Method for constructing ozone pollution cause path based on causal inference and application

Abstract

The application discloses an ozone pollution cause path construction method based on causal inference, which comprises the following steps: step 1) taking the conventional influencing factors in ozone pollution cause analysis as explanatory variables of the ozone pollution cause path; step 2) performing significance test on each explanatory variable and defining confounding explanatory variables; step 3) determining key explanatory variables and intermediate variables by using Granger causality test, and initially building an intermediate path; step 4) performing test and correction on the intermediate path; step 5) determining a regulating variable by performing group difference test, and initially building a regulated intermediate path; and step 6) performing statistical test and correction on the regulated intermediate path, and evaluating the regulating effect of the regulating variable on the intermediate path. The ozone pollution cause path construction method provides theoretical guidance for clarifying the ozone pollution formation mechanism under a complex pollution system and proposing effective and feasible ozone pollution prevention and control measures.

Description

Technical Field

[0001] This invention relates to the field of ozone pollution causation analysis technology, and in particular to a method for constructing ozone pollution causal pathways based on causal inference, and the application of this ozone pollution causal pathway. Background Technology

[0002] Ozone (O3) is a typical secondary pollutant. Understanding the causes of ozone pollution is key to solving ozone pollution and is also the main focus of continuously improving air quality.

[0003] The causes of ozone pollution are highly complex. Existing ozone pollution causal analysis models mainly describe the combined effects of multiple factors on ozone within a complex pollution system, failing to accurately describe the interaction processes and mechanisms between these variables. Therefore, although ozone pollution causal analysis studies have been conducted in various regions, the key processes of ozone pollution formation remain unclear due to the interdependence and mutual influence of multiple variables influencing ozone, hindering the development of effective new strategies for ozone pollution control. Therefore, establishing a novel ozone pollution causal pathway, identifying factors significantly impacting ozone formation, and revealing the causal mechanisms of these factors within a complex pollution system are important directions for current ozone pollution research. Summary of the Invention

[0004] This invention provides a method for constructing the causal pathway of ozone pollution based on causal inference, which addresses the problems in the prior art, such as the unclear key pathway of ozone pollution formation due to the interaction between multiple factors, and the inability of targeted pollution control measures to effectively control ozone pollution due to non-causal spurious correlations. It provides theoretical guidance for clarifying the ozone pollution formation mechanism under a complex pollution system and proposing effective and feasible ozone pollution prevention and control measures.

[0005] The method for constructing ozone pollution causal pathways based on causal inference of the present invention includes the following steps:

[0006] Step 1) Using the conventional influencing factors in the analysis of ozone pollution causes as explanatory variables for the ozone pollution cause pathway, obtain the time series data of each explanatory variable;

[0007] Step 2) The significance of each explanatory variable is tested by stepwise regression to obtain the explanatory variables that have a significant impact on ozone, which are used as significant explanatory variables in the ozone pollution causal path. Explanatory variables other than significant explanatory variables and other unconventional influencing factors that may affect ozone are defined as confounding explanatory variables. The confounding explanatory variables are represented by the residuals of stepwise regression simulation of ozone.

[0008] Step 3) analyze the causal relationship between significant explanatory variables, significant explanatory variables and ozone by using Granger causality test, select key explanatory variables and intermediate variables, and preliminarily build an ozone pollution cause path with intermediate effect based on the intermediate effect model, which is called intermediate path;

[0009] Step 4) chemical test and correction of the intermediate path by using MCM box model; statistical test and correction of the intermediate path by using Bootstrap method, and further evaluation of the direct effect and indirect effect of the intermediate path by using Bootstrap method, that is, evaluation of the influence strength of the intermediate path on ozone and the proportion of each path influence;

[0010] Step 5) group difference test of other significant explanatory variables and mixed explanatory variables except key explanatory variables and intermediate variables; the significant explanatory variables in the group difference are determined as adjustment variables, and the intermediate path with adjustment is preliminarily built based on the adjustment effect model;

[0011] Step 6) chemical test and correction of the intermediate path with adjustment by using MCM box model; statistical test and correction of the intermediate path with adjustment by using Bootstrap method, and further evaluation of the adjustment effect of the adjustment variable on the intermediate path by using Bootstrap method.

[0012] The present application explores the ozone pollution cause path under the joint action of multiple factors in the big data background, determines the key explanatory variables and intermediate variables based on the causal relationship and chemical mechanism, evaluates the direct effect and indirect effect of the intermediate path on ozone by using the intermediate effect model, determines the adjustment variables through group difference test, evaluates the influence of the adjustment variables on the intermediate path by using the adjustment effect model, so as to determine the possible causal action mechanism and pollution cause path between the influencing factors and ozone, improve the utilization efficiency of monitoring data in the big data era, and effectively mine the potential influence of multiple factors on ozone, and provide technical support for ozone pollution control.

[0013] At present, two commonly used ozone pollution cause analysis models are chemical transport model and statistical model. The chemical transport model can explore the process of generating ozone pollution by physical and chemical reactions of multiple influencing factors, but is greatly limited by the chemical mechanism embedded in the model, such as different simulation results caused by different chemical mechanisms embedded in the model, and the problem that the common chemical mechanism does not consider multiphase reaction; in addition, the data quality requirement of the chemical transport model is high, and it is difficult to obtain high-resolution emission inventory data.

[0014] Statistical models are based on observational data, which effectively avoids the problem of data acquisition difficulty. Statistical models mainly explore the statistical correlation between various influencing factors and ozone, and according to the correlation, targeted emission reduction measures can be proposed to achieve the purpose of reducing ozone concentration. For example, PM 10 Usually presents a negative correlation with ozone, which is mainly due to the increase in particulate matter concentration, which increases the aerosol optical thickness, leading to a decrease in the near-surface photolysis rate of NO2 and O3. According to this correlation, PM 10 can be regulated to achieve the purpose of controlling ozone pollution. However, in a complex atmospheric pollution system, only relying on correlation to determine the causes of ozone pollution may lead to illusory correlation and misleading conclusions. For example, PM 2.5 has a significant positive correlation with ozone, which is not due to PM 2.5 promoting the generation of ozone, but rather due to the presence of common precursors between PM 2.5 and ozone. At this time, regulating PM 2.5 has little effect on controlling ozone pollution. Therefore, we need to introduce causal inference methods in the analysis of the causes of ozone pollution, and according to the causal relationship between the explanatory variables and the changes in ozone, we can analyze the real causes of ozone pollution.

[0015] The definition of causal relationship is that the cause is the phenomenon that causes a certain phenomenon; the result is the phenomenon that is caused by the action of the cause and is linked in series. In other words, the occurrence of A leads to the occurrence of B, A is the cause of B, and B is the result of A. However, this definition is not clear in the field of ozone. The occurrence of A leads to the occurrence of B can be understood as A generating B through chemical reaction (direct generation), or A affecting the reaction process leading to changes in the concentration of B (participating in the ozone generation process). Therefore, the introduction of causal inference first needs to clarify who is the cause and who is the result.

[0016] According to the characteristics of the cause and effect: the cause is first, the result is later, the appearance of the former leads to the latter. The cause and effect is often judged according to the order of time occurrence. In the field of ozone, the cause and effect exist at the same time (the precursors of ozone and ozone exist at the same time in the actual atmosphere), and there is a reverse cause and effect (for example, NO2 is a precursor of ozone, it can generate ozone through photolysis reaction, and ozone can also affect the generation of NO2 by affecting the oxidation of the atmosphere). Therefore, the present application takes the time sequence in the statistical cause and effect test and the generation relationship in the chemical generation mechanism as the standard for determining the cause and effect. The Granger cause and effect test and the MCM box model are combined to determine the cause and effect. The Granger cause and effect test can test the time sequence of the variables in statistics, that is, the time of A and B occurrence is used to determine the cause and effect of A and B. This method is simple in algorithm and has low requirements for data. However, this relationship is not a real cause and effect, so it must be further combined with the chemical reaction mechanism in the MCM box model for judgment. According to the chemical reaction in which A and B participate in the actual atmosphere, the reactants that will generate ozone are determined as the cause (including the reactants that directly produce ozone through photolysis reaction and the reactants that indirectly produce ozone through chain reaction, etc.). It should be noted that theoretically, the MCM box model can also be used alone to determine the cause and effect, but in the actual atmospheric environment, multiple physical and chemical reaction processes occur simultaneously and are coupled with each other, so it is difficult to determine the cause and effect by using the MCM box model alone.

[0017] The core content of the ozone pollution cause path is to build an intermediate path and a mediated intermediate path with adjustment. Among them, the intermediate path building contains three parts: the determination of key explanatory variables and intermediate variables, the test and correction of the intermediate path, and the evaluation of the intermediate effect; the mediated intermediate path contains three parts: the determination of the adjustment variable, the test and correction of the mediated intermediate path, and the evaluation of the adjustment effect.

[0018] The method for determining the key explanatory variables and the intermediate variables of the present application is the combination of statistical Granger cause and effect test and chemical mechanism, that is, the cause and effect determined by the Granger cause and effect test and the MCM box model are used to select the key explanatory variables and the intermediate variables. The method for determining the adjustment variable of the present application is the combination of statistical difference test and reaction process in chemical generation mechanism. The difference test between groups can test whether there is a significant difference between the high group and the low group of the explanatory variable on the influence of ozone, but this significant difference only proves that the influence of the explanatory variable on ozone is strong or weak, and cannot prove that the explanatory variable will affect the cause and effect, which needs to be further combined with the MCM box model to judge whether there is a chemical reaction mechanism that the explanatory variable affects the cause and effect. Due to the complexity of the chemical mechanism, it is impossible to determine the adjustment variable by using the MCM box model alone.

[0019] The inspection and correction of the mediation path mainly includes chemical mechanism inspection and correction and statistical inspection and correction. The chemical mechanism inspection and correction is to determine whether the initially built mediation path conforms to the ozone generation mechanism by using the MCM box model, and to correct the path that does not conform to the ozone generation mechanism. The statistical inspection and correction is to determine whether the initially built mediation path meets the statistical requirements by using the Bootstrap inspection, and to correct the statistically insignificant path.

[0020] The inspection and correction of the mediation path mainly includes chemical mechanism inspection and correction and statistical inspection and correction. The chemical mechanism inspection and correction is to determine whether the initially built mediation path conforms to the ozone generation mechanism by using the MCM box model, and to correct the path that does not conform to the ozone generation mechanism. The statistical inspection and correction is to determine whether the initially built mediation path meets the statistical requirements by using the Bootstrap inspection, and to correct the statistically insignificant path.

[0021] As a limitation of the above technical solutions, the conventional influencing factors in the ozone pollution cause analysis are the factors recognized in the field of ozone pollution cause analysis that have a greater impact on ozone, specifically including meteorological factors such as ultraviolet radiation intensity, relative humidity, temperature, air pressure, wind speed, and wind direction; and atmospheric pollutant factors such as PM 10 , PM 2.5 , NO2, SO2, CO, NMVOCs, etc. The unconventional influencing factors are factors recognized in the field of ozone pollution cause analysis that have a relatively small impact on ozone or are difficult to obtain data, including meteorological factors such as rainfall, cloud cover, and boundary layer height, pollutant factors such as HONO and VOCs species, and other factors such as population density, ground roughness, and land use type.

[0022] As a limitation of the above technical solutions, the selection of the significant explanatory variable needs to meet the following conditions:

[0023] The significant explanatory variable is selected one by one from large to small according to the normalized coefficient of the explanatory variable, and it is required that the significant explanatory variable has a higher degree of explanation of ozone than the threshold value of the stepwise regression model; at least three significant explanatory variables are selected, which are one key explanatory variable, one mediation variable, and one adjustment variable; the number of significant explanatory variables should be adjusted according to the path inspection results and actual application requirements.

[0024] As a limitation of the above technical solutions, the calculation formula of the confounding explanatory variable is:

[0025]

[0026] Wherein, x k represents the time series data of the significant explanatory variable; β k represents the regression coefficient. ε represents the time series data of ozone simulated by stepwise regression; O3 represents the time series data of ozone; ε represents the time series data of confounding explanatory variables.

[0027] As a limitation on the above technical solution, the mediation effect model is as follows:

[0028] Y = cX + e1

[0029] M = aX + e2

[0030] Y = c′X + bM + e3

[0031] Where X is the key explanatory variable, Y is ozone, M is the mediating variable, and a, b, c, and c′ are the coefficients of the mediation effect test results, where c is the total effect, c′ is the direct effect, and ab is the indirect effect; e1, e2, and e3 are the constants of the test results.

[0032] As a limitation of the above technical solution, the Granger causality test is used to determine the causal relationship between significant explanatory variables and between significant explanatory variables and ozone, to determine the position of significant explanatory variables in the ozone pollution causal pathway, and to identify the key explanatory variable X and the mediating variable M.

[0033] As a limitation on the above technical solution, X and M must meet the following requirements: X is the cause of M and Y, and M is the cause of Y.

[0034] Identifying key explanatory variables, mediating variables, and moderating variables is fundamental to constructing the causal pathways of ozone pollution. The determination of key explanatory and mediating variables is based on the definition of mediating effects. A mediating effect refers to a cause influencing a result through one or more intermediate variables; these intermediate variables are called mediating variables. For example... Figure 1 The diagram illustrates the mediating effect, where Y represents ozone, X represents the key explanatory variable, and M represents the mediating variable. X and M are determined based on the following condition: X is a cause of both M and Y, and M is a cause of Y.

[0035] The intermediate path must satisfy both statistical and chemical tests.

[0036] As a limitation of the above technical solution, the direct and indirect effects in the mediation path are evaluated according to the following steps:

[0037] Step 4.1) Test the total effect c using the Bootstrap method. If it is significant, consider it a mediating effect; otherwise, consider it a masking effect. Regardless of whether it is significant or not, proceed to the next step of testing.

[0038] Step 4.2) Test coefficients a and b using the Bootstrap method. If both are significant, the indirect effect is significant, and proceed to step 4.4); if at least one is not significant, proceed to step 4.3.

[0039] Step 4.3) test the indirect effect ab by Bootstrap method; if significant, the indirect effect is significant, proceed to step 4.4); otherwise the indirect effect is not significant, stop the analysis;

[0040] Step 4.4) after determining the existence of the mediation effect, test the direct effect c' by Bootstrap method; if not significant, the direct effect is not significant, indicating that there is only mediation effect; if significant, the direct effect is significant, proceed to step 4.5);

[0041] Step 4.5) compare the signs of ab and c; if the same, it belongs to partial mediation effect, report the proportion of mediation effect to total effect ab / c; if different, it belongs to masking effect, report the absolute value of the ratio of indirect effect to direct effect |ab / c|.

[0042] As a limitation of the above technical solution, the adjustment effect model is:

[0043] Y = β0 + β1X + β2W + β3WX + ε

[0044] Where X is the key explanatory variable, Y is ozone, W is the adjustment variable, β is the coefficient of the adjustment effect test result, and ε is the constant of the test result.

[0045] The mediated path with adjustment includes three parts: determination of the adjustment variable, test and correction of the mediated path with adjustment, and evaluation of the adjustment effect.

[0046] As a limitation of the above technical solution, the inter-group difference test includes the following steps:

[0047] Step 5.1) import time series data of other significant explanatory variables except the key explanatory variable and the mediation variable and corresponding ozone time series data, import time series data of confounding explanatory variables and corresponding ozone time series data, and analyze each explanatory variable one by one;

[0048] Step 5.2) sort the explanatory variables to be analyzed according to their numerical values, and divide them into two groups according to the sample size, i.e. take the median of all data as the standard, define the values greater than or equal to the standard as the high value group, and the values less than the standard as the low value group;

[0049] Step 5.3) perform Mann-Whitney U test on the ozone time series data corresponding to the high group and the low group, respectively; if the test result is significant, it indicates that there is a significant difference between the ozone of the high group and the ozone of the low group; if the test result is not significant, it indicates that there is no significant difference between the ozone of the high group and the ozone of the low group.

[0050] As a limitation of the above technical solution, the adjustment variable W needs to meet the following requirements: Y corresponding to the high group and the low group of W has a significant difference.

[0051] The mediated path with adjustment needs to meet the statistical test and the chemical test at the same time.

[0052] The evaluation method of the adjustment effect is: if the effect value of the adjustment variable β3 is greater than 0, the positive influence of X on Y is enhanced with the increase of W, or the negative influence is weakened with the increase of M; if the effect value of the adjustment variable β3 is less than 0, the positive influence of X on Y is weakened with the increase of W, or the negative influence is enhanced with the increase of M.

[0053] The determination of the adjustment variable is based on the definition of the adjustment effect, and the adjustment effect refers to the influence intensity of the cause on the result which varies due to individual characteristics or environmental conditions, and such characteristics or conditions are called adjustment variables. Figure 2 As shown in the adjustment effect schematic diagram, Y represents ozone, X represents a key explanatory variable, and W represents an adjustment variable. The determination of W needs to meet the following conditions: the causal relationship of X→Y is adjusted by W, that is, there is a significant difference between the influence of X on Y in the high W group and the low W group.

[0054] As a limitation of the above technical solution, the stepwise regression is realized by the step method in SPSS; the Granger causality test is fitted by using STATA; the difference test is realized by using the non-parametric test in SPSS; and the mediated path and the mediated path with adjustment are realized by using the Process plug-in in SPSS.

[0055] The ozone pollution cause path construction method based on causal inference provided by the application is different from the traditional statistical model for analyzing the correlation between the influence factors and ozone and the chemical transmission model for analyzing the chemical process of the influence factors and ozone. The application is based on causal inference, combines statistical algorithms and chemical mechanisms, overcomes the defect that the chemical transmission model requires high basic data, and overcomes the problem that the numerical relationship has hidden nature due to the fact that the statistical model does not fully consider the chemical reaction mechanism, and deeply analyzes the real influence of the influence factors on ozone.

[0056] Meanwhile, the application also provides the application of the ozone pollution cause path based on causal inference as described above, which is to use the ozone pollution cause path based on causal inference obtained above to guide emission reduction, so that the ozone concentration can be effectively reduced.

[0057] Specifically, the intermediary path explains how the key explanatory variable affects ozone. When the intermediary path is complete mediation, the effect of the key explanatory variable on ozone is completely transmitted by the intermediary variable, and controlling the intermediary variable can effectively control the occurrence of ozone pollution. When the intermediary path is partial mediation, the key explanatory variable affects ozone through direct action and indirect action through the intermediary variable, and the key explanatory variable and the intermediary variable need to be controlled at the same time to control the occurrence of ozone pollution. The mediated path with adjustment provides the optimal condition for the selection of ozone pollution control measures. When the adjustment variable has a positive effect on the intermediary path, increasing the adjustment variable can enhance the emission reduction effect, and when the adjustment variable has a negative effect on the intermediary path, increasing the adjustment variable can weaken the emission reduction effect. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 , Schematic diagram of mediation effect;

[0059] Figure 2 , Schematic diagram of adjustment effect;

[0060] Figure 3 , SPSS stepwise regression variable model screening process summary;

[0061] Figure 4 , Possible ozone pollution cause path, left side is path 1, right side is path 2;

[0062] Figure 5 , SPSS mediation effect model running result interface;

[0063] Figure 6 , Shijiazhuang City ozone pollution cause path based on mediation effect, ** in the figure indicates p<0.05;

[0064] Figure 7 , SPSS adjustment effect model running result interface;

[0065] Figure 8 , Shijiazhuang City ozone pollution cause path based on mediated path with adjustment;

[0066] Figure 9 , Flowchart of ozone pollution cause path construction method based on causal inference; DETAILED DESCRIPTION

[0067] The technical solutions of the present application will be described clearly and completely below in conjunction with embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0068] EMBODIMENT

[0069] Step 1) Obtain the time series data of the explanatory variables related to the change of ozone concentration in the winter ozone pollution process in 2020 (2020.1.25-2020.2.9), including meteorological factors: UVB (ultraviolet radiation intensity), RH (relative humidity), T (temperature), P (atmospheric pressure), WS (wind speed), WD (wind direction) and the like; pollution factors: 12 kinds of inhalable particulate matter (PM 10 ), fine particulate matter (PM 2.5 ), nitrogen dioxide (NO2), sulfur dioxide (SO2), carbon monoxide (CO), non-methane total hydrocarbons (NMVOCs).

[0070] In this embodiment, the time series data of meteorological factors (T, P, RH, WS and WD) are obtained from the national ground meteorological monitoring station (https: / / q-weather.info / weather), and the hourly observation data of UVB are obtained from the Shijiazhuang UV1000-H solar radiation observation system. The hourly observation data of the six pollution factors (CO, O3, NO2, PM 2.5 , PM 10 , SO2) are obtained from the China National Environmental Monitoring Center (http: / / www.cnemc.cn), and the NMVOCs data are obtained by the Shijiazhuang VOCs automatic online monitoring instrument AirmOzone analysis system (ASS), and part of the data is shown in Table 1.

[0071] Table 1: Pollution process data display (part) of Shijiazhuang

[0072]

[0073]

[0074] Step 2) Use the stepwise method of stepwise regression to perform significance test on the obtained meteorological and pollution factors, and determine the significant explanatory variables that have significant impact on ozone change from the various explanatory variables, i.e. the significant explanatory variables in the ozone pollution cause path.

[0075] The screening method of significant explanatory variables is not unique, and the stepwise method used in the present application combines accuracy and simplicity. In addition, the mapminmax function is used to normalize the explanatory variables when the stepwise method is used in this embodiment, which ensures that the impact of the significant explanatory variables on ozone is comparable, and the larger the normalization coefficient, the more important the impact on ozone.

[0076] To ensure the simplicity of the ozone pollution cause path, the selection of significant explanatory variables should meet the following conditions: the selection of significant explanatory variables should be selected one by one according to the normalized coefficient of the explanatory variable from large to small; the degree of explanation of significant explanatory variables to ozone should be higher than the threshold value of the stepwise regression model, R 2 =0.6; at least 3 significant explanatory variables are required, including 1 key explanatory variable, 1 intermediate variable and 1 moderator variable; the number of significant explanatory variables should be adjusted according to the path test results and actual application requirements.

[0077] In this embodiment, this step is realized by the step-by-step method in the linear regression of SPSS, and the process summary of the variable selection of SPSS is shown in Table 1. Figure 3 In Shijiazhuang City, 10 variables were screened out as significant variables affecting ozone, including RH, T, NO2, SO2, PM 10 , P, WD, WS, PM 2.5 and CO. To establish the simplest ozone pollution cause path, the three variables with the greatest impact on ozone were first selected: NO2, PM 2.5 and T. However, since T does not pass the statistical test of the mediated path with moderation in the subsequent actual modeling process, although it meets the selection criteria of the moderator variable, the variable is selected again according to the ranking order of the normalized coefficient until the established mediated path with moderation can pass the test. The final actual selection of significant explanatory variables in this embodiment is shown in Table 2, and the significant explanatory variables of the ozone pollution process in Shijiazhuang City are CO, NO2, PM 2.5 , T and RH, among which NO2 has the greatest impact on ozone and RH has the smallest impact on ozone. These 5 explanatory variables can collectively explain 75% of the ozone changes during the study period.

[0078] Table 2 Screening results of significant explanatory variables of ozone pollution process in Shijiazhuang City

[0079]

[0080] However, in the actual atmospheric environment, in addition to the above 5 significant explanatory variables, other influencing factors will also affect ozone. To comprehensively consider the influence of various influencing factors on ozone, in this embodiment, the explanatory variables other than the significant explanatory variables and the unconventional influencing factors are included in the pollution cause path construction in the form of confounding explanatory variables. Part of the calculation results of the confounding explanatory variables are shown in Table 3.

[0081] Table 3 Calculation results of confounding explanatory variables of ozone pollution process in Shijiazhuang City (part)

[0082]

[0083]

[0084] Step 3) Determine the causal relationship between significant explanatory variables, between significant explanatory variables and ozone based on Granger causality test, and select key explanatory variables and intermediate variables based on this. The causal chain between key explanatory variables, intermediate variables and ozone should be met: key explanatory variables are the cause of intermediate variables, and key explanatory variables and intermediate variables are the cause of ozone. And based on the intermediate effect model, the formation path of ozone pollution with intermediate effect is preliminarily built, which is called intermediate path; taking the simplest scenario as an example, one key explanatory variable and one intermediate variable are selected according to the causal chain relationship, at this time, the intermediate path built by combining the intermediate effect model is as shown in Figure 1 .

[0085] In this embodiment, the Granger causality test results between significant explanatory variables and between significant explanatory variables and ozone are shown in Table 4. CO and NO2 are the cause of ozone (p<0.05), so CO and NO2 may be key explanatory variables and intermediate variables in the formation path of ozone pollution. However, since CO and NO2 are causally related (p<0.05), and ozone has a reverse causal relationship with NO2 (p<0.05), it indicates that the position of significant explanatory variables in the intermediate effect model is not uniquely determined. The possible paths are as shown in Figure 4 , path 1: CO is the key explanatory variable, and NO2 is the intermediate variable; path 2: NO2 is the key explanatory variable, and CO is the intermediate variable.

[0086] Table 4 Granger causality test results of ozone pollution process in Shijiazhuang

[0087]

[0088] Step 4) Use the MCM box model to chemically test and correct the intermediate path, and use the Bootstrap test to statistically test and correct the intermediate path, to determine the intermediate path after testing and correction. At this time, the intermediate path meets the chemical mechanism of ozone generation and also meets the statistical requirements. And further evaluate the direct effect and indirect effect of the intermediate path in the Bootstrap test process.

[0089] According to the preliminary inference of the possible intermediate path, the corresponding intermediate effect model is the fourth intermediate effect model in SPSS-Process. The SPSS intermediate effect model running result interface is as shown in Figure 5 . After sorting, the intermediate path after testing and correction in this embodiment is as shown in Figure 6The key explanatory variable affecting the change of ozone concentration is NO2, and the intermediate variable is CO. There are two main ways to affect ozone, Pathway 1: NO2 directly affects ozone (p<0.05, R1-R3), Pathway 2: NO2 affects ozone by affecting CO (p<0.05, NO2→CO is R5-R7 in Table 5, and CO→O3 is R10).

[0090] Table 5 Main chemical reaction processes involved in the cause path of ozone pollution (extracted from MCM box model)

[0091]

[0092]

[0093] The evaluation results of the mediation path are shown in Table 6. The total effect c, the direct effect c', and the indirect effect ab of the mediation path are all significant, so the mediation effect is partial mediation. And ab and c have opposite signs, and the mediation effect shows a masking effect. That is, it indicates that the effect of NO2 on ozone in the process of ozone pollution in Shijiazhuang is transmitted through two pathways, the first is that NO2 directly affects ozone, and the second is that NO2 indirectly affects ozone by affecting CO. Among them, the indirect effect accounts for 28.87% of the direct effect.

[0094] Table 6 Evaluation results of mediation path in Shijiazhuang

[0095]

[0096] Step 5) Determine the adjustment variable from the other significant explanatory variables (in this embodiment, specifically T, RH, P, WS, WD, UVB, PM 2.5 , PM 10 , SO2, NMVOCs) and mixed explanatory variables, a total of 11 kinds of explanatory variables, through the group difference test method. The adjustment variable W needs to meet the following conditions: the causal relationship of X→Y is adjusted by W, that is, there is a significant difference in the effect of X on Y between the high W group and the low W group. And based on the adjustment effect model, a mediation path with adjustment is preliminarily built; taking the simplest scenario as an example, according to the group difference test result, select 1 adjustment variable, at this time, combined with the adjustment effect model, the adjustment path built as shown in Figure 2 .

[0097] Here, PM 2.5 is taken as an example for illustration. First, according to PM 2.5The time series data of ozone were sorted by numerical value and divided into two groups by sample size, i.e. the median of all data was taken as the standard, and the values greater than or equal to the standard were taken as the high value group, and the values less than the standard were taken as the low value group. Mann-Whitney U test was performed on the ozone time series data corresponding to the high and low groups, and the test results are shown in Table 7. The ozone in the high PM 2.5 group was significantly different from the ozone in the low PM 2.5 group (p = 0.000), i.e. PM 2.5 may be a regulatory variable of the path of ozone pollution causes. In this embodiment, the preliminary determination result of the regulatory variable that has a significant difference in the influence on ozone between the high group and the low group is T, RH, WS, UVB, PM 2.5 and NMVOCs.

[0098] Table 7 Mann-Whitney U test summary of PM 2.5

[0099]

[0100] Step 6) Chemical test and correction of the intermediate path by MCM box model, statistical test and correction of the intermediate path by Bootstrap test, and final determination of the tested and corrected intermediate path with regulation, which meets both the chemical mechanism of ozone generation and the statistical requirements. The regulatory effect of the regulatory variable on the intermediate path is further evaluated in the Bootstrap test process.

[0101] The preliminary constructed intermediate path with regulation corresponds to the regulatory effect model 21 in SPSS-Process. The running result interface of the SPSS regulatory effect model is shown in Figure 7 After sorting, the intermediate path with regulation after test and correction in this embodiment is shown in Figure 8 , and the regulatory variables are PM 2.5 and RH. The evaluation results of the regulatory effect are shown in Table 8, and PM 2.5 has a significant regulatory effect on NO2→CO, and RH has a significant regulatory effect on CO→O3. Moreover, the effect value β3 of PM 2.5 and CO is positive, i.e. PM 2.5 has a positive regulatory effect on NO2→CO, and RH has a positive regulatory effect on CO→O3.

[0102] Table 8 Evaluation results of regulatory effect

[0103]

[0104] ​In summary, the ozone pollution cause path constructed by the present application (see Figure 8 ), NO2 is a key explanatory variable affecting ozone, CO is an intermediate variable, PM 2.5 and RH are adjustment variables. The changes of NO2 and CO are the main reasons for the occurrence of ozone pollution in Shijiazhuang City. The high and low of PM 2.5 can significantly adjust the influence of NO2 on CO (the adjustment direction is positive adjustment), and RH can adjust the influence of CO on ozone (the adjustment direction is positive adjustment). That is, in terms of guiding ozone emission reduction, controlling the concentrations of NO2 and CO can effectively reduce the ozone concentration level, and on the basis of controlling NO2 and CO, adjusting the levels of PM 2.5 and RH can significantly improve the efficiency of ozone pollution prevention and control.

[0105] To verify the effectiveness of the present application in ozone pollution prevention and control, the emission reduction effects of different ozone pollution cause analysis models are compared here. The statistical model is taken as an example, and the chemical transmission model is taken as an example of WRF-CMAQ model for comparison. For the same ozone pollution process, the ozone pollution causes obtained by different pollution cause analysis models are different. The ozone pollution cause obtained by the ozone pollution cause path of the present application is CO and NO2, the ozone pollution cause obtained by the statistical model is CO, NO2, PM 2.5 and RH, and the ozone pollution cause obtained by the chemical transmission model is T, NO2 and VOCs. In view of the particularity of the environmental field, in order to compare the effects of different models in guiding ozone pollution prevention and control, the box model based on observation (OBM) is used to simulate the emission reduction scenarios for the ozone pollution causes obtained by different models, and the results are shown in Table 9.

[0106] Table 9 Different emission reduction scenario simulation

[0107]

[0108] According to the emission reduction of the ozone pollution cause path constructed by the present application, the ozone concentration can be effectively reduced. Controlling the key explanatory variable and the intermediate variable can effectively control the ozone concentration level. Among them, controlling the key explanatory variable can achieve the optimal effect of ozone pollution prevention and control (the ozone concentration is reduced by 7.50 μg / m 3 ), and the effect of controlling the intermediate variable is second (the ozone concentration is reduced by 3.50 μg / m 3 ). On this basis, adjusting the adjustment variable can effectively improve the efficiency of ozone pollution prevention and control (the ozone concentration is reduced by 11.75 and 4.10 μg / m 3 ).

[0109] Based on the results of the statistical model and the chemical transmission model, the emission reduction simulation results show that PM 2.5The ozone concentration value and the chemical transmission process of VOCs participating in the ozone concentration change have significant influence, but it is not the real cause of the ozone pollution, and the ozone prevention and control effect obtained by controlling it is not high, and even rebound (the ozone concentration increases by 1.52 mu g / m 3 and decreases by 0.20 mu g / m 3 ) respectively.

[0110] Obviously, due to the complex relationship between the influencing factors, there are simple one-way influence relationship and mutual influence relationship, the traditional statistical model reflects the direct relationship between the independent variable and the dependent variable, and it is difficult to clearly express the complex relationship between the independent variables. Based on the cause and effect inference of the ozone pollution cause path, the complex relationship can be effectively analyzed out step by step through the construction process as shown in Figure 9 , which is more scientific and effective in guiding the prevention and control of ozone pollution.

Claims

1. A method for constructing ozone pollution causal pathways based on causal inference, characterized in that: The construction method includes the following steps: Step 1) Using the conventional influencing factors in the analysis of ozone pollution causes as explanatory variables for the ozone pollution cause pathway, obtain the time series data of each explanatory variable; Step 2) The significance of each explanatory variable is tested by stepwise regression to obtain the explanatory variables that have a significant impact on ozone, which are used as significant explanatory variables in the ozone pollution causal path. Explanatory variables other than significant explanatory variables and other unconventional influencing factors that may affect ozone are defined as confounding explanatory variables. The confounding explanatory variables are represented by the residuals of stepwise regression simulation of ozone. Step 3) Use Granger causality test to test the causal relationships between significant explanatory variables and between significant explanatory variables and ozone, identify key explanatory variables and mediating variables, and preliminarily build a mediating ozone pollution causal path based on the mediation effect model, which is called the mediating path; Step 4) Use the MCM box model to perform chemical tests and corrections on the intermediate pathways; use the Bootstrap method to perform statistical tests and corrections on the intermediate pathways, and further use the Bootstrap method to evaluate the direct and indirect effects of the intermediate pathways, in order to assess the intensity of the impact of the intermediate pathways on ozone and the proportion of each pathway's impact. Step 5) Test the differences between groups for other significant explanatory variables besides the key explanatory variables and mediating variables, as well as confounding explanatory variables; identify the explanatory variables with significant differences between groups as moderating variables, and initially construct a moderated mediation path based on the moderating effect model; Step 6) Use the MCM box model to perform chemical tests and corrections on the moderated mediation pathways; use the Bootstrap method to perform statistical tests and corrections on the moderated mediation pathways, and further use the Bootstrap method to evaluate the moderating effect of the moderating variables on the mediation pathways.

2. The method for constructing ozone pollution causal pathways based on causal inference according to claim 1, characterized in that: Conventional influencing factors in ozone pollution causation analysis are those factors that are generally recognized as having a significant impact on ozone, including meteorological factors and air pollutant factors. Unconventional influencing factors are those factors that are generally recognized as having a relatively small impact on ozone or whose data are difficult to obtain, including meteorological elements, pollutant factors, and other factors.

3. The method for constructing ozone pollution causal pathways based on causal inference according to claim 1, characterized in that, The selection of the significant explanatory variables must meet the following conditions: Select the explanatory variables one by one from largest to smallest normalization coefficient, while requiring that the explanatory power of significant explanatory variables for ozone is higher than the threshold of the stepwise regression model. Select at least 3 significant explanatory variables.

4. The method for constructing ozone pollution causal pathways based on causal inference according to claim 1, characterized in that, The formula for calculating the confounding explanatory variables is as follows: In the formula, x k Time series data representing significant explanatory variables; β k Represents the regression coefficient; ε represents the time series data of ozone simulated by stepwise regression; O3 represents the time series data of ozone; ε represents the time series data of confounding explanatory variables.

5. The method for constructing ozone pollution causal pathways based on causal inference according to claim 1, characterized in that, The mediation effect model is as follows: Y = cX + e1 M = aX + e2 Y = c′X + bM + e3 Where X is the key explanatory variable, Y is ozone, M is the mediating variable, and a, b, c, and c′ are the coefficients of the mediation effect test results, where c is the total effect, c′ is the direct effect, and ab is the indirect effect; e1, e2, and e3 are the constants of the test results.

6. The method for constructing ozone pollution causal pathways based on causal inference according to claim 5, characterized in that: By using Granger causality tests, we can determine the causal relationships between significant explanatory variables and between significant explanatory variables and ozone, identify the position of significant explanatory variables in the ozone pollution causal pathway, and determine the key explanatory variable X and the mediating variable M.

7. The method for constructing ozone pollution causal pathways based on causal inference according to claim 6, characterized in that: X and M must satisfy the following requirements: X is the cause of M and Y, and M is the cause of Y.

8. The method for constructing ozone pollution causal pathways based on causal inference according to claim 5, characterized in that, The direct and indirect effects in the mediation path are evaluated using the following steps: Step 4.1) Test the total effect c using the Bootstrap method. If it is significant, consider it a mediating effect; otherwise, consider it a masking effect. Regardless of whether it is significant or not, subsequent steps are performed for testing; Step 4.2) Test coefficients a and b using the Bootstrap method. If both are significant, the indirect effect is significant, and proceed to step 4.4). If at least one is not significant, proceed to step 4.3); Step 4.3) Test the indirect effect ab using the Bootstrap method; if significant, the indirect effect is significant, proceed to step 4.4); Otherwise, the indirect effect is not significant, so the analysis should be stopped. Step 4.4) After confirming the existence of the mediating effect, the direct effect c' is tested using the Bootstrap method. If it is not significant, that is, the direct effect is not significant, indicating that there is only a mediating effect; if it is significant, that is, the direct effect is significant, proceed to step 4.5). Step 4.5) Compare the signs of ab and c. If they are the same, it is a partial mediation effect, and the mediating effect accounts for the proportion of the total effect, ab / c. If they are different, it is a masking effect, and the absolute value of the proportion of the indirect effect to the direct effect, |ab / c|, is reported.

9. The method for constructing ozone pollution causal pathways based on causal inference according to claim 1, characterized in that, The moderating effect model is as follows: Y = β0 + β1X + β2W + β3WX + ε Where X is the key explanatory variable, Y is ozone, W is the moderating variable, β is the coefficient of the moderating effect test result, and ε is the constant of the test result.

10. The method for constructing ozone pollution causal pathways based on causal inference according to claim 9, characterized in that, The intergroup difference test includes the following steps: Step 5.1) Import the time series data of other significant explanatory variables except for the key explanatory variables and mediating variables, and the corresponding ozone time series data. Import the time series data of confounding explanatory variables and the corresponding ozone time series data. Analyze each explanatory variable one by one. Step 5.2) Sort the explanatory variables to be analyzed according to their numerical values, and divide them into two groups according to the sample size. That is, based on the median of all data, those greater than or equal to the median are defined as the high value group, or simply the high group, and those less than the median are defined as the low value group, or simply the low group. Step 5.3) Perform the Mann-Whitney U test on the ozone time series data corresponding to the high group and the low group respectively. If the test result is significant, it indicates that there is a significant difference between the ozone in the high group and the ozone in the low group; if the test result is not significant, it indicates that there is no significant difference between the ozone in the high group and the ozone in the low group.

11. The method for constructing ozone pollution causal pathways based on causal inference according to claim 10, characterized in that, The moderating variable W must meet the following requirements: there must be a significant difference in Y between the high and low W groups.

12. The method for constructing ozone pollution causal pathways based on causal inference according to claim 9, characterized in that, The method for assessing the moderating effect is as follows: if the effect value β3 of the moderating variable is greater than 0, then the positive effect of X on Y increases with the increase of W, or the negative effect decreases with the increase of M; if the effect value β3 of the moderating variable is less than 0, then the positive effect of X on Y decreases with the increase of W, or the negative effect increases with the increase of M.

13. The application of ozone pollution causal pathways based on causal inference, using the ozone pollution causal pathways based on causal inference obtained from any one of claims 1 to 12 to guide emission reduction, can effectively reduce ozone concentration.

Images