Response surface model construction method and device for atmospheric pollutant concentration analysis

Abstract

The application discloses a response surface model construction method and device for atmospheric pollutant concentration analysis, wherein the method comprises the following steps: performing precursor emission simulation on a research area based on a Lagrangian diffusion model to generate precursor emission concentration data; using a statistical method to establish a response relationship of precursor concentration to precursor emission change based on the precursor emission concentration data; using a polynomial response surface method to establish a response relationship of atmospheric pollutant concentration to precursor concentration change based on multi-scenario simulation of a three-dimensional air quality model; and coupling the response relationship of precursor concentration to precursor emission change with the response relationship of atmospheric pollutant concentration to precursor concentration change to obtain an atmospheric pollutant concentration response surface model to precursor emission change. The application can reduce the cost and time of response surface model modeling, and quickly quantifies the influence of small-scale regional emission source change on atmospheric pollutant concentration.

Description

Technical Field

[0001] This invention relates to the field of atmospheric environment technology, and in particular to a method and apparatus for constructing a response surface model for analyzing atmospheric pollutant concentrations. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] Air pollution is a significant environmental problem, PM2.5 2.5 Ozone (O3) and PM2.5, as major atmospheric pollutants, pose a significant threat to human health. To achieve continuous improvement in air quality, a series of air pollution reduction measures have been implemented. This requires assessing air pollution and the impact of these reduction measures on PM2.5. 2.5 The impact of O3 concentration is crucial for the formulation of pollution control measures. Three-dimensional air quality models can predict PM2.5 concentrations in different regions under various emission scenarios. 2.5 While O3 concentration can be measured, its calculation is complex, computationally intensive, and time-consuming, limiting its effectiveness when dealing with numerous emission scenarios. To efficiently assess the impact of massive emission reduction measures on air quality, existing research utilizes statistical methods to simplify three-dimensional air quality models (such as CMAQ, CAMx, and WRF-Chem) into response surface models (RSM) based on multi-sampling scenario simulations. Compared to three-dimensional air quality models, RSM can achieve PM2.5 concentration measurement while saving over 90% of computation time and cost. 2.5 Predicting the response of O3 concentration to precursor emissions is an important tool to support the formulation of air pollution reduction plans.

[0004] Existing response surface modeling methods, utilizing deep learning technology, can control the number of sampling scenarios for a 3D air quality model to approximately twice the emission reduction target. This means that building a response surface model for a target region requires two sampling scenarios, thus achieving PM2.5 reduction. 2.5This method rapidly assesses the response of O3 concentration to changes in average emissions at large scales such as provincial and regional levels. However, when applied to smaller-scale response surface models, it requires generating two separate emission change scenarios for each small-scale area, such as a park / facility / factory production process. This limits the number of three-dimensional air quality model scenarios to thousands or tens of thousands, significantly increasing computation time and cost. Therefore, there is an urgent need to develop a novel response surface model construction method to achieve rapid response of air pollutant concentrations to changes in precursor emissions at small-scale regional levels such as parks / facilities / production processes with low computation time and cost. This would address the limitations of response surface models in fine-scale applications and meet the needs of refined air quality management for control. Summary of the Invention

[0005] This invention provides a method for constructing a response surface model for atmospheric pollutant concentration analysis, which reduces the cost and time of response surface model building and rapidly quantifies the impact of small-scale regional emission source changes on atmospheric pollutant concentrations. The method includes:

[0006] The study area was simulated using the Lagrange diffusion model to generate precursor emission concentration data; the precursor emission concentration data included the precursor concentrations after emissions from different precursor emission sources.

[0007] Based on precursor emission concentration data, statistical methods were used to establish the response relationship between precursor concentration and changes in precursor emissions.

[0008] Based on a three-dimensional air quality model, a scenario simulation of precursor emissions was conducted in the study area to generate scenario simulation result data; the scenario simulation result data includes the concentration of air pollutants after the emission of different precursors;

[0009] Based on the polynomial response surface methodology, and using scenario simulation data, the response relationship between atmospheric pollutant concentration and precursor concentration is established.

[0010] By coupling the response relationship of precursor concentration to changes in precursor emissions with the response relationship of air pollutant concentration to changes in precursor concentration, a surface model of the response of air pollutant concentration to changes in precursor emissions is obtained.

[0011] This invention also provides a response surface model construction device for atmospheric pollutant concentration analysis, wherein the device includes:

[0012] The diffusion model processing module is used to simulate precursor emissions in the study area based on the Lagrange diffusion model and generate precursor emission concentration data. The precursor emission concentration data includes the precursor concentrations after emissions from different precursor emission sources. Based on the precursor emission concentration data, statistical methods are used to establish the response relationship between precursor concentration and changes in precursor emissions.

[0013] The three-dimensional air quality model processing module is used to simulate precursor emission scenarios in the study area based on the three-dimensional air quality model and generate scenario simulation result data. The scenario simulation result data includes the concentration of air pollutants after different precursor emissions. Based on the polynomial response surface method, the response relationship between air pollutant concentration and precursor concentration changes is established using the scenario simulation result data.

[0014] The coupling module is used to couple the response relationship of precursor concentration to changes in precursor emissions with the response relationship of air pollutant concentration to changes in precursor concentration, so as to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions.

[0015] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for constructing a response surface model for atmospheric pollutant concentration analysis.

[0016] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for constructing a response surface model for atmospheric pollutant concentration analysis.

[0017] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method for constructing a response surface model for atmospheric pollutant concentration analysis.

[0018] This invention combines Lagrange diffusion modeling, three-dimensional air quality modeling, and polynomial response surface methodology to establish a high-resolution response surface model. Compared with existing technologies, this invention effectively reduces the direct application scope of complex three-dimensional air quality models. While ensuring simulation accuracy, it significantly reduces the computation time and resource costs required for constructing the response surface model, thereby significantly improving the modeling efficiency. The high-resolution response surface model in this invention can quickly quantify the impact of small-scale emission source changes on PM2.5 in small-scale areas such as industrial parks, specific facilities, and specific production processes. 2.5 The impact of atmospheric pollutant concentrations such as O3. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0020] Figure 1 This is a flowchart illustrating the method for constructing a response surface model for analyzing atmospheric pollutant concentrations in an embodiment of the present invention.

[0021] Figure 2 This is a specific example of a response surface model construction method for atmospheric pollutant concentration analysis in an embodiment of the present invention;

[0022] Figure 3 This is another specific example of the response surface model construction method for atmospheric pollutant concentration analysis in this invention.

[0023] Figure 4 This is a schematic diagram illustrating the process of establishing the response relationship between precursor concentration and precursor emission changes in an embodiment of the present invention.

[0024] Figure 5 This is another specific example of the response surface model construction method for atmospheric pollutant concentration analysis in this invention.

[0025] Figure 6 This is a schematic diagram of a response surface model construction device for analyzing atmospheric pollutant concentrations in an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0027] To facilitate a clear description of the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish the same or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order.

[0028] Figure 1 This is a flowchart illustrating the method for constructing a response surface model for analyzing atmospheric pollutant concentrations in an embodiment of the present invention. Figure 1 As shown, the method includes:

[0029] Step 101: Simulate precursor emissions in the study area based on the Lagrange diffusion model to generate precursor emission concentration data; the precursor emission concentration data includes the precursor concentrations after emissions from different precursor emission sources;

[0030] Step 102: Based on precursor emission concentration data, establish the response relationship between precursor concentration and changes in precursor emissions using statistical methods;

[0031] Step 103: Simulate precursor emission scenarios in the study area based on a three-dimensional air quality model, and generate scenario simulation result data; the scenario simulation result data includes the concentration of air pollutants after different precursor emissions;

[0032] Step 104: Based on the polynomial response surface methodology, and using scenario simulation results data, establish the response relationship between atmospheric pollutant concentration and precursor concentration changes.

[0033] Step 105: Couple the response relationship of precursor concentration to changes in precursor emissions with the response relationship of air pollutant concentration to changes in precursor concentration to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions.

[0034] This invention proposes a technique for constructing PM2.5 using a coupled Lagrange diffusion model and a three-dimensional air quality model. 2.5 A method for high-resolution response relationship between O3 atmospheric pollutant concentration and precursor emissions is proposed. First, based on the computationally inexpensive and time-efficient Lagrange diffusion model, a high-resolution "precursor concentration-precursor emission" response relationship is constructed using statistical methods. Second, based on a three-dimensional air quality model, a low-resolution "atmospheric pollutant concentration-precursor concentration" response relationship is constructed using polynomial response surface methodology. By coupling these two response relationships, a high-resolution response surface model of atmospheric pollutant concentration to precursor emission changes is constructed, enabling rapid prediction of atmospheric pollutant response to precursor emission reduction. This high-resolution response surface model of atmospheric pollutant concentration to precursor emission changes not only significantly reduces modeling costs and time but also rapidly quantifies the impact of small-scale regional emission source changes on atmospheric pollutant concentrations, thus providing a scientific basis for precise pollution control strategies.

[0035] Step 101: Simulate precursor emissions in the study area based on the Lagrange diffusion model to generate precursor emission concentration data.

[0036] During implementation, several precursor emission sampling scenarios were designed, and the precursor emission sources in each fine grid cell area within the study region were simulated using the Lagrange diffusion model to generate high-precision precursor concentration datasets under several emission scenarios.

[0037] In this embodiment of the invention, the Lagrange diffusion model is suitable for simulating SO2 and NO on complex terrains and at small scales. X One PM 2.5 Physical processes such as transport, diffusion, and deposition of atmospheric precursors such as VOCs and NH3.

[0038] In one embodiment, simulating precursor emissions in the study area based on the Lagrange diffusion model to generate precursor emission concentration data may include: simulating precursor emissions in each grid within the study area based on the Lagrange diffusion model to generate a precursor emission concentration dataset; the precursor emission concentration dataset reflects the precursor concentration of each grid at a set time after only a single grid cell emits and no other grid cells emit.

[0039] Figure 2 This is a specific example diagram of the response surface model construction method for atmospheric pollutant concentration analysis in an embodiment of the present invention, as shown in the figure. Figure 2 The diagram illustrates the specific steps involved in generating precursor emission concentration data based on the Lagrange diffusion model, including:

[0040] Step 201: Setting up the simulation domain and preprocessing data for the Lagrange diffusion model;

[0041] In a preferred embodiment, the Lagrange diffusion model includes a meteorological module and a Lagrange-Gaussian smoke cloud module. The meteorological module is used to generate a temporally and spatially varying meteorological field using topographic elevation data and land use data; the Lagrange-Gaussian smoke cloud module is used to simulate the diffusion and transformation processes along the transport path of smoke clouds released from precursor emission sources by using the meteorological field generated by the meteorological module.

[0042] This embodiment uses gridded meteorological field data output by a mesoscale numerical weather prediction model (WRF) as the initial guess field for the meteorological module in the diffusion model. The topographic elevation data used by the meteorological module is GEBCO data (General Bathymetric Chart of the Oceans), with a resolution of 15 arcseconds (approximately 500m); the land use data is high-resolution data of the study area for a specific year, with a resolution of 30m. The geodetic reference system selected is the 1984 World Geodetic System (WGS-84), and the projected coordinate system is the Universal Transverse Mercator (UTM). The Lagrange diffusion model requires dividing the study area into several grid regions horizontally and several layers vertically. In this embodiment, the grid resolution is set to 100m × 100m, dividing the study area into several small 100m × 100m regions. The vertical simulation range extends from the ground to a height of 3000m, divided into 10 unequal layers with top heights of 20, 40, 80, 160, 300, 600, 1000, 1500, 2200, and 3000m, respectively. It should be noted that the simulation domain settings and input data of the Lagrange diffusion model should be determined based on the specific study area and the pollutants of interest for emission reduction; no limitations are imposed in this embodiment.

[0043] Step 202: Setting key parameters for the Lagrange diffusion model;

[0044] In this embodiment of the invention, the core module of the Lagrange diffusion model is the Lagrange-Gaussian plume module. This model utilizes the temporally and spatially varying meteorological field generated by the meteorological module to simulate the advection transport of plumes released from emission sources and their diffusion and transformation processes along the transport path. Depending on the input data, the Lagrange-Gaussian plume module provides different diffusion calculation options. In this embodiment, the default diffusion scheme is selected, and horizontal and vertical diffusion parameters are calculated based on similarity theory using the micrometeorological parameters output by the meteorological module. The pollution source emission precursors in the Lagrange-Gaussian plume module include SO2 and NO. X One PM 2.5 For VOCs and NH3, the dry and wet deposition module of the precursors was enabled in the simulation. The dry and wet deposition module is mainly used to simulate and calculate the deposition process of pollutant precursors in the atmosphere.

[0045] Regarding the emission sources of precursors, these sources include one or any combination of the following pollution source types: area source, point source, and line source. For example, mobile line sources, power plants, and industrial point sources, as well as industrial process sources and agricultural area sources. Time allocation coefficients are set for different months of different emission sectors. It should be noted that the key parameters of the Lagrange diffusion model should be set according to the specific study area and the pollutants of interest for emission reduction; this embodiment of the invention does not impose any limitations.

[0046] Step 203: Design several precursor emission sampling scenarios and generate a high-precision precursor emission concentration dataset.

[0047] Figure 3 This is another specific example of the response surface model construction method for atmospheric pollutant concentration analysis in this invention, as shown in the figure. Figure 3 As shown, precursor emissions are simulated on each grid within the study area based on the Lagrange diffusion model, generating a precursor emission concentration dataset, which may include:

[0048] Step 301: Configure a first baseline scenario and multiple precursor discharge zero placement scenarios; the first baseline scenario means that the emission source strength of each grid area is the actual emission source strength of the baseline year, and the precursor discharge zero placement scenarios mean that each grid area and each emission source sector are respectively set up with a precursor discharge zero placement scenario;

[0049] Step 302: Input the first baseline scenario into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations and obtain the first precursor concentration matrix; the first precursor concentration matrix includes the precursor concentrations in the region where each grid is located in the baseline year;

[0050] Step 303: Input each precursor placement zero scenario into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations, and obtain multiple second precursor concentration matrices; the second precursor concentration matrix includes the precursor concentration of each grid corresponding to the precursor placement zero scenario;

[0051] Step 304: Based on the first precursor concentration matrix and multiple second precursor concentration matrices, a precursor emission concentration dataset is formed; the precursor emission concentration dataset includes concentration information from different emission sources, different regional locations, and different time resolutions.

[0052] In this embodiment of the invention, different types of pollution sources require specific data formats and input requirements when inputting into the Lagrange diffusion model. This embodiment uses a sampling scenario of area source precursors as an example to provide a detailed description of the specific implementation method.

[0053] The aforementioned precursor emission sampling scenarios include one first baseline scenario and several precursor emission zero scenarios.

[0054] The aforementioned first baseline scenario refers to setting the emission source intensity of each grid region to the actual emission source intensity of the baseline year, expressed in tons per year per square kilometer. The baseline emission scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations, yielding the simulated precursor concentrations for each grid region in the baseline year, which can be represented in the form of a precursor concentration matrix.

[0055] The aforementioned zero-scenarios for precursor material placement refer to setting zero-scenarios for precursor material placement for each emission source sector and each grid area. There are R area source emission sectors, I total grid rows within the study area, and J total grid columns within the study area. Therefore, the total number of zero-scenarios for precursor material placement for each area source emission sector is (R × I × J). The emission source strength matrix for each zero-scenarios for precursor material placement is as follows:

[0056]

[0057] In the formula, E rn E_p represents the emission source strength matrix of the study area for the r-th emission source sector and the n-th sampling scenario; ij E_p represents the emission source intensity of precursor p in the region where the grid in row i and column j is located, in tons per year per square kilometer; when (i-1)×J+j=n, E_p ij = 1 ton / year / square kilometer, while the emission source intensity of precursor p in the remaining grid areas is 0; p represents the precursor type, and in this embodiment of the invention, the precursor types include SO2 and NO. X One PM 2.5 VOCs and NH3.

[0058] The above precursors are placed in a zero-case scenario and input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations, resulting in several precursor concentration matrices, denoted as the second precursor concentration matrix. The second precursor concentration matrix is ​​shown below:

[0059]

[0060] In the formula, Q trnp Q represents the concentration matrix of precursor p calculated using the Lagrange diffusion model for the nth sample of the rth emission source at time t, with the sample set to zero. ij represents the simulated concentration of precursors in the region where the grid in row i and column j is located; p represents the type of precursor, including SO2 and NO. X One PM 2.5 VOCs and NH3.

[0061] Finally, the above hourly-resolution precursor concentration matrix is ​​processed into precursor concentration matrices with different time resolutions, such as 24-hour average, monthly average, and annual average, through a post-processing program.

[0062] Through the above steps, a precursor emission concentration dataset with kilometer-level resolution was generated for the study area, covering concentration information from different emission source sectors, different regional locations, and different time resolutions. Based on this dataset, it is possible to quickly obtain SO2 and NO concentration data for each grid region at a specific time after emissions from a single grid have ceased in other grids. X One PM 2.5 The concentrations of VOCs and NH3 provide fundamental data support for constructing a rapid response relationship between precursor concentrations and changes in precursor emissions.

[0063] Step 102: Based on precursor emission concentration data, establish the response relationship between precursor concentration and changes in precursor emissions using statistical methods.

[0064] During implementation, based on the high-precision precursor concentration dataset generated by the Lagrange diffusion model, statistical methods were used to establish a high-resolution response relationship between precursor concentration and changes in precursor emissions.

[0065] Figure 4 This is a schematic diagram illustrating the process of establishing the response relationship between precursor concentration and changes in precursor emissions in this embodiment of the invention. Taking the impact of annual changes in precursor emissions on the annual average concentration of precursors as an example, this embodiment details the process of establishing a high-resolution rapid response relationship between precursor concentration and changes in precursor emissions, specifically including:

[0066] Step 401: Select a linear regression model to establish the response relationship between precursor concentration and changes in precursor emissions;

[0067] The Lagrange diffusion model focuses on the physical transport and diffusion processes of precursors, neglecting or only minimally considering the chemical transformation processes between different precursors. Therefore, this embodiment of the invention assumes a substantially linear relationship between precursor emissions and precursor concentration. The basic framework expression for the response relationship between precursor concentration and changes in precursor emissions is as follows:

[0068] Qreduce=K×Ereduce+M(Formula 2-1)

[0069] In the formula, the independent variable is Erreduce, which represents the change in precursor emissions; the dependent variable is Qreduce, which represents the change in precursor concentration; and K and M are regression model parameters to be confirmed.

[0070] To achieve a high-resolution response of precursor concentrations to changes in precursor emissions, it is necessary to establish a mathematical relationship between the changes in precursor concentrations in the receiver grid and the changes in precursor emissions in the source grids within the study area. Based on atmospheric chemical transport theory, when precursor emissions from any source grid within the study area change, the change will affect the precursor concentration distribution throughout the entire area through chemical reactions and physical transport processes. Therefore, the change in precursor concentration in a particular receiver grid is not only related to its own emission changes but also to the emission changes in other source grids. The relationship between the changes in precursor concentrations in the receiver grid and the changes in precursor emissions in the source grids can be modified based on Equation 2-1.

[0071] In one embodiment, establishing a response relationship between precursor concentration and changes in precursor emissions based on precursor emission concentration data using statistical methods may include:

[0072] Based on precursor emission concentration data, the following regression model is used to establish the response relationship between precursor concentration and changes in precursor emissions using statistical methods:

[0073]

[0074] Among them, Qreduce ij This represents the change in precursor concentration in the region of the receptor grid in row i and column j; Ereduce ab The value represents the change in precursor emissions in the region where the source grid in row a and column b is located; I represents the total number of grid rows in the study area, and J represents the total number of grid columns in the study area. The number of receptor grids and source grids is the same, which is I×J; K and M are regression model parameters.

[0075] Step 402: Based on the precursor concentration dataset generated in Step 101 for several zero-sampling scenarios, confirm the parameter K of the linear regression model.

[0076] Based on the precursor emission concentration dataset generated in step 101, the annual average simulated precursor concentration matrix for all receiver grids can be directly obtained when the precursor emission source intensity of a single grid within the study area is 1 ton / year / km² (assuming no emissions from other grids). That is, the response matrix of the emission source intensity of a single grid to the annual average precursor concentration of all receiver grids, based on the precursor emission concentration dataset, is as follows:

[0077]

[0078] In the formula, Qyear ab This represents the annual average precursor concentration matrix for all receiver grids when the source grid in row a and column b has a precursor emission source intensity of 1 ton / year / km². Qyear ab_ijThe value represents the annual average precursor concentration in the receptor grid in the i-th row and j-th column when the source intensity of the source grid in the a-th row and b-th column is equal to the unit emission amount. The unit is micrograms per cubic meter.

[0079] In the regression model, parameter K represents the change in precursor concentration in the region of the receiver grid (row a, column b) caused by a change in precursor emissions in the region of the source grid (row a, column b). Therefore, parameter K is Qyear. ab_ij The relationship between the change in precursor concentration in the receptor grid and the change in precursor emissions in the source grid can be further quantified as follows:

[0080]

[0081] In the formula, Qreduce ij This represents the change in precursor concentration in the region of the receptor grid in row i and column j; Ereduce ab The value represents the change in precursor emissions in the region where the source grid in row a and column b is located; I represents the total number of grid rows in the study area, and J represents the total number of grid columns in the study area. The number of receptor grids and source grids is the same, and the number of each grid is (I×J); K and M are regression model parameters to be confirmed.

[0082] Step 403: Based on the first precursor concentration matrix generated in the first baseline scenario in step 302, confirm the parameter M of the regression model.

[0083] By analyzing the relationship between precursor emissions and precursor concentrations under the baseline scenario, the parameters M of the regression model can be determined. This is when the precursor emission reduction in the source grid region is equal to the baseline year's emission amount Ebase. ab At that time, the change in precursor concentration in the region where the receptor grid is located is the precursor concentration Qbase under the baseline scenario. ij The specific expression is as follows:

[0084] Given:

[0085] Ereduce ab =Ebase ab (Equation 2-5)

[0086] Qreduce ij =Qbase ij (Equation 2-6)

[0087] Substituting equations 2-5 and 2-6 into equation 2-4, we get:

[0088]

[0089] Substituting Equation 2-7 into Equation 2-4, the relationship between the change in precursor concentration in the receptor grid and the change in precursor emissions in the source grid can be further quantified as follows:

[0090]

[0091] In the formula, Qreduce ij The dependent variable represents the change in precursor concentration in the region where the receptor grid in row i and column j is located; Ereduce ab The independent variable is Qyear, which represents the change in precursor emissions in the region where the source grid in row a and column b is located. ab_ij Ebase ab and Qbase ij All are constants, Qyear ab_ij Ebase represents the change in the annual average concentration of precursors in the receiver grid (row a, column b) resulting from a unit reduction in precursor emissions in the region containing the source grid (row a, column b). ab This represents the baseline annual emissions of precursors for the region where the source grid in row a and column b is located; Qbase ij The reference concentration of the precursor is indicated by the region where the receptor grid in row i and column j is located; I represents the total number of grid rows in the study area, and J represents the total number of grid columns in the study area.

[0092] Step 404: Based on actual pollution prevention and control needs, construct a high-resolution response relationship between precursor concentration and changes in precursor emissions.

[0093] In a preferred embodiment, based on precursor emission concentration data, a statistical method is used to establish the response relationship between precursor concentration and changes in precursor emissions, including:

[0094] The following formula is used to predict changes in output precursor concentrations based on the precursor reduction ratio of the emission source sector:

[0095]

[0096] In the formula, Qreduce p_ij K is the dependent variable, representing the predicted concentration change of precursor p in the region where the receptor grid in the i-th row and j-th column is located; p_r_ab The independent variable is Qyear, which represents the emission reduction ratio of precursor p in the region where the source grid in row a and column b is located in the pollution prevention and control measures. p_r_ab_ij Ebase p_r_ab and Qbase p_r_ij All are constants, Qyear p_r_ab_ij Ebase represents the change in the annual average concentration of precursors in the receptor grid (row i, column j) caused by a unit reduction in the emission of precursor p in the emission source sector r of the region where the source grid (row a, column b) is located.p_r_ab This represents the baseline annual emissions of precursor p from emission source sector r in the region where the source grid in row a and column b is located; Qbase p_r_ij Let R represent the baseline concentration of precursor p from emission source sector r in the region where the receptor grid is located (row i, column j); R represents the total number of emission source sectors; I represents the total number of grid rows in the study area; and J represents the total number of grid columns in the study area. The number of source grids and receptor grids is the same. Given that the Lagrange diffusion model has relatively low computational resource requirements, the study area can be infinitely subdivided to a certain extent by setting the grid resolution of the Lagrange diffusion model in step 101, thereby realizing a high-resolution response relationship between the change in precursor concentration in the receptor grid and the change in precursor emission in the source grid.

[0097] This invention establishes a response relationship between changes in target pollutant concentration and changes in precursor emissions, as shown in Equations 2-9. Equations 2-9 are applicable to precursor concentration change analysis under general conditions. Furthermore, for special grids where nonlinear relationships may exist, such as highly polluted areas or areas with complex terrain, local polynomial fitting or piecewise linear functions can be introduced based on Equations 2-9. The fitting parameters can then be optimized using statistical methods such as least squares to minimize errors and ensure that the established regression model accurately reflects the impact of precursor emission changes on precursor concentrations. Using the high-resolution precursor concentration response relationship to precursor emissions constructed in this invention, the precursor concentration in each grid area within the study region can be quickly predicted when different emission reduction measures are implemented.

[0098] Step 103: Simulate precursor emission scenarios in the study area based on a three-dimensional air quality model, and generate scenario simulation result data.

[0099] In one embodiment, the precursor emission scenario simulation of the study area based on the three-dimensional air quality model and the generation of scenario simulation result data may include: performing precursor emission scenario simulation of each large-scale unit divided into the study area based on the three-dimensional air quality model and generating scenario simulation result data; the number of large-scale units is less than the number of grids.

[0100] During implementation, the study area will be divided into H large-scale units, each of which will serve as a "PM". 2.5The smallest unit of precursor emission change in the equation “Response relationship of O3 atmospheric pollutant concentration to changes in precursor concentration” is used. It's worth noting that in the polynomial response surface methodology, the smallest unit of precursor concentration change determines the number of scenarios simulated in the 3D air quality model. Because the 3D air quality model meticulously depicts the physical and chemical evolution of atmospheric precursors, including atmospheric transport processes such as advection, turbulence, and convection; gaseous chemical processes such as the influence of nitrogen oxides on O3 formation; aerosol chemical processes such as the influence of sulfur dioxide oxidation on sulfate formation; and physical processes such as dry and wet deposition, it suffers from high computational costs and long computation times. Therefore, when defining the smallest unit of precursor concentration change, it is necessary to balance computational resources with research needs; the total number of large-scale units after the division should not be too large, typically H is less than 50. The number of large-scale units is much smaller than the number of grids in the diffusion model.

[0101] In one embodiment, precursor emission scenario simulations are performed on various large-scale units divided into study areas based on a three-dimensional air quality model to generate scenario simulation result data, which may include:

[0102] Configure precursor emission sampling scenarios; precursor emission sampling scenarios include a second baseline scenario, a gaseous precursor emission control scenario, or multiple gaseous precursor emission zeroing scenarios; the second baseline scenario represents the actual emission amount of precursors in each large-scale unit as the baseline year, the gaseous precursor emission control scenario is a control scenario in which the precursor emission amount in each large-scale unit is set to be several times the precursor emission amount in the second baseline scenario, and the gaseous precursor emission zeroing scenario is a scenario in which the precursor emission in each large-scale unit is set to zero;

[0103] The emission sampling scenario of precursors is input into a three-dimensional air quality model to calculate the spatiotemporal distribution of pollutant concentrations and obtain scenario simulation results data.

[0104] Step 104: Based on the polynomial response surface methodology, and using scenario simulation results data, establish the response relationship between atmospheric pollutant concentration and precursor concentration changes.

[0105] In implementation, based on the polynomial response surface methodology and using scenario simulation data, the response relationship between atmospheric pollutant concentration and precursor concentration changes is established. This can include: constructing a nonlinear polynomial expression for atmospheric pollutant concentration and precursor concentration based on the polynomial response surface methodology and using scenario simulation data. In the nonlinear polynomial expression, the dependent variable is the atmospheric pollutant concentration in the region where each grid is located, and the independent variable is the change in multiple precursor concentrations in each large-scale cell. The nonlinear polynomial expression includes multiple undetermined polynomial coefficients, which are determined through nonlinear regression fitting or deep learning methods.

[0106] During implementation, multi-scenario simulations based on a three-dimensional air quality model were conducted, and low-resolution PM2.5 data were established using polynomial response surface methodology. 2.5 / Response relationship between O3 concentration and changes in precursor concentration.

[0107] PM 2.5 The relationship between O3 concentration and precursor concentration exhibits a highly nonlinear relationship, which is observed in the polynomial response surface methodology for PM2.5. 2.5 The mathematical abstract expression for the response of O3 concentration to changes in precursor concentration is as follows:

[0108] CONCreduce ij =

[0109] f(Qreduce_VOCs h Qreduce_PPM h ,Qreduce_NH3 h Qreduce_SO2 h Qreduce_NOX h )

[0110] (Equation 3-1)

[0111] In the formula: Qreduce_VOCs h Qreduce_PPM h Qreduce_NH3 h Qreduce_SO2 h Qreduce_NOX h These are the independent variables, representing the precursor VOCs of the large-scale unit h and the primary PM2.5, respectively. 2.5 NH3, SO2 and NO X Concentration change; CONCreduce ij The dependent variable is the PM2.5 concentration in the region where the grid in row i and column j is located. 2.5 The predicted concentration change of / O3, i = 1, 2, ..., I, j = 1, 2, ..., J, where I represents the total number of grid rows in the study area and J represents the total number of grid columns in the study area, CONCreduce h It is made by Qreduce_VOCs h Qreduce_PPM h Qreduce_NH3 h Qreduce_SO2 h and Qreduce_NOX h The resulting polynomial has a fixed form, and the coefficients of the polynomial need to be determined using multi-scenario simulation results from a three-dimensional air quality model.

[0112] The coefficients of the polynomial in Equation 3-1 can be determined by two methods: nonlinear regression fitting and deep learning. In the specific implementation, one of these methods can be selected.

[0113] The specific implementation method for determining polynomial coefficients using nonlinear regression fitting is as follows:

[0114] First, we designed (23H+1) three-dimensional air quality model precursor emission sampling scenarios (H represents dividing the study area into H large-scale units), including: one second baseline scenario, where the precursor emissions in each large-scale unit are the actual emissions in the baseline year; and 20H gaseous precursor emission control scenarios, which use Hamersley sequence sampling to set 20 SO2 and NO emission control scenarios for each large-scale unit. X The control scenario involves setting emissions of four gaseous precursors—VOCs, NH3, and PM2.5—to 0 to 2 times the second baseline scenario emissions; and setting H gaseous precursor emissions to zero, meaning that the emissions of all four gaseous precursors in each large-scale unit are set to zero; 2H primary PM2.5 emissions are also included. 2.5 Control scenarios, that is, setting a PM once for each large-scale unit. 2.5 Emissions of 0 and primary PM 2.5 The scenario involves emissions that are twice the amount of the second baseline scenario. The (23H+1) precursor emission sampling scenarios are input into a three-dimensional air quality model to calculate the spatiotemporal distribution of pollutant concentrations, obtaining the different precursor concentrations for each large-scale unit and the PM2.5 concentrations for each fine-grid area. 2.5 A multi-scenario dataset of O3 concentration was generated. Finally, the multi-scenario dataset generated by the three-dimensional air quality model was used to fit polynomial 3-1, obtaining the response equation coefficients in Equation 3-1, thereby establishing a low-resolution PM2.5 concentration dataset for the study area. 2.5 / Response relationship between O3 concentration and changes in precursor concentration.

[0115] The specific implementation method for determining polynomial coefficients using deep learning is as follows:

[0116] First, (2H+3) precursor emission sampling scenarios are designed, including: one second baseline scenario, where the precursor emissions for each large-scale unit are the actual emissions in the baseline year; and H+1 zero-emission scenarios for gaseous precursor emissions, where SO2 and NO are set for each large-scale unit respectively. X The levels of the four gaseous precursors—VOCs, NH3, and H+1—were all zero, as were the levels of all four gaseous precursors throughout the entire study area; 2.5 Precursor placement in a zero scenario, i.e., setting a single PM for each large-scale unit. 2.5 Emissions were zero, and a single PM2.5 concentration was set for the entire study area. 2.5Emissions are 0. The above (2H+3) precursor emission sampling scenarios are input into a three-dimensional air quality model to calculate the spatiotemporal distribution of pollutant concentrations, obtaining the PM2.5 concentration for each large-scale unit. 2.5 / O3 concentration and "PM 2.5 A multi-scenario dataset of "O3 concentration-precursor concentration response" indicator species concentrations was finally input into an existing polynomial response surface training model based on deep learning methods to directly obtain the response equation coefficients in Equation 3-1, thereby establishing a low-resolution PM2.5 concentration model for the study area. 2.5 / Response relationship between O3 concentration and changes in precursor concentration.

[0117] Step 105: Couple the response relationship of precursor concentration to changes in precursor emissions with the response relationship of air pollutant concentration to changes in precursor concentration to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions.

[0118] During implementation, the response relationship between high-resolution precursor concentrations and precursor emissions will be compared with that of low-resolution PM2.5 concentrations. 2.5 The response relationship between O3 concentration and precursor concentration was coupled to construct a PM2.5 concentration-produced model. 2.5 A high-resolution response surface model of / O3 concentration to changes in precursor emissions is used to screen for refined precursor emission reduction strategies.

[0119] In one embodiment, the response relationship of precursor concentration to changes in precursor emissions is coupled with the response relationship of air pollutant concentration to changes in precursor concentration to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions, including:

[0120] The response relationship between atmospheric pollutant concentration and precursor concentration is obtained by replacing the precursor concentration change in the response relationship between precursor concentration and precursor emission with the precursor concentration change in the response relationship between precursor concentration and precursor emission.

[0121] Based on the response relationship between atmospheric pollutant concentration and precursor emission changes, a surface model of atmospheric pollutant concentration response to precursor emission changes is formed.

[0122] The study area was divided into (I×J) fine grid regions in the Lagrange diffusion model and into H large-scale units in the three-dimensional air quality model, where I×J is much larger than H. Therefore, each large-scale unit h includes several fine grid regions. This embodiment of the invention assumes that within the same large-scale unit, changes in precursor concentration affect PM2.5 concentrations. 2.5The effect of O3 concentration is consistent across all fine-grid regions; that is, the response relationships of several fine-grid regions within the same large-scale cell h all follow the pattern of the average response of the large-scale cell h. Therefore, the PM2.5 concentration in each grid region established in the three-dimensional air quality model is processed... 2.5 Equation 3-1, which describes the response of O3 concentration to changes in precursor concentration over a large area of ​​unit h, can be extended to the PM concentration response of each grid region. 2.5 The response of O3 concentration to changes in precursor concentration in each grid area is given by equation 2-9, which can be used to replace the changes in precursor concentration in each grid area. 2.5 The high-resolution response relationship between O3 concentration and precursor emissions is shown below:

[0123]

[0124]

[0125] In the formula, CONCreduce ij The dependent variable is the PM2.5 concentration in the region where the receptor grid in the i-th row and j-th column is located. 2.5 Change in O3 concentration; Qreduce_VOCs ij QPreduce_PPM ij Qreduce_NH3 ij Qreduce_SO2 ij and Qreduce_NOX ij These represent the precursor VOCs and primary PM2.5 in the region where the grid in row i and column j is located, respectively. 2.5 NH3, SO2 and NO X Concentration change; the change in precursor concentration is related to the change in precursor emissions, K_VOCs r_ab K_PPM r_ab K_NH3 r_ab K_SO2 r_ab and K_NOX r_ab These represent the precursors VOCs and primary PM2.5 in the region where the source grid in row a and column b of the pollution control measures is located. 2.5 NH3, SO2 and NO X The emission reduction ratio; Qyear_VOCs r_ab_ij Qyear_PPM r_ab_ij Qyear_NH3 r_ab_ij Qyear_SO2 r_ab_ij and Qyear_NOX r_ab_ij These represent the precursors VOCs and primary PM2.5 in the emission source sector r of the region where the source grid in row a and column b is located, respectively. 2.5NH3, SO2 and NO X The change in annual average concentration of precursors in the receptor grid at row i and column j caused by a unit reduction in emissions; Ebase_VOCs r_ab Ebase_PPM r_ab Ebase_NH3 r_ab Ebase_SO2 r_ab and Ebase_NOX r_ab f represents the precursor VOCs and primary PM2.5 in the emission source sector r of the region where the source grid in row a and column b is located, respectively. 2.5 NH3, SO2 and NO X Baseline annual emissions; Qbase_VOCs r_ij Qbase_PPM r_ij Qbase_NH3 r_ij Qbase_SO2 r_ij and Qbase_NOX r_ij These represent the precursors VOCs and primary PM2.5 in the emission source sector r of the region where the source grid in row a and column b is located, respectively. 2.5 NH3, SO2 and NO X The change in the annual average concentration of precursors in the receptor grid (row i, column j) caused by a unit reduction in emissions; R represents the total number of emission source sectors; I represents the total number of grid rows in the study area; J represents the total number of grid columns in the study area. The number of source grids and receptor grids is the same. Given that the Lagrange diffusion model has relatively low computational resource requirements, the study area can be infinitely subdivided to a certain extent by setting the grid resolution of the Lagrange diffusion model, thereby realizing the response relationship between the change in precursor concentration in the receptor grid and the change in precursor emissions in the source grid at a high resolution.

[0126] The aforementioned "high-resolution PM" 2.5 The "response relationship between O3 concentration and changes in precursor emissions" can abstractly express the complex three-dimensional air quality model in a mathematical form.

[0127] In one embodiment, after obtaining the surface model of the response of atmospheric pollutant concentration to changes in precursor emissions, the method of this embodiment may further include: using the surface model of the response of atmospheric pollutant concentration to changes in precursor emissions to screen emission reduction measures from a plurality of emission reduction measures to be implemented that have air quality benefits exceeding a set threshold.

[0128] To ensure the accuracy of this high-resolution response surface model, this embodiment of the invention compares and verifies it with the simulation results of another three-dimensional air quality model, as follows:

[0129] This embodiment sets up 30 multi-grid emission sampling scenarios and 30 single-grid sampling scenarios for the validation of the high-resolution response surface model. The 30 multi-grid emission reduction scenarios are implemented as follows: First, the number of emission reduction grids in 30 groups is determined by random sampling, where the number of grids in each group ranges from 2 to the total number of grids in the study area. The emission reduction ratios of the randomly determined emission reduction grids are set using the Latin hypercube sampling method, specifically including: 10 emission reduction scenarios with an average emission rate of 0.5 and a variance of 0.4; and 20 emission reduction scenarios with an average emission rate of 0.1 and a variance of 0.03. The 30 single-grid emission reduction scenarios are implemented as follows: First, 30 different individual emission reduction grids are determined by random sampling. Specifically, the emission reduction ratios were set for randomly determined emission reduction grids using the Latin hypercube sampling method, including: 10 emission reduction scenarios with an average emission rate of 0.5 and a variance of 0.4; and 20 emission reduction scenarios with an average emission rate of 0.1 and a variance of 0.03.

[0130] By rationally designing the distribution characteristics of emission reduction scenarios using the aforementioned scientific sampling methods, it is possible to effectively cover combinations of emission reduction intensities and regions. The 30 multi-grid emission reduction scenarios and 30 single-grid emission reduction scenarios are then input into a high-resolution response surface model and a three-dimensional air quality model for comparison, respectively, for calculation to obtain the PM2.5 concentration in each grid region under the corresponding scenario. 2.5 / O3 simulated concentration.

[0131] The verification results show that the results of the high-resolution response surface model and the three-dimensional air quality model are close, indicating that the embodiments of the present invention have successfully constructed PM. 2.5 A high-resolution response prediction model for O3 concentration to precursor emissions.

[0132] Figure 5 This is another specific example of the response surface model construction method for atmospheric pollutant concentration analysis in this invention, as shown in the figure. Figure 5 As shown, using a surface model of the response of atmospheric pollutant concentration to changes in precursor emissions, emission reduction measures that exceed a set threshold in terms of air quality benefits can be selected from multiple proposed emission reduction measures. These measures may include:

[0133] Step 501: Obtain data on emission reduction measures to be implemented from multiple emission source sectors;

[0134] Step 502: Calculate the precursor emission reduction ratio for each emission reduction measure in each emission source sector using the emission factor method;

[0135] Step 503: Input the precursor emission reduction ratio corresponding to each emission reduction measure of each emission source sector into the atmospheric pollutant concentration response surface model to precursor emission changes, and calculate and output the atmospheric pollutant concentration change corresponding to each emission reduction measure.

[0136] Step 504: Based on the change in atmospheric pollutant concentration corresponding to each emission reduction measure, screen emission reduction measures whose air quality benefits exceed the set threshold.

[0137] During implementation, based on the massive emission reduction measures provided by relevant enterprises, the emission factor method is used to calculate the emission reduction ratio of atmospheric precursors corresponding to each emission reduction measure for each emission source sector. The emission reduction ratio of each sector and pollutant is then input into response equation 4-1 to quickly obtain the PM2.5 reduction ratio corresponding to the massive emission reduction measures. 2.5 The change in O3 concentration is used to screen for emission reduction measures that offer the greatest benefit to air quality. The following are some specific application scenarios:

[0138] The method of this invention can be applied to specific scenarios involving the elimination of highly polluting enterprises in cities. First, the contribution ratio of each enterprise to precursor emissions in a city is calculated using the emission factor method. Then, the method of this invention is used to construct a high-resolution response surface model of the city, thereby rapidly calculating the contribution of each enterprise to PM2.5 emissions. 2.5 The contribution of O3 concentration. Through this process, the contribution of PM2.5 in a city can be identified. 2.5 The main enterprises that have a significant impact on O3 concentration are identified, thus clarifying the high-pollution enterprises that need to be prioritized for elimination in air pollution prevention and control.

[0139] The method of this invention can be applied to specific scenarios of green transformation analysis in industrial parks. First, the contribution ratio of each production process to precursor emissions is calculated using the emission factor method. Then, the method of this invention is used to construct a high-resolution response surface model of the industrial park, thereby rapidly calculating the individual contribution of hundreds or thousands of production processes in the park to PM2.5 emissions. 2.5 The contribution of O3 concentration. Through this process, the contribution of PM2.5 to PM2.5 concentration can be identified. 2.5 The study identifies key production processes that have a significant impact on O3 concentration, thus clarifying the critical production links that need to be prioritized and improved during the green transformation process.

[0140] The embodiments of the present invention have the following beneficial technical effects:

[0141] (1) This invention combines Lagrange diffusion model technology, three-dimensional air quality model technology, and polynomial response surface technology to propose a novel method for constructing high-resolution response surface models. This method effectively reduces the direct application scope of complex three-dimensional air quality models, and significantly reduces the computation time and resource costs required for constructing response surface models while ensuring simulation accuracy, thereby significantly improving the modeling efficiency of response surface models.

[0142] (2) The embodiments of the present invention construct a high-resolution PM 2.5 The response surface model of O3 concentration to changes in precursor emissions can quickly quantify the impact of precursor emission changes on PM2.5 in small-scale areas such as industrial parks, specific facilities, and specific production processes. 2.5 The impact of O3 concentration. This high-resolution response surface model provides a scientific basis for the formulation and implementation of refined pollution prevention and control measures such as "one policy per city" and "one policy per enterprise," which helps to achieve more precise and effective air pollution control.

[0143] This invention also provides a response surface model construction device for atmospheric pollutant concentration analysis, as described in the following embodiments. Since the principle behind this device is similar to the response surface model construction method for atmospheric pollutant concentration analysis, its implementation can refer to the implementation of the response surface model construction method for atmospheric pollutant concentration analysis; repeated details will not be elaborated further.

[0144] Figure 6 This is a schematic diagram of a response surface model construction device for atmospheric pollutant concentration analysis in an embodiment of the present invention, as shown below. Figure 6 As shown, the device 600 includes:

[0145] The diffusion model processing module 601 is used to simulate precursor emissions in the study area based on the Lagrange diffusion model and generate precursor emission concentration data. The precursor emission concentration data includes the precursor concentrations after emissions from different precursor emission sources. Based on the precursor emission concentration data, statistical methods are used to establish the response relationship between precursor concentration and changes in precursor emissions.

[0146] The three-dimensional air quality model processing module 602 is used to simulate precursor emission scenarios in the study area based on the three-dimensional air quality model and generate scenario simulation result data. The scenario simulation result data includes the concentration of air pollutants after different precursor emissions. Based on the polynomial response surface method, the scenario simulation result data is used to establish the response relationship between air pollutant concentration and precursor concentration changes.

[0147] The coupling module 603 is used to couple the response relationship of precursor concentration to changes in precursor emissions with the response relationship of atmospheric pollutant concentration to changes in precursor concentration to obtain a surface model of atmospheric pollutant concentration response to changes in precursor emissions.

[0148] In one embodiment, the diffusion model processing module 601 is specifically used to: simulate precursor emissions for each grid divided within the study area based on the Lagrange diffusion model, and generate a precursor emission concentration dataset; the precursor emission concentration dataset reflects the precursor concentration of each grid at a set time after only a single grid cell emits and no other grid cells emit.

[0149] In one embodiment, the Lagrange diffusion model includes a meteorological module and a Lagrange-Gaussian smoke cloud module;

[0150] The meteorological module is used to generate time-varying and space-varying meteorological fields using topographic elevation data and land use data;

[0151] The Lagrange Gaussian plume module is used to simulate the diffusion and transformation processes along the transport path of plumes released from precursor emission sources by using the meteorological field generated by the meteorological module.

[0152] In one embodiment, the diffusion model processing module 601 is specifically used for:

[0153] Configure a first baseline scenario and multiple precursor discharge zero placement scenarios; the first baseline scenario represents that the emission source strength of each grid region is the actual emission source strength of the baseline year, and the precursor discharge zero placement scenarios represent that each grid region and each emission source sector are respectively set up with precursor discharge zero placement scenarios;

[0154] The first baseline scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations, resulting in the first precursor concentration matrix. The first precursor concentration matrix includes the precursor concentrations in the region where each grid is located in the baseline year.

[0155] Each precursor placement zero scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentration, resulting in multiple second precursor concentration matrices. The second precursor concentration matrix includes the precursor concentration of each grid corresponding to the precursor placement zero scenario.

[0156] A precursor emission concentration dataset is formed based on a first precursor concentration matrix and multiple second precursor concentration matrices; the precursor emission concentration dataset includes concentration information from different emission sources, different regional locations, and different time resolutions.

[0157] In one embodiment, the diffusion model processing module 601 is specifically used for:

[0158] Based on precursor emission concentration data, the following regression model is used to establish the response relationship between precursor concentration and changes in precursor emissions using statistical methods:

[0159]

[0160] Among them, Qreduce ij This represents the change in precursor concentration in the region of the receptor grid in row i and column j; Ereduce ab The value represents the change in precursor emissions in the region where the source grid in row a and column b is located; I represents the total number of grid rows in the study area, and J represents the total number of grid columns in the study area. The number of receptor grids and source grids is the same, which is I×J; K and M are regression model parameters.

[0161] In one embodiment, the regression model parameter M is determined using a first precursor concentration matrix, and the regression model parameter K is determined using multiple second precursor concentration matrices.

[0162] In one embodiment, the diffusion model processing module 601 is specifically used for:

[0163] The following formula is used to predict changes in output precursor concentrations based on the precursor reduction ratio of the emission source sector:

[0164]

[0165] In the formula, Qreduce p_ij K is the dependent variable, representing the predicted concentration change of precursor p in the region where the receptor grid in the i-th row and j-th column is located; p_r_ab The independent variable is Qyear, which represents the emission reduction ratio of precursor p in the region where the source grid in row a and column b is located in the pollution prevention and control measures. p_r_ab_ij Ebase p_r_ab and Qbase p_r_ij All are constants, Qyear p_r_ab_ij Ebase represents the change in the annual average concentration of precursors in the receptor grid (row i, column j) caused by a unit reduction in the emission of precursor p in the emission source sector r of the region where the source grid (row a, column b) is located. p_r_ab This represents the baseline annual emissions of precursor p from emission source sector r in the region where the source grid in row a and column b is located; Qbase p_r_ij The reference concentration of precursor p from emission source sector r is represented in the region where the receptor grid in row i and column j is located; R represents the total number of emission source sectors; I represents the total number of grid rows in the study area; J represents the total number of grid columns in the study area; the number of source grids and receptor grids is the same.

[0166] In one embodiment, the precursor emission source includes one or any combination of the following pollution source types:

[0167] Surface source, point source, line source.

[0168] In one embodiment, the three-dimensional air quality model processing module 602 is specifically used for:

[0169] Based on a three-dimensional air quality model, precursor emission scenarios are simulated for each large-scale unit in the study area, generating scenario simulation result data; the number of large-scale units is less than the number of grids.

[0170] In one embodiment, the three-dimensional air quality model processing module 602 is specifically used for:

[0171] Configure precursor emission sampling scenarios; precursor emission sampling scenarios include a second baseline scenario, a gaseous precursor emission control scenario, or multiple gaseous precursor emission zeroing scenarios; the second baseline scenario represents the actual emission amount of precursors in each large-scale unit as the baseline year, the gaseous precursor emission control scenario is a control scenario in which the precursor emission amount in each large-scale unit is set to be several times the precursor emission amount in the second baseline scenario, and the gaseous precursor emission zeroing scenario is a scenario in which the precursor emission in each large-scale unit is set to zero;

[0172] The emission sampling scenario of precursors is input into a three-dimensional air quality model to calculate the spatiotemporal distribution of pollutant concentrations and obtain scenario simulation results data.

[0173] In one embodiment, the three-dimensional air quality model processing module 602 is specifically used to: construct a nonlinear polynomial expression for the concentration of air pollutants and precursors based on the polynomial response surface method and using scenario simulation result data. In the nonlinear polynomial expression, the dependent variable is the concentration of air pollutants in the region where each grid is located, and the independent variable is the change in the concentration of multiple precursors in each large-scale unit. The nonlinear polynomial expression includes multiple polynomial coefficients to be determined. The polynomial coefficients to be determined are determined by a nonlinear regression fitting method or a deep learning method.

[0174] In one embodiment, the coupling module 603 is specifically used for:

[0175] The response relationship between atmospheric pollutant concentration and precursor concentration is obtained by replacing the precursor concentration change in the response relationship between precursor concentration and precursor emission with the precursor concentration change in the response relationship between precursor concentration and precursor emission.

[0176] Based on the response relationship between atmospheric pollutant concentration and precursor emission changes, a surface model of atmospheric pollutant concentration response to precursor emission changes is formed.

[0177] In one embodiment, the device 600 further includes:

[0178] The emission reduction measure screening and analysis module is used to screen emission reduction measures whose air quality benefits exceed a set threshold from multiple emission reduction measures to be implemented after the atmospheric pollutant concentration response surface model to the change in precursor emissions is obtained by the coupling module 603.

[0179] In one embodiment, the emission reduction measure screening and analysis module is specifically used for:

[0180] Obtain data on emission reduction measures to be implemented from multiple emission source sectors;

[0181] The emission factor method was used to calculate the precursor emission reduction ratio for each emission reduction measure in each emission source sector;

[0182] The emission reduction ratio of precursors for each emission reduction measure for each emission source sector is input into the surface model of atmospheric pollutant concentration response to changes in precursor emissions, and the change in atmospheric pollutant concentration for each emission reduction measure is calculated and output.

[0183] Based on the change in atmospheric pollutant concentration corresponding to each emission reduction measure, emission reduction measures that have air quality benefits exceeding a set threshold are selected.

[0184] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for constructing a response surface model for atmospheric pollutant concentration analysis.

[0185] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for constructing a response surface model for atmospheric pollutant concentration analysis.

[0186] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method for constructing a response surface model for atmospheric pollutant concentration analysis.

[0187] This invention combines Lagrange diffusion modeling, three-dimensional air quality modeling, and polynomial response surface methodology to establish a high-resolution response surface model. Compared with existing technologies, this invention effectively reduces the direct application scope of complex three-dimensional air quality models. While ensuring simulation accuracy, it significantly reduces the computation time and resource costs required for constructing the response surface model, thereby significantly improving the modeling efficiency. The high-resolution response surface model in this invention can quickly quantify the impact of small-scale emission source changes on PM2.5 in small-scale areas such as industrial parks, specific facilities, and specific production processes. 2.5The impact of atmospheric pollutant concentrations such as O3.

[0188] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0189] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0190] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0191] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0192] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing a response surface model for atmospheric pollutant concentration analysis, characterized in that, include: Precursor emissions were simulated in the study area based on the Lagrange diffusion model, and precursor emission concentration data were generated. The precursor emission concentration data includes the precursor concentrations after emissions from different precursor emission sources; Based on precursor emission concentration data, statistical methods were used to establish the response relationship between precursor concentration and changes in precursor emissions. Based on a three-dimensional air quality model, a scenario simulation of precursor emissions was conducted in the study area to generate scenario simulation result data; the scenario simulation result data includes the concentration of air pollutants after the emission of different precursors; Based on the polynomial response surface methodology, and using scenario simulation data, the response relationship between atmospheric pollutant concentration and precursor concentration changes is established. The response relationship of precursor concentration to changes in precursor emissions is coupled with the response relationship of air pollutant concentration to changes in precursor concentration to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions. The process involves simulating precursor emissions in the study area based on the Lagrange diffusion model to generate precursor emission concentration data. This includes: simulating precursor emissions in each grid within the study area based on the Lagrange diffusion model to generate a precursor emission concentration dataset. The precursor emission concentration dataset reflects the precursor concentration of each grid at a set time after only a single grid cell emits and no other grid cells emit. The Lagrange diffusion model includes a meteorological module and a Lagrange-Gaussian smoke cloud module. The meteorological module is used to generate a time-varying and spatially varying meteorological field using topographic elevation data and land use data. The Lagrange-Gaussian smoke cloud module is used to use the meteorological field generated by the meteorological module to simulate the diffusion and transformation process along the transport path of the smoke cloud released from the precursor emission source by advection. Among them, precursor emissions were simulated for each grid within the study area based on the Lagrange diffusion model, generating a precursor emission concentration dataset, including: Configure a first baseline scenario and multiple precursor discharge zero placement scenarios; the first baseline scenario represents that the emission source strength of each grid region is the actual emission source strength of the baseline year, and the precursor discharge zero placement scenarios represent that each grid region and each emission source sector are respectively set up with precursor discharge zero placement scenarios; The first baseline scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations, resulting in the first precursor concentration matrix. The first precursor concentration matrix includes the precursor concentrations in the region where each grid is located in the baseline year. Each precursor placement zero scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentration, resulting in multiple second precursor concentration matrices. The second precursor concentration matrix includes the precursor concentration of each grid corresponding to the precursor placement zero scenario. A precursor emission concentration dataset is formed based on a first precursor concentration matrix and multiple second precursor concentration matrices; the precursor emission concentration dataset includes concentration information from different emission sources, different regional locations, and different time resolutions.

2. The method as described in claim 1, characterized in that, Based on precursor emission concentration data, statistical methods were used to establish the response relationship between precursor concentration and changes in precursor emissions, including: Based on precursor emission concentration data, the following regression model is used to establish the response relationship between precursor concentration and changes in precursor emissions using statistical methods: in, This represents the change in precursor concentration in the region where the receptor grid in row i and column j is located; The value represents the change in precursor emissions in the region where the source grid in row a and column b is located; I represents the total number of grid rows in the study area, and J represents the total number of grid columns in the study area. The number of receptor grids and source grids is the same, which is I×J; K and M are regression model parameters.

3. The method as described in claim 2, characterized in that, The regression model parameters M are determined using the first precursor concentration matrix, and the regression model parameters K are determined using multiple second precursor concentration matrices.

4. The method as described in claim 2, characterized in that, Based on precursor emission concentration data, statistical methods were used to establish the response relationship between precursor concentration and changes in precursor emissions, including: The following formula is used to predict changes in output precursor concentrations based on the precursor reduction ratio of the emission source sector: In the formula, The dependent variable represents the predicted concentration change of precursor p in the region where the receptor grid in the i-th row and j-th column is located. The independent variable represents the emission reduction ratio of precursor p in the region where the source grid in row a and column b is located in the pollution prevention and control measures. , and All are constants. This represents the change in the annual average concentration of precursors in the receptor grid (row i, column j) caused by a unit reduction in the emission of precursor p in the emission source sector r of the region where the source grid (row a, column b) is located. This represents the baseline annual emissions of precursor p from emission source sector r in the region where the source grid in row a and column b is located. The reference concentration of precursor p from emission source sector r is represented in the region where the receptor grid in row i and column j is located; R represents the total number of emission source sectors; I represents the total number of grid rows in the study area; J represents the total number of grid columns in the study area; the number of source grids and receptor grids is the same.

5. The method according to any one of claims 1 to 4, characterized in that, The precursor emission sources include one or any combination of the following pollution source types: Surface source, point source, line source.

6. The method as described in claim 4, characterized in that, Based on a three-dimensional air quality model, a scenario simulation of precursor emissions was conducted in the study area, generating scenario simulation result data, including: Based on a three-dimensional air quality model, precursor emission scenarios are simulated for each large-scale unit in the study area, generating scenario simulation result data; the number of large-scale units is less than the number of grids.

7. The method as described in claim 6, characterized in that, Based on a three-dimensional air quality model, precursor emission scenarios were simulated for various large-scale units within the study area, generating scenario simulation result data, including: Configure precursor emission sampling scenarios; precursor emission sampling scenarios include a second baseline scenario, a gaseous precursor emission control scenario, or multiple gaseous precursor emission zeroing scenarios; the second baseline scenario represents the actual emission amount of precursors in each large-scale unit as the baseline year, the gaseous precursor emission control scenario is a control scenario in which the precursor emission amount in each large-scale unit is set to be several times the precursor emission amount in the second baseline scenario, and the gaseous precursor emission zeroing scenario is a scenario in which the precursor emission in each large-scale unit is set to zero; The emission sampling scenario of precursors is input into a three-dimensional air quality model to calculate the spatiotemporal distribution of pollutant concentrations and obtain scenario simulation results data.

8. The method as described in claim 7, characterized in that, Based on the polynomial response surface methodology, and using scenario simulation data, the response relationship between atmospheric pollutant concentrations and precursor concentration changes is established, including: Based on the polynomial response surface method, a nonlinear polynomial expression for the concentration of air pollutants and precursors is constructed using scenario simulation data. In the nonlinear polynomial expression, the dependent variable is the concentration of air pollutants in the region where each grid is located, and the independent variable is the change in the concentration of multiple precursors in each large-scale cell. The nonlinear polynomial expression includes multiple polynomial coefficients to be determined. The polynomial coefficients to be determined are determined by nonlinear regression fitting or deep learning methods.

9. The method as described in claim 1, characterized in that, The response relationship between precursor concentration and changes in precursor emissions is coupled with the response relationship between air pollutant concentration and changes in precursor concentration to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions, including: The response relationship between atmospheric pollutant concentration and precursor concentration is obtained by replacing the precursor concentration change in the response relationship between precursor concentration and precursor emission with the precursor concentration change in the response relationship between precursor concentration and precursor emission. Based on the response relationship between atmospheric pollutant concentration and precursor emission changes, a surface model of atmospheric pollutant concentration response to precursor emission changes is formed.

10. The method as described in claim 1, characterized in that, After obtaining the surface model of the response of atmospheric pollutant concentration to changes in precursor emissions, it also includes: By using a surface model of the response of atmospheric pollutant concentration to changes in precursor emissions, emission reduction measures with air quality benefits exceeding a set threshold can be selected from multiple emission reduction measures to be implemented.

11. The method as described in claim 10, characterized in that, Using a surface model of the response of atmospheric pollutant concentrations to changes in precursor emissions, emission reduction measures with air quality benefits exceeding a set threshold are selected from multiple proposed emission reduction measures, including: Obtain data on emission reduction measures to be implemented from multiple emission source sectors; The emission factor method was used to calculate the precursor emission reduction ratio for each emission reduction measure in each emission source sector; The emission reduction ratio of precursors for each emission reduction measure for each emission source sector is input into the surface model of atmospheric pollutant concentration response to changes in precursor emissions, and the change in atmospheric pollutant concentration for each emission reduction measure is calculated and output. Based on the change in atmospheric pollutant concentration corresponding to each emission reduction measure, emission reduction measures that have air quality benefits exceeding a set threshold are selected.

12. A response surface model construction device for atmospheric pollutant concentration analysis, characterized in that, include: The diffusion model processing module is used to simulate precursor emissions in the study area based on the Lagrange diffusion model and generate precursor emission concentration data. The precursor emission concentration data includes the precursor concentrations after emission from different precursor emission sources; based on the precursor emission concentration data, statistical methods are used to establish the response relationship between precursor concentration and changes in precursor emissions; The three-dimensional air quality model processing module is used to simulate precursor emission scenarios in the study area based on the three-dimensional air quality model and generate scenario simulation result data. The scenario simulation result data includes the concentration of air pollutants after different precursor emissions. Based on the polynomial response surface method, the response relationship between air pollutant concentration and precursor concentration changes is established using the scenario simulation result data. The coupling module is used to couple the response relationship of precursor concentration to changes in precursor emissions with the response relationship of air pollutant concentration to changes in precursor concentration, so as to obtain a surface model of the response of air pollutant concentration to changes in precursor emissions. Specifically, the diffusion model processing module is used to: simulate precursor emissions for each grid within the study area based on the Lagrange diffusion model, and generate a precursor emission concentration dataset; the precursor emission concentration dataset reflects the precursor concentration of each grid at a set time after only a single grid cell emits and no other grid cells emit; The Lagrange diffusion model includes a meteorological module and a Lagrange-Gaussian smoke cloud module. The meteorological module is used to generate time-varying and space-varying meteorological fields using topographic elevation data and land use data; The Lagrange Gaussian plume module is used to simulate the diffusion and transformation processes along the transport path of plumes released from precursor emission sources by using the meteorological field generated by the meteorological module. Specifically, the diffusion model processing module is used for: Configure a first baseline scenario and multiple precursor discharge zero placement scenarios; the first baseline scenario represents that the emission source strength of each grid region is the actual emission source strength of the baseline year, and the precursor discharge zero placement scenarios represent that each grid region and each emission source sector are respectively set up with precursor discharge zero placement scenarios; The first baseline scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentrations, resulting in the first precursor concentration matrix. The first precursor concentration matrix includes the precursor concentrations in the region where each grid is located in the baseline year. Each precursor placement zero scenario is input into the Lagrange diffusion model to calculate the spatiotemporal distribution of pollutant concentration, resulting in multiple second precursor concentration matrices. The second precursor concentration matrix includes the precursor concentration of each grid corresponding to the precursor placement zero scenario. A precursor emission concentration dataset is formed based on a first precursor concentration matrix and multiple second precursor concentration matrices; the precursor emission concentration dataset includes concentration information from different emission sources, different regional locations, and different time resolutions.

13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 11.

15. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 11.

Images