Air quality coherent change detection method and system based on detection satellite

Abstract

The invention discloses an air quality coherent change detection method and system based on a detection satellite, and the method comprises the steps: obtaining the literature data of a research region about the air quality, carrying out the screening of the research region according to the literature data, obtaining the original data of the air quality, obtaining the raster images, collected by Sentinel-5P and MODIS satellites, of the research region, and carrying out the detection of the coherent change of the air quality. The method comprises the following steps: acquiring an air quality original data of a research area, calculating grating data of a grating image by adopting a Google Earth Engine (GEE), collecting a Sentinel-1 SAR image of the research area, generating a coherence graph according to the grating data and the Sentinel-1 SAR image, and performing statistical analysis and spatial analysis on the coherence graph according to the air quality original data to obtain a coherence change detection result of the research area. By integrating data collected by S5-P and MODIS sensors from ultraviolet, visible and short-wave infrared wavelengths and radar data detected by an S1 satellite from the other side, optical and synthetic aperture radar (SAR) data are combined to detect air quality, and the accuracy of air quality detection is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of air quality detection, and in particular relates to a method and system for detecting air quality coherent changes based on a detection satellite. Background Art

[0002] One of the main consequences of armed conflict is air pollution, which can be defined as any substance or concentration that can harm humans, plants, or animals by increasing mortality and the incidence of dangerous diseases. There are many ways to monitor air pollution, including the use of ground-based air quality monitors and monitoring systems such as satellite imagery, which can provide extensive and detailed information about the concentration of pollutants in the atmosphere. Ground-based observation systems have some limitations in terms of the discrete number of spatiotemporal observations, which affects their use in geospatial space. In recent years, many satellites have been used to monitor atmospheric conditions, but few have been used to effectively detect air pollution caused by armed conflict. Therefore, there is an urgent need to provide a method and system for detecting coherent changes in air quality based on detection satellites, which can effectively detect the air quality in the study area. Summary of the invention

[0003] In view of this, the present invention provides a method and system for detecting coherent changes in air quality based on detection satellites, which can conduct air quality detection at different periods in research areas where armed conflicts occur, thereby improving the accuracy and efficiency of air quality detection, and is specifically implemented using the following technical solutions.

[0004] In a first aspect, the present invention provides a method for detecting air quality coherent changes based on a detection satellite, comprising the following steps: Obtain literature data on air quality in the study area, and screen the study area based on the literature data to obtain original air quality data; Acquire raster images of the study area collected by Sentinel-5P and MODIS satellites, and use Google Earth Engine GEE to calculate raster data of the raster images, wherein the raster data includes gas pollution concentration collected by Sentinel-5P and particulate matter concentration collected by MODIS; collecting Sentinel-1 SAR images of the study area, and generating a coherence map based on the raster data and the Sentinel-1 SAR images; The coherence map is statistically analyzed and spatially analyzed according to the original air quality data to obtain the coherence change detection result of the study area.

[0005] As a preferred embodiment of the above technical solution, collecting Sentinel-1 SAR images of the study area, and generating a coherence map according to the grating data and the Sentinel-1 SAR images, comprises: Using SAR interferometric coherence to measure the similarity of two backscattered signals with different phases obtained at different times in the study area, wherein SAR stands for synthetic aperture radar; The SAR coherence between each two images in the Sentinel-1 SAR image is calculated according to the similarity of the backscattered signals, and a coherence map is generated from the Sentinel-1 SAR image according to formula (1): (1) in, represents the resulting coherence, N is the number of neighboring pixels to be estimated, M is the complex primary image, S is the complex secondary image, represents the complex conjugate, and the result of formula (1) ranges between 0 and 1, where 1 represents total coherent pixels and 0 represents absolutely incoherent; The ideal coherence value is determined according to formula (2): (2) in, is the temporal coherence, represents spatial coherence, Thermal coherence.

[0006] As a preferred embodiment of the above technical solution, temporal coherence is an interesting component, and temporal coherence is used to measure the change of target characteristics over time. The coherence estimation process includes: registering multiple secondary images to a single primary image and creating a coherence map for each pair of images; by The coherence of the coherence map is estimated using a coherence window size of , correcting for flat earth and terrain effects, and using The multi-view coherence results are corrected to obtain the coherence map; The coherence maps were geocoded and exported in GeoTIFF format, and the pixel size of the coherence maps was resampled to fit the gas pollution concentrations corresponding to the raster data.

[0007] As a preferred embodiment of the above technical solution, statistical analysis and spatial analysis are performed on the coherence map according to the original air quality data to obtain the coherence change detection result of the study area, including: The correlation between the coherence loss and the gas pollution concentration is analyzed by using a statistical analysis algorithm, wherein the correlation is used to show the degree of association between two or more variables; The Pearson coefficient is used to perform panel regression analysis on the correlation, wherein the Pearson coefficient is used to calculate the strength of the linear relationship between the two variables in the correlation analysis corresponding to the correlation, and the strength of the linear relationship is represented by r, which is a value between +1 and -1. The Pearson correlation coefficient is used to measure the direction and strength of the relationship between the two variables. Panel regression analysis is used to estimate the functional relationship between the two variables. The expression for calculating the Pearson regression is: (3) Where r is the Pearson correlation coefficient, is the value of the x variable in the sample, is the mean of the x variable, is the value of the y variable in the sample, is the mean of the y variable.

[0008] As a preferred embodiment of the above technical solution, Kendall's tau coefficient is used to estimate the relationship between the coherence change and the gas pollution concentration, wherein Kendall's tau coefficient is used to measure the monotonic correlation between two variables; Monotonic associations include two variables being consistent, inconsistent, or tied together, Kendall's tau coefficient It can be expressed as: (4) Where c represents the number of consistent pairs and d represents the number of inconsistent pairs.

[0009] As a preferred embodiment of the above technical solution, a GWR model is used to handle the coexistence of spatial connectivity and spatial heterogeneity, and spatial non-stationarity is ensured between a dependent variable and a set of independent variables in the GWR model, wherein the GWR model is a geographically weighted regression model; Local linear regression is used to presuppose that the statistical relationship between the independent and dependent variables will not change in space. The GWR model embeds the spatial location into the regression parameters to reflect the spatial non-stationarity of the parameters in different regions. The parameter values ​​are spatially changed according to the spatial location of the data. The calculation formula of the GWR model is: (5) in, represents the dependent variable, express The matrix of explanatory variables, represents the parameter vector that needs to be estimated, is a function of the position, k represents the number of parameters to be estimated, i represents the i-th region, represents the random error term of the ith area in the study area.

[0010] As a preferred embodiment of the above technical solution, the GWR model analysis is applied to the internal local spatial distribution of multiple gas pollution concentrations in the study area as explanatory variables, and the quantile algorithm is used to classify the coherent change detection results.

[0011] As a preferred embodiment of the above technical solution, the coherence change detection results include spatiotemporal analysis of gas pollution concentration, key spatial distribution of gas pollution concentration and fine resolution related to coherence change detection.

[0012] As a preferred embodiment of the above technical solution, the gas pollution concentrations collected by Sentinel-5P include sulfur dioxide, nitrogen dioxide, carbon monoxide, ozone, methane and formaldehyde concentrations and aerosol index UVAI. Sentinel-5P is used to measure air quality and pollution in the atmosphere, and MODIS on Terra and Aqua satellites is used to retrieve from the blue to thermal infrared part of the spectrum. Among them, MODIS is a moderate-resolution imaging spectrometer with a data resolution of 1 km.

[0013] In a second aspect, the present invention further provides an air quality coherent change detection system based on a detection satellite, which is applied to the above-mentioned air quality coherent change detection method based on a detection satellite, comprising: A data acquisition unit, used to acquire literature data on air quality in a study area, and screen the study area according to the literature data to obtain original air quality data; An image collection unit, used to obtain raster images of the study area collected by Sentinel-5P and MODIS satellites, and calculate raster data of the raster images using Google Earth Engine GEE, wherein the raster data includes gas pollution concentration collected by Sentinel-5P and particulate matter concentration collected by MODIS; a coherence map generating unit, configured to collect Sentinel-1 SAR images of the study area and generate a coherence map based on the grating data and the Sentinel-1 SAR images; A coherent change detection unit is used to perform statistical analysis and spatial analysis on the coherent map according to the original air quality data to obtain a coherent change detection result of the study area.

[0014] The present invention provides a detection method and system for air quality coherent change based on a detection satellite. The method obtains literature data on air quality in a study area, screens the study area according to the literature data to obtain original air quality data, obtains raster images of the study area collected by Sentinel-5P and MODIS satellites, calculates raster data of the raster images by using Google Earth Engine GEE, collects Sentinel-1 SAR images of the study area, generates coherence graphs according to the raster data and the Sentinel-1 SAR images, performs statistical analysis and spatial analysis on the coherence graphs according to the original air quality data to obtain coherence change detection results of the study area, integrates data collected from ultraviolet, visible and short-wave infrared wavelengths by S5-P and MODIS sensors, and radar data detected from the other side by the S1 satellite, combines optical and synthetic aperture radar data SAR to detect air quality, thereby improving the accuracy of air quality detection, analyzing the impact of wars and conflicts on air quality, and providing a favorable basis for the analysis of changing climate. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 The present invention is a flow chart of the air quality coherent change detection method based on the detection satellite; Figure 2 The present invention is a structural block diagram of the air quality coherent change detection system based on detection satellite. DETAILED DESCRIPTION

[0017] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0018] See also Figure 1 The present invention provides a method for detecting air quality coherent changes based on a detection satellite, comprising the following steps: S1: Obtain literature data on air quality in the study area, and screen the study area based on the literature data to obtain original air quality data; S2: obtaining raster images of the study area collected by Sentinel-5P and MODIS satellites, and calculating raster data of the raster images using Google Earth Engine GEE, wherein the raster data includes gas pollution concentration collected by Sentinel-5P and particulate matter concentration collected by MODIS; S3: collecting Sentinel-1 SAR images of the study area, and generating a coherence map according to the raster data and the Sentinel-1 SAR images; S4: Performing statistical analysis and spatial analysis on the coherence map according to the original air quality data to obtain a coherence change detection result of the study area.

[0019] In this embodiment, collecting the Sentinel-1 SAR image of the study area, and generating a coherence map according to the grating data and the Sentinel-1 SAR image, includes: Using SAR interferometric coherence to measure the similarity of two backscattered signals with different phases obtained at different times in the study area, wherein SAR stands for synthetic aperture radar; The SAR coherence between each two images in the Sentinel-1 SAR image is calculated according to the similarity of the backscattered signals, and a coherence map is generated from the Sentinel-1 SAR image according to formula (1): (1) in, represents the resulting coherence, N is the number of neighboring pixels to be estimated, M is the complex primary image, S is the complex secondary image, represents the complex conjugate, and the result of formula (1) ranges between 0 and 1, where 1 represents total coherent pixels and 0 represents absolutely incoherent; The ideal coherence value is determined according to formula (2): (2) in, is the temporal coherence, represents spatial coherence, Thermal coherence.

[0020] It should be noted that temporal coherence is the component of interest. Temporal coherence is used to measure the change of target features over time. The coherence estimation process includes: registering multiple secondary images to a single primary image and creating a coherence map for each pair of images; by The coherence of the coherence map is estimated using a coherence window size of , correcting for flat earth and terrain effects, and using The multi-view coherence results are corrected to obtain the coherence map; The coherence maps were geocoded and exported in GeoTIFF format, and the pixel size of the coherence maps was resampled to fit the gas pollution concentrations corresponding to the raster data.

[0021] Among them, the coherence change detection results include the spatiotemporal analysis of gas pollution concentrations, gas pollution concentrations related to coherence change detection, and key spatial distributions with fine resolution. Sentinel-5P's gas pollution concentrations include sulfur dioxide, nitrogen dioxide, carbon monoxide, ozone, methane and formaldehyde concentrations and aerosol index UVAI. Sentinel-5P is used to measure air quality and pollution in the atmosphere. MODIS on Terra and Aqua satellites is used to retrieve from the blue to thermal infrared part of the spectrum. MODIS is a moderate resolution imaging spectrometer with a data resolution of 1 km.

[0022] Specifically, the study area was selected and a preliminary screening was conducted, and all available papers related to world wars, conflicts, and air quality were collected; eight raster images of air pollution concentrations were collected through Google Earth Engine (GEE) and S-5P and MODIS satellites, and Sentinel-1 SAR images were collected for the study area to generate coherence maps; ArcGIS Pro 2.5 software environment, Microsoft Excel and SPSS were used for spatial analysis (GWR) and statistical analysis (GWR); among them, the spatial and statistical analysis processing included the spatiotemporal analysis of gas pollution concentrations, gas concentrations related to coherence values, and the application of GWR to analyze the spatial distribution of fine-resolution significance. Data from three different sensors were used: Sentinel-5P (S-5P), MODIS, and Sentinel-1 (S-1). The main goal of the S-5P satellite mission is to conduct atmospheric measurements with relatively high temporal and spatial resolutions for measuring air quality and pollution in the atmosphere. The S-5P mission includes a spectrometer system operating in the ultraviolet to shortwave infrared range with seven different spectral bands. The MODIS instrument is able to retrieve data in 36 spectral bands from the blue to the thermal infrared part of the spectrum (0.41-14.4 μm). In our study, the MODIS data used had a resolution of 1 km.

[0023] Specifically, in order to obtain the data of S-5P and MODIS, GEE was used to calculate the monthly and daily average concentrations of NO2, SO2, O3, HCHO, CH4, CO and UV aerosol index (UVAI) of S-5P, as well as the concentration of particulate matter (PM2.5) of MODIS. GEE is a cloud-based computing platform, which is used to analyze and process geospatial data. It provides a flexible environment for processing a wide range of geospatial, remote sensing, GIS and satellite data. The main advantage of using GEE to analyze S-5P data is that it provides powerful functions and flexibility for processing geospatial and temporal data and performing complex analysis tasks. Among them, the average concentrations of eight pollutant measurements during the six-month war period from October 2023 to March 2024, and the six months from October 2022 to March 2023 (peacetime), then, 18 images covering the period from August 30, 2023 to April 2, 2024 were downloaded, and the average concentrations every 12 days were matched with the S-1 images.

[0024] Specifically, the S-1 is a C-band SAR sensor that provides all-weather, all-sky radar imagery data with global coverage. SAR is a coherent active microwave imaging method used to map the scattering properties of the Earth's surface. 19 images of the Single Look Complex (SLC) product were downloaded from the S-1A's ascending orbit. The images cover the period from August 30, 2023 to April 2, 2024, covering the month before the war and the first six months of the war in the study area.

[0025] It should be understood that eight different pollution indicators were averaged using GEE. InSAR coherence information was also used to understand the changes caused by the war in the study area. The changes in the concentrations of seven pollutants and one pollution indicator index were tested between the first six months of the war in the study area and the same period last year. Compared with the peacetime period of the same period last year, the pollutants So2, O3, CH4 and CO increased during the war, while the pollutants NO2, PM2.5 and HCHO decreased. The UV index showed an overall change in the study area from UV absorbing pollutants in peacetime to non-absorbing pollutants in wartime. Statistical and spatiotemporal methods were used to explain the relationship between land changes caused by the war and pollutant indicators in 18 periods, of which 3 periods were before the war and 15 periods were after the war. The correlation between coherence and PM2.5 and CO changed from positive correlation in the three periods before the war to negative correlation during the war. By obtaining literature data on air quality in the study area and screening the study area according to the literature data to obtain the original air quality data, raster images of the study area collected by Sentinel-5P and MODIS satellites are obtained, and the raster data of the raster images are calculated using Google Earth Engine GEE, Sentinel-1SAR images of the study area are collected, and coherence maps are generated based on the raster data and the Sentinel-1 SAR images. The coherence map is statistically analyzed and spatially analyzed based on the original air quality data to obtain the coherence change detection results of the study area. By integrating the data collected by S5-P and MODIS sensors from ultraviolet, visible and short-wave infrared wavelengths, and the radar data detected by the S1 satellite from the other side, optical and synthetic aperture radar data SAR are combined to detect air quality, which improves the accuracy of air quality detection, analyzes the impact of war and conflict on air quality, and provides a favorable basis for the analysis of the changing climate.

[0026] Optionally, performing statistical analysis and spatial analysis on the coherence graph according to the original air quality data to obtain a coherence change detection result of the study area includes: The correlation between the coherence loss and the gas pollution concentration is analyzed by using a statistical analysis algorithm, wherein the correlation is used to show the degree of association between two or more variables; The Pearson coefficient is used to perform panel regression analysis on the correlation, wherein the Pearson coefficient is used to calculate the strength of the linear relationship between the two variables in the correlation analysis corresponding to the correlation, and the strength of the linear relationship is represented by r, which is a value between +1 and -1. The Pearson correlation coefficient is used to measure the direction and strength of the relationship between the two variables. Panel regression analysis is used to estimate the functional relationship between the two variables. The expression for calculating the Pearson regression is: (3) Where r is the Pearson correlation coefficient, is the value of the x variable in the sample, is the mean of the x variable, is the value of the y variable in the sample, is the mean of the y variable.

[0027] In this embodiment, Kendall's tau coefficient is used to estimate the relationship between the coherence change and the gas pollution concentration, wherein Kendall's tau coefficient is used to measure the monotonic correlation between two variables; monotonic correlation includes that the two variables are consistent, inconsistent or linked together, and Kendall's tau coefficient It can be expressed as: (4) Where c represents the number of consistent pairs and d represents the number of inconsistent pairs.

[0028] Specifically, Optionally, a GWR model is used to handle the coexistence of spatial connectivity and spatial heterogeneity, and spatial non-stationarity is ensured between a dependent variable and a set of independent variables in the GWR model, wherein the GWR model is a geographically weighted regression model; Local linear regression is used to presuppose that the statistical relationship between the independent and dependent variables will not change in space. The GWR model embeds the spatial location into the regression parameters to reflect the spatial non-stationarity of the parameters in different regions. The parameter values ​​are spatially changed according to the spatial location of the data. The calculation formula of the GWR model is: (5) in, represents the dependent variable, express The matrix of explanatory variables, represents the parameter vector that needs to be estimated, is a function of the position, k represents the number of parameters to be estimated, i represents the i-th region, represents the random error term of the ith area in the study area.

[0029] In this embodiment, the GWR model analysis is applied to the concentrations of multiple gas pollution with local spatial distribution inside the study area as explanatory variables, and the quantile algorithm is used to classify the coherent change detection results.

[0030] Specifically, Geographically Weighted Regression (GWR) is a method that is well suited for spatial level data. The model involves the coexistence of spatial connectivity and spatial heterogeneity. In the GWR model, spatial non-stationarity is ensured between the dependent variable and a set of independent variables. Local linear regression is preferably applied because it is better than global linear regression, which assumes that the statistical relationship between the independent and dependent variables does not change in space. The GWR model embeds spatial location into the regression parameters, reflecting the spatial non-stationarity of the parameters in different regions. The coherence map is used to illustrate the destruction of the built-up areas in the study area in the first six months of the war. The coherence value ranges from 0 to 1, where 0 indicates absolute incoherence and 1 indicates a total coherent radar signal. In the first three periods from August 30 to October 5, 2023, the coherence value was only affected by the temporal coherence changes in the vegetation-land cover area, while in the built-up area, the coherence remained relatively high. These three periods were before the start of the war on October 7, 2023.

[0031] To get a better idea of ​​the extent of the disruption during the study period, the coherences of the first three periods before the war were averaged and only the coherent pixels were extracted, considering the average coherence equal to or above 0.6 as coherent pixels, and then the time series of these extracted pixels were calculated to estimate the coherence mean values ​​for the five provinces of the study area during the 18 periods studied. It was mentioned that in the 7th, 13th and 14th periods, the coherence mean values ​​of all provinces dropped suddenly, and the investigation of war-related events did not reveal any large-scale actions that could have caused such a drop during these periods. It is assumed that this could have been due to natural causes, such as rain and / or snow. In the 7th period, continuous rainfall was found from November 12 to 21, 2023, during which the rainfall totaled 23.26 mm, and severe and continuous record deviations were observed from January 22, 2023 to February 2, 2024, totaling 212.32 mm, which led to a drop in the coherence mean value in the 13th period. Subsequently, a limited correction was made in the 14th period with 11.84 mm. Therefore, these periods were excluded from this analysis. In particular, period 13, which was the largest during the study period, is expected to reduce detectable levels of pollutants in the region's air.

[0032] In the three periods before the start of the war, the average coherence of the studied pixels in all provinces was higher than 0.7. In the fourth period after the start of the war, the coherence of the North Province suddenly dropped significantly. This drop exceeded 0.14 times the average coherence. The drop in the average coherence value represents the huge destruction of buildings in the North Province during this period. A similar trend with a smaller impact was also seen in the East Province, where the average coherence loss of coherent pixels was about 0.07. Although the average consistent coherence loss in the North Province was relatively stable between 0.6 and 0.63 from the 4th to the 10th period, this indicates that the destruction rate of buildings during this period was stable and continuous. During the same period, the destruction rate of buildings in the East Province gradually increased, and the average coherence decreased from 0.7 in the 4th period to 0.61 in the 10th period. Compared with the average coherence value of the province before the war, the coherence loss in the 10th period was equal to 0.16. During the 11-year period, the situation in the North and East Provinces stabilized, and the average coherence of the two provinces returned to the average level before the war. The average coherence of the two provinces was 0.73. However, from the 12th period, building damage and continuity loss reappeared and gradually increased in both provinces until the end of the study period, reaching 0.62 and 0.63 in the North and East provinces, respectively, during the 18-year period.

[0033] The coherence of the Western Province was relatively stable during the first six periods. During this period, the coherence fluctuated between 0.7 and 0.76; then, after the sixth period, the coherence gradually decreased, with the coherence decreasing from 0.73 in the 6th period to 0.59 in the 18th period. This gradually decreasing trend indicates that the Southern Province has been increasingly overtime, resulting in serious damage to the buildings in the province. While the destruction in the Southern Province was found to be gradually increasing, the situation was different in Provinces A and B, where the coherence of these two provinces dropped significantly during the 8th period. This was similar to the decrease in coherence in the Northern Province during the fourth period. In Province A, the average decrease in coherence between the 6th and 8th periods was estimated to be 0.15, and in Province B to be 0.14. In addition, it was estimated that the two provinces had the highest degree of building destruction during the 10-year period, with the average values ​​of coherence of the pixels being 0.54 and 0.57, respectively. The most destructive period for these two provinces during the study period of the war. In the last few periods of the study time, the damage levels were relatively similar, with coherence averages slightly above 0.6 in periods 15 and 18.

[0034] The average values ​​of pollutant gas concentrations were investigated during the first six months of the war, from October 2023 to March 2024. This period is defined here as "wartime", while the same period in the previous year, from October 2022 to March 2023, is defined as "peacetime". Certain concentrations mostly increased during the war, such as carbon monoxide (CO), methane (CH4), ozone (O3), and sulfur dioxide (SO2). However, nitrogen dioxide (NO2), particulate matter (PM2.5), ultraviolet aerosol index (UVAI), and formaldehyde (HCHO) decreased during the war. The concentrations of SO2, O3, CH4, and CO increased during the war in the five provinces of the study area. At the same time, the concentrations of NO2, PM2.5, and UVIR decreased. Formaldehyde concentrations decreased in most provinces during the war. Except for Province A, the concentrations were similar in both periods, and the increases in CO, CH4, O3, and SO2 were relatively low compared to the typical range of pollutant gas concentrations, and did not exceed the typical range of concentrations during wartime. On the other hand, the concentration levels of PM2.5 and UVAI exceeded the typical ranges of these two pollutants. Similarly, the changes in NO2 concentrations during war and peacetime remained within the typical range of this pollutant, and the UVAI index was positive in Provinces A and B and negative in the North, East, and South provinces during both peacetime and wartime. However, the mean value of the index was different across the study area, with the mean value of UVAI being positive during peacetime and switching to negative values ​​during wartime.

[0035] Specifically, two different statistical methods were used to evaluate the correlation between the consistent coherence values ​​and the concentrations of eight polluting gases in 18 periods in the first six months of the war in the study area, namely Pearson and Kendall's tau. Among all pollutant concentrations, PM2.5 showed a clear effect since the outbreak of the war, with a significantly positive correlation with the coherence value in the first three periods before the war. Since the start of the war, the correlation between PM2.5 and coherence has become negative in all subsequent periods except the thirteenth period, and the correlation is most significant during the war. This result indicates that PM2.5 concentrations were significantly correlated with high coherence values ​​before the war, and also with low coherence values ​​due to the large-scale destruction of many built-up areas during the war. Similar characteristics were found in the correlation between coherence values ​​and CO concentrations. Although there were almost no outliers in the correlation between CO and coherence in the eighteen periods, this correlation also showed a direct transition from a positive correlation in the three periods before the war to a negative correlation in the subsequent periods after the start of the war.

[0036] However, the significant periods are much smaller compared to the correlation between PM2.5 and coherence. Similarly, the war had a direct impact on the correlation between coherence values ​​and CO concentrations, as the correlation between coherence values ​​and CO concentrations was reversed after the start of the war. The war did not have a certain impact on the exposure to sulfur dioxide and ozone in the study area, and the correlation between coherence values ​​and both concentrations fluctuated before and after the start of the war. However, UVAI was significantly positively correlated with coherence, except for four periods from December 4, 2023 to January 21, 2024, which mostly showed a significant negative correlation with the coherence values ​​in the study area, which can be explained by the common relationship between high coherence values ​​and positive UVAI values. Among these four periods, there was a period of war, when positive UVAI values ​​were associated with low coherence values.

[0037] Although the statistical correlation between NO2 and coherence fluctuated before the war, it showed a clear, significant correlation during the war in the first three periods. The statistical correlation between periods 4 and 6 was significantly negative, which turned into a significant positive correlation between periods 7 and 12. This was followed by an almost significant negative correlation in periods 13 and 15. A positive correlation here means that high concentrations of NO2 are associated with coherent areas, while a negative correlation means that areas with low coherence values ​​have high NO2 concentrations. Similarly, CH4 showed an irregular coherence correlation behavior in the first three periods before the war, with a statistically significant positive correlation between CH4 and coherence in the first four periods of the war, namely periods 4 and 7. By excluding two outliers, this correlation was converted to a negative correlation until period 17, when only 50-60% of these periods showed a significant correlation between coherence and CH4. High concentrations of CH4 were associated with coherent areas in the study area from periods 4 to 7, and this relationship changed for most of the following periods, being associated with areas of low coherence.

[0038] Specifically, the spatial distribution of the significant impact of the war on the concentrations of seven pollutants in the study area during the first six months of the war was performed using a geographically weighted regression model (GWR). UVAI was excluded from this analysis to avoid any misinterpretation due to the nature of this index, where positive values ​​indicate the presence of UV-absorbing aerosols and negative values ​​indicate the presence of non-absorbing aerosols. The correlation between coherence and PM2.5 in grey is not significant, pink represents areas of negative correlation, while areas shown in peridotite color are positively correlated with PM2.5 concentrations. Despite the absence of significant correlation between coherence and PM2.5 concentrations, most urban areas in the study area showed positive correlations in their regional distributions during the first three periods before the war. In the 4th and 5th periods after the start of the war, the distribution of insignificant areas was mainly concentrated in the study area, with the exception of the fifth period occurring in the Northern Province, where statistically relatively large areas of negative correlation were present. In contrast to the three periods before the war, the areas with statistically negative correlations increased over time during the war, with the correlation coefficients reaching high levels in the 11th and 12th periods, at -0.30 and -0.41, respectively. In addition, starting from the ninth period, the distribution of regions with significant negative correlation began to concentrate in the southern part of Province B and Province A. In addition, the last four periods showed that many areas in the northern part of the study area had a significant negative correlation.

[0039] In summary, in assessing the environmental impact of the war on the study area from the perspective of air pollution, this objective was achieved by integrating data collected by the S5-P and MODIS sensors in UV, visible and shortwave infrared wavelengths, and radar data sensed from the other side by the S1 satellite. No previous study has combined optical and synthetic aperture radar data SAR to analyze the impact of the war on air quality, coherence coherence and changes in pollutant concentrations compared to the main war actions and pollutants. In the fourth period, the decrease in coherence values ​​in the North and East provinces coincides with the news recorded in Wikipedia, and in the following periods, there were limited actions in other provinces of the study area. The first period of coherence decrease appeared in the eighth period, which represents the impact of the strike on other provinces in the region during this period, which coincides with the large-scale actions carried out throughout the study area from December 1, 2023, after the end of the only fire since the start of the war from November 24 to 30, 2023. The matching of war actions and coherence losses in the study area highlights the important rule of SAR coherence information in the application of remote sensing war-related change detection.

[0040] When the difference between the mean values ​​of NO2 concentrations during the six months before the war (wartime) and the same period of the previous year (peacetime) was examined, a slight decrease of 0.000012 in the mean value of NO2 concentrations during the wartime was found. The slight change in the mean value of NO2 concentrations in the study area during the wartime could be due to the cessation of fuel combustion emissions from power plants and transport vehicles, as these fuel combustion emissions are the main source of NO2 in the atmosphere. The sudden cessation of human activities, such as factories and related industries, car exhaust and human-related activities, leads to a decrease in nitrogen dioxide gas, especially in cities. However, the correlation results between coherence and NO2 concentrations during the 18 study periods during the six months before the war showed a changing temporal pattern. In general, high NO2 emissions were associated with low coherence values ​​during most of the study period, especially between periods 4-6 and 13-15 at the beginning of the war. This correlation was reversed between periods 7-12, with NO2 emissions associated with high coherence values. This temporary behavior could be due to the nature of the war operations in the study area.

[0041] Similar to NO2, PM2.5 concentrations decreased during the war compared to the peacetime period of the previous year, with average concentrations above the typical range for both periods. These solid particles and droplets with a diameter of 2.5 μm and below are mainly the result of complex reactions of chemicals such as sulfur dioxide and nitrogen oxides. These particles are associated with high coherence values ​​in the first three periods before the war and with low coherence values ​​in the war period starting from the fourth period. This may be due to the difference in the sources of PM2.5 before and after the start of the war. The sources before the war were chemical reactions in the industrial and transportation sectors in urban areas, and the sources in the destroyed areas changed, with chemical reactions of NO2 and SO2 from particles of destroyed buildings, chimneys, and fires. The SO2 concentration increased by about 62% on average during the war compared to the peacetime period. Mainly due to the increase in fossil fuel consumption in the study area during the war, there was no certain correlation trend between SO2 and coherence values ​​in the 18 study periods. However, the CH4 results showed a slight increase of CH4 in the war compared to the peacetime period, about 2.4 moles. By monitoring the correlation between CH4 and coherence, it was found that high CH4 concentrations were associated with the coherent area between study periods 4 and 7. After the resumption of operations on December 1, 2023, increased building damage in the study area led to increased methane concentrations.

[0042] Compared with the peacetime average, the wartime average O3 increased by 0.007 mol / m 2. Overall, O3 did not show any specific correlation with coherence values ​​during the first six months of the war, perhaps due to the winter weather conditions during the study period. O3 is the chemical reaction of nitrogen oxides (NOx) and volatile organic components (VOCs) in sunlight. Urban areas are most likely to reach unhealthy levels during the hot summer months. Formaldehyde concentrations in the study area decreased slightly during the war period compared to the peace period, by 0.000054 mol / m2. The results of the two correlation coefficient tests were mostly different in the 18 periods studied, except for the period between 9 and 13, when high concentrations of formaldehyde were associated with areas of low coherence, which may be due to the oxidation of CH4 emitted by the strip fires. UVAI is an indicator for detecting UV absorbing / non-absorbing aerosols, with positive values ​​indicating the presence of UV absorbing aerosols such as dust and smoke, and negative values ​​indicating the presence of non-absorbing aerosols, including sulfate aerosols. When comparing UVAI across all provinces, there was no change in UV absorbing / non-absorbing aerosols in each province between the peace and war periods, and this change generally refers to a decrease in UVAI values ​​across all provinces. However, the average value of UVAI was positive during peacetime with an index value of 0.0218 and became negative with a value of -0.0066 during the war period, indicating that the overall average of aerosols in the study area changed from UV-absorbing aerosols to non-absorbing aerosols during the war. When correlated with coherence, UVAI revealed the presence of dust and smoke in the coherent areas during most of the 18 periods studied, with this correlation changing during the intense actions of periods 9 and 12, with coherent pixels in these periods being associated with non-absorbing aerosols or mainly sulfate aerosols.

[0043] The changes in the concentrations of seven pollutants and one pollution index were tested between the first six months of the war in the study area and the same period last year. Compared with the same period last year in peace, the pollutants So2, O3, CH4 and CO increased during the war, while the pollutants NO2, PM2.5 and HCHO decreased. The UV index showed an overall change in the study area from UV absorbing pollutants in peacetime to non-absorbing pollutants in wartime. Statistical and spatiotemporal methods were used to explain the relationship between war-induced land changes and pollutant indicators in 18 periods, of which 3 periods were before the war and 15 periods were after the war. The correlation between coherence and PM2.5 and CO changed from positive correlation in the three periods before the war to negative correlation during the war. The correlation between coherence and UVAI revealed the change of pollutants associated with high coherence values ​​from UV absorbing pollutants to non-absorbing pollutants in the same period.

[0044] See also Figure 2 The present invention also provides an air quality coherent change detection system based on a detection satellite, which is applied to the above-mentioned air quality coherent change detection method based on a detection satellite, comprising: A data acquisition unit, used to acquire literature data on air quality in a study area, and screen the study area according to the literature data to obtain original air quality data; An image collection unit, used to obtain raster images of the study area collected by Sentinel-5P and MODIS satellites, and calculate raster data of the raster images using Google Earth Engine GEE, wherein the raster data includes gas pollution concentration collected by Sentinel-5P and particulate matter concentration collected by MODIS; a coherence map generating unit, configured to collect Sentinel-1 SAR images of the study area and generate a coherence map based on the grating data and the Sentinel-1 SAR images; A coherent change detection unit is used to perform statistical analysis and spatial analysis on the coherent map according to the original air quality data to obtain a coherent change detection result of the study area.

[0045] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limiting, and thus other examples of the exemplary embodiments may have different values.

[0046] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0047] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for detecting air quality coherent changes based on detection satellites, characterized in that: The following steps are involved: Obtain literature data on air quality in the study area, and screen the study area based on the literature data to obtain original air quality data; Acquire raster images of the study area collected by Sentinel-5P and MODIS satellites, and use Google Earth Engine GEE to calculate raster data of the raster images, wherein the raster data includes gas pollution concentration collected by Sentinel-5P and particulate matter concentration collected by MODIS; collecting Sentinel-1 SAR images of the study area, and generating a coherence map based on the raster data and the Sentinel-1 SAR images; The coherence map is statistically analyzed and spatially analyzed according to the original air quality data to obtain the coherence change detection result of the study area.

2. The air quality coherent change detection method based on detection satellite according to claim 1 is characterized in that: Collecting Sentinel-1 SAR images of the study area and generating a coherence map based on the raster data and the Sentinel-1 SAR images, including: Using SAR interferometric coherence to measure the similarity of two backscattered signals with different phases obtained at different times in the study area, wherein SAR stands for synthetic aperture radar; The SAR coherence between each two images in the Sentinel-1 SAR image is calculated according to the similarity of the backscattered signals, and a coherence map is generated from the Sentinel-1 SAR image according to formula (1): （1） in, represents the resulting coherence, N is the number of neighboring pixels to be estimated, M is the complex primary image, S is the complex secondary image, represents the complex conjugate, and the result of formula (1) ranges between 0 and 1, where 1 represents total coherent pixels and 0 represents absolutely incoherent; The ideal coherence value is determined according to formula (2): （2） in, is the temporal coherence, represents spatial coherence, Thermal coherence.

3. The air quality coherent change detection method based on detection satellite according to claim 2 is characterized in that: Temporal coherence is the component of interest. Temporal coherence is used to measure the change of target characteristics over time. The coherence estimation process includes: registering multiple secondary images to a single primary image and creating a coherence map for each pair of images; by The coherence of the coherence map is estimated using a coherence window size of , correcting for flat earth and terrain effects, and using The multi-view coherence results are corrected to obtain the coherence map; The coherence maps were geocoded and exported in GeoTIFF format, and the pixel size of the coherence maps was resampled to fit the gas pollution concentrations corresponding to the raster data.

4. The air quality coherent change detection method based on detection satellite according to claim 3 is characterized in that: The coherence map is statistically analyzed and spatially analyzed according to the original air quality data to obtain the coherence change detection result of the study area, including: The correlation between the coherence loss and the gas pollution concentration is analyzed by using a statistical analysis algorithm, wherein the correlation is used to show the degree of association between two or more variables; The Pearson coefficient is used to perform panel regression analysis on the correlation, wherein the Pearson coefficient is used to calculate the strength of the linear relationship between the two variables in the correlation analysis corresponding to the correlation, and the strength of the linear relationship is represented by r, which is a value between +1 and -1. The Pearson correlation coefficient is used to measure the direction and strength of the relationship between the two variables. Panel regression analysis is used to estimate the functional relationship between the two variables. The expression for calculating the Pearson regression is: （3） Where r is the Pearson correlation coefficient, is the value of the x variable in the sample, is the mean of the x variable, is the value of the y variable in the sample, is the mean of the y variable.

5. The air quality coherent change detection method based on detection satellite according to claim 4 is characterized in that: The relationship between the coherence change and the gas pollution concentration was estimated using Kendall's tau coefficient, which is used to measure the monotonic association between two variables. Monotonic associations include two variables being consistent, inconsistent, or tied together, Kendall's tau coefficient It can be expressed as: （4） Where c represents the number of consistent pairs and d represents the number of inconsistent pairs.

6. The air quality coherent change detection method based on detection satellite according to claim 5 is characterized in that: The GWR model is used to deal with the coexistence of spatial connectivity and spatial heterogeneity, and the spatial non-stationarity is ensured between a dependent variable and a set of independent variables in the GWR model, where the GWR model is a geographically weighted regression model; Local linear regression is used to presuppose that the statistical relationship between the independent and dependent variables will not change in space. The GWR model embeds the spatial location into the regression parameters to reflect the spatial non-stationarity of the parameters in different regions. The parameter values ​​are spatially changed according to the spatial location of the data. The calculation formula of the GWR model is: （5） in, represents the dependent variable, express The matrix of explanatory variables, represents the parameter vector that needs to be estimated, is a function of the position, k represents the number of parameters to be estimated, i represents the i-th region, represents the random error term of the ith area in the study area.

7. The air quality coherent change detection method based on detection satellite according to claim 6 is characterized in that: Also includes: The GWR model analysis was applied to the interior of the study area with local spatial distribution of multiple gas pollution concentrations as explanatory variables, and the quantile algorithm was used to classify the coherent change detection results.

8. The air quality coherent change detection method based on detection satellite according to claim 1 is characterized in that: The coherence change detection results include spatiotemporal analysis of gas pollution concentrations, key spatial distributions of gas pollution concentrations and fine resolution associated with coherence change detection.

9. The air quality coherent change detection method based on detection satellite according to claim 1 is characterized in that: The gas pollution concentrations collected by Sentinel-5P include sulfur dioxide, nitrogen dioxide, carbon monoxide, ozone, methane and formaldehyde concentrations and the aerosol index UVAI. Sentinel-5P is used to measure air quality and pollution in the atmosphere. The MODIS on the Terra and Aqua satellites is used to retrieve from the blue to the thermal infrared part of the spectrum. Among them, MODIS is a moderate-resolution imaging spectrometer with a data resolution of 1 km.

10. An air quality coherent change detection system based on detection satellite, characterized in that: The method for detecting air quality coherent changes based on a detection satellite as described in any one of claims 1 to 9 comprises: A data acquisition unit, used to acquire literature data on air quality in a study area, and screen the study area according to the literature data to obtain original air quality data; An image collection unit, used to obtain raster images of the study area collected by Sentinel-5P and MODIS satellites, and calculate raster data of the raster images using Google Earth Engine GEE, wherein the raster data includes gas pollution concentration of Sentinel-5P and particle concentration of MODIS; a coherence map generating unit, configured to collect Sentinel-1 SAR images of the study area and generate a coherence map based on the grating data and the Sentinel-1 SAR images; A coherent change detection unit is used to perform statistical analysis and spatial analysis on the coherent map according to the original air quality data to obtain a coherent change detection result of the study area.

Images