Atmospheric pollution tracing method based on regional distribution identification and screening

Abstract

The invention belongs to the technical field of atmospheric pollution monitoring, and discloses an atmospheric pollution traceability method based on regional distribution identification and screening, which comprises the following steps of: performing interpolation processing on atmospheric pollution concentration data by using a diffusion interpolation method containing obstacles, performing overall aggregation analysis on the data after interpolation processing, and determining the concentration of atmospheric pollution. And obtaining a result that the concentration of the target pollutants in the to-be-detected area is high-value aggregation or low-value aggregation in space, taking the area corresponding to the high-value aggregation result as a hot spot area, and tracing the hot spot area. Compared with the process of directly tracing the source of the target pollutants in the to-be-detected area, the method has the advantages that the time for tracing the source of the area with low concentration value or low concentration degree is saved, the tracing efficiency is further improved, and the hot spot area is divided according to the pollutant concentration and the concentration degree, so that the accuracy of tracing the source is improved. Therefore, the accuracy of the obtained traceability result can be ensured in the traceability process based on the division result.

Description

Technical Field

[0001] The present invention relates to the technical field of air pollution monitoring, and in particular to an air pollution source tracing method based on regional distribution identification and screening. Background Art

[0002] Air pollution has significant negative impacts on human health and social development, and air pollution control is of great significance. Local-scale air quality control has become a new stage and new demand for my country's atmospheric environmental management. To achieve local-scale air quality control, it is necessary to trace the source of air pollution within a small scale. However, due to the randomness of the location of air pollution sources, the uncertainty of time, and the complexity of gas types, the identification, screening, and tracing of hot spots (i.e., areas with high pollutant concentrations) are very complicated. The country has now established an environmental monitoring network with several fixed monitoring stations distributed in various cities. There is a strong dependence between the three characteristics of air pollution data: time, space, and multivariate. If we can make full use of these monitoring data and explore the laws behind the data, this will be of great significance for tracing the source.

[0003] In summary, the identification and screening of air pollution areas and the tracing of air pollution sources are very complex and difficult problems. It is difficult to provide effective solutions relying solely on the background knowledge of a single discipline. Therefore, identification, screening and tracing require the integration of data science methods and atmospheric field knowledge. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide an air pollution source tracing method based on regional distribution identification and screening, so as to solve the problem of difficulty in identifying and screening existing air pollution areas.

[0005] To achieve the above objectives, the method of the present invention studies a new source tracing method based on mathematical analysis to achieve the identification and screening of hotspot areas and the tracing of the pollution process. Specifically, the method includes the following steps:

[0006] 1) Obtain historical data of target pollutant concentrations in the area to be detected as a sample data set;

[0007] 2) Using the diffusion interpolation method with obstacles to interpolate the sample data set, a complete data set including the predicted value of the target pollutant concentration at the target location in the detection area is obtained;

[0008] 3) Based on the data of the complete data set and using the global spatial autocorrelation analysis method, the target pollutant concentrations of each sub-area of ​​interest in the area to be tested within the same time period are analyzed for overall clustering, and the first cluster analysis result is obtained. When the first cluster analysis result is clustered in space, it is determined whether the target pollutant concentrations in each sub-area of ​​interest are clustered in space at high values ​​or low values;

[0009] 4) The sub-area of ​​concern corresponding to the high-value aggregation result is the hotspot area, and the target pollutants in the hotspot area are traced to obtain the tracing results.

[0010] The method of the present invention has the following advantages: First, the present invention takes into account that the impact of atmospheric pollutant sources on the environment weakens with increasing distance, and the degree of aggregation also weakens with increasing distance. Therefore, in order to accurately and efficiently carry out the atmospheric pollution source tracing process, the present invention first identifies and screens the atmospheric pollution area to determine the area with high atmospheric pollutant concentration and aggregation as the hot spot area, and then traces the source of the hot spot area. Compared with the process of directly tracing the target pollutants in the detection area, this process saves the time of tracing the source of those areas with low concentration values ​​or sparse aggregation, thereby improving the efficiency of tracing the source. In addition, the division of the hot spot area in the present invention is based on the division of pollutant concentration and aggregation degree, which conforms to the law of diffusion, and then the process of tracing the source based on the division result can ensure the accuracy of the tracing result.

[0011] Secondly, in order to ensure the accuracy of regional division in the present invention, the diffusion interpolation method is first used to supplement the acquired sample data set, so as to realize the prediction of pollutant concentration at the position that does not appear in the sample data set, and then the predicted value of pollutant concentration at the target position is supplemented into the sample data set to form a complete data set, so that the data of the area to be detected is more substantial. In addition, the present invention takes into account that the diffusion of pollutants will be affected by obstacles, that is, when encountering obstacles, the diffusion path will be changed, and the propagation mode of attribute values ​​is no longer a simple isotropic diffusion, but diffusion along the edge of the obstacle or bypassing the obstacle to be more in line with the actual geographical situation. Therefore, the diffusion interpolation method with obstacles is adopted in the present invention. The diffusion interpolation method with obstacles can provide interpolation results that are more in line with the actual situation, thereby improving the accuracy of the predicted value, thereby ensuring the accuracy of regional division while ensuring the accuracy of the tracing results.

[0012] The regional division in the present invention adopts the global spatial autocorrelation analysis method to obtain whether the measured object has autocorrelation in space, and then judge whether there is aggregation characteristic according to the autocorrelation situation, and perform the first cluster analysis to obtain whether the target pollutant concentration has aggregation characteristics in space, that is, to obtain the result of whether the concentration data is clustered, dispersed or random in space, and then when there is aggregation characteristic, it is specifically distinguished whether it is high-value aggregation or low-value aggregation. This process obtains the result that includes both the degree of aggregation and the high and low concentration values, which is conducive to the accurate division of the region.

[0013] Based on the above-mentioned regional division process of the present invention, accurate regional division results are obtained. Based on this accurate regional division result, the tracing results obtained by tracing the target pollutants whose division results are hot spots are accurate. In addition, the method of first dividing the region and then tracing the key areas based on the present invention also simplifies the overall tracing process, thereby improving the efficiency of tracing.

[0014] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following specifically cites a preferred embodiment and describes it in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the atmospheric pollution source tracing method based on regional distribution identification and screening of the present invention.

[0016] Figure 2 Schematic diagram of the distribution of regional air quality monitoring stations used in this embodiment.

[0017] Figure 3 PM in Beijing-Tianjin-Hebei region from 2015 to 2022 was obtained by using the method of this embodiment. 2.5 Spatial distribution of mean concentration.

[0018] Figure 4 This is a spatial distribution map of the mean O3 concentration in the Beijing-Tianjin-Hebei region from 2015 to 2022 obtained using the method of this embodiment.

[0019] Figure 5 The PM of Beijing-Tianjin-Hebei from 2015 to 2022 was obtained by using the method of this embodiment. 2.5 Spatial distribution heat map.

[0020] Figure 6 This is a hotspot map of O3 spatial distribution in Beijing-Tianjin-Hebei from 2015 to 2022 obtained by using the method of this embodiment.

[0021] Figure 7 This is a result map of the pollution source area obtained by using the method of this embodiment.

[0022] Figure 8 This is a traceability result diagram of the pollution source locations and names obtained by using the method of this embodiment. DETAILED DESCRIPTION

[0023] The technical scheme of the present invention will be clearly and completely described below in conjunction with specific embodiments, but those skilled in the art should understand that the embodiments described below are only used to illustrate the present invention and should not be regarded as limiting the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] Example of air pollution source tracing method based on regional distribution identification and screening

[0025] The present invention aims to solve the problem of air pollution source tracing. It improves the existing tracing methods and studies new tracing methods based on mathematical analysis, so as to realize the identification and screening of hot spots and the tracing of pollution processes, and provide technical support for the refined control of air pollution.

[0026] The present invention collects atmospheric pollutant concentration data within a given time period in the area to be tested, and pre-processes the data. The pre-processed atmospheric pollutant concentration data is used for interpolation calculation and spatiotemporal distribution calculation. The interpolation calculated data is used for autocorrelation data analysis, cluster analysis is performed on the data, hot and cold area calculations are performed, hotspot areas are identified and screened, and atmospheric pollution sources are traced for the identified and screened hotspot areas. This method can realize automated operation based on network data, real-time automatic identification, screening and tracing. This method can greatly improve the level of refined control of pollutant concentration in hotspot areas, and improve the accuracy of real-time hotspot area screening and tracing of pollutants.

[0027] like Figure 1 Shown is a schematic flow diagram of the method of the present invention, comprising the following steps:

[0028] 1) Obtain the target pollutant concentration data of the area to be detected.

[0029] This embodiment takes into account that an environmental monitoring network has been established in the existing environment, and multiple fixed monitoring sites are distributed in various cities, such as Figure 2 As shown, the target pollutant data of the area to be detected can be obtained based on the data obtained at this fixed site, thereby avoiding the process of re-investing monitoring equipment in the area to be detected to obtain data, thereby improving the efficiency of data acquisition.

[0030] Based on the above analysis, the pollutant concentration data of this embodiment mainly comes from network crawling. The hourly atmospheric pollutant concentration data for many consecutive years are crawled from the network monitoring platform to obtain the target pollutant concentration data of the area to be detected, including AQI, PM 2.5 、PM 10, O3, NO2, SO2 and CO and other atmospheric pollutants. In addition, this embodiment ensures the real-time nature of the sample data by capturing data hourly, that is, based on different time periods, the time intervals for acquiring data are different, ensuring the real-time update of the data set, thereby ensuring the real-time nature of the sample data.

[0031] In addition, in order to ensure the accuracy of the data set, it is also considered that there may be missing data and outliers in the directly captured data. Therefore, as another implementation method, after the data is directly captured, the data is also pre-processed such as cleaning, screening and outliers to obtain an accurate sample data set. The process of pre-processing the data includes counting missing values ​​and outliers, and filling them with knn nearest neighbor mean interpolation (i.e., knn interpolation method). The knn interpolation method is used to predict the concentration of air pollutants by collecting monitoring data of air pollutant concentrations at known locations, and using these data to train the model, and then using the model to estimate or predict the concentration of air pollutants at unknown locations. Usually, in the process of evaluating and comparing indicators, in order to facilitate the weighting or comparison of indicators of different units and magnitudes, the unit of the original analysis data is often discarded and converted into a dimensionless level value. The process of scaling the original analysis data to a smaller range after scaling it according to a certain ratio is called data normalization. In this embodiment, the data is normalized to the interval [0,1].

[0032] 2) The target pollutant concentration data is interpolated using the diffusion interpolation method with obstacles.

[0033] The target pollutant concentration data of the area to be detected obtained in step 1) is interpolated to predict the pollutant concentration at the location that does not appear in the sample data set, and then the predicted pollutant concentration value at the target location is added to the sample data set to form a complete data set. The data of the complete data set realizes that there are data at all the required locations in the area to be detected, which is conducive to the subsequent accurate division of the area based on the complete data set. In addition, the data prediction process at the required location avoids the need to increase the hardware investment such as monitoring equipment at the required location, saving costs and improving the efficiency of data acquisition.

[0034] This embodiment uses a diffusion interpolation method with obstacles to perform interpolation operations. This method calculates the shortest path distance from the data position of each sample data set to the target position to bypass obstacles, and combines the results of the concentration diffusion of the data of all sample data sets at the corresponding shortest path distance to obtain a comprehensive prediction value of the target pollutant concentration at the target position based on the data of all sample data sets. Unlike ordinary interpolation methods, it can identify and process obstacle elements in geographic space, such as mountains, rivers, buildings and other obstacles that hinder the diffusion of attribute values. When encountering obstacles, the diffusion path will be changed, and the propagation mode of attribute values ​​is no longer a simple isotropic diffusion, but diffusion along the edge of the obstacle or around the obstacle to be more in line with the actual geographical situation.

[0035] Diffusion interpolation refers to the fundamental solution of the heat equation, which describes how heat or particles diffuse over time in a homogeneous medium. The predictions obtained using this method flow slowly around obstacles.

[0036] In the absence of obstacles, the predictions obtained by diffusion interpolation are roughly the same as those obtained by kernel interpolation using a Gaussian kernel.

[0037] Diffusion interpolation uses a replicated distance metric defined by a cost surface, which is a common raster function that calculates the cost of traveling from one raster cell to another. Diffusion interpolation generates predictions for automatically selected grids (cells).

[0038] If a cost surface is not specified, the distance between adjacent cell centers is the Euclidean distance. If a cost surface is specified, the distance between cells on the cost surface is defined using one of the following formulas:

[0039] 1. Additional barriers: (average cost value of adjacent pixels)*(distance between pixel centers);

[0040] 2. Cumulative barrier: (difference in cost values ​​between adjacent cells) + (distance between cell centers);

[0041] 3. Flow barrier: indicator (cost value to adjacent pixel > cost value from adjacent pixel) * (cost value to adjacent pixel - cost value from adjacent pixel) + (distance between pixel centers);

[0042] Among them, Indicator(True)=1, Indicator(False)=0.

[0043] Flow barriers can be used to interpolate data using the primary direction of data variation.

[0044] The barrier-based diffusion interpolation method is based on diffusion theory, which is similar to the diffusion phenomenon of matter in space. It assumes that the interpolated attribute value (such as a certain pollutant concentration, temperature value, etc.) diffuses from a known sampling point to the surrounding space.

[0045] The diffusion process is affected by the spatial distance. The closer the area is to the sampling point, the greater the influence of the attribute value of the sampling point. As the distance increases, the influence gradually decreases, which conforms to the general law of distance attenuation.

[0046] Conceptually, the value of the interpolation result at a certain location is the combined result of the influence of all sampling points on that location. For each sampling point, its influence on the target location is proportional to the attribute value of the sampling point and inversely proportional to the distance from the sampling point to the target location (effective distance after considering obstacles), and this relationship is adjusted by a diffusion coefficient. The formula is roughly as follows (this is a simplified illustration, and the actual calculation is much more complicated):

[0047]

[0048] Where I(x,y) is the interpolation result at the coordinate (x,y), n is the number of sampling points, P i is the attribute value of the i-th sampling point, d i (x, y) is the effective distance from the i-th sampling point to the (x, y) position after considering obstacles. Calculate the distance d from the sampling point to the target position i When calculating the distance between the sampling point and the target location (x, y), the existence of obstacles needs to be considered, which usually involves complex spatial analysis algorithms to determine the shortest path length from the sampling point to the target location that bypasses the obstacle. For example, if there is a river as an obstacle, the straight-line distance cannot be calculated directly, but the distance must be calculated along a path that bypasses the river. k is the diffusion coefficient, which controls the rate and degree of diffusion of the attribute value. It may be a fixed value or may be adjusted according to the specific research object and spatial characteristics. That is, the diffusion coefficient can be a fixed value set artificially for the calculation of any substance. In order to more accurately analyze the diffusion process of different substances and make the interpolation more accurate, different fixed values ​​of the diffusion coefficient can also be set based on different substances. For example, for some substances that diffuse faster (such as gases), the diffusion coefficient may be larger, and for substances that diffuse slower (such as certain components in soil), the diffusion coefficient may be smaller.

[0049] When determining the distance d from the sampling point to the target position i When (x,y), absolute feature barriers use a "non-Euclidean" distance method instead of a "line of sight" method. The "line of sight" method requires that the straight line between the measured location and the location to be predicted does not intersect the barrier feature. If the distance around the barrier is within the searching neighborhood specification, the "non-Euclidean" distance method will be used for it.

[0050] The processing time depends on the complexity of the barrier feature class geometry. New feature classes can be created by smoothing or deleting features using tools in the Generalization toolset.

[0051] For the Input additional barrier rasters parameter, its value must be greater than or equal to 1. A value of 1 indicates no barrier.

[0052] The order of the values ​​in the Input cumulative barrier raster parameter should be the same as the (x,y) coordinate order. If adjacent cells have the same value, it means there is no barrier at that location.

[0053] The order of the values ​​in the Input flow barrier raster parameter should be the same as the (x,y) coordinate order. If the values ​​of adjacent cells are the same, it means there is no barrier at that location. On the other hand, if the values ​​go from high to low, it means there is no barrier.

[0054] The presence of NoData in any of the optional raster barrier inputs is equivalent to including the input absolute barrier features.

[0055] The barrier-free diffusion interpolation method of this embodiment is used to interpolate the annual average concentration of the hourly air quality monitoring stations in the Beijing-Tianjin-Hebei region, and the PM 2.5 Visual simulation of mass concentration of O3, PM 2.5 The spatial distribution of the mean concentration is as follows Figure 3 As shown in Figure 2, the spatial distribution of the mean O3 concentration is as follows: Figure 4 shown.

[0056] 3) Perform cluster analysis on the interpolated data.

[0057] This embodiment takes into account that the impact of atmospheric pollutant sources on the environment weakens with increasing distance, and the degree of aggregation also weakens with increasing distance. Therefore, in the process of regional division, the global spatial autocorrelation analysis method is first used to obtain whether the measured object has autocorrelation in space, and then based on the autocorrelation situation, it is judged whether there is aggregation characteristic, and the first cluster analysis is performed to obtain whether the target pollutant concentration has aggregation characteristics in space, that is, to obtain the result of whether the concentration data is clustered, dispersed or random in space. Then, when there is an aggregation characteristic, it is specifically distinguished whether it is high-value aggregation or low-value aggregation. This process obtains a result that includes both the degree of aggregation and the high and low concentration values, which is conducive to the accurate division of the region.

[0058] When performing cluster analysis in this embodiment, first, the global spatial autocorrelation analysis method is used to perform an overall cluster analysis on the target pollutant concentrations of each sub-area of ​​interest in the area to be tested, and the first clustering result is obtained. Spatial autocorrelation can measure whether things have autocorrelation in space. High autocorrelation represents the existence of clustering of things in space. During this analysis process, the global Moran index is calculated based on the target pollutant concentration values ​​of each sub-area of ​​interest and the average value of the target pollutant concentrations of all sub-areas of interest. The global Moran index is expressed as:

[0059]

[0060] in, i and j represent the codes of different sub-areas of concern; n represents the number of all sub-areas of concern; x represents the target pollutant concentration value of each sub-area of ​​concern; is the average value of the target pollutant concentration in all sub-areas of concern; W ij Spatial weight matrix; The value range of the global Moran's index I is (-1~1). When I>0, it indicates positive spatial correlation, which indicates spatial agglomeration; when I<0, it indicates negative spatial correlation, which indicates spatial dispersion; when I=0, it indicates random distribution in space. The closer the I value is to 1 or -1, the more intense the degree of spatial autocorrelation is and the higher the degree of spatial agglomeration is.

[0061] In addition, the result of the global Moran's index is further statistically tested to obtain the first clustering result, which is expressed as:

[0062]

[0063] Among them, Var(I) is the theoretical variance of I, E(I) is the theoretical expectation, and Z(I) is the first clustering result.

[0064] Then, since the global Moran index is used to measure the clustering of things in space, it cannot determine whether things are clustered at high values ​​or low values ​​in space. Therefore, based on the first clustering result, it is necessary to further determine whether the target pollutant concentration in each sub-area of ​​interest is clustered at high values ​​or low values ​​in space, so as to make the subsequent regional division more accurate through the subdivision process of the clustering type. This embodiment uses a high / low clustering analysis method (i.e., applying Getis-OrdGeneral G analysis) to determine whether the target pollutant concentration in each sub-area of ​​interest is clustered at high values ​​or low values ​​in space. During this analysis process, the target pollutant concentration values ​​of each sub-area of ​​interest are counted to obtain the second clustering result, which is expressed as:

[0065]

[0066] Where G(d) is the second clustering result, d is the distance, W ij (d) is the spatial weight defined by the distance rule, i and j are the codes of different sub-areas of interest, n is the number of all sub-areas of interest in the area to be tested, and x is the target pollutant concentration value of each sub-area of ​​interest.

[0067] Calculate the expected value of the second clustering result under the assumption that the space is not clustered:

[0068] Calculate the statistical test value of the second clustering result under the condition of normal distribution:

[0069] When the second clustering result G(d) is higher than the expected value E(G) and the statistical test value Z(I) is significant, it indicates that there is a high-value aggregation in space, and the judgment result is a high-value aggregation. When the second clustering result G(d) is lower than the expected value E(G) and the statistical test value Z(I) is significant, the judgment result is a low-value aggregation. When the second clustering result G(d) tends to the expected value E(G), it indicates that there is a random distribution feature in space.

[0070] The global spatial autocorrelation analysis method used in this embodiment is based on the acquisition of the global Moran's index to determine whether the phenomenon has clustering characteristics in space. The global spatial autocorrelation analysis method of the global Moran's index is used to measure the similarity of spatial unit attribute values. It takes into account the similarity between each unit and its neighbors and calculates a value between -1 and 1, where 1 represents a complete positive correlation, -1 represents a complete negative correlation, and 0 represents no correlation. As other implementation methods, other global spatial autocorrelation analysis methods can also be used to analyze the overall distribution characteristics of spatial data and determine whether the phenomenon has clustering characteristics in space.

[0071] Based on the method of this embodiment, PM data from 2015 to 2022 are obtained. 2.5 The global Moran index I of O3 and its test results are shown in Table 1. Getis-OrdGeneral G is used to analyze whether the clustering is high-value clustering or low-value clustering. Table 2 shows Getis-Ord General G and its test results. It can be seen that PM 2.5 and O3 belong to high-value aggregation patterns in space.

[0072] Table 1 Global Moran's index I and its test results

[0073]

[0074] Table 2 Getis-Ord General G and its test results

[0075]

[0076] 4) Identify hot spots.

[0077] Considering that the impact of air pollutant sources on the environment weakens with increasing distance, and the degree of aggregation also weakens with increasing distance, in order to accurately and efficiently carry out the process of tracing the source of air pollution, the hotspot area is divided according to the concentration of pollutants and the degree of aggregation, which conforms to the law of diffusion, and then the process of tracing the source based on the division result can ensure the accuracy of the tracing result. Therefore, based on the above analysis, this embodiment takes the area corresponding to the high-value aggregation as the hotspot area.

[0078] In order to further ensure the accuracy of the division of hotspots in this embodiment, the locations where high-value or low-value elements are clustered in space are obtained through hotspot analysis, so that we can intuitively understand where high and low values ​​are clustered and how densely they are clustered. Specifically, in this process, the high-value clusters or low-value clusters of the target pollutant concentration in each sub-area of ​​interest are statistically identified to identify the high-value points and low-value points of each sub-area of ​​interest. In this process, the score value of each cluster position of the target pollutant is calculated based on the target pollutant concentration value of each sub-area of ​​interest and the average value of the target pollutant concentration of all sub-areas of interest. The score value is expressed as:

[0079]

[0080] in, i and j are the codes of different sub-areas of interest, n is the number of all sub-areas of interest in the area to be tested, and x is the target pollutant concentration value of each sub-area of ​​interest. is the average value of the target pollutant concentration in all sub-areas of interest in the area to be tested, W i,j is the defined spatial weight matrix. Score value G i * The higher the value, the tighter the clustering of high values, and the score G i * The lower the score, the tighter the clustering of low values. According to the relationship between the score value and different preset thresholds, the distribution of points of different levels is determined, wherein when the score value exceeds the first preset threshold, it is identified as a high-value point, and when the score value exceeds the second preset threshold, it is identified as a low-value point.

[0081] Through the above-mentioned further division process of high-value clusters and low-value clusters in this embodiment, the locations where high-value or low-value elements are clustered in space are clarified, which is beneficial to improving the accuracy of subsequent tracing results of hot spot areas.

[0082] like Figure 5 , Figure 6The PM of Beijing-Tianjin-Hebei from 2015 to 2022 obtained by the method of this embodiment 2.5 Spatial distribution hotspot map and O3 spatial distribution hotspot map of Beijing-Tianjin-Hebei from 2015 to 2022. From the map, we can clearly see the concentration distribution of pollutants. The concentration is divided into 7 levels. The red area indicates high concentration concentration (hot spot area), and the blue area indicates low concentration concentration (cold spot area). From the year-on-year situation, PM 2.5 There is little change in the high-value and low-value concentration clustering areas. The hot spots are mainly located in the southern region represented by Baoding, Shijiazhuang, Xingtai, Handan and Hengshui, and the cold spots are mainly located in the northern region represented by Zhangjiakou and Chengde. The distribution of O3 concentrations has changed greatly. From 2015 to 2022, the hot spots will gradually shift from the central region represented by Beijing and Tianjin to the southern region. In 2015 and 2016, the cold spots were mainly located in Xingtai, Handan, Tianjin and Qinhuangdao. From 2017 to 2022, the cold spots were mainly located in Zhangjiakou and Chengde in the northern region. Analysis of the causes of the agglomeration characteristics of high-value areas may be affected by multiple factors such as local anthropogenic source emissions, meteorological and topographic conditions, and regional transmission.

[0083] 5) Trace the source of hot spots.

[0084] Based on the above-mentioned regional distribution identification and screening process of this embodiment, the hotspot area is accurately identified. Therefore, in the process of tracing the source of the target pollutants in the detection area, the target pollutants in the detection area are directly traced. Compared with the process of directly tracing the target pollutants in the detection area, the time for tracing the source of those areas with low concentration values ​​or sparse concentration is saved, thereby improving the efficiency of tracing the source. In addition, this embodiment is based on the division of hotspot areas according to the concentration and degree of concentration of pollutants, which conforms to the law of diffusion, and then the process of tracing the source based on the division result can ensure the accuracy of the tracing results. By real-time screening of high-value points and conducting traceability analysis, the level of refinement of pollutant concentration control at high-value points can be greatly improved, the real-time prediction accuracy of pollutants can be improved, and the increase in hardware investment such as monitoring equipment can be avoided.

[0085] In this embodiment, the tracing of the hot spot area is calculated using the FLEXPART mode, which is a Lagrangian particle diffusion model that characterizes the diffusion, transmission, attenuation, dry and wet deposition of tracers in the atmosphere by calculating the trajectory of particles released by emission sources. Emission sources can be point sources, line sources, surface sources, and volume sources. This mode can perform forward trajectory calculations as well as backward trajectory calculations. The forward trajectory can simulate the changes over time after the tracer is released, and characterize the transport and diffusion of tracers with meteorological conditions after release from the source area. It is often used to determine the evolution of pollutants after special conditions occur (such as leakage of atmospheric pollutants). The calculation of backward trajectories can determine the distribution of potential source areas that contribute to the target point or area.

[0086] The meteorological field is simulated by WRF model, and the meteorological field results are used to drive the FLEXPART model to run backward to obtain the temporal and spatial distribution of potential source areas of atmospheric pollutants in the target area and target time period. Figure 7 As shown, Figure 7 The part circled by black lines is the polluted area. On this basis, the emission intensity of the pollution source point is used as the weight to calculate the contribution rate of the "source" at any time and location to the selected atmospheric pollutants (i.e., target pollutants) in the target area during the target period, and the concentration contribution of different sources to the hot spot area is calculated to identify and screen the main pollution sources, providing support for targeted governance and emission reduction, such as Figure 8 The part circled with black lines is the location and name of the pollution source in the traceability results.

[0087] The formula for calculating the contribution rate of the "source" at any time and location to the selected air pollutants in the target area during the target period is:

[0088]

[0089] Where (i, j) and t represent a certain spatial position and time respectively, P is the sensitivity coefficient of the source at that time and position, and E is the corresponding representative tracer emission intensity. t1 and t2 are the start and end times of the period to be studied, and -∞ represents the earliest time of the research target; (N1, S1) and (N2, S2) represent the starting and ending positions of the cumulative contribution respectively. When (0, 0) and (N, S) and t∈(-∞, 0], R=1; (N, S) is the maximum position of the model area, and when E is 1, the result is the potential contribution rate.

[0090] In the present invention, the process of first identifying and screening the air pollution area is used to determine the area with high concentration and concentration of air pollutants as the hot spot area, and then the hot spot area is traced. Compared with the process of directly tracing the target pollutants in the detection area, this process saves the time of tracing the source of those areas with low concentration values ​​or sparse concentration, thereby improving the efficiency of tracing the source. In addition, the division of the hot spot area in the present invention is based on the division of pollutant concentration and concentration, which conforms to the diffusion law, and then the process of tracing the source based on the division result can ensure the accuracy of the tracing result.

[0091] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A method for tracing the source of air pollution based on regional distribution identification and screening, characterized in that: The steps include: 1) Obtain historical data of target pollutant concentrations in the area to be detected as a sample data set; 2) Using the diffusion interpolation method with obstacles to interpolate the sample data set, a complete data set including the predicted value of the target pollutant concentration at the target location in the detection area is obtained; 3) Based on the data of the complete data set and using the global spatial autocorrelation analysis method, the target pollutant concentrations of each sub-area of ​​interest in the area to be tested within the same time period are analyzed for overall clustering, and the first cluster analysis result is obtained. When the first cluster analysis result is clustered in space, it is determined whether the target pollutant concentrations in each sub-area of ​​interest are clustered in space at high values ​​or low values; 4) The sub-area of ​​concern corresponding to the high-value aggregation result is the hotspot area, and the target pollutants in the hotspot area are traced to obtain the tracing results.

2. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 1 is characterized in that: In step 3), when the target pollutant concentration is spatially high-value clustered, the clustering degree of each sub-area of ​​interest of the high-value cluster is determined, so as to divide the sub-areas of interest of the high-value cluster into different clustering degree groups according to the clustering degree.

3. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 2 is characterized in that: The score value of the clustering position corresponding to each sub-area of ​​interest of the target pollutant is calculated by the target pollutant concentration value of each sub-area of ​​interest and the average value of the target pollutant concentration of all sub-areas of interest in the tested area, and the degree of aggregation of each sub-area of ​​interest with high values ​​is determined based on the score value.

4. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 3 is characterized in that: The formula for calculating the score value of the cluster location corresponding to each sub-area of ​​concern of the target pollutant is: in, i and j are the codes of different sub-areas of interest, n is the number of all sub-areas of interest in the area to be tested, and x is the target pollutant concentration value of each sub-area of ​​interest. is the average value of the target pollutant concentration in all sub-areas of interest in the area to be tested, W i,j is the defined spatial weight matrix.

5. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 4 is characterized in that: When the score value is higher than the first preset threshold, the sub-area of ​​interest corresponding to the score value is a high-value point.

6. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 1 is characterized in that: In step 3), the global spatial autocorrelation analysis method is: using the target pollutant concentration values ​​of each sub-area of ​​interest, calculate the global Moran's index of the area to be detected, and obtain the first cluster analysis result based on the global Moran's index.

7. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 6 is characterized in that: In step 3), the method for determining whether the concentration of target pollutants in each sub-area of ​​interest is spatially clustered at high values ​​or low values ​​is as follows: The second clustering result is obtained, where G(d) is the second clustering result, d is the distance, and W ij (d) is the spatial weight defined by the distance rule, i and j are the codes of different sub-areas of interest, n is the number of all sub-areas of interest in the area to be tested, and x is the target pollutant concentration value of each sub-area of ​​interest; Calculate the expected value of the second clustering result under the assumption that the space is not clustered, and calculate the statistical test value of the second clustering result under the condition of obeying the normal distribution; When the second clustering result is higher than the expected value and the statistical test value is significant, the judgment result is a high-value clustering; when the second clustering result is lower than the expected value and the statistical test value is significant, the judgment result is a low-value clustering.

8. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 1 is characterized in that: In step 1), historical data of target pollutant concentrations in the area to be detected are obtained by capturing hourly atmospheric pollutant concentration data for consecutive years from a network monitoring platform.

9. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 8, characterized in that: The historical data is preprocessed data obtained by removing abnormal values ​​from the hourly atmospheric pollutant concentration data and supplementing the missing values ​​in the data using the knn interpolation method.

10. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 1, characterized in that: In step 2), the diffusion interpolation method with obstacles is: The shortest path distance from the data location of each sample data set to the target location bypassing obstacles is calculated respectively, and the concentration diffusion results of the data of all sample data sets under the corresponding shortest path distance are combined to obtain the comprehensive prediction value of the target pollutant concentration at the target location from the data of all sample data sets.

11. The method for tracing the source of air pollution based on regional distribution identification and screening according to claim 1, characterized in that: In step 4), the method for tracing the source of the target pollutants in the hot spot area is as follows: the meteorological field is simulated by the WRF model, and the meteorological field results are used to drive the FLEXPART model to run backward to obtain the spatiotemporal distribution of the potential source areas of the target pollutants in the target area and target time period, and the emission intensity of the pollution source point is used as the weight to calculate the contribution rate of the source at any time and location to the target pollutants in the target area within the target period, and the concentration contribution of different sources at any time and location to the hot spot area is calculated, and the main pollution sources are identified and screened out according to the concentration contribution.

Images