A method for estimating daily atmospheric pollutant concentrations at a fine scale in cities

Abstract

The present invention relates to a method for estimating daily atmospheric pollutant concentrations at a fine scale in a city, including: obtaining multi-source data in the urban area and performing multi-scale spatio-temporal matching processing; constructing a global estimation model for atmospheric pollutants and training it; obtaining the annual global atmospheric pollutant concentrations at the fine scale in the urban area and converting them into global atmospheric pollutant concentration weights; calculating the daily local atmospheric pollutant concentration estimation residuals; and estimating the daily atmospheric pollutant concentrations at the fine scale in the city. The beneficial effects of the present invention are as follows: The present invention comprehensively considers global characteristics and local changes, effectively improves the spatial resolution of existing data products within the city, obtains a high-precision daily atmospheric pollutant data product at the hundred-meter level, can more accurately locate polluted areas, and thus more scientifically guide the formulation and implementation of precise prevention and control measures for urban air pollution, having important application value.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental monitoring, and more precisely, it relates to a method for estimating the daily atmospheric pollutant concentration at a fine scale in a city. Background Art

[0002] Dense human activities in cities lead to the emission of a large amount of pollutants, resulting in a significant decline in the atmospheric environmental quality. A large number of studies have shown that atmospheric pollutants, especially fine particulate matter PM2.5, will have various adverse effects on human health. In order to achieve the precise prevention and control of atmospheric pollutants in urban areas, timely and accurate understanding of the spatio-temporal distribution of atmospheric pollutants within the city is the key to solving the problem.

[0003] At present, China has established a relatively complete ground air quality monitoring network, which can monitor the air quality of key urban areas in real time. However, the spatial distribution of limited ground monitoring stations is still relatively sparse compared to the entire urban area, and it is difficult to reflect the distribution of atmospheric pollutants in the whole region. With the development of artificial intelligence and satellite remote sensing technology, a large number of studies have jointly estimated the distribution of atmospheric pollutants on a large scale by combining ground station monitoring and satellite remote sensing technology, making up for the deficiencies of ground station monitoring. However, due to satellite hardware limitations, the current spatial resolution of satellite observations is relatively low, only reaching the kilometer scale, and it is difficult to reflect the fine distribution of atmospheric pollutants within the city. Therefore, there is an urgent need for a method for estimating the concentration distribution of atmospheric pollutants at a fine scale in the city to meet the actual needs of urban air pollution control. Summary of the Invention

[0004] The purpose of the present invention is to propose a method for estimating the daily atmospheric pollutant concentration at a fine scale in a city in view of the deficiencies of the prior art.

[0005] In a first aspect, a method for estimating the daily atmospheric pollutant concentration at a fine scale in a city is provided, including:

[0006] Step 1: Obtain multi-source data of the urban area, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data, and ground station monitoring data;

[0007] Step 2: Perform multi-scale spatio-temporal matching processing on the multi-source data of the urban area to obtain a processed multi-scale spatio-temporal data set; the multi-scale spatio-temporal data set includes a low-resolution scale spatio-temporal data set and a fine-scale spatio-temporal data set;

[0008] Step 3: Based on the low-resolution scale spatio-temporal data set, construct a global estimation model for atmospheric pollutants and train it, and iteratively optimize the model parameters;

[0009] Step 4: Using the trained global estimation model and combining with the fine-scale spatio-temporal dataset, obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area, and convert it into the global atmospheric pollutant concentration weight;

[0010] Step 5: Based on the daily satellite remote sensing inversion data at the low-resolution scale, calculate the daily local atmospheric pollutant concentration estimation residual;

[0011] Step 6: Combining the fine-scale spatio-temporal dataset, the global atmospheric pollutant concentration weight, and the daily local atmospheric pollutant concentration estimation residual, estimate the daily atmospheric pollutant concentration at the fine scale of the city.

[0012] Preferably, the satellite remote sensing inversion data is the daily low-resolution atmospheric pollutant fine particulate matter PM 2.5 remote sensing inversion data; the land use data is the annual classification data at the fine scale, covering water bodies, trees, flooded vegetation, crops, built-up areas, bare land, and pasture types; the ground digital elevation model data is the raster data at the fine scale; the road network data is the annual line vector data, and the road network includes highways, main lines, first-class highways, second-class highways, and third-class highways; the ground station monitoring data is the hourly real-time monitoring data of the national ground air quality monitoring network

[0013] Preferably, step 2 includes:

[0014] Step 2.1: Taking the satellite remote sensing inversion data as the benchmark, construct a low-resolution scale standard grid. Under the unified spatial coordinate system, count the proportion of the area occupied by each land use type in the low-resolution scale standard grid, the mean value of the ground digital elevation model data, and the total length of each road, and at the same time calculate the annual average value of the atmospheric fine particulate matter PM 2.5 remote sensing inversion data to obtain the low-resolution scale spatio-temporal dataset;

[0015] Step 2.2: Construct a fine-scale standard grid under the unified spatial range and coordinate system, count the proportion of the area occupied by each land use type in the fine-scale standard grid, the mean value of the ground digital elevation model data, and the total length of each road, and use the nearest neighbor resampling method for the low-resolution daily PM 2.5 remote sensing inversion data to obtain the fine-scale dataset of daily PM 2.5 ;

[0016] Preferably, in step 3, the global estimation model of the atmospheric pollutant is a supervised machine learning model, and the supervised machine learning model is any one of the random forest regression model, the Cubist regression model, and the XGBoost regression model; the formula of the supervised machine learning model is expressed as:

[0017] PM2.5 global= F(DEM, Road, LU water , LU veg , LU wet , LU agri , LU build , LU bare , LU farm )

[0018] In the above formula, PM2.5 global represents the global estimation result of PM 2.5 , F represents any supervised machine learning model, DEM represents digital terrain elevation, Road represents the total length of various types of roads, LU represents the proportion of each land use type, and the subscripts water, veg, wet, agri, build, bare, and farm represent water body, trees, flooded vegetation, crops, built-up area, bare land, and pasture respectively.

[0019] Preferably, in step 4, the spatio-temporal data set of elevation, road length, and proportion of each land use type at the fine scale constructed in step 2.2 is used as the input of the global estimation model to obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area; and within the range of the low-resolution scale standard grid, the global PM 2.5 concentration weight is calculated and standardized, and its formula is expressed as:

[0020]

[0021] In the above formula, represents the standardized PM 2.5 concentration weight of the i-th fine-scale grid within the range of a low-resolution scale standard grid, is the corresponding global PM 2.5 concentration. After standardized calculation, the average value of all weights within the range of each low-resolution scale standard grid is 1.

[0022] Preferably, step 5 includes:

[0023] Step 5.1: According to the daily satellite remote sensing inversion data at the low-resolution scale, the inverse distance weighting method is used to obtain the spatial distribution of daily PM 2.5 in the urban area at the fine scale;

[0024] Step 5.2: Subtract the spatial distribution of the daily PM 2.5 in the urban area at the fine scale from the daily PM 2.5 data obtained by the nearest neighbor resampling method during the spatio-temporal matching process;

[0025] Step 5.3: Within each low-resolution scale standard grid range, perform standardized calculation on the residuals to obtain the daily local atmospheric pollutant concentration estimation residuals, and its formula is expressed as:

[0026]

[0027] In the above formula, represents the local residual of the PM 2.5 concentration after standardized calculation for the i-th fine-scale grid within a low-resolution scale standard grid range, and respectively represent the fine-scale PM 2.5 data obtained based on the inverse distance weighting method and the nearest neighbor resampling method, and mean and σ represent the mean and standard deviation of the local residuals of the PM 2.5 concentrations for all fine-scale grids within the low-resolution scale standard grid range.

[0028] Preferably, the calculation formula for Step 6 is:

[0029]

[0030] In the above formula, represents the estimated concentration of PM 2.5 for the k-th fine-scale grid in the urban area.

[0031] In a second aspect, a system for estimating the daily atmospheric pollutant concentration at a fine scale in a city is provided, which is used to execute any of the methods in the first aspect, and includes:

[0032] A first acquisition module, which is used to acquire multi-source data in the urban area, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data, and ground station monitoring data;

[0033] A matching module, which is used to perform multi-scale spatio-temporal matching processing on the multi-source data in the urban area to obtain a processed multi-scale spatio-temporal data set; the multi-scale spatio-temporal data set includes a low-resolution scale spatio-temporal data set and a fine-scale spatio-temporal data set;

[0034] A construction module, which is used to construct a global estimation model of atmospheric pollutants based on the low-resolution scale spatio-temporal data set and perform training, and iteratively optimize the model parameters;

[0035] A second acquisition module, which is used to use the trained global estimation model, combine it with the fine-scale spatio-temporal data set, obtain the annual global atmospheric pollutant concentration at a fine scale in the urban area, and convert it into the global atmospheric pollutant concentration weight;

[0036] A calculation module, configured to calculate the daily local atmospheric pollutant concentration estimation residuals based on the daily satellite remote sensing inversion data at a low-resolution scale;

[0037] An estimation module, configured to estimate the daily urban fine-scale atmospheric pollutant concentration by combining the fine-scale spatio-temporal dataset, the global atmospheric pollutant concentration weights, and the daily local atmospheric pollutant concentration estimation residuals.

[0038] In a third aspect, a computer storage medium is provided, in which a computer program is stored; when the computer program runs on a computer, the computer is made to execute the method according to any one of the first aspects.

[0039] In a fourth aspect, an electronic device is provided, including:

[0040] A memory, configured to store the computer program;

[0041] A processor, configured to execute the computer program to implement the method according to any one of the first aspects.

[0042] The beneficial effects of the present invention are as follows: First, based on the low-resolution scale spatio-temporal dataset, the present invention constructs a global estimation model of atmospheric pollutants, and based on the scale invariance hypothesis, obtains the global atmospheric pollutant concentration trend distribution weights at the urban fine scale. Further considering the daily local variation of atmospheric pollutants, the daily atmospheric pollutant estimation residuals are calculated, and the daily atmospheric pollutant concentration calculated by the global trend weights is corrected by using the residuals, and finally the daily urban fine-scale atmospheric pollutant concentration is obtained. Compared with the existing kilometer-level atmospheric pollutant products, the present invention comprehensively considers the global characteristics and local variations, effectively improves the spatial resolution of the existing data products within the city, obtains a high-precision daily atmospheric pollutant data product at the hundred-meter level, can more accurately locate the pollution area, and thus can more scientifically guide the formulation and implementation of precise urban air pollution prevention and control measures, and has important application value. Description of the Drawings

[0043] Figure 1 It is a flow chart of the method for estimating the daily urban fine-scale atmospheric pollutant concentration provided by the embodiment of the present invention;

[0044] Figure 2 It is a scatter plot for verifying the method for estimating the daily urban fine-scale atmospheric pollutant concentration provided by the embodiment of the present invention;

[0045] Figure 3 It is a spatial distribution diagram of the method for estimating the daily urban fine-scale atmospheric pollutant concentration provided by the embodiment of the present invention. Detailed Embodiments

[0046] The present invention will be further described below in conjunction with embodiments. The description of the following embodiments is only for helping to understand the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several modifications can still be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

[0047] Embodiment 1:

[0048] To solve the problems of the prior art, Embodiment 1 of the present application provides a method for estimating the daily atmospheric pollutant concentration at a fine scale in a city, as Figure 1 shown, including:

[0049] Step 1: Obtain multi-source data of the urban area, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data, and ground station monitoring data.

[0050] Specifically, the satellite remote sensing inversion data is the daily atmospheric pollutant fine particulate matter PM 2.5 remote sensing inversion data with a low resolution (1 km); the land use data is the annual classification data at a fine scale (10 m), covering water bodies, trees, flooded vegetation, crops, built-up areas, bare land, and pasture types; the ground digital elevation model data is the raster data at a fine scale (30 m); the road network data is the annual line vector data, and the road network includes expressways, main lines, first-class highways, second-class highways, and third-class highways; the ground station monitoring data is the hourly real-time monitoring data of the national ground air quality monitoring network.

[0051] Step 2: Perform multi-scale spatio-temporal matching processing on the multi-source data of the urban area to obtain a processed multi-scale spatio-temporal data set; the multi-scale spatio-temporal data set includes a low-resolution scale spatio-temporal data set and a fine-scale spatio-temporal data set.

[0052] Step 2 includes:

[0053] Step 2.1: Taking the satellite remote sensing inversion data as a reference, construct a standard grid at a low resolution scale (1 km). Under the unified spatial coordinate system, count the proportion of the area occupied by each land use type, the mean value of the ground digital elevation model data, and the total length of each road in the low-resolution scale standard grid, and at the same time calculate the annual average value of the atmospheric fine particulate matter PM 2.5 remote sensing inversion data to obtain a low-resolution scale spatio-temporal data set;

[0054] Step 2.2: On the basis of the constructed standard grid at the low-resolution scale, further construct a standard grid at the fine scale (100 m) within its unified spatial range and coordinate system, and calculate the proportion of the area occupied by each land use type within the 100-m standard grid (the total number of land use type grids in each 10-m grid within the 100-m standard grid / 100), the mean value of the digital elevation model data on the ground (the mean value of all 30-m digital elevation model grids covered within the 100-m standard grid), and the total length of each road (the sum of the lengths of all roads covered within the 100-m standard grid). For the daily PM at low resolution 2.5 The remotely sensed inversion data uses the nearest neighbor resampling method to obtain the daily PM 2.5 dataset with a resolution of 100 m.

[0055] Step 3: Based on the low-resolution scale spatio-temporal dataset, construct a global estimation model for air pollutants and train it, and iteratively optimize the model parameters.

[0056] In Step 3, the global estimation model for air pollutants is a supervised machine learning model, and the supervised machine learning model is any one of the random forest regression model, the Cubist regression model, and the XGBoost regression model; the formula of the supervised machine learning model is expressed as:

[0057] PM2.5 global = F(DEM, Road, LU water , LU veg , LU wet , LU agri , LU build , LU bare , LU farm )

[0058] In the above formula, PM2.5 global represents the global estimation result of PM 2.5 , F represents any supervised machine learning model, DEM represents the digital ground elevation, Road represents the total length of various roads, LU represents the proportion of each land use type, and the subscripts water, veg, wet, agri, build, bare, and farm represent water body, trees, flooded vegetation, crops, built-up area, bare land, and pasture respectively.

[0059] Step 4: Use the trained global estimation model, combined with the fine-scale spatio-temporal dataset, to obtain the annual global air pollutant concentration at the fine scale of the urban area and convert it into the global air pollutant concentration weight.

[0060] Step 5: Based on the daily satellite remotely sensed inversion data at the low-resolution scale, calculate the daily local air pollutant concentration estimation residual.

[0061] Step 6: Combine the fine-scale spatio-temporal dataset, the global atmospheric pollutant concentration weight, and the estimated residuals of the daily local atmospheric pollutant concentration to estimate the daily fine-scale atmospheric pollutant concentration of the city.

[0062] Example 2:

[0063] Based on Example 1, Example 2 of the present application provides a more specific method for estimating the daily fine-scale atmospheric pollutant concentration of the city, including:

[0064] Step 1: Obtain multi-source data of the urban area, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data, and ground station monitoring data.

[0065] Step 2: Perform multi-scale spatio-temporal matching processing on the multi-source data of the urban area to obtain a processed multi-scale spatio-temporal dataset; the multi-scale spatio-temporal dataset includes a low-resolution scale spatio-temporal dataset and a fine-scale spatio-temporal dataset.

[0066] Step 3: Based on the low-resolution scale spatio-temporal dataset, construct a global estimation model for atmospheric pollutants and train it, and iteratively optimize the model parameters.

[0067] Step 4: Use the trained global estimation model, combine the fine-scale spatio-temporal dataset, to obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area, and convert it into the global atmospheric pollutant concentration weight.

[0068] In Step 4, based on the scale invariance hypothesis, that is, the relationship between the PM 2.5 concentration and the dependent variables (elevation, road length, proportion of each land use type) at the low-resolution scale remains unchanged and applicable at the fine scale. Therefore, the global estimation model trained at the low-resolution scale can be used, and the spatio-temporal dataset of elevation, road length, and proportion of each land use type constructed at the fine scale is used as the model input to obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area; further considering that the global concentration is the annual trend result, reflecting the distribution trend of the PM 2.5 concentration of each 100-meter grid in the urban area, it needs to be converted into the global PM 2.5 concentration weight in the subsequent calculation. Therefore, to ensure the reliability of the data, within each 1-kilometer standard grid range, the global PM 2.5 concentration weight is calculated by standardization, and its formula is expressed as:

[0069]

[0070] In the above formula, It represents the PM after standardized calculation of the i-th fine-scale grid within the range of a low-resolution scale standard grid. 2.5 Concentration weight, For the corresponding PM 2.5 Global concentration. After standardized calculation, the average value of all weights within each low-resolution scale standard grid range is 1.

[0071] Step 5: Calculate the daily local atmospheric pollutant concentration estimation residuals based on the daily satellite remote sensing inversion data at the low-resolution scale.

[0072] In Step 5, the purpose of the daily local atmospheric pollutant concentration estimation residuals is to correct the uncertainty in the estimation based on the global trend and obtain a more reliable daily atmospheric pollutant concentration estimation result.

[0073] Specifically, Step 5 includes:

[0074] Step 5.1: Considering the spatial distribution of the daily satellite remote sensing inversion PM 2.5 at the low-resolution scale, based on the first law of geography, that is, geographical things or attributes are mutually related in spatial distribution, and the correlation between similar things is greater, the inverse distance weighting method is used to obtain the spatial distribution of the daily fine-scale urban area PM 2.5 ;

[0075] Step 5.2: Subtract the spatial distribution of the daily fine-scale urban area PM 2.5 from the daily PM 2.5 data obtained by the nearest neighbor resampling method during the spatio-temporal matching process;

[0076] Step 5.3: Within each low-resolution scale standard grid range, perform standardized calculation on the residuals to obtain the daily local atmospheric pollutant concentration estimation residuals, and its formula is expressed as:

[0077]

[0078] In the above formula, represents the local residual of the PM 2.5 concentration after standardized calculation of the i-th 100-meter fine-scale grid within the range of a 1-kilometer scale standard grid, and respectively represent the fine-scale PM 2.5 data obtained by the inverse distance weighting method and the nearest neighbor resampling method, and mean and σ represent the average value and standard deviation of all local residuals of the PM 2.5 concentration within the range of a 1-kilometer scale standard grid.

[0079] Step 6: Combine the fine-scale spatio-temporal dataset, the global atmospheric pollutant concentration weights, and the daily local atmospheric pollutant concentration estimation residuals to estimate the daily fine-scale atmospheric pollutant concentrations in the city.

[0080] The calculation formula for Step 6 is:

[0081]

[0082] In the above formula, represents the estimated concentration of PM 2.5 in the kth fine-scale grid in the urban area. Further verify it with the monitoring results of ground stations to ensure the accuracy of the estimation results.

[0083] The effects of the present invention are further analyzed through specific experimental results as follows:

[0084] First, collect the data from January 1 to December 31, 2019 in the central urban area of Ningbo, including the daily PM 2.5 data with a 1-km spatial resolution in the Chinese High-Resolution High-Quality Near-Surface Air Pollutant CHAP dataset, the 10-m spatial resolution land use data of ESRI, the road network vector data of Open Street Map, the 30-m spatial resolution DEM data of the US Geological Survey (USGS), and the PM 2.5 concentration data of the ground air quality monitoring stations of the National Environmental Monitoring Center. Perform multi-scale spatio-temporal matching according to the foregoing method of the present invention to obtain the processed multi-scale spatio-temporal dataset.

[0085] Then, according to the foregoing method of the present invention, based on the low-resolution spatio-temporal dataset, construct a global atmospheric pollutant estimation model based on random forest regression for training. Then, combined with the high-resolution spatio-temporal dataset, apply it to the 100-m fine scale, calculate the standardized global PM 2.5 concentration weights, and then estimate the daily global trend results. At the same time, calculate the daily local residuals to correct the results. The comparison between the model estimation results and the verification of the ground stations is as Figure 2 shown. The results show that the R 2.5 of the PM 2 estimated by the model of the present invention at the fine scale and the RMSE are 0.98 and 2.46 μg / m 3 respectively. The slope of the fitting line reaches 0.98, and the estimation accuracy is high. In addition, as Figure 3 shown, compared with the original atmospheric pollutant data with a 1-km spatial resolution, the model of the present invention can achieve higher spatial resolution PM 2.5 estimation, more accurately depict the spatial distribution details of atmospheric pollutants within the city, and can better promote the precise and in-depth treatment of urban air pollution.

[0086] It should be noted that the same or similar parts in this embodiment and Embodiment 1 can be referred to each other, and will not be described in detail in this application.

[0087] Embodiment 3:

[0088] Based on Embodiments 1 and 2, Embodiment 3 of this application provides a daily urban fine-scale atmospheric pollutant concentration estimation system, including:

[0089] A first acquisition module, configured to acquire multi-source data of the urban area, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data, and ground station monitoring data;

[0090] A matching module, configured to perform multi-scale spatio-temporal matching processing on the multi-source data of the urban area to obtain a processed multi-scale spatio-temporal data set; the multi-scale spatio-temporal data set includes a low-resolution scale spatio-temporal data set and a fine-scale spatio-temporal data set;

[0091] A construction module, configured to construct a global estimation model of atmospheric pollutants based on the low-resolution scale spatio-temporal data set and perform training, and iteratively optimize the model parameters;

[0092] A second acquisition module, configured to use the trained global estimation model, combine with the fine-scale spatio-temporal data set, obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area, and convert it into the global atmospheric pollutant concentration weight;

[0093] A calculation module, configured to calculate the daily local atmospheric pollutant concentration estimation residual based on the low-resolution scale daily satellite remote sensing inversion data;

[0094] An estimation module, configured to estimate the daily urban fine-scale atmospheric pollutant concentration by combining the fine-scale spatio-temporal data set, the global atmospheric pollutant concentration weight, and the daily local atmospheric pollutant concentration estimation residual.

[0095] Specifically, the system provided in this embodiment is the system corresponding to the methods provided in Embodiments 1 and 2. Therefore, the same or similar parts in this embodiment and Embodiments 1 and 2 can be referred to each other, and will not be described in detail in this application.

[0096] To sum up, the present invention provides a method for estimating the daily urban fine-scale atmospheric pollutant concentration, which can achieve accurate estimation of PM at high spatial resolution in the city 2.5 and has important application value.

Claims

1. A method for estimating daily urban air pollutant concentrations at a fine scale, characterized in that: include: Step 1: Obtain multi-source data of urban areas, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data and ground site monitoring data; Step 2: Perform multi-scale spatiotemporal matching processing on the multi-source data of the urban area to obtain a processed multi-scale spatiotemporal dataset; the multi-scale spatiotemporal dataset includes a low-resolution spatiotemporal dataset and a fine-scale spatiotemporal dataset; Step 3: Based on the low-resolution spatiotemporal dataset, a global estimation model of atmospheric pollutants is constructed and trained, and model parameters are iteratively optimized; Step 4: Using the trained global estimation model and the fine-scale spatiotemporal dataset, the annual global atmospheric pollutant concentration at the fine scale of the urban area is obtained, and it is converted into the global atmospheric pollutant concentration weight; In step 4, the fine-scale spatiotemporal dataset of elevation, road length, and proportion of each land use type constructed in step 2.2 is used as the input of the global estimation model to obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area; The global PM is calculated by standardization within the low-resolution standard grid. 2.5 Concentration weight, its formula is expressed as: In the above formula, Represents the PM calculated after normalization of the i-th fine-scale grid within a low-resolution standard grid. 2.5 Concentration weight, For the corresponding PM 2.5 The global concentration is calculated by standardization, and the average value of all weights within each low-resolution scale standard grid is 1; Step 5: Calculate the daily local atmospheric pollutant concentration estimation residual based on the low-resolution daily satellite remote sensing inversion data; Step 5 includes: Step 5.1: Based on the low-resolution daily satellite remote sensing inversion data, the inverse distance weighted method is used to obtain the fine-scale daily urban area PM 2.5 Spatial distribution; Step 5.2: Convert the fine-scale daily urban area PM 2.5 Daily PM obtained by nearest neighbor resampling method in the process of spatial distribution and spatiotemporal matching 2.5 Subtract the data; Step 5.3: Within each low-resolution scale standard grid, the residuals are standardized to obtain the daily local atmospheric pollutant concentration estimation residuals, which are expressed as follows: In the above formula, Represents the PM calculated after normalization of the i-th fine-scale grid within a low-resolution standard grid. 2.5 The local residual of the concentration, and Represent the fine-scale PM obtained based on the inverse distance weighting method and the nearest neighbor resampling method respectively. 2.5 Data, mean and σ represent all fine-scale PM within the low-resolution standard grid. 2.5 mean and standard deviation of local residuals of concentration; Step 6: Combine the fine-scale spatiotemporal dataset, the global air pollutant concentration weights and the daily local air pollutant concentration estimation residuals to estimate the fine-scale daily air pollutant concentrations in the city.

2. The method for estimating daily urban fine-scale atmospheric pollutant concentrations according to claim 1 is characterized in that: The satellite remote sensing inversion data is a low-resolution daily atmospheric pollutant fine particulate matter PM 2.5 Remote sensing inversion data; land use data is annual classification data with fine scale, covering water bodies, trees, flooded vegetation, crops, built-up areas, bare land and pasture types; ground digital elevation model data is raster data with fine scale; road network data is annual line vector data, and the road network includes expressways, trunk roads, first-class highways, second-class highways and third-class highways; The ground station monitoring data is the real-time hourly monitoring data of the national ground air quality monitoring network.

3. The method for estimating daily urban fine-scale atmospheric pollutant concentrations according to claim 2 is characterized in that: Step 2 includes: Step 2.1: Based on the satellite remote sensing inversion data, a low-resolution scale standard grid is constructed. In a unified spatial coordinate system, the proportion of the area occupied by each land use type in the low-resolution scale standard grid, the mean of the ground digital elevation model data, and the total length of each road are counted, and the atmospheric fine particulate matter PM is calculated at the same time. 2.5 The annual mean of remote sensing inversion data is used to obtain a low-resolution spatiotemporal dataset; Step 2.2: Construct a fine-scale standard grid in a unified spatial range and coordinate system, and calculate the proportion of each land use type in the fine-scale standard grid, the mean of the ground digital elevation model data, and the total length of each road. 2.5 The remote sensing inversion data was obtained by using the nearest neighbor resampling method to obtain the daily PM 2.5 A fine-scale dataset.

4. The method for estimating daily urban air pollutant concentrations at a fine scale according to claim 3 is characterized in that: In step 3, the global estimation model of air pollutants is a supervised machine learning model, and the supervised machine learning model is any one of a random forest regression model, a Cubist regression model, and an XGBoost regression model; the formula of the supervised machine learning model is expressed as follows: PM2.5 global =F(DEM,Road,LU water ,LU veg ,LU wet ,LU agri ,LU build ,LU bare ,LU farm ) In the above formula, PM2.5 global Indicates PM 2.5 where F represents any supervised machine learning model, DEM represents digital surface elevation, Road represents the total length of various roads, LU represents the proportion of each land use type, and the subscripts water, veg, wet, agri, build, bare and farm represent water, trees, flooded vegetation, crops, built-up areas, bare land and pasture, respectively.

5. The method for estimating daily urban fine-scale atmospheric pollutant concentrations according to claim 4 is characterized in that: The calculation formula for step 6 is: In the above formula, represents the PM of the kth fine-scale grid in the urban area 2.5 Estimate the concentration.

6. A system for estimating the concentration of daily atmospheric pollutants at a fine scale in a city, characterized in that: The method for executing any one of claims 1 to 5 comprises: The first acquisition module is used to acquire multi-source data of urban areas, including satellite remote sensing inversion data, land use data, ground digital elevation model data, road network data and ground site monitoring data; A matching module, used for performing multi-scale spatiotemporal matching processing on the multi-source data of the urban area to obtain a processed multi-scale spatiotemporal dataset; the multi-scale spatiotemporal dataset includes a low-resolution spatiotemporal dataset and a fine-scale spatiotemporal dataset; A construction module is used to construct and train a global estimation model of atmospheric pollutants based on the low-resolution spatiotemporal dataset, and iteratively optimize model parameters; The second acquisition module is used to use the trained global estimation model and the fine-scale spatiotemporal data set to obtain the annual global atmospheric pollutant concentration at the fine scale of the urban area and convert it into the global atmospheric pollutant concentration weight; A calculation module is used to calculate the residuals of the daily local atmospheric pollutant concentration estimation based on the low-resolution daily satellite remote sensing inversion data; The estimation module is used to combine the fine-scale spatiotemporal data set, the global atmospheric pollutant concentration weights and the daily local atmospheric pollutant concentration estimation residuals to estimate the fine-scale daily atmospheric pollutant concentrations in the city.

7. A computer storage medium, characterized in that: The computer storage medium stores a computer program; when the computer program is executed on a computer, the computer executes any one of the methods described in claims 1 to 5.

8. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the method according to any one of claims 1 to 5.

Images