory of the satellite pixel. When the overall height of a wildfire plume is unknown, multiple different starting heights are set within the possible height range of the plume, and the backward trajectory of satellite pixels corresponding to each starting height is established.

6. The wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation according to claim 5, characterized in that: When the overall height of the wildfire plume is unknown, the proportion of satellite pixels whose backward trajectory matches the candidate fire point to the total number of satellite pixels involved in the processing is calculated for each starting height. The starting height where the proportion reaches a set threshold is determined as the effective height layer, and the initial emission quality of satellite pixels is obtained based on the backward trajectory corresponding to each effective height layer. When multiple effective altitude layers exist, the average value of the initial emission mass obtained by the same satellite pixel at each of the multiple effective altitude layers is taken as the final initial emission mass of that satellite pixel.

7. The wildfire plume attenuation correction method based on satellite remote sensing and trajectory simulation according to claim 1, characterized in that: For satellite pixels whose backward trajectory does not match a candidate fire point, a reference pixel with a spatial distance is selected from the satellite pixels whose backward trajectory has matched a candidate fire point within the plume boundary. A quality attenuation correction coefficient is determined based on the initial emission quality and observation quality of each reference pixel. The quality attenuation correction coefficient of the satellite pixel that does not match a candidate fire point is determined based on the quality attenuation correction coefficients of multiple reference pixels, and the observation quality of the satellite pixel is corrected accordingly.

8. A wildfire plume attenuation correction system based on satellite remote sensing and trajectory simulation, characterized in that, Includes data acquisition components, memory, and processor; The data acquisition component acquires wildfire plume boundary data, target reactive gas satellite observation data, satellite active fire point data, and meteorological data, and stores them in the memory. The processor identifies satellite pixels within the plume boundary that have invalid or missing target reactive gas concentration measurements, and completes the target reactive gas concentration measurements when such satellite pixels exist. Based on the target reactive gas satellite concentration data, the processor obtains the observation quality of satellite pixels within the plume boundary, establishes a backward trajectory for each pixel, and determines candidate fire points based on the spatiotemporal correspondence between the backward trajectory and active fire points on satellites. For satellite pixels that match candidate fire points, the processor obtains the corresponding transport time for each candidate fire point. When there are multiple candidate fire points for a corresponding satellite pixel, the processor determines the contribution weight of each candidate fire point based on the fire radiation power and the retention ratio of the target reactive gas, and allocates the satellite pixel observation quality accordingly. Then, it inverts the initial emission quality corresponding to each candidate fire point based on the retention ratio of each candidate fire point, and sums them to obtain the initial emission quality of the corresponding satellite pixel. When there is only one candidate fire point for a corresponding satellite pixel, the processor directly inverts the initial emission mass of the corresponding satellite pixel based on the retention ratio of the target reactive gas corresponding to the candidate fire point. The processor aggregates the initial emission quality of each satellite pixel within the plume boundary that has obtained its initial emission quality, and determines the initial emission quality of the wildfire plume.

Images