Air quality assessment method and system based on multi-source data

Abstract

This application relates to the field of data processing technology and discloses an air quality assessment method and system based on multi-source data. The method includes: collecting and correcting multi-source monitoring data; dividing the data into grids and integrating environmental characteristics; generating air quality distribution maps through weighted calculations; predicting future concentrations based on historical and meteorological data; tracking pollution transmission to determine the contribution of source areas; and assessing risks based on population distribution and generating early warning recommendations. This application achieves accurate interpolation of pollutant concentrations, future trend predictions, and quantitative analysis of pollution source contribution rates based on environmental characteristics when monitoring points are unevenly distributed, thereby supporting accurate health risk warnings and management decisions.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to an air quality assessment method and system based on multi-source data. Background Art

[0002] Traditional air quality monitoring relies primarily on the construction of a fixed monitoring station network, which collects regional air pollutant data by deploying a certain number of high-precision monitoring stations. These monitoring stations are usually equipped with standardized sampling and analysis equipment, capable of accurately measuring the concentrations of conventional pollutants. Existing air quality assessment methods often use a single data source, such as using only national control station data for time series analysis and trend forecasting, or using only satellite remote sensing data for large-scale spatial distribution assessments. In terms of pollution source tracing, traditional methods mainly rely on pollution source lists and simple meteorological conditions analysis, making it difficult to accurately quantify the contribution of different source areas to the target area. Early warning mechanisms are mostly based on a single pollutant concentration threshold, adopting unified control measures across the city, and lacking temporal and spatial precision and specificity.

[0003] However, existing technologies have significant shortcomings. First, the high construction and operation and maintenance costs of fixed monitoring stations, coupled with a limited number of stations, result in insufficient spatial coverage and fail to fully reflect the spatial differences in air quality within a region. Second, single-data source analysis methods struggle to balance both temporal and spatial resolution, and fail to fully consider the complex relationship between meteorological factors and the diffusion and transmission of pollutants. Third, traditional pollution source tracing methods rely heavily on empirical judgment, making it difficult to quantitatively analyze the contribution of different pollution sources and unable to accurately trace specific pollution incidents. Finally, early warning mechanisms lack consideration of population exposure risks, and control measures are crude, leading to insufficient prevention and control in high-risk areas and unnecessary socioeconomic losses in low-risk areas. These shortcomings severely limit the scientific nature and precision of air quality management. Summary of the Invention

[0004] This application provides an air quality assessment method and system based on multi-source data, which is used to achieve accurate interpolation of pollutant concentrations, future trend prediction and quantitative analysis of pollution source contribution rates in combination with environmental characteristics when monitoring points are unevenly distributed, thereby supporting accurate health risk warning and control decisions.

[0005] In a first aspect, the present application provides an air quality assessment method based on multi-source data, the air quality assessment method based on multi-source data comprising: collecting pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors, and performing outlier identification and correction processing on the data to obtain a multi-source air quality original data set; dividing the target area into fine urban grids and large-scale suburban grids based on the multi-source air quality original data set, and forming an environmental feature grid database by combining terrain, land use, roads, traffic, population, and pollution source information; performing weighted calculations on the air quality data in the environmental feature grid database according to data accuracy, environmental similarity, and spatial distance to generate an interpolated distribution map of air quality for the entire region; calculating predicted pollutant concentration values ​​for future time periods based on the interpolated distribution map of air quality for the entire region, in combination with historical data for the same period and meteorological forecast information, to form a spatiotemporal dynamic prediction result; utilizing the spatiotemporal dynamic prediction result, in combination with wind field data and pollution source distribution, tracing the pollutant transmission path, and determining the pollution contribution value of each source area to the target area; calculating the health risk level of different regions based on the pollution contribution value, in combination with population distribution and activity patterns, and generating graded warning information and control measure recommendations.

[0006] In a second aspect, the present application provides an air quality assessment system based on multi-source data, the air quality assessment system based on multi-source data comprising:

[0007] The correction module is used to collect pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors, and perform outlier identification and correction on the data to obtain a multi-source air quality raw data set;

[0008] a partitioning module for dividing the target area into fine urban grids and large suburban grids based on the multi-source air quality raw data set, and forming an environmental feature grid database by combining topography, land use, road, traffic, population, and pollution source information;

[0009] A weighting module is used to perform weighted calculation on the air quality data in the environmental feature grid database according to data accuracy, environmental similarity and spatial distance to generate an interpolation distribution map of air quality for the entire region;

[0010] A prediction module is used to calculate the predicted value of pollutant concentration in the future period based on the interpolated distribution map of air quality in the entire region, combined with historical data of the same period and meteorological forecast information, to form a spatiotemporal dynamic prediction result;

[0011] A tracking module is used to use the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, to track the pollutant transmission path and determine the pollution contribution of various source areas to the target area;

[0012] The control module is used to calculate the health risk level of different areas based on the pollution contribution value, combined with population distribution and activity patterns, and generate graded warning information and control measure recommendations.

[0013] A third aspect of the present invention provides a computer device comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned air quality assessment method based on multi-source data.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, enable the computer to execute the above-mentioned air quality assessment method based on multi-source data.

[0015] The technical solution provided in this application breaks through the limitations of traditional single data sources by collecting multi-source pollutant concentration data from fixed air monitoring stations, micro-environmental stations and portable sensors, and performing outlier identification and correction processing, significantly improving the integrity and reliability of data and laying a solid data foundation for air quality assessment; combining terrain, land use, roads, traffic, population and pollution source information to divide the target area into fine grids in urban areas and large-scale grids in suburban areas to form an environmental feature grid database, achieving high-resolution spatial expression of environmental elements and fully considering the heterogeneous characteristics of the regional environment; generating an interpolation distribution map of air quality in the entire region by weighted calculation according to data accuracy, environmental similarity and spatial distance, solving the problem of uneven spatial distribution of monitoring points and achieving The pollutant concentration field was accurately reconstructed; based on the interpolation distribution map of air quality in the entire region, the predicted pollutant concentration values ​​for future time periods were calculated in combination with historical data for the same period and meteorological forecast information, thus realizing the dynamic prediction of the spatiotemporal distribution of pollutants and providing a time window for early prevention and control; using the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, the pollutant transmission path was tracked, and the pollution contribution value of various source areas to the target area was determined, thus realizing the quantitative and scientific tracing of pollution sources and providing direction for precise pollution control; according to the pollution contribution value, combined with population distribution and activity patterns, the health risk levels of different regions were calculated, and graded warning information and control measures were recommended, thus realizing a closed loop from air quality assessment to health risk management, and providing precise support for public health protection and pollution prevention and control decisions. It is particularly worth emphasizing that this solution utilizes artificial intelligence algorithms in multiple stages. For example, the improved geographically weighted regression algorithm used in the spatial fusion and interpolation of multi-source data can adaptively learn the complex nonlinear relationship between environmental characteristics and pollutant concentrations; the random forest algorithm used in the spatiotemporal dynamic prediction model can handle the high-dimensional interaction between meteorological factors and pollutant concentrations; the Lagrangian particle model and source contribution matrix algorithm used in pollution source analysis can accurately simulate the complex transmission process of pollutants under different meteorological conditions; and the multi-factor risk model used in health risk assessment can comprehensively consider population density, activity patterns, and the distribution characteristics of sensitive populations. The application of these algorithms enables this solution to process large amounts of heterogeneous data, capture complex spatiotemporal patterns and causal relationships, and achieve intelligent and precise air quality assessment and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1This is a schematic diagram of an embodiment of an air quality assessment method based on multi-source data in an embodiment of the present application;

[0018] Figure 2 This is a schematic diagram of an embodiment of an air quality assessment system based on multi-source data in an embodiment of the present application;

[0019] Figure 3 It is a schematic block diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The embodiments of the present application provide an air quality assessment method and system based on multi-source data. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or that are inherent to these processes, methods, products or devices.

[0021] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In one embodiment of the present application, an air quality assessment method based on multi-source data includes:

[0022] Step S101: Collect pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors, and perform outlier identification and correction on the data to obtain a multi-source air quality raw data set;

[0023] Step S102: Based on the multi-source air quality original dataset, the target area is divided into fine urban grids and large suburban grids, and an environmental feature grid database is formed by combining terrain, land use, road, traffic, population, and pollution source information;

[0024] Step S103: performing weighted calculation on the air quality data in the environmental feature grid database according to data accuracy, environmental similarity, and spatial distance to generate an interpolated distribution map of air quality for the entire region;

[0025] Step S104: Calculate the predicted pollutant concentration for the future period based on the interpolated distribution map of air quality for the entire region, combined with historical data for the same period and weather forecast information, to form a spatiotemporal dynamic prediction result;

[0026] Step S105: Using the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, trace the pollutant transmission path and determine the pollution contribution of various source areas to the target area;

[0027] Step S106: Calculate the health risk levels of different areas based on the pollution contribution value, combined with population distribution and activity patterns, and generate graded warning information and control measures recommendations.

[0028] It is understandable that the execution subject of this application can be an air quality assessment system based on multi-source data, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0029] Specifically, fixed air monitoring stations, as high-precision equipment, usually collect pollutant indicators once an hour, including data on six major pollutants: fine particulate matter, inhalable particulate matter, sulfur dioxide, nitrogen dioxide, ozone, and carbon monoxide. Miniature environmental stations, as medium-precision equipment, collect data every thirty minutes, while portable sensors, as low-precision equipment, collect data every ten minutes. For all collected data, outlier identification processing is performed by calculating the deviation of the data from the seasonal historical average. When the data of a monitoring point exceeds three times the standard deviation of the seasonal historical average, it is marked as abnormal data in the time dimension. At the same time, the percentage deviation between each monitoring point and the average value of the five adjacent monitoring points is calculated. When the deviation exceeds 75%, it is marked as an abnormal data point in the spatial dimension. Data points marked as both abnormal in the time dimension and abnormal in the spatial dimension are removed, and the remaining data points are retained. Subsequently, cross-validation is performed on devices of different precisions in adjacent distributions to establish a data conversion relationship between devices of different precisions. For example, a fixed monitoring station in the northern area of ​​a city measured a fine particulate matter concentration of 75 micrograms per cubic meter, while a portable sensor in the same area measured 85 micrograms per cubic meter. Through multiple comparisons, a correction factor of 0.88 was formed.

[0030] The target area was gridded based on the multi-source air quality raw data set. Due to the dense population and complex pollution sources, urban areas used a fine grid cell of 500 meters by 500 meters, while suburban areas used a larger grid cell of 2,000 meters by 2,000 meters. Digital elevation data was extracted from the geographic information system, and the average altitude and terrain relief of each grid cell were calculated to generate terrain characteristic data. Land use classification data was also extracted from the dominant land type and its proportion within each grid cell to generate land use characteristic data. Road characteristic data was obtained by calculating the total length of roads and the number of intersections within each grid cell. Traffic characteristic data recorded the traffic volume and congestion level within each grid cell. Population characteristic data included the residential population density and floating population density within each grid cell. Pollution source characteristic data identified the number of industrial enterprises and emission levels within the grid cell. These characteristic data of different dimensions were linked to the grid cells through spatial overlay analysis to form an environmental characteristic grid database.

[0031] Air quality interpolation calculation is performed based on the environmental feature grid database. First, different weights are assigned to monitoring equipment according to accuracy: the data weight of fixed air monitoring stations is 1.0, the data weight of micro environmental stations is 0.6, and the data weight of portable sensors is 0.3. Then, the environmental similarity between the grid to be interpolated and the known monitoring points is calculated, and the environmental feature vectors are compared using the cosine similarity method to obtain the environmental similarity weight. The actual reachable distance is calculated based on the road network data to form the spatial distance weight. For each grid to be interpolated, the monitoring point data within a range of 5 kilometers is extracted. If the grid is located downwind of the monitoring point, the weight of the monitoring point is enhanced. The accuracy weight, environmental similarity weight and spatial distance weight are normalized and multiplied together to obtain a comprehensive weight coefficient, which is used for the weighted average calculation of the monitoring point data to generate an air quality interpolation distribution map for the entire region.

[0032] Forecasts are made using interpolated regional air quality distribution maps. First, historical data for the past three years is extracted to construct a database of pollutant concentrations for the same period. Weather forecast data for the next 72 hours is obtained, including wind direction, wind speed, temperature, humidity, air pressure, and precipitation probability. Time series decomposition is performed on the existing pollutant concentration data to separate long-term trends from cyclical patterns. Seasonal variations are determined based on historical data for the same period, generating a basic forecast curve. The basic forecast is adjusted using weather forecast data, taking into account the impact of pollution conditions in upwind areas to generate a spatiotemporal dynamic forecast. Based on the spatiotemporal dynamic forecast results, pollutant transmission paths are tracked. First, 72-hour wind field data for the target area is obtained. Wind direction is divided into 16 sectors and wind speed into 6 levels, creating a wind field data grid. The location and emission intensity of pollution source areas are obtained from the environmental protection department. Backtracking is performed using the Lagrangian particle method to generate a pollutant transmission path map. Spatial overlay analysis is used to calculate the impact frequency of source areas. The pollution contribution of each source area is determined by combining emission intensity and distance attenuation.

[0033] Use pollution contribution values ​​to generate health risk assessments and early warnings. First, obtain population distribution data and construct a population distribution map. Extract urban resident activity patterns from social survey data and identify activity patterns during different time periods. Obtain respiratory disease patient data from medical institutions to generate a distribution map of sensitive populations. Combining pollution contribution values, population distribution, activity patterns, and the distribution of sensitive populations, calculate the exposure of each grid cell at different time periods. Based on this, health risk levels are set and targeted early warning information and control measures are generated.

[0034] In the embodiment of the present application, by collecting multi-source pollutant concentration data from fixed air monitoring stations, micro-environmental stations and portable sensors, and performing outlier identification and correction processing, the limitations of the traditional single data source are broken through, the integrity and reliability of the data are significantly improved, and a solid data foundation is laid for air quality assessment; combining terrain, land use, roads, traffic, population and pollution source information to divide the target area into fine grids in urban areas and large-scale grids in suburbs to form an environmental feature grid database, achieving high-resolution spatial expression of environmental factors and fully considering the heterogeneous characteristics of the regional environment; by performing weighted calculation according to data accuracy, environmental similarity and spatial distance to generate an interpolation distribution map of air quality in the entire region, the problem of uneven spatial distribution of monitoring points is solved, and the problem of uneven spatial distribution of monitoring points is achieved. Accurate reconstruction of pollutant concentration fields; based on the interpolation distribution map of air quality in the entire region, combined with historical data for the same period and meteorological forecast information, the predicted pollutant concentration values ​​for future time periods are calculated, realizing the dynamic prediction of the spatiotemporal distribution of pollutants, providing a time window for early prevention and control; using the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution to track the pollutant transmission path, determine the pollution contribution value of various source areas to the target area, realize the quantitative and scientific tracing of pollution sources, and provide direction for precise pollution control; according to the pollution contribution value, combined with population distribution and activity patterns, calculate the health risk level of different regions, generate graded warning information and control measures, realize the closed loop from air quality assessment to health risk management, and provide precise support for public health protection and pollution prevention and control decision-making. It is particularly worth emphasizing that this solution utilizes artificial intelligence algorithms in multiple stages. For example, the improved geographically weighted regression algorithm used in the spatial fusion and interpolation of multi-source data can adaptively learn the complex nonlinear relationship between environmental characteristics and pollutant concentrations; the random forest algorithm used in the spatiotemporal dynamic prediction model can handle the high-dimensional interaction between meteorological factors and pollutant concentrations; the Lagrangian particle model and source contribution matrix algorithm used in pollution source analysis can accurately simulate the complex transmission process of pollutants under different meteorological conditions; and the multi-factor risk model used in health risk assessment can comprehensively consider population density, activity patterns, and the distribution characteristics of sensitive populations. The application of these algorithms enables this solution to process large amounts of heterogeneous data, capture complex spatiotemporal patterns and causal relationships, and achieve intelligent and precise air quality assessment and management.

[0035] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0036] (1) Obtaining concentration data of fine particulate matter, inhalable particulate matter, sulfur dioxide, nitrogen dioxide, ozone, and carbon monoxide air pollutants and equipment geographic coordinate information from monitoring equipment of different accuracy levels;

[0037] (2) Establish a monitoring equipment accuracy rating system, classifying fixed air monitoring stations as high-precision equipment, micro-environmental stations as medium-precision equipment, and portable sensors as low-precision equipment;

[0038] (3) Calculate the deviation of the collected pollutant concentration data from the seasonal historical average value. When the deviation exceeds three times the standard deviation, it is marked as an abnormal data point in the time dimension;

[0039] (4) Calculate the percentage deviation between each monitoring point and the average value of the five adjacent monitoring points. When the percentage exceeds 75%, mark it as an abnormal data point in the spatial dimension;

[0040] (5) Remove the data points that are marked as both time dimension anomalies and space dimension anomalies, and retain the remaining data points;

[0041] (6) Cross-validate the data of adjacently placed equipment with different accuracy levels and establish data conversion equations between equipment with different accuracy levels;

[0042] (7) Linear interpolation is used to fill in missing time series data, and inverse distance weighted interpolation is used to fill in missing spatial distribution data, forming a multi-source air quality original data set.

[0043] Specifically, data on various pollutant concentrations is collected from monitoring equipment with varying levels of accuracy. Fixed air monitoring stations are built and maintained by environmental protection departments and equipped with standardized sampling and analysis equipment. They accurately measure the concentrations of six major air pollutants: fine particulate matter, respirable particulate matter, sulfur dioxide, nitrogen dioxide, ozone, and carbon monoxide, and record the precise geographic coordinates of the equipment. Micro-environmental stations, typically deployed by district and county-level environmental protection departments or research institutions, are smaller and more affordable. They can measure key pollutants but are slightly less accurate than fixed stations. Portable sensors, deployed by communities, schools, or citizen science projects, are low-cost but have limited accuracy, primarily measuring common pollutants such as fine particulate matter. These three types of equipment form a multi-tiered monitoring network with coverage ranging from point-to-surface to high-to-low accuracy. The accuracy grading system for monitoring equipment is established to rationally utilize data from different sources. Fixed air monitoring stations, as high-precision equipment, collect complete data every hour, and their results serve as a reference standard. Micro-environmental stations, classified as medium-precision equipment, collect data every 30 minutes. Portable sensors, classified as low-precision equipment, collect data every 10 minutes. The classification of accuracy levels is based on the calibration frequency, instrument accuracy and maintenance level of the equipment, and is directly related to the weight distribution in subsequent data processing.

[0044] Identifying outliers in collected pollutant concentration data is a key step in ensuring data quality. First, the deviation from the seasonal historical average must be calculated, that is, the difference between the current measurement value and the average value of the same month and time period in the past three years, and then compared with the standard deviation. The standard deviation reflects the fluctuation range of historical data. When the deviation value exceeds three times the standard deviation, it means that the current value is far away from the normal fluctuation range. At this time, the data point is marked as an abnormal data point in the time dimension. For example, the PM2.5 concentration at a monitoring station at 2:00 p.m. on June 15 was 120 micrograms / cubic meter, while the average value of the station at 2:00 p.m. on June 15 in the past three years was 35 micrograms / cubic meter, with a standard deviation of 15 micrograms / cubic meter. The deviation value is 85, which exceeds three times the standard deviation of 45, and is therefore marked as a time dimension anomaly.

[0045] Spatial anomaly identification is achieved by calculating the percentage deviation between each monitoring point and the average of its five adjacent monitoring points. Adjacent monitoring points are the five closest monitoring points in spatial proximity, determined by calculating the Euclidean distance between them. When calculating the percentage deviation, the average of the measurements taken at the five adjacent monitoring points at the same moment is first calculated. The percentage difference between the current monitoring point and this average is then calculated as a percentage of the average. If the percentage exceeds 75%, it indicates a significant difference between the current monitoring point's data and the surrounding environment, and the point is marked as a spatial anomaly.

[0046] Removing data points marked as anomalous in both the temporal and spatial dimensions is a dual verification mechanism. Data is removed only when it is identified as anomalous in both the temporal and spatial dimensions. This eliminates obviously erroneous data while retaining unique but authentic data points, such as localized pollution incidents. The retained data points include normal data points, data points with only temporal anomalies, and data points with only spatial anomalies, forming the basis for subsequent processing.

[0047] Cross-validating data from adjacently deployed devices of varying accuracy is a crucial approach to addressing system errors. By analyzing the measurement results of closely located devices of varying accuracy over the same period (typically within a 100-meter radius), a linear regression equation is established as a data conversion equation. The conversion equation takes the form Y = aX + b, where Y represents the data from the high-precision device, X represents the data from the low-precision device, and a and b are the regression coefficients. This allows the low-precision device data to be calibrated to a level comparable to that of the high-precision device.

[0048] For missing data in time series, linear interpolation is used. This assumes a linear relationship between the two known data points before and after the missing point, and the missing value is calculated by taking the weighted average of the two data points before and after. For missing data in spatial distribution, inverse distance weighted interpolation is used. This calculates a weighting coefficient based on the distance from the known point to the interpolated point, with closer distances giving a greater weight. Using these two interpolation methods, we fill in all missing data and create a multi-source air quality raw data set.

[0049] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0050] (1) Divide the urban built-up area in the target area into fine-scale urban grid cells of 500 meters by 500 meters, and divide the suburban area in the target area into large-scale suburban grid cells of 2,000 meters by 2,000 meters;

[0051] (2) Extract digital elevation data from the geographic information system, calculate the average altitude and terrain relief of each grid cell, and obtain terrain feature data;

[0052] (3) Extract the dominant land type and land type composition ratio of each grid cell from the land use classification data to form land use characteristic data;

[0053] (4) Obtain road network data from the traffic management department, calculate the road length density and the number of road intersections in each grid unit, and generate road feature data;

[0054] (5) Obtain traffic data from the traffic flow monitoring system, calculate the average vehicle flow and traffic congestion index of each grid unit, and obtain traffic characteristic data;

[0055] (6) Extract population distribution information from the census data, calculate the resident population density and floating population density of each grid cell, and form population characteristic data;

[0056] (7) Obtain a list of fixed pollution sources from the environmental protection department, mark the number of industrial enterprises and emission levels in each grid unit, and generate pollution source characteristic data;

[0057] (8) Through spatial overlay analysis, terrain feature data, land use feature data, road feature data, traffic feature data, population feature data and pollution source feature data are associated with grid cells to form an environmental feature grid database.

[0058] Specifically, based on urban planning data, the boundaries of urban built-up areas are identified and divided into fine-scale urban grid cells measuring 500 meters by 500 meters. This scale not only reflects the spatial variation of pollutants within urban areas but also matches the spatial resolution of commonly used traffic and population data. For suburban areas, where pollution sources are relatively sparsely distributed and change slowly, a larger grid cell size of 2,000 by 2,000 meters is used, reducing computational complexity while ensuring the rationality of spatial representation. The grid is divided using the UTM projected coordinate system to ensure the accuracy of area calculations. Each grid cell is assigned a unique code consisting of row and column numbers. For example, urban grid cell C0512 represents the urban grid cell in row 5 and column 12.

[0059] After the grid division is completed, the terrain feature data of each grid is extracted. The digital elevation model (DEM) data is obtained from the geographic information system, usually with a resolution of 30 meters or 90 meters. For each grid cell, the elevation values ​​of all DEM points within its range are extracted, and the arithmetic mean is calculated as the average altitude of the grid. The terrain undulation is obtained by calculating the standard deviation of the DEM points in the grid. The larger the standard deviation, the more drastic the terrain changes. For example, a grid contains 100 DEM points with elevation values ​​ranging from 45 meters to 75 meters. The calculated average altitude is 60 meters and the standard deviation is 8 meters. These two values ​​together constitute the terrain feature data of the grid.

[0060] Land use feature data is extracted based on remote sensing image classification results or urban planning data. Land use is typically categorized into residential, commercial, industrial, green, water, and transportation types. For each grid cell, the proportion of each land type within the grid is calculated. The type with the largest proportion is defined as the dominant land type for that grid cell. The composition ratio of each type is also calculated to form a proportion vector. For example, if the land use composition of a grid cell is: residential 40%, commercial 30%, green 15%, and transportation 15%, and the dominant land type is residential, the land type composition ratio vector is [0.4, 0.3, 0, 0.15, 0, 0.15].

[0061] Road characteristic data is extracted from road network data obtained from traffic management departments. Road network data contains the spatial location and attribute information of roads at all levels, such as road grade, number of lanes, etc. Road length density refers to the total length of roads per square kilometer. It is calculated by adding the lengths of all roads in the grid and dividing it by the grid area. The number of road intersections is obtained by identifying road intersections. Intersections usually mean the changing points of traffic flow and are closely related to pollutant emissions. For example, the grid unit area of ​​an urban area is 0.25 square kilometers, the total length of various roads contained in it is 3 kilometers, and the number of intersections is 8. Then its road length density is 12 kilometers / square kilometer and the road intersection density is 32 / square kilometer.

[0062] Traffic characteristic data needs to be obtained from the traffic flow monitoring system. Modern urban traffic monitoring systems collect road traffic flow information through intersection surveillance cameras, electronic police systems or floating car data. For each grid unit, first identify the main roads inside, and then extract the average traffic flow data of these roads, usually expressed in PCU (passenger car equivalent unit). The traffic congestion index is calculated by comparing the ratio of actual driving speed to the free flow speed of the road. The closer the value is to 1, the smoother the traffic, and the closer it is to 0, the more serious the congestion. For roads without direct monitoring data, the average value of similar road types and location conditions is used for estimation.

[0063] Population characteristics data are derived from census or survey data. Resident population density reflects nighttime population distribution and is calculated by dividing the number of residents within a grid by the grid area. Mobile population density reflects the intensity of daytime population activity and is derived from data such as mobile phone signaling, public transportation card swipe data, or indirect indicators such as commercial point-of-interest (POI) density. Population density data is closely related to health impact assessments of air pollution because it reflects the size of the population exposed to pollutants.

[0064] Pollution source characteristic data requires obtaining a list of stationary pollution sources from the environmental protection department. Stationary pollution sources primarily include point source emission facilities such as industrial enterprises, boiler rooms, and waste disposal sites. For each grid cell, the number of industrial enterprises within it is counted and classified into three tiers based on their emissions: large, medium, and small. Emission tiers are determined based on a company's total annual pollutant emissions. For example, annual emissions of Class I air pollutants exceeding 100 tons are classified as large. This data directly impacts the local pollution contribution of a grid cell.

[0065] Through spatial overlay analysis, these various feature data are associated with grid cells to form an environmental feature grid database. Spatial overlay analysis is a fundamental function of geographic information systems (GIS). It integrates attribute information from different layers into a unified spatial unit based on spatial location relationships. Specific operations include determining whether a point is within a surface, performing line-surface intersection operations, and analyzing surface-surface overlap. The data structure of the environmental feature grid database consists of grid identifiers and their associated feature attributes, forming a "one grid, one record" relational database that facilitates subsequent query and analysis.

[0066] Take a fine grid unit in an urban area in a certain city center as an example: the grid is numbered C0305, with an area of ​​0.25 square kilometers. The average altitude calculated from the DEM data is 45 meters, and the terrain relief is 2 meters. Land use analysis shows that the dominant type of land is commercial land, accounting for 60%, followed by residential land, accounting for 30%, and transportation land, accounting for 10%. Road network analysis shows that there are 2.8 kilometers of main and secondary roads in the area, with 6 intersections, and a road density of 11.2 kilometers per square kilometer. Traffic monitoring data shows that the average daytime traffic volume on weekdays is 2,000 PCU / hour, and the traffic congestion index is 0.6. Population data shows that the permanent population is 5,000, with a permanent population density of 20,000 people per square kilometer, and the daily floating population increase is 8,000 people, with a floating population density of 32,000 people per square kilometer. The pollution source inventory shows that there are two medium-sized industrial enterprises in the area, which emit 20 tons of fine particulate matter annually. These detailed features are associated with grid C0305 through spatial overlay analysis, becoming a complete record in the environmental feature grid database, providing multi-dimensional environmental background information for subsequent air quality interpolation calculations.

[0067] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0068] (1) Assign an accuracy weight coefficient to the data of each monitoring device, where the weight of fixed air monitoring station data is 1, the weight of micro-environmental station data is 0.6, and the weight of portable sensor data is 0.3;

[0069] (2) Calculate the environmental similarity between the grid cell to be interpolated and the known monitoring points, and obtain the environmental similarity weight by calculating the vector cosine similarity of terrain, land use, road, traffic, population and pollution source characteristics;

[0070] (3) Calculate the actual reachable distance between the grid cell to be interpolated and the monitoring point based on the road network data to form a spatial distance weight;

[0071] (4) For each grid cell to be interpolated, extract the data of all monitoring points within a radius of 5 kilometers;

[0072] (5) When the grid cell to be interpolated is located downwind of the monitoring point, the influence weight of the monitoring point is enhanced according to the wind field data;

[0073] (6) Normalize the accuracy weight coefficient, environmental similarity weight, and spatial distance weight and multiply them together to obtain the comprehensive weight coefficient;

[0074] (7) Using the comprehensive weight coefficient, perform weighted average calculation on the pollutant concentration data of the monitoring points to obtain the pollutant concentration value of the grid cell to be interpolated, and calculate the interpolation uncertainty index;

[0075] (8) Calculate the pollutant concentration value of each grid cell one by one according to the division of urban fine grid and suburban large-scale grid to form the spatial distribution data of pollutant concentration;

[0076] (9) The spatial distribution data of pollutant concentrations are visualized using the contour line method to generate an interpolated distribution map of air quality for the entire region.

[0077] Specifically, each monitoring device's data is assigned an accuracy weighting factor based on its measurement accuracy, calibration frequency, and maintenance level. Fixed air monitoring stations, maintained by professional technicians using instruments that meet national standards, have the highest accuracy and are assigned a weight of 1.0. Micro-environmental stations have the second-highest accuracy and are assigned a weight of 0.6. Portable sensors, due to their low cost and infrequent calibration, have lower accuracy and are assigned a weight of 0.3. This weighting ensures that high-precision devices play a greater role in the interpolation calculations.

[0078] For the calculation of environmental similarity between the grid cell to be interpolated and the known monitoring point, the vector cosine similarity method is used. First, the terrain, land use, road, traffic, population and pollution source characteristics of the grid cell are constructed into a feature vector , and similarly, the features of the grid where the monitoring point is located are formed into a vector , and then calculate the cosine similarity of the two vectors:

[0079] ;

[0080] in Representation Grid With monitoring points The environment similarity ranges from [0,1]. The larger the value, the more similar the environment. Represents the total number of characteristic dimensions, which in this method include average altitude, terrain relief, proportion of various land uses, road density, traffic flow, population density, pollution source intensity and other dimensions.

[0081] The actual reachable distance calculated based on road network data is more consistent with the law of pollutant transmission than the straight-line distance. The road network topology is established, and the road network is regarded as a weighted directed graph, where the weight of the edge is the road length. Then, the Dijkstra shortest path algorithm is used to calculate the shortest path length from the center point of the interpolated grid to the monitoring point as the actual reachable distance. Spatial distance weight Calculated by distance decay function:

[0082] ;

[0083] in and is the attenuation parameter, usually taken as =0.5, =2, that is, inverse square attenuation is adopted.

[0084] For each grid cell to be interpolated, it is not necessary to consider all monitoring point data. Instead, the data of monitoring points within a radius of 5 kilometers are extracted for calculation. This range is set based on the balance between the diffusion characteristics of pollutants and calculation efficiency. In specific implementation, the straight-line distance between the grid center point and all monitoring points is calculated, and the point set with a distance less than 5 kilometers is selected. :

[0085] ;

[0086] in Represents the Euclidean distance calculation function.

[0087] Wind direction has a significant impact on pollutant transmission. When the grid cell to be interpolated is located downwind of the monitoring point, the pollutant concentration at the monitoring point has a greater impact on the grid. Calculate the wind direction correction coefficient based on the wind field data. :

[0088] ;

[0089] in From the monitoring point Pointing Grid The azimuth of is the current wind direction angle, is the wind direction enhancement coefficient, which is set to 0.5. When the grid is located directly downwind of the monitoring point, , the maximum wind direction correction coefficient is 1.5; when the grid is located directly upwind of the monitoring point, the minimum correction coefficient is 0.5.

[0090] The precision weight coefficient, environmental similarity weight, spatial distance weight and wind direction correction coefficient are normalized and multiplied to obtain the comprehensive weight coefficient. :

[0091] ;

[0092] The pollutant concentration data of the monitoring points are weighted averaged using the comprehensive weight coefficient to obtain the pollutant concentration value of the grid unit to be interpolated. :

[0093] ;

[0094] in Indicates monitoring point Measured pollutant concentration values. Interpolation uncertainty index is also calculated , reflecting the credibility of the interpolation result:

[0095] ;

[0096] in Indicates the number of monitoring points involved in the calculation. The value range is [0,1], and larger values ​​indicate higher uncertainty.

[0097] Following the division of fine urban grids and large-scale suburban grids, pollutant concentrations were calculated for each grid cell, forming a spatial distribution matrix C of pollutant concentrations. Finally, the spatial distribution of pollutant concentrations was visualized using the contouring method to generate an interpolated air quality distribution map for the entire region. This contouring method uses a triangular irregular network (TIN) to construct a continuous surface, then extracts contour lines at specific concentrations to create a visually clear pollutant concentration distribution map.

[0098] For example, a city has eight fixed air monitoring stations, 15 micro-stations, and 50 portable sensors, with data collected at 10:00 AM. Interpolation calculations are performed on a grid cell located in a commercial district. First, monitoring equipment within a 5-kilometer radius is extracted, including two fixed stations, four micro-stations, and 12 portable sensors. When calculating environmental similarity, the eigenvector for this grid cell is [45, 2, 0.6, 0.3, 0.1, 0, 0, 0, 11.2, 6, 2000, 0.6, 20000, 32000, 2, 20], indicating an average altitude of 45 meters, a terrain relief of 2 meters, and a commercial land area of ​​60%. The calculated environmental similarity with the nearest fixed station is 0.92, indicating similar environmental conditions. When calculating the spatial distance weight, the actual road distance from this grid cell to the nearest fixed station is 2.3 kilometers, resulting in a spatial distance weight of 0.32. The wind direction on that day was northwesterly, and the grid was located southeast of one of the fixed stations, downwind, with a wind direction correction factor of 1.35. A comprehensive calculation determined that the fixed station's combined weight for the grid was 0.28. Similar calculations were performed on all surrounding monitoring points and normalized, resulting in a weighted average PM2.5 concentration of 85 micrograms per cubic meter for this grid. The interpolation uncertainty index was 0.15, indicating a relatively reliable result. Similar calculations were performed on all grids across the entire region, generating a PM2.5 concentration map covering the entire city. This map clearly shows high-concentration areas around the commercial center and major traffic arteries, as well as low-concentration areas in the suburbs.

[0099] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0100] (1) Extract the interpolation distribution map of the air quality of the target area over the past three years and build a database of pollutant concentrations over the same period;

[0101] (2) Obtain data on wind direction, wind speed, temperature, humidity, air pressure, and precipitation probability for the next 72 hours from the meteorological department to form a weather forecast data set;

[0102] (3) Perform time series decomposition on the interpolated distribution map of air quality across the entire region to separate the long-term trend and periodic variation patterns of pollutant concentrations;

[0103] (4) Based on the historical pollutant concentration database, determine the seasonal variation pattern of pollutant concentration and generate a basic prediction curve;

[0104] (5) Using the meteorological forecast data set, adjust the basic prediction curve to form the meteorological factor correction value;

[0105] (6) Considering the impact of pollution conditions in the upwind area, the meteorological factor correction value is spatially transmitted to generate spatiotemporal dynamic prediction results.

[0106] Specifically, an interpolated distribution map of regional air quality is extracted for the target area over the past three years to construct a historical pollutant concentration database. Using this interpolated distribution map as the base data, data from the same period is traced back three years in time. The interpolated distribution map of regional air quality is generated by weighting the air quality data in the environmental characteristic grid database based on data accuracy, environmental similarity, and spatial distance. For example, when extracting pollutant data for a city in July each year from 2022 to 2024, the concentration values ​​of multiple pollutants for the same period are extracted for each grid cell (500m x 500m in urban areas and 2km x 2km in suburban areas), forming a multidimensional database containing time stamps, spatial coordinates, and concentrations of multiple pollutants.

[0107] Wind direction, wind speed, temperature, humidity, air pressure, and precipitation probability data for the next 72 hours are obtained from the meteorological department to form a weather forecast dataset. This weather data is stored in a spatial grid, with a spatial index consistent with the environmental feature grid. Each grid cell contains hourly weather forecast data for the next 72 hours. Wind direction data is expressed as an angle (0-359 degrees), wind speed in meters per second, temperature in degrees Celsius, humidity in percentage, air pressure in hectopascals, and precipitation probability in percentage. This data is matched with air quality data to provide input variables for subsequent prediction models.

[0108] Time series decomposition of the interpolated regional air quality distribution map was performed to isolate the long-term trend and cyclical variation patterns of pollutant concentrations. This decomposition employed a classic additive model, breaking down the original time series data into three components: a trend term, a seasonal term, and a residual term. For each grid cell's pollutant concentration data, a moving average method was used to extract the long-term trend, and a Fourier transform was used to identify 24-hour (diurnal), 168-hour (weekly), and seasonal cycles. This decomposition effectively identifies the underlying patterns of pollutant concentration variation, providing a foundation for forecasting.

[0109] Based on the historical pollutant concentration database, seasonal variations in pollutant concentrations are determined. Generating a basic forecast curve involves analyzing seasonal patterns in historical data for the same period, calculating daily and weekly averages and standard deviations, and constructing a forecast curve based on these patterns. Specifically, for each grid cell, hourly pollutant concentration data for the same period over the past three years is extracted, and the average and range of variation for the same period are calculated to form the basic forecast curve. This curve reflects the natural variation in pollutant concentrations, excluding the influence of unusual weather conditions.

[0110] Using weather forecast datasets, the basic forecast curve is adjusted to generate meteorological corrections. This step, based on correlation analysis between meteorological factors and pollutant concentrations, constructs a regression model to quantify the impact of meteorological conditions on pollutant concentrations. The model considers the effects of temperature on photochemical reaction rates, humidity on particulate matter deposition, wind speed on pollutant dispersion, and the influence of pressure systems on regional transport. Using multivariate linear regression or random forest algorithms, correction coefficients based on future meteorological conditions are calculated to adjust the basic forecast curve.

[0111] Taking into account the impact of pollution conditions in the upwind area, the meteorological factor correction value is spatially transmitted to generate a spatiotemporal dynamic prediction result. This step introduces the spatial lag effect driven by the wind field. According to the wind direction and speed data, the pollutant transmission path and transmission time are determined, and the contribution of pollutants in the upwind area to the target grid unit is calculated. The specific implementation is to quantify the influence relationship between different grid cells by constructing a spatial weight matrix. For example, when the dominant wind direction is northwest wind and the wind speed is 3 meters per second, the pollutants in the grid cell located 10 kilometers northwest will take about 1 hour to be transmitted to the target area. At this time, the current pollution concentration of the grid cell will be used as the pollution contribution factor of the target grid cell 1 hour later. Through the data processing of the above six steps, it is possible to accurately predict the pollutant concentration of each grid cell in the next 72 hours. For example, when applying this method, the PM2.5 forecast results for a certain city showed that based on the analysis of historical data for the same period, the predicted value of the PM2.5 basic concentration in the first week of July was 35 micrograms / cubic meter. Considering that the weather forecast for the next 72 hours showed that there would be a cooling and humidifying process, the meteorological factor correction reduced the predicted value to 28 micrograms / cubic meter. Further considering the planned emission activities in the upwind industrial areas, the spatial transmission correction increased the predicted value for the specific area to 42 micrograms / cubic meter.

[0112] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0113] (1) Obtain wind direction and wind speed data for the target area within 72 hours, divide the wind direction into 16 sectors, and the wind speed into 6 levels, and construct a wind field data grid;

[0114] (2) Extract the location information, types of pollutants emitted, and their hourly average emission intensity of industrial parks, high-density traffic areas, heating areas, and surrounding urban transmission areas from the pollution source list of the environmental protection department;

[0115] (3) Based on the wind field data grid, the Lagrangian particle method is used to treat each grid cell as the starting point, and the pollutant particles are tracked in reverse according to the wind direction. The position of the pollutant particles is recorded every hour to form a 72-hour pollutant transmission path map;

[0116] (4) Perform spatial overlay analysis on the pollutant transmission path map, count the number of times each source area is crossed by the particle path, and divide it by the total number of particles to obtain the source area impact frequency value;

[0117] (5) Classify and label the source areas according to industrial sources, transportation sources, domestic sources, and regional transmission sources, and sum the source area impact frequency values ​​within each source area to obtain the comprehensive impact frequency of each source area;

[0118] (6) Multiply the comprehensive impact frequency of each source area by its emission intensity, and set the attenuation coefficient according to the distance between the source area and the target area to obtain the pollution contribution value of each source area to the target area.

[0119] Specifically, the implementation of multi-scenario pollution source tracing and contribution rate analysis begins with obtaining wind direction and speed data for the target area within 72 hours. The wind direction is divided into 16 sectors and the wind speed is divided into six levels to construct a wind field data grid. Specifically, the wind direction data obtained from the meteorological department is divided according to the angle value, and the 360-degree angle is evenly divided into 16 sectors, each covering a range of 22.5 degrees. For example, the first sector is 348.75°-11.25°, the second sector is 11.25°-33.75°, and so on. The wind speed is divided into six levels based on the actual distribution: 0-1.5m / s (no wind or light breeze), 1.5-3.3m / s (light breeze), 3.3-5.4m / s (gentle breeze), 5.4-7.9m / s (strong wind), 7.9-10.7m / s (strong wind), and above 10.7m / s (strong wind or storm). Each spatial grid unit has a corresponding wind direction sector number and wind speed level number at each time point, forming a four-dimensional wind field data grid (three dimensions in space and one dimension in time).

[0120] Next, we extracted the location information, pollutant types, and hourly average emission intensities for industrial parks, high-density traffic areas, heating areas, and peripheral urban transmission areas from the environmental protection department's pollution source list. Industrial parks refer to areas with concentrated industrial enterprises; high-density traffic areas are defined as arterial roads and transportation hubs with heavy traffic; heating areas are defined as areas with centralized heating or decentralized coal-fired or gas-fired heating during the winter heating season; and peripheral urban transmission areas are defined as urban built-up areas in the direction of surrounding cities where pollutants could be transported to the target area. For each pollution source area, we recorded its center coordinates and shape (polygon or circle), along with the main pollutant types emitted and their hourly average emission intensities (in kilograms per hour).

[0121] Based on a grid of wind field data, the Lagrangian particle method is used to track pollutant particles in reverse according to wind direction, with each grid cell as the starting point. The particle positions are recorded every hour, creating a 72-hour pollutant transport map. The Lagrangian particle method is a numerical method for simulating atmospheric pollutant transport. Its basic principle is to release a large number of virtual particles into the air and track their trajectories as they move with the airflow. In this method, a certain number (e.g., 100) of virtual particles are released into each grid cell. The particle's movement direction and speed are then determined based on wind field data. Because the tracking is reversed, the particle's movement direction is opposite to the wind direction, and its speed is proportional to the wind speed. The specific algorithm is as follows: For each particle, the hourly displacement vector under the influence of the wind field is calculated based on the wind direction and speed of the current grid cell. The particle's position is then updated and recorded. This process is repeated 72 times, recording the complete trajectory of each particle over 72 hours to form a pollutant transport map for the entire region.

[0122] A spatial overlay analysis is performed on the pollutant transmission path map. The number of times each source region is crossed by a particle path is counted and divided by the total number of particles to obtain the source region impact frequency value. Spatial overlay analysis involves overlaying the pollutant particle trajectory data with the source region spatial distribution data and calculating the intersection of each particle trajectory with each source region. When a particle trajectory passes through a source region, the number of times that source region is crossed is increased by 1. After the statistics are completed, the number of times each source region is crossed is divided by the total number of particles (the number of grid cells multiplied by the number of particles released per cell) to obtain the source region impact frequency value, which reflects the potential impact of the source region on the air quality of the target area.

[0123] Source areas are categorized and labeled as industrial, transportation, household, and regional transmission sources. The impact frequency values ​​within each source area are summed to obtain the combined impact frequency of each source area. Industrial sources include industrial parks and factories; transportation sources include major roads, highways, and transportation hubs; household sources include residential areas and heating areas; and regional transmission sources include surrounding cities and areas with long-distance transmission. The combined impact frequency of source areas within the same type is summed to obtain the combined impact frequency, reflecting the relative contribution of different pollution source types to the air quality of the target area.

[0124] The pollution contribution of each source area to the target area is calculated by multiplying the combined impact frequency of each source area by its emission intensity and setting an attenuation coefficient based on the source area's distance from the target area. Emission intensity refers to the amount of pollutants emitted from the source area, measured in kilograms per hour. The attenuation coefficient reflects the decrease in pollutant concentration with increasing transmission distance and can be set as the inverse of distance or an exponential decay function. By combining the three factors of impact frequency, emission intensity, and attenuation coefficient, the pollution contribution of each source area to the target area is calculated, providing a quantitative basis for scientific pollution control.

[0125] The data processing process of this method is illustrated using a city-based PM2.5 source apportionment as an example. First, wind direction and speed data for the city and surrounding areas over a 72-hour period were obtained. The data was divided into 16 wind direction sectors and six wind speed levels, and a wind field data grid was constructed. The location and emission data for four industrial parks, three high-density traffic areas, two large residential heating areas, and two surrounding cities were extracted from a pollution source list obtained from the environmental protection department. The average PM2.5 emission intensity for each industrial park was 5 kg / hour, 3 kg / hour for high-density traffic areas, 2 kg / hour for heating areas, and 8 kg / hour for surrounding cities. Using the Lagrangian particle method, 100 virtual particles were released at the city center and traced back to their sources based on the 72-hour wind field data. The tracking results showed that within 72 hours, 35 of the 100 particles passed through the northern industrial park, 25 through the eastern traffic artery, 15 through the southern heating area, and 20 through the surrounding cities in the west. The trajectories of five particles did not intersect any known source areas. The calculated frequency of industrial source impact is 0.35, the frequency of traffic source impact is 0.25, the frequency of household source impact is 0.15, and the frequency of regional transmission source impact is 0.20. Considering that the northern industrial park is 15 kilometers from the city center, an attenuation coefficient of 0.8 is assumed; the eastern traffic artery is 10 kilometers away, with an attenuation coefficient of 0.9; the southern heating area is 5 kilometers away, with an attenuation coefficient of 0.95; and the western surrounding city is 30 kilometers away, with an attenuation coefficient of 0.6. The calculated contribution rate of industrial sources is 0.35 × 5 × 0.8 = 1.4, the contribution rate of traffic sources is 0.25 × 3 × 0.9 = 0.675, the contribution rate of household sources is 0.15 × 2 × 0.95 = 0.285, and the contribution rate of regional transmission sources is 0.20 × 8 × 0.6 = 0.96. According to the order of contribution rate, the pollutant sources are industrial, regional transmission, traffic, and household.

[0126] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0127] (1) Obtain the resident population density and floating population density of the target area from the census database and construct a population distribution data map;

[0128] (2) Based on social survey data, the activity patterns of urban residents are extracted, and the day is divided into commuting peak hours, working hours, leisure hours, and night rest hours to form a time-based activity feature database;

[0129] (3) For the respiratory disease patient data received from medical institutions, mark the patient's residential area and generate a sensitive population distribution map;

[0130] (4) Calculate the population exposure of each grid unit at different time periods by combining the pollution contribution value, population distribution data map, time period activity feature database, and sensitive population distribution map;

[0131] (5) Set health risk thresholds based on population exposure and divide risk values ​​into four health risk levels: low risk, medium risk, high risk, and extremely high risk;

[0132] (6) Based on the health risk level, generate graded warning information and control measures recommendations, including warning period, impact range, protective measures and control measures recommendations for major pollution source areas.

[0133] Specifically, the resident population density and floating population density of the target area are obtained from the census database to construct a population distribution data map. The resident population density refers to the number of people who live permanently within a square kilometer, while the floating population density refers to the number of people who live non-permanently but are frequently active within a square kilometer. Census data usually contain demographic information down to the street or community level, which needs to be further processed to the same grid scale as the air quality assessment. The specific processing method is to distribute the population data at the street or community level to overlapping grid cells according to the area ratio, forming a population distribution data map with the same spatial resolution as the pollutant concentration distribution map. The urban area uses a fine grid of 500 meters × 500 meters, and each grid cell records the number of permanent residents and floating population; the suburban area uses a larger grid of 2 kilometers × 2 kilometers, and also records two types of population data.

[0134] Based on social survey data, the activity patterns of urban residents are extracted and divided into peak commuting hours, working hours, leisure hours, and nighttime rest periods, creating a database of time-based activity characteristics. Peak commuting hours typically refer to the morning hours of 7:00-9:00 and the evening hours of 17:00-19:00, with residents primarily confined to major traffic arteries and public transportation hubs. Working hours are from 9:00-17:00, with residents primarily confined to workplaces (such as offices, industrial areas, and commercial districts) on weekdays and to commercial and leisure areas on non-working days. Leisure hours are from 19:00-23:00, with residents primarily confined to residential areas, commercial areas, and leisure venues. Nighttime rest periods are from 23:00-7:00, with residents primarily confined to residential areas. Based on this temporal categorization, an activity intensity coefficient is assigned to each grid cell during different time periods, reflecting the percentage of residents active within that grid cell during that specific time period. For example, commercial areas experience higher activity intensity during working hours and leisure hours, while residential areas experience higher activity intensity around peak commuting hours and during nighttime rest periods.

[0135] For the data on patients with respiratory diseases received from medical institutions, the patient's residential area is marked to generate a distribution map of sensitive populations. Patients with respiratory diseases include those with asthma, chronic bronchitis, emphysema, cor pulmonale, and pulmonary fibrosis, and these people are particularly sensitive to air pollutants. The patient data provided by medical institutions contains information about the patient's residential address or community. The address information is converted into geographic coordinates through geocoding technology and then aggregated into grid cells. The number of sensitive people in each grid cell and their proportion of the total population are calculated. In addition to patients with respiratory diseases, people over 65 and children under 14 are also marked as sensitive populations. The generated distribution map of sensitive populations is based on grids, recording the number and proportion of sensitive people in each grid cell.

[0136] The population exposure of each grid unit at different time periods is calculated by combining pollution contribution values, population distribution data maps, time period activity feature libraries, and sensitive population distribution maps. Population exposure is a comprehensive function of air pollutant concentration, population size, activity intensity, and sensitivity. For each grid unit in each time period, the predicted pollutant concentration value of the grid is first obtained, then multiplied by the population size of the grid and the activity intensity coefficient of the time period, and then weighted according to the proportion of sensitive population to calculate the population exposure. The weighting factor for sensitive populations is usually set to 1.5-3 times, reflecting their higher sensitivity to pollutants. This calculation method takes into account the spatiotemporal dynamic characteristics of air quality, population distribution, and activity patterns, and can more accurately reflect the actual impact of pollutants on human health.

[0137] Health risk thresholds are set based on population exposure, and risk values ​​are divided into four health risk levels: low risk, medium risk, high risk, and extremely high risk. The setting of health risk thresholds is based on epidemiological research and health standards, taking into account the health impact characteristics of different pollutants and the exposure-response relationship curve. For example, for PM2.5, risk thresholds can be set based on the World Health Organization and national ambient air quality standards, combined with local population characteristics. When the population exposure is below the first threshold, it is low risk; when it is above the first threshold but below the second threshold, it is medium risk; when it is above the second threshold but below the third threshold, it is high risk; and when it is above the third threshold, it is extremely high risk. Different risk levels correspond to different health impact severity and prevention and control urgency.

[0138] The final step is to generate graded warning information and control measures based on the health risk level, including warning period, impact area, protective measures, and recommended control measures for major pollution source areas. The warning period is the duration of the health risk exceeding a certain level; the impact area is the spatial area where the health risk exceeds a certain level; protective measures are health protection recommendations for the general public and sensitive groups; and control measures are pollution reduction measures for major pollution source areas. At the low risk level, information dissemination and protection recommendations for sensitive groups are primarily provided; at the medium risk level, public protection recommendations and recommendations for controlling some pollution sources are added; at the high risk level, strict public protection measures and emission limits for major pollution sources are implemented; at the extremely high risk level, the emergency plan is activated, and the most stringent regional control measures are implemented.

[0139] The complete process of this method is illustrated using a city's winter heavy pollution weather warning as an example. First, population density data for each area of ​​the city was obtained from a census database. The average density in the city center is 20,000 people / square kilometer, and in the outer suburbs, it is 2,000 people / square kilometer. Based on social survey data, the activity intensity of grid cells on urban arterial roads during peak commuting hours (7:00-9:00 and 17:00-19:00) was determined to be 1.8. The activity intensity of grid cells in commercial and industrial areas during working hours (9:00-17:00) was 1.5. The activity intensity of grid cells in commercial and residential areas during leisure hours (19:00-23:00) was 1.3. The activity intensity of residential areas during the nighttime hours (23:00-7:00) was 1.0. Data from medical institutions showed that approximately 8% of the city's population suffers from respiratory diseases, primarily concentrated around industrial areas and in the old city. Combined with the pollution contribution values ​​determined in the previous step, it is estimated that PM2.5 concentrations in the northern part of the city will reach 150 μg / m³ within the next 48 hours, overlapping with high-density residential areas and areas with high concentrations of sensitive populations. Calculations indicate that during peak commuting hours, the exposure of a grid cell in the northern region will reach 2.5 times the risk assessment standard, placing it at a high risk level. Based on this, an early warning message is generated: the warning period is tomorrow from 7:00 AM to 7:00 PM, affecting six streets in the northern part of the city. Sensitive populations are advised to stay indoors and use air purification equipment, and the general public is advised to limit outdoor activities. Temporary control measures will be implemented in the northern industrial zone, limiting production by 30%, and temporary traffic restrictions will be implemented for vehicles entering and leaving the area.

[0140] The above describes the air quality assessment method based on multi-source data in the embodiment of the present application. The following describes the air quality assessment system based on multi-source data in the embodiment of the present application. Figure 2 In one embodiment of the present application, an air quality assessment system based on multi-source data includes:

[0141] The correction module is used to collect pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors, and perform outlier identification and correction on the data to obtain a multi-source air quality raw data set;

[0142] a partitioning module for dividing the target area into fine urban grids and large suburban grids based on the multi-source air quality raw data set, and forming an environmental feature grid database by combining topography, land use, road, traffic, population, and pollution source information;

[0143] A weighting module is used to perform weighted calculation on the air quality data in the environmental feature grid database according to data accuracy, environmental similarity and spatial distance to generate an interpolation distribution map of air quality for the entire region;

[0144] A prediction module is used to calculate the predicted value of pollutant concentration in the future period based on the interpolated distribution map of air quality in the entire region, combined with historical data of the same period and meteorological forecast information, to form a spatiotemporal dynamic prediction result;

[0145] A tracking module is used to use the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, to track the pollutant transmission path and determine the pollution contribution of various source areas to the target area;

[0146] The control module is used to calculate the health risk level of different areas based on the pollution contribution value, combined with population distribution and activity patterns, and generate graded warning information and control measure recommendations.

[0147] Through the collaborative cooperation of the above components, by collecting multi-source pollutant concentration data from fixed air monitoring stations, micro-environmental stations and portable sensors, and performing outlier identification and correction processing, the limitations of the traditional single data source have been broken through, the integrity and reliability of the data have been significantly improved, and a solid data foundation has been laid for air quality assessment; combining terrain, land use, roads, traffic, population and pollution source information to divide the target area into fine grids in urban areas and large-scale grids in suburban areas to form an environmental feature grid database, achieving high-resolution spatial expression of environmental factors and fully considering the heterogeneous characteristics of the regional environment; by weighted calculation according to data accuracy, environmental similarity and spatial distance, an interpolation distribution map of air quality for the entire region is generated, solving the problem of uneven spatial distribution of monitoring points. It achieved accurate reconstruction of the pollutant concentration field; based on the interpolation distribution map of air quality in the entire region, combined with historical data for the same period and meteorological forecast information, the predicted pollutant concentration values ​​for future time periods were calculated, realizing dynamic prediction of the spatiotemporal distribution of pollutants, providing a time window for early prevention and control; using the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, the pollutant transmission path was tracked, and the pollution contribution value of various source areas to the target area was determined, realizing the quantitative and scientific tracing of pollution sources, and providing direction for precise pollution control; based on the pollution contribution value, combined with population distribution and activity patterns, the health risk levels of different regions were calculated, and graded warning information and control measures were generated, realizing a closed loop from air quality assessment to health risk management, and providing precise support for public health protection and pollution prevention and control decisions. It is particularly worth emphasizing that this solution utilizes artificial intelligence algorithms in multiple stages. For example, the improved geographically weighted regression algorithm used in the spatial fusion and interpolation of multi-source data can adaptively learn the complex nonlinear relationship between environmental characteristics and pollutant concentrations; the random forest algorithm used in the spatiotemporal dynamic prediction model can handle the high-dimensional interaction between meteorological factors and pollutant concentrations; the Lagrangian particle model and source contribution matrix algorithm used in pollution source analysis can accurately simulate the complex transmission process of pollutants under different meteorological conditions; and the multi-factor risk model used in health risk assessment can comprehensively consider population density, activity patterns, and the distribution characteristics of sensitive populations. The application of these algorithms enables this solution to process large amounts of heterogeneous data, capture complex spatiotemporal patterns and causal relationships, and achieve intelligent and precise air quality assessment and management.

[0148] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3As shown. The computer device includes a processor, memory, display screen, input device, network interface and database connected via a system bus. The processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the above method is implemented.

[0149] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0150] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which implements the above-described method when executed by a processor. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0151] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware using a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media provided herein and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM.

[0152] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0153] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0154] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An air quality assessment method based on multi-source data, characterized in that: The method comprises: Collect pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors, identify and correct outliers on the data, and obtain a multi-source air quality raw data set; Based on the multi-source air quality original dataset, the target area is divided into fine urban grids and large suburban grids, and an environmental characteristic grid database is formed by combining topography, land use, road, traffic, population and pollution source information; The air quality data in the environmental feature grid database are weighted according to data accuracy, environmental similarity and spatial distance to generate an interpolation distribution map of air quality in the entire region, including: assigning an accuracy weight coefficient to the data of each monitoring device, wherein the weight of fixed air monitoring station data is 1, the weight of micro-environmental station data is 0.6, and the weight of portable sensor data is 0.3; calculating the environmental similarity between the grid unit to be interpolated and the known monitoring point, and obtaining the environmental similarity weight by performing vector cosine similarity calculation on the terrain, land use, road, traffic, population and pollution source characteristics; calculating the actual reachable distance between the grid unit to be interpolated and the monitoring point based on the road network data to form a spatial distance weight; for each grid unit to be interpolated, extracting the surrounding The data of all monitoring points within a radius of five kilometers are collected; when the grid unit to be interpolated is located downwind of the monitoring point, the influence weight of the monitoring point is enhanced according to the wind field data; the precision weight coefficient, environmental similarity weight and spatial distance weight are normalized and multiplied together to obtain a comprehensive weight coefficient; the pollutant concentration data of the monitoring points are weighted averaged using the comprehensive weight coefficient to obtain the pollutant concentration value of the grid unit to be interpolated, and the interpolation uncertainty index is calculated; the pollutant concentration value of each grid unit is calculated one by one according to the division of urban fine grids and suburban large-scale grids to form the spatial distribution data of pollutant concentration; the spatial distribution data of pollutant concentration is visualized by the contour line method to generate an interpolation distribution map of air quality in the entire region; According to the interpolation distribution map of air quality in the entire region, combined with historical data for the same period and meteorological forecast information, the predicted values ​​of pollutant concentrations in future time periods are calculated to form a spatiotemporal dynamic prediction result, including: extracting the interpolation distribution map of air quality in the entire region of the target area for the same period in the past three years, and constructing a historical pollutant concentration database for the same period; obtaining wind direction, wind speed, temperature, humidity, air pressure, and precipitation probability data for the next seventy-two hours from the meteorological department to form a meteorological forecast data set; performing time series decomposition on the interpolation distribution map of air quality in the entire region to separate the long-term trend and periodic variation pattern of pollutant concentrations; determining the seasonal variation pattern of pollutant concentrations based on the historical pollutant concentration database for the same period to generate a basic prediction curve; using the meteorological forecast data set, adjusting the basic prediction curve to form a meteorological factor correction value; considering the influence of pollution conditions in the upwind area, performing spatial transmission correction on the meteorological factor correction value to generate a spatiotemporal dynamic prediction result; Using the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, the pollutant transmission path is tracked to determine the pollution contribution of various source areas to the target area; Based on the pollution contribution value, combined with population distribution and activity patterns, the health risk levels of different areas are calculated, and graded warning information and control measures recommendations are generated.

2. The air quality assessment method based on multi-source data according to claim 1, characterized in that: The pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors are collected, and outlier identification and correction processing are performed on the data to obtain a multi-source air quality original data set, including: Obtain concentration data of fine particulate matter, inhalable particulate matter, sulfur dioxide, nitrogen dioxide, ozone, carbon monoxide and equipment geographic coordinate information from monitoring equipment of different accuracy levels; Establish a monitoring equipment accuracy rating system, classifying fixed air monitoring stations as high-precision equipment, micro-environmental stations as medium-precision equipment, and portable sensors as low-precision equipment; Calculate the deviation of the collected pollutant concentration data from the seasonal historical average. When the deviation exceeds three times the standard deviation, it is marked as an abnormal data point in the time dimension. Calculate the percentage deviation between each monitoring point and the average value of the five adjacent monitoring points. When the percentage exceeds 75%, it is marked as an abnormal data point in the spatial dimension. Remove the data points that are marked as both time dimension anomalies and space dimension anomalies, and retain the remaining data points; Cross-validate the data of adjacently placed equipment with different accuracy levels and establish data conversion equations between equipment with different accuracy levels; The missing data of time series are filled by linear interpolation method, and the missing data of spatial distribution are filled by inverse distance weighted interpolation method to form a multi-source air quality original data set.

3. The air quality assessment method based on multi-source data according to claim 1, characterized in that: Based on the multi-source air quality original dataset, the target area is divided into fine urban grids and large suburban grids, and combined with terrain, land use, road, traffic, population and pollution source information to form an environmental feature grid database, including: Divide the urban built-up area in the target area into fine-scale urban grid cells of 500 meters by 500 meters, and divide the suburban area in the target area into large-scale suburban grid cells of 2,000 meters by 2,000 meters; Extract digital elevation data from the geographic information system, calculate the average altitude and terrain relief of each grid cell, and obtain terrain feature data; Extract the dominant land type and land type composition ratio of each grid unit from the land use classification data to form land use characteristic data; Obtain road network data from the traffic management department, calculate the road length density and the number of road intersections for each grid unit, and generate road feature data; Obtain traffic data from the traffic flow monitoring system, calculate the average vehicle flow and traffic congestion index of each grid unit, and obtain traffic characteristic data; Extract population distribution information from census data, calculate the resident population density and floating population density of each grid cell, and form population characteristic data; Obtain a list of fixed pollution sources from the environmental protection department, mark the number of industrial enterprises and emission levels in each grid unit, and generate pollution source characteristic data; Through spatial overlay analysis, the terrain feature data, land use feature data, road feature data, traffic feature data, population feature data and pollution source feature data are associated with grid units to form an environmental feature grid database.

4. The air quality assessment method based on multi-source data according to claim 1, characterized in that: The spatiotemporal dynamic prediction results are used in combination with wind field data and pollution source distribution to track pollutant transmission paths and determine the pollution contribution of various source areas to the target area, including: Obtain wind direction and speed data for the target area within 72 hours, divide the wind direction into 16 sectors, and the wind speed into six levels, and construct a wind field data grid; Extract the location information, types of pollutants emitted, and their hourly average emission intensity of industrial parks, high-density traffic areas, heating areas, and surrounding urban transmission areas from the pollution source list of the environmental protection department; Based on the wind field data grid, the Lagrangian particle method is used to treat each grid cell as a starting point, and pollutant particles are tracked in reverse according to wind direction. The position of pollutant particles is recorded every hour to form a 72-hour pollutant transmission path map; Performing spatial overlay analysis on the pollutant transmission path map, counting the number of times each source area is crossed by the particle path, and dividing the result by the total number of particles to obtain the source area impact frequency value; The source areas are classified and marked according to industrial sources, transportation sources, domestic sources, and regional transmission sources, and the source area impact frequency values ​​within each type of source area are summed to obtain the comprehensive impact frequency of each type of source area; The comprehensive impact frequency of each type of source area is multiplied by its emission intensity, and the attenuation coefficient is set according to the distance between the source area and the target area to obtain the pollution contribution value of each type of source area to the target area.

5. The air quality assessment method based on multi-source data according to claim 1, characterized in that: The health risk levels of different areas are calculated based on the pollution contribution values, combined with population distribution and activity patterns, to generate graded warning information and control measures, including: Obtain the resident population density and floating population density of the target area from the census database and construct a population distribution data map; Based on social survey data, we extract the activity patterns of urban residents and divide the day into commuting peak hours, working hours, leisure hours, and nighttime rest hours to form a time-based activity feature database. For respiratory disease patient data received from medical institutions, mark the patient's residential area and generate a sensitive population distribution map; Calculate the population exposure of each grid unit at different time periods by combining the pollution contribution value, population distribution data map, time period activity feature library and sensitive population distribution map; Set health risk thresholds based on the population's exposure and categorize risk values ​​into four health risk levels: low risk, medium risk, high risk, and extremely high risk; Based on the health risk level, graded warning information and control measures recommendations are generated, including warning period, impact range, protective measures and control measures recommendations for major pollution source areas.

6. An air quality assessment system based on multi-source data, for implementing the air quality assessment method based on multi-source data according to any one of claims 1 to 5, characterized in that: The air quality assessment system based on multi-source data includes: The correction module is used to collect pollutant concentration data from fixed air monitoring stations, micro-environmental stations, and portable sensors, and perform outlier identification and correction on the data to obtain a multi-source air quality raw data set; a partitioning module for dividing the target area into fine urban grids and large suburban grids based on the multi-source air quality raw data set, and forming an environmental feature grid database by combining topography, land use, road, traffic, population, and pollution source information; A weighting module is used to perform weighted calculation on the air quality data in the environmental feature grid database according to data accuracy, environmental similarity and spatial distance to generate an interpolation distribution map of air quality for the entire region; A prediction module is used to calculate the predicted value of pollutant concentration in the future period based on the interpolated distribution map of air quality in the entire region, combined with historical data of the same period and meteorological forecast information, to form a spatiotemporal dynamic prediction result; A tracking module is used to use the spatiotemporal dynamic prediction results, combined with wind field data and pollution source distribution, to track the pollutant transmission path and determine the pollution contribution of various source areas to the target area; The control module is used to calculate the health risk level of different areas based on the pollution contribution value, combined with population distribution and activity patterns, and generate graded warning information and control measure recommendations.

7. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, the air quality assessment method based on multi-source data according to any one of claims 1 to 5 is implemented. 8 . A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor is enabled to perform the air quality assessment method based on multi-source data according to any one of claims 1 to 5 .

Images