Research on the Impact of Atmospheric Environment Self-Purification Capacity on Fog-Haze Pollution

Abstract

Why is fog-haze pollution very serious in Hebei province, where there are many pollution-intensive industries, and in Guangdong province, where it is not so serious? This paper uses the spatial Durbin model, the threshold effect model, and relevant local city data, etc., to explore the effect of the atmospheric environment’s self-purification capacity on haze pollution from the perspective of green technology innovation. We found that the great haze outbreak in China is due to the large amount of ultrafine-particle low-cost emissions caused by the haze detection by weight method implemented in 2011 and 2012. This study also found that haze pollution in China has a significant impact on the atmospheric environment’s self-purification capacity. The atmospheric environment’s self-purification capacity has an inhibitory effect on haze pollution. When green technology innovation reaches the first threshold, the atmospheric self-purification capacity can significantly reduce the impact of haze pollution. When green technology innovation reaches the second threshold, the atmospheric self-purification capacity to reduce haze pollution is significantly enhanced. China’s local haze pollution is serious due to the industrial layout being unreasonable, caused by high-pollution industries emitting particles beyond the limits of atmospheric environment self-purification capacity. Industries in Hebei Province and Guangdong Province are more pollution-intensive, and haze pollution in Hebei Province is serious due to the weak self-purification capacity of the atmospheric environment. Guangdong Province’s atmospheric environment self-purification capacity is strong, and its haze pollution is not serious. Given the scientific use of atmospheric environment self-purification capacity and regional differences in green technology innovation, the development of targeted green input and atmospheric self-purification capacity enhancement policies in areas with serious air pollution, along with green technology innovations based on a region with less pollution, would be beneficial. To increase the amount of green technology innovation investment in regions where the atmospheric environment is not seriously polluted and green technology innovation is based on a bad region, more green funds should be invested in the atmospheric environment’s self-purification capacity. In regions where the atmospheric environment is not seriously polluted and the foundation of green technology innovation needs improvement, more green funds should be invested into atmospheric environment self-purification capacity to fully harness its inhibition of haze pollution. This should be accompanied by scientific planning and adjustments to the high-pollution industrial layout, etc., to effectively enhance the self-purification capacity of the regional atmospheric environment. In addition, the gradient transfer of high-pollution industries should be implemented based on atmospheric environment self-purification capacity to effectively reduce the impact of haze pollution.

1. Introduction

The reason why severe fog-haze pollution occurred three times in most parts of northern China from the end of January to February 2020 when the novel coronavirus pandemic began is that air pollution-intensive industries such as thermal power and steel did not shut down [1]. Air pollution cannot be solved without green technology [2] and atmospheric environment self-purification capacity [3]. Therefore, can the current level of green technology innovation effectively curb haze pollution through atmospheric environment self-purification capacity? Under what conditions can atmospheric self-purification capacity play a better role in preventing environmental pollution? How can we scientifically enhance the atmospheric self-purification capacity?

Academics have conducted many studies on haze pollution, atmospheric environment self-purification capacity, and green technology innovation.

In their research on the causes of haze pollution, academics have mainly explored the natural and anthropogenic causes of haze pollution. In terms of natural causes, more features of urban haze pollution are analyzed through meteorological causes, and factors such as pollutant concentration, the meteorological factor index, and the air quality index are considered to have an important influence on the formation of haze pollution [4]. Aerosol optical thickness (AOD) inversion is closely related to PM2.5 [5]. In terms of anthropogenic causes, some scholars focus on exploring the causes of haze pollution formation from macroscopic factors such as large-scale coal combustion, unrestricted vehicle exhaust emissions, massive industrial pollution emissions, and massive dust outputs from the construction industry [6]. Among them, both environmental regulations and industrial structures can reduce haze pollution [7]. Some scholars have also analyzed the micro-, social, and economic factors within cities, and the effects of population size and structural reform on the intensity of urban PM concentration are obvious in cities with different levels of population and economic development, while foreign investment and science and technology can significantly reduce the concentration of urban PM emissions [8].

Relevant studies on the atmospheric self-purification capacity and fog-haze pollution have shown that the self-purification capacity index of the atmospheric environment is negatively correlated with the air pollution index [9,10]. When the index of atmospheric environment self-purification capacity is low, the atmospheric diffusion capacity is weak, which is not conducive to the removal and diffusion of pollutants, and the air pollution index is higher; when the index of atmospheric environment self-purification capacity is large, it is conducive to the removal of pollutants through precipitation and the diffusion capacity of ventilation [11]. Enhancing the atmospheric self-purification capacity through the use of dilution purification by wind, wet and dry deposition purification by rain and snow, etc., and the appropriate use of human-designed purification can effectively reduce environmental pollutants [12]. Most heavy pollution events occur when the atmospheric self-purification capacity is low, which also shows that there is a relationship between the atmospheric self-purification capacity and the concentration of PM2.5 particulate matter [13].

In the study on green technology innovation and fog-haze pollution, green technology innovation, as one of the important measures to prevent and control air pollution, has an impact on environmental pollution, which has received more and more attention. Technology research and development can positively affect green economic development [14]. Green growth, eco-innovation, environmental taxes, renewable energy, and other factors related to green technology innovation play a crucial role in reducing fog-haze pollution [15]. In addition, green technology innovation also has a mediating role among environmental information disclosure, economic development, and fog-haze pollution [16].

Although the academic community has conducted various studies on the causes of haze, few studies have paid attention to atmospheric environment self-purification capacity, the rationality of the layout of polluting industries, and other anthropogenic factors, as well as the amount of particulate matter. Now, relevant studies have posited that atmospheric environment self-purification capacity is negatively correlated with air pollution, and improving atmospheric environment self-purification capacity is conducive to improving air quality. In addition, related studies show that green technology innovation helps reduce fog-haze pollution, but there are very few studies exploring how green innovation technology has an impact on environmental pollution in relation to atmospheric environment self-purification capacity, and most existing studies are based on regional samples and exploring real-world problems or the empirical examination of provinces and districts [17]. Fewer studies have been conducted with a national scope, and the significance and future suggestions in the conclusions of these studies have been restricted, to a certain extent. The impact of the self-purification capability of the atmospheric environment on fog-haze pollution has not been sufficiently explored in the current research from a green technology innovation perspective. Therefore, in the present day, with China vigorously promoting high-quality development, it is of great practical and theoretical significance to investigate the impact of atmospheric self-purification capacity on environmental pollution from the perspective of green technology innovation.

2. Theoretical Analysis and Research Hypotheses

Fog-haze is a visual blurring phenomenon that occurs due to the combination of a large number of particulate matters with vapor in the air in the case of sufficient particulate matters, sufficient humidity, and static weather. In September 2011, PM2.5 particulate matters were detected by the gravimetric method, while in 2012, it was also detected the same way in key protected cities. Although the gravimetric method is widely used, it still has drawbacks in detecting haze. For example, for spherical particulate matters, the weight of 1 PM2.5 ≈ the weight of 125 PM0.5 ≈ the weight of 15,625 PM0.1 (assumed in the same density). Furthermore, 100 PM0.5 is equal to 80% of the weight of 1 PM2.5, and 100 PM0.5 is divided into four layers of 25. Each layer of 25 PM0.5 and four layers of PM0.5 can block the line of sight in an area, and the average thickness of the PM0.5 is much larger than 1 PM2.5, while the weight is reduced by 20% of 100 PM0.5, leading to a line-of-sight-blocking effect much larger than that with 1 PM2.5. Therefore, using the weight method for detecting haze is a serious mistake. The implementation of a PM2.5 detection policy has seriously misled China in its fog-haze prevention and control, resulting in the emission of a large amount of vapor and ultra-fine particulate matters with no repercussions. To reduce the weight of particulate matters in the air, northern China continued to spray vapor into the air (the severer the fog-haze was, the more vapor would be sprayed). The water used for spraying was tap water and reclaimed water; as a result, a large amount of ultra-fine particles in the water were sprayed into the air. To reduce the weight of particulate matters, polluting enterprises applied electrostatic dust collection to absorb particulate matters by making them negatively charged. The larger particulate matters are, the more negatively charged they are and the easier they are to adsorb. However, this also meant that tens to hundreds of times more ultra-fine particulate matters were discharged than was standard. In 2012, flue gas heat exchangers used for desulfurization were completely abolished in enterprises; as a result, the flue gas emitted by thermal power enterprises (these enterprises used circulating water to saturate particulate matters) did not need to be dehydrated, and these vapors and a large amount of ultra-fine particles created by vapors or dissolved during later recycling were discharged into the air with no repercussions. Moreover, during a period of stagnant weather in 2012, a concentrated fog-haze outbreak occurred in China due to these vapors and extremely small airborne particles. After China implemented stringent online monitoring and punishment measures based on the gravimetric method on 1 January 2013, desulfurization, denitration, and dust removal facilities for enterprises causing high fog-haze pollution, such as thermal power plants, started to work rapidly, leading to more ultra-fine particles and vapors being discharged into the air. In addition, due to high standards around denitration emissions and other reasons, related enterprises sprayed ammonia excessively to meet emission standards, which caused severe ammonia escape and a sharp increase in the number of secondary ultra-fine particulate matters, so that the fog-haze pollution in China became increasingly severe in 2013. After 2014, China set environmentally friendly electricity prices for thermal power companies to encourage them to build or renovate desulfurization, denitration, and dust removal facilities. Such newly built and renovated facilities were gradually put in place to offset the impact of China’s fog-haze prevention and control measures. Thus, effective countermeasures to prevent and control fog-haze pollution include innovation in desulfurization, denitration, and dust removal methods, improving the atmospheric self-purification capacity, and stringently control the quantity and quality of vapor emissions. Moreover, due to the influence of natural conditions such as humidity, ventilation, and precipitation, coupled with human factors such as the location of polluting industries and urban planning and arrangement, there will be a certain correlation between environmental pollution and space. The atmospheric self-purification capacity mainly depends on the diffusion and dilution of the atmosphere, and there is a certain spatial overflow between regions. Differences in green technology innovation will also have an impact on the effect of atmospheric environment self-purification capacity in controlling environmental pollution. Based on the above analysis, for this paper, we studied the impact of fog-haze pollution on atmospheric self-purification capabilities from the perspective of green technology innovation.

2.1. Analysis on Regional Variability in Atmospheric Environment Self-Purification Capacity

Previous studies have shown that there is regional variability in atmospheric environment self-purification capacity. This is due to the fact that the resource environmental bearing capacity, which is the degree of tolerance of the resource environment to pressure, has a very different index among different regions [18]. The ecological environmental carrying capacity can vary greatly depending on geographic location [19]. The meteorological potential of atmosphere (MPA), a criterion for assessing atmospheric self-purification, varies from region to region depending on the characteristic properties of atmospheric circulation, wind regimes, fog, and atmospheric precipitation [20].

Based on the above analysis, for this article, we posited hypothesis 1: Atmospheric environment self-purification capacity has obvious regional differences.

2.2. Threshold Characterization of Atmospheric Environment Self-Purification Capacity for Fog-Haze Pollution

Relevant studies have shown that when environmental regulation reaches a certain level, green technology innovation can effectively curb atmospheric environmental pollution, and it has a significant inhibitory and spatial spillover effect on atmospheric environmental pollution [21]. It can be seen that atmospheric environment self-purification capacity has an influence on fog-haze pollution in relation to green technology innovation threshold characteristics. The level of green technology innovation will control the concentration and quantity of pollutants discharged within the scope of atmospheric environment self-purification capacity, fully harnessing the function of atmospheric environment self-purification capacity to remove fog-haze pollution. By enhancing the level of green technology innovation, the self-purification capacity of the regional atmospheric environment will be improved.

Based on the above analysis, for this article, we posited hypothesis 2: When green technology innovation reaches a certain level, atmospheric environment self-purification capacity can significantly improve fog-haze pollution.

2.3. Influence of Atmospheric Environment Self-Purification Capacity on Fog-Haze Pollution

In recent years, due to industrial structures and urban layouts, environmental pollution emissions have continued to increase, exacerbating air quality [22]. When the emissions of pollutants reach the limit for allowable emission levels considering atmospheric environmental protection targets, the pollution emission level at this time represents the atmospheric self-purification capacity [23]. The effect of atmospheric environment self-purification capacity on the governance mechanisms behind pollution mainly lies in two aspects. On the one hand, air pollutants can be removed through temperature, near-surface wind speeds, precipitation, and other factors [24,25]. On the other hand, the policy of improving the self-purification capacity of the environment is an important means of controlling air pollution [26], atmospheric environment self-purification capacity can promote urban green planning in the reverse direction [27].

Based on the above analysis, for this article, we posited hypothesis 3: Improving the atmospheric self-purification capacity is helpful in alleviating regional environmental pollution.

2.4. Analysis of Spatial Effects

PM2.5 pollution intensifies locally as neighboring areas increase [28]. The mechanisms behind the spatial effects of atmospheric self-purification capacity on fog-haze pollution are as follows: first, there is the positive externality characteristic of the enhancement of atmospheric self-purification capacity [26]; second, there is the positive externality characteristic of fog-haze pollution management; and third, there is the “free-riding” phenomenon in the fog-haze pollution management process [22].

This leads to hypothesis 4 of this study: The improvement of atmospheric environment self-purification capacity in neighboring areas is sufficient to reduce the level of local fog-haze pollution.

3. Research Design

3.1. Framework and Variables

This paper draws on the STIRPAT model and the research framework of Huo et al. (2020) [23], which is an extensible and stochastic model for environmental impact assessment (EIA). The STIRPAT model can also be used for environmental and economic impact assessment. By summarizing previous studies on fog-haze pollution, selecting control variables, such as how built-up an area is, the population density, industrial structures, and the economic development level, and choosing the green technology innovation capacity as a threshold variable, for this article, we constructed the model as follows:

$$lnyit=\beta 0+\beta 1lnxit+\beta 2lnkit+\beta 3lnmit+\epsilon it$$

(1)

In particular, i and t represent the city and year, respectively, β0 represents the constant term, β1-β4 represent the estimation coefficient of each factor, x represents the explaining variable, y represents the explained variable, k represents the threshold variable, m represents the control variable, and ε represents the random error term.

Explained variable: Fog-haze pollution (y1). This paper follows the practice of Shao et al. (2016) [24]. In the robustness test, this paper agrees with Qu et al. (2015) [25] and replaces fog-haze data with the emissions of industrial sulfur dioxide.

Explaining variable: Atmospheric environment self-purification capacity (x). On the basis of the research results of Zhu et al. (2018) [27], Dong et al. (2018) [26], and Yu et al. (2017) [28], according to the standardized index value of the maximum wind speed, average temperature, maximum daily precipitation, and average pressure, the standardized score for atmospheric environment self-purification capacity is calculated. The specific calculation process is expressed as follows:

$$\begin{gathered} APIit=\beta 1\left(ATit\right)+\beta 2\left(SHit\right)+\beta 3\left(ARHit\right)+\beta 4\left(AAit\right)+\beta 5\left(MWSit\right)+\beta 6\left(AWSit\right)+\beta 7\left(MDPit\right) \\ +\beta 8\left(APit\right)+\beta 9\left(DPit\right)+\beta 10\left(BUAit\right)+\beta 11\left(PDit\right)+\beta 12\left(GRBit\right)+\beta 13\left(WAit\right) \end{gathered}$$

(2)

In particular, β represents the weight of different indicators, i represents the city, and t represents the year. The values in brackets are the standardized values for different indicator data.

Control variable: This paper draws on the calculation method for population density indicators by Huo et al. (2020) [23] and He (2023) [29], the calculation method for built-up area indicators by Li et al. (2019) [30], the calculation method for economic development level indicators by Zhang et al. (2020) [31], the three-factor indicator calculation method by Yu et al.(2017) [28], and the practice for determining the influence of industrial structures on pollution haze by Lei et al. (2017) [32].

Threshold variable: Green-patent licensing (K2). Environmental pollution can be reduced by improving the level of green technology innovation and optimizing the structure of employment skills [33]. When the level of green technology innovation reaches a certain level, its inhibition effect on sulfur dioxide emissions in air pollution-intensive industries increases significantly [34]. This paper follows the practice of Fan et al. (2013) [35], using the number of green patents in different years in different cities and regions to measure the level of green technology innovation.

3.2. Measurement Model and Method

Atmospheric environment self-purification capacity can reduce local and surrounding environmental pollution through green technology innovation. Therefore, for this paper, we took regional correlation into consideration while conducting empirical analyses and chose spatial autoregressive models to analyze the spatial correlation of dependent variables, as well as designing the space lag and time lag of environmental pollution in SAR. In this paper, the estimation model for spatial correlation is expressed as follows:

$$\begin{gathered} lny{1}_{it}={\beta }_0+{\beta }_{1}lnk{1}_{it}\left(lnk{2}_{it}<\mathsf{\gamma }\right)+{\beta }_{2}lnk{1}_{it}\left(lnk{2}_{it}\ge \mathsf{\gamma }\right)+{\beta }_{3}ln{x}_{it}+{\beta }_{4}lnm{1}_{it}+{\beta }_{5}lnm{2}_{it}+{\beta }_{6}lnm{3}_{it} \\ +{\beta }_{7}lnm{4}_{it}+{\epsilon }_{it} \end{gathered}$$

(3)

In particular, $\rho$ represents the spatial lag (autoregression) coefficient, $W_{it}$represents the element in row i and column j of a standardized non-negative spatial weight matrix W in the N ∗ N dimension, and${\mu }_i$ and${\lambda }_i$ represent the spatial (individual) effect and temporal effect, respectively.

The Spatial Durbin Model (SDM) constructed for this study is as follows:

$$\begin{gathered} lnAE{P}_{it}=\rho {{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lnAE{P}_{it}+{\beta }_{1}wlnAP{I}_{it}+{\beta }_{2}lne{r}_{it}+{\beta }_{3}lnde{n}_{it}+{\beta }_{4}lnbu{a}_{it}+{\beta }_{5}lne{y}_{it}+{\beta }_{6}lni{n}_{it}+ \\ {\beta }_7{{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}w\mathrm{API}+{\beta }_8{{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lne{r}_{it}+{\beta }_9{{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lnde{n}_{it}+{\beta }_{10}{{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lnbu{a}_{it}+{\beta }_{11}{{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lne{y}_{it}+ \\ {\beta }_{12}{{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lni{n}_{it}+{\mu }_i+{\lambda }_i+{\epsilon }_{it} \end{gathered}$$

(4)

Green technology innovation and atmospheric environment self-purification capacity probably have a threshold effect on fog-haze pollution. In this study, the threshold characteristic is modeled as follows:

$$\begin{gathered} lnAE{P}_{it}={\beta }_0+{\beta }_{1}lne{r}_{it}\left(lne{r}_{it}<\mathsf{\gamma }\right)+{\beta }_{2}lne{r}_{it}\left(lne{r}_{it}\ge \mathsf{\gamma }\right)+{\beta }_{3}wlnAP{I}_{it}+{\beta }_{4}lnde{n}_{it}+{\beta }_{5}lnbu{a}_{it}+{\beta }_{6}lne{y}_{it} \\ +{\beta }_{7}lni{n}_{it}+{\epsilon }_{it} \end{gathered}$$

(5)

Multiple threshold estimation equations are extended from the above equation.

3.3. Data Sources and Descriptive Statistics

As China used administrative means to shut down and limit production of hundreds of thousands of scattered and polluting enterprises in 2017 and 2018, considering the impact of this extraordinary event, for this study, we selected the annual data at the prefecture level before 2018 (for data availability, excluding the data of municipalities in Tibetan provinces and some incomplete data, we finally selected 286 municipalities and provincial-level municipalities as samples). Data sources cover China Regional Economy Statistical Yearbook, China Meteorological Data Service Center and National Intellectual Property Administration, China Urban Statistical Yearbook, China Urban Construction Statistical Yearbook, statistical yearbooks for some provinces, etc. This paper analyzes the data with STATA16.0 software. Since some values are 0, when we take the natural logarithm, the variable containing 0 is shifted by 0.1 unit, or the variable is added by 0.1. Those missing samples are filled in with provincial average value. The descriptive statistics of variables are given in

Table 1. Descriptive statistics of variables.

Variable	Obs	Mean	Std. Dev.	Min	Max
y1	4004	43.66	19.15	3.13	110.12
y2	4004	55,335.50	77,136.63	65.00	3,044,570.00
x	4004	0.84	0.21	0.27	1.34
k2	4004	211.65	706.59	0.00	13,337.00
m1	4004	3751.58	2837.54	27.00	20,335.00
m2	4004	126.34	185.30	6.55	2919.09
m3	4004	41,480.24	105,792.50	99.00	6,421,762.00
m4	4004	48.25	10.92	9.00	90.97
lny1	4004	3.67	0.49	1.14	4.70
lny2	4004	10.42	1.09	4.17	14.93
lnx	4004	−0.21	0.27	−1.30	0.29
lnk2	4004	3.50	2.03	−2.30	9.50
lnm1	4004	7.94	0.81	3.30	9.92
lnm2	4004	4.37	0.87	1.88	7.98
lnm3	4004	10.31	0.77	4.60	15.68
lnm4	4004	3.85	0.25	2.20	4.51

.

4. Measurement Estimation and Analysis

4.1. Analysis of Regional Variability in Atmospheric Environment Self-Purification Capacity

Atmospheric environment self-purification capacity refers to an inherent ability of the environment to automatically eliminate pollutants. The atmospheric environment achieves the function of purifying pollutants through physical and chemical reactions, and the concentration and toxicity of pollutants naturally gradually dilute and reduce, or even disappear, restoring these elements in the environment to their original clean state. Regional differences in atmospheric self-purification capacity will be analyzed in this article.

4.1.1. Selection of Indicators for Evaluating Atmospheric Environment Self-Purification Capacity

On the basis of the comprehensive consideration of meteorological factors, urban economic development status, and urban pollution status according to the theoretical screening of indicators and the principles of availability, realism, and practicability in indicator data, this article applies a comprehensive calculation method with multiple indicators and refers to relevant studies in the existing literature to screen out indicators of natural purification and the rationality of anthropogenic design and to construct a standard for atmospheric environment self-purification capacity for the present study. There is an indicator system for classifying atmospheric environment self-purification capacity in this study. Indicators reflecting the purification capacity of the natural environment include the maximum wind speed, average 2 min wind speed, and cumulative precipitation; indicators reflecting the role of anthropogenic design rationality on the purification capacity of the air environment include the level of urban built-up area, the population density, and the green coverage rate in built-up areas. The related measurement indicator system for this article is shown in

Table 2. Measurement indicator system for classification of atmospheric environment self-purification capacity.

Indicator Type	Indicator	Indicator Type	Indicator
Natural purification indicators	Maximum wind speed (m/s)	Indicators of rationality of human design	Urban built-up area (square kilometers)
	Average 2 min wind speed (m/s)		Population density (people/square kilometer)
	Cumulative precipitation (mm)		Green coverage rate in built-up areas (%)
	Maximum daily precipitation (mm)
	Daily precipitation ≥0.1 mm days (days)
	Average air pressure (hectopascal)
	Average air temperature (°C)

:

The maximum wind speed means the maximum value of the 10 min average wind speed for a given time period, which is an average value. The average 2 min wind speed means the sum of wind speeds observed at a given point in space, divided by the number of observations during a given time period, and is also taken as the average wind speed. The cumulative precipitation means the depth to which the liquid or solid (after melting) water that has landed on the ground has accumulated on the ground. The maximum daily precipitation means the maximum amount of precipitation on a single day. The average air pressure means the arithmetic mean of barometric pressure values observed at various times during a certain period of time. The average air temperature means the temperature in a louvered box 1.5 m above the ground. The average annual temperature means the average monthly temperature over 12 months. Sunshine hours mean the number of hours that the sun actually shines on the ground. The urban built-up area means the area included in the built-up area. The population density means the number of people per unit of land area. The green coverage in built-up areas means the percentage of green coverage in the urban built-up area.

4.1.2. Setting the Indicator Weights

This study adopts the Delphi method. Relevant experts have discussed selected indicators according to the existing research and their own knowledge, determined the weights of different indicators based on the degree of purification of atmospheric pollutants, and obtained the average value of different results. The final results for the weighting design of atmospheric self-purification capacity indicators in this study are shown in

Table 3. Indicator weights for classification of atmospheric environment self-purification capacity.

Indicators	Weights
Built-up area	0.02 (negative)
Population density	0.04 (negative)
Urban green space	0.08 (positive)
Average temperature	0.01 (positive)
Sunshine hours	0.01 (positive)
Average relative humidity	0.02 (negative)
Average air pressure	0.01 (negative)
Maximum wind speed	0.1 (positive)
Average 2 min wind speed	0.25 (positive)
Maximum daily precipitation	0.05 (positive)
Cumulative precipitation	0.18 (positive)
Daily precipitation ≥ 0.1 mm	0.23 (positive)

.

4.1.3. Specific Analysis Results for Atmospheric Environment Self-Purification Capacity in Prefecture-Level Administrative Regions

Based on the theoretical analysis of atmospheric environment self-purification capacity combined with the collection of data on indicators of atmospheric environment self-purification capacity from a variety of sources, we calculated the level of atmospheric environment self-purification capacity of 331 prefecture-level administrative regions in China from 2005 to 2018. Due to the huge number of data, for the article, we adopted a short table to present them:

Table 4. Number of regions with various levels of atmospheric environment self-purification capacity in prefecture-level administrative districts, 2005–2018.

2005	2010	2015	2018
17	18	14	17
102	116	120	131
115	102	106	96
90	89	76	84
7	6	5	3

. However, due to limitations in research funding and time, the indicators for the rationality of anthropogenic design, such as the urban built-up area, population density, green coverage in built-up areas, rationality of layout for more important pollution sources, and wetland and vegetation coverage, were not analyzed, leading to the incomplete collection of relevant data, which may have had a certain impact on our results.

As can be seen from Table 4, from 2005 to 2018, among 331 prefecture-level administrative regions in China, vertically, the number of regions with level II and III self-purification capacity is the largest, accounting for 68.58% of the total, with an increase from 65.56% in 2005 to 68.58% in 2018, while that with level I and V self-purification capacity is the smallest. This indicates that the distribution of atmospheric self-purification capacity level is relatively reasonable among prefecture-level administrative regions but appears to be obviously uneven. Horizontally, the number of second-ranked cities in prefecture-level administrative regions is rising year by year, from 102 in 2005 to 131 in 2018, and the number of fifth-ranked prefecture-level administrative regions also shows a decreasing trend, from 7 in 2005 to 3 in 2018. The area of second- and higher-ranked prefecture-level cities has expanded from prefecture-level cities in Guangdong, Guangxi, Fujian, Jiangxi, and Hunan provinces in 2005 to some prefecture-level cities in Sichuan, Hubei, Anhui, and Jiangsu provinces in 2018, as well as the northernmost cities in Weihai and Shandong provinces. The area covered by fifth-ranked cities has not changed much over time and is mainly distributed in Heihe City, Heilongjiang Province, Daxinganling region, Hulunbeier City, and Xing’anmeng in Inner Mongolia and Nagchu region in Tibet, while fourth-ranked prefectures are mainly located in Hebei and Henan provinces, as well as other areas with more serious fog-haze pollution.

4.2. Threshold Characterization

The threshold effect test for green technology innovation is shown in

Table 5. Single and double threshold analysis table for green technology innovation with 95% confidence intervals.

	Threshold	95% CI
Single Model	3.5863	3.5437	3.6136
Double Model
Ito1	3.5863	3.5437	3.6136
Ito2	6.865	6.7522	6.9306

. The single threshold of green technology innovation is 3.5863, and the double threshold is 3.5863 and 6.865, while the 95% confidence intervals are detailed in Table 5. The analysis results for the threshold characteristics of green technology innovation are shown in

Table 6. Table of results for the threshold characterization of green technology innovation.

Model	Threshold	F	p-Value	BS-Reps	10%	5%	1%
Single Threshold	Single	83.39	0.000	300	23.8978	27.5915	41.5049
Double Threshold	Single	83.39	0.000	300	25.0726	31.8267	43.6277
	Double	51.55	0.000	300	18.0618	21.5921	30.4893

.

As shown by the statistical results in

Table 7. Estimation of threshold effect.

	(1)	(2)
	lny1	lny1
lnx	−0.570 ***	−0.593 ***
	(0.077)	(0.077)
lnm1	0.040 ***	0.040 ***
	(0.005)	(0.005)
lnm2	−0.053 ***	−0.046 ***
	(0.012)	(0.012)
lnm3	−0.009	−0.006
	(0.010)	(0.010)
lnm4	0.242 ***	0.217 ***
	(0.022)	(0.022)
0._cat#c.lnk2	−0.006	−0.004
	(0.004)	(0.004)
1._cat#c.lnk2	−0.025 ***	−0.024 ***
	(0.004)	(0.004)
2._cat#c.lnk2		−0.039 ***
		(0.004)
_cons	2.688 ***	2.724 ***
	(0.112)	(0.112)
N	4004	4004

Standard errors in parentheses. *** p < 0.001.

, when the level of green technology innovation is less than 3.5863, the regression coefficient of lnk2 to lny1 proves insignificant and reaches −0.004; when the level ranges from 3.5863 to 6.865, its regression coefficient to fog-haze pollution proves significant and reaches −0.024; and when the level is higher than 6.865, its regression coefficient to fog-haze pollution also proves significant and reaches −0.039. This indicates that when the level of green technology innovation reaches a higher first threshold, the atmospheric self-purification capacity can significantly reduce the impact of fog-haze pollution, and when it reaches the second threshold, the atmospheric self-purification capacity to reduce fog-haze pollution is significantly enhanced. This is possibly because when the level of local green technology innovation is not high, the production at this stage is dominated by rough development, and the layout of pollution sources shows serious irrationality. The pollutants emitted from the production of high-pollution industries exceed the scope of the atmospheric environment self-purification capacity of the region, resulting in a limited self-purification capacity to reduce fog-haze pollution; with continuous improvements in levels of green technology innovation, regions can enhance the technological capability of haze pollutant treatment by introducing the foreign capital of clean technology and stimulating independent innovation in green technology in the region. When the level of green technology innovation is higher than the first threshold, the irrationality of the layouts of pollution sources is obviously reduced. Then, pollutants emitted due to production in high-pollution industries are lowered to the range of the atmospheric environment self-purification capability, which can significantly reduce the influence of fog-haze pollution. When the level of green technology innovation reaches the second threshold, the irrationality of the layout of pollution sources appears to be significantly reduced, and the haze pollutants emitted due to production in highly polluting industries are reduced to the range that can be controlled by atmospheric environment self-purification capacity, which can also significantly reduce the impact of fog-haze pollution.

4.3. Estimation Results and Analytical Statistics for Spatial Correlation Effects

To verify hypotheses 3 and 4, we analyzed the spatial effects, and the results of our empirical analysis are shown in

Table 8. Analysis of spatial correlation effects.

	Total Effect		Direct Effect		Indirect Effect
Model	(1) SAR	(2) SDM	(1) SAR	(2) SDM	(1) SAR	(2) SDM
Variable	lny1	lny1	lny1	lny1	lny1	lny1
Main	−0.390 ***	−1.308 ***	−0.224 ***	−0.007	−0.166 ***	−1.301 ***
lnx	(−3.078)	(−7.289)	(−3.021)	(−0.099)	(−3.098)	(−7.991)
	−0.027 ***	0.016	−0.015 ***	−0.018 ***	−0.012 ***	0.034 ***
lnk2	(−4.726)	(1.644)	(−4.736)	(−5.603)	(−4.524)	(3.884)
	0.059 ***	0.052 ***	0.034 ***	0.029 ***	0.025 ***	0.023 **
lnm1	(7.043)	(4.150)	(7.183)	(6.257)	(6.343)	(2.003)
	−0.104 ***	−0.146 ***	−0.060 ***	−0.054 ***	−0.045 ***	−0.092 ***
lnm2	(−5.309)	(−3.897)	(−5.420)	(−5.012)	(−4.927)	(−2.665)
	0.002	−0.033	0.001	−0.032 ***	0.001	−0.002
lnm3	(0.174)	(−1.488)	(0.171)	(−3.268)	(0.177)	(−0.085)
	0.205 ***	0.615 ***	0.117 ***	0.075 ***	0.087 ***	0.540 ***
lnm4	(5.885)	(11.559)	(5.635)	(3.574)	(5.857)	(11.192)
	3990	3990	3990	3990	3990	3990
N	0.017	0.000	0.017	0.000	0.017	0.000
R2	−0.390 ***	−1.308 ***	−0.224 ***	−0.007	−0.166 ***	−1.301 ***

t statistics in parentheses. ** p < 0.05, *** p < 0.01.

. The spatial lag parameter of fog-haze pollution is significant, indicating that there is a spatial spillover effect of fog-haze pollution.

Table 8 shows that the total, direct, and indirect effects of atmospheric environment self-purification capacity on fog-haze pollution are significantly negatively correlated. For example, in China’s coastal provinces such as Fujian, Guangdong, and Zhejiang, precipitation has played an important role in purifying environmental pollution [36].

In terms of other factors, green technology innovation (lnk2) passed the significance test on fog-haze pollution in one region and its neighboring regions, indicating that a level of green technology innovation helps prevent fog-haze pollution. The result for population density (lnm1) suggests that the population factor has an exacerbating effect on fog-haze pollution. For example, an increase in population density places higher pressure on land bearing capacity and raises demands for land, which causes resource problems (e.g., a reduction in forest area and land degradation) and exacerbates environmental pollution. There is a significant negative correlation between built-up area and fog-haze pollution. There is an insignificant positive correlation for economic development (lnm3). This is mainly because when the level of economic development remains low, the extensive economic growth model aggravates atmospheric pollution. The factor of industrial structures (lnm4) is shown to be significantly positive, which is because regions generally rely on industrial development to achieve high-speed economic growth, thereby exacerbating environmental pollution.

4.4. Robustness Test

Drawing on the robustness testing practices of Qu (2015) [25], the explaining variable PM2.5 was replaced, and we adopted the variable of emissions of industrial sulfur dioxide to measure the degree of atmospheric environmental pollution. The higher the concentration of industrial sulfur dioxide emissions (lny2) in the air, the more severe the atmospheric environmental pollution will be, and vice versa. As Table 6 and Table 7 suggest, in terms of the threshold effect of the robustness test, this paper firstly gives the single and double thresholds of green-patent licensing (lnk2) and their 95% confidence intervals. At the 95% confidence interval, its single threshold (6.9306) and double threshold (6.9306 and 3.5863) are shown in

Table 9. Threshold effect values and 95% confidence intervals of variables.

	Threshold	95% CI
Single Model	6.9306	6.8281	7.0259
Double Model
Ito1	6.9306	6.8281	7.0259
Ito2	3.5863	3.5289	3.6136

below.

As shown in

Table 10. Results of self-sampling test for threshold effect of variables.

Model	Threshold	F	p-Value	BS-Reps	10%	5%	1%
Single Threshold	Single	83.98	0.000	300	31.8563	37.9136	51.93
Double Threshold	Single	83.98	0.000	300	26.724	30.9479	44.2525
	Double	60.53	0.003	300	23.0755	27.8131	44.3331

, the green technology innovation capacity is characterized by the double threshold (both the single and double thresholds have passed the 1% significance test).

In

Table 11. Single and double threshold characterization of green technology innovation capabilities.

	(1)	(2)
	lny2	lny2
lnx	−1.733 ***	−1.729 ***
	(0.282)	(0.280)
lnm1	0.017	0.007
	(0.019)	(0.019)
lnm2	−0.219 ***	−0.183 ***
	(0.045)	(0.045)
lnm3	−0.303 ***	−0.250 ***
	(0.034)	(0.035)
lnm4	1.438 ***	1.328 ***
	(0.081)	(0.082)
0._cat#c.lnk2	−0.030 *	0.006
	(0.013)	(0.014)
1._cat#c.lnk2	−0.104 ***	−0.053 ***
	(0.015)	(0.014)
2._cat#c.lnk2		−0.129 ***
		(0.016)
_cons	8.603 ***	8.416 ***
	(0.409)	(0.407)
N	4004	4004

Standard errors in parentheses. * p < 0.05, *** p < 0.001.

below, it can be seen that after the explaining variables are replaced, atmospheric environment self-purification capacity also has more obvious single and double threshold characteristics for fog-haze pollution.

As

Table 12. Robustness test for spatial correlation.

	Total Effect		Direct Effect		Indirect Effect
Model	(1) SAR	(2) SDM	(1) SAR	(2) SDM	(1) SAR	(2) SDM
Variable	lny2	lny2	lny2	lny2	lny2	lny2
Main
lnx	−1.743 ***	−0.588	−1.157 ***	−0.926 ***	−0.586 ***	0.338
	(−4.281)	(−0.897)	(−4.231)	(−3.255)	(−4.139)	(0.568)
lnk2	−0.030	−0.129 ***	−0.020	−0.024 *	−0.010	−0.105 ***
	(−1.605)	(−3.806)	(−1.604)	(−1.879)	(−1.591)	(−3.372)
lnm1	0.004	0.040	0.002	−0.010	0.001	0.051
	(0.139)	(0.885)	(0.139)	(−0.591)	(0.139)	(1.203)
lnm2	−0.350 ***	−0.221	−0.232 ***	−0.229 ***	−0.118 ***	0.008
	(−5.517)	(−1.639)	(−5.611)	(−5.555)	(−4.925)	(0.068)
lnm3	−0.271 ***	−0.100	−0.180 ***	−0.255 ***	−0.091 ***	0.155 **
	(−5.916)	(−1.231)	(−5.832)	(−6.883)	(−5.512)	(2.064)
lnm4	1.683 ***	2.572 ***	1.116 ***	1.053 ***	0.567 ***	1.519 ***
	(14.967)	(13.304)	(14.224)	(13.065)	(10.217)	(8.726)
N	3990	3990	3990	3990	3990	3990
R2	0.027	0.001	0.027	0.001	0.027	0.001

t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

suggests, the total effect, direct effect, and indirect effect proved significantly negative. As evidenced, in terms of spatial spillover, after the fog-haze is replaced with industrial sulfur dioxide emissions, there is still a significant negative correlation between the self-purification capacity of local and adjacent regions and air pollution.

4.5. Heterogeneity Analysis

Table 13. Regression results for Spatial Durbin Model heterogeneity in five regions.

		(1) The Yangtze River Delta	(2) The Pearl River Delta	(3) Beijing–Tianjin–Hebei Region	(4) Northeast China Region	(5) Chengdu–Chongqing Economic Circle
		lny1	lny1	lny1	lny1	lny1
Total Effect	LR_Total
	lnx	−1.105 ***	−1.332 ***	0.041	0.769	−1.535
		(−3.574)	(−2.951)	(0.051)	(1.189)	(−1.555)
	lnk2	−0.064 ***	−0.146 ***	−0.051	0.041	0.020
		(−3.737)	(−6.027)	(−0.889)	(0.938)	(0.675)
	lnm1	0.061 *	0.065 *	−0.099	0.149 ***	0.013
		(1.654)	(1.805)	(−1.153)	(3.600)	(0.352)
	lnm2	0.058 *	−0.036	−0.221	−0.525 **	−0.312 ***
		(1.787)	(−0.435)	(−1.065)	(−2.379)	(−2.642)
	lnm3	0.065	0.007	0.302 **	0.293 ***	−0.081 *
		(1.422)	(0.137)	(2.314)	(3.295)	(−1.827)
	lnm4	0.885 ***	0.344 ***	0.993 ***	0.064	0.308
		(7.654)	(2.621)	(2.870)	(0.378)	(1.617)
	N	574	126	182	490	266
	R2	0.041	0.450	0.296	0.092	0.093
Direct Effect	LR_Direct
	lnx	−0.395 ***	−0.781 ***	0.014	0.217	−0.677
		(−3.486)	(−2.958)	(0.051)	(1.225)	(−1.568)
	lnk2	−0.023 ***	−0.086 ***	−0.018	0.012	0.008
		(−4.008)	(−5.288)	(−0.919)	(0.967)	(0.683)
	lnm1	0.022 *	0.038 *	−0.034	0.042 ***	0.006
		(1.672)	(1.869)	(−1.190)	(4.101)	(0.353)
	lnm2	0.021 *	−0.021	−0.078	−0.148 **	−0.136 ***
		(1.836)	(−0.429)	(−1.090)	(−2.546)	(−2.809)
	lnm3	0.023	0.004	0.106 **	0.083 ***	−0.036 *
		(1.468)	(0.150)	(2.463)	(3.477)	(−1.889)
	lnm4	0.314 ***	0.200 ***	0.350 ***	0.018	0.136
		(9.325)	(2.794)	(3.023)	(0.392)	(1.631)
	N	574	126	182	490	266
	R2	0.041	0.450	0.296	0.092	0.093
Indirect Effect	LR_Indirect
	lnx	−0.710 ***	−0.551 ***	0.027	0.552	−0.858
		(−3.481)	(−2.593)	(0.051)	(1.163)	(−1.504)
	lnk2	−0.041 ***	−0.060 ***	−0.033	0.030	0.011
		(−3.459)	(−4.745)	(−0.865)	(0.921)	(0.662)
	lnm1	0.039	0.027	−0.064	0.107 ***	0.007
		(1.627)	(1.637)	(−1.119)	(3.290)	(0.348)
	lnm2	0.038 *	−0.016	−0.143	−0.377 **	−0.176 **
		(1.740)	(−0.435)	(−1.038)	(−2.268)	(−2.396)
	lnm3	0.042	0.002	0.196 **	0.210 ***	−0.046 *
		(1.385)	(0.118)	(2.168)	(3.108)	(−1.726)
	lnm4	0.571 ***	0.144 **	0.644 ***	0.045	0.172
		(6.137)	(2.215)	(2.655)	(0.370)	(1.559)
	N	574	126	182	490	266
	R2	0.041	0.450	0.296	0.092	0.093

t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

shows that in Chengdu-Chongqing, Pearl River Delta, and Yangtze River Delta, there is a significant difference in the inhibition of fog-haze pollution by atmospheric environment self-purification capacity. The total effect in the Pearl River Delta and Yangtze River Delta passes the significance test, and atmospheric environment self-purification capacity has a very significant inhibitory effect on environmental pollution, while the effect is not obvious in the Chengdu-Chongqing region. This is mainly due to there being relatively higher rainfall and wind in the YRD and PRD regions, which have a relatively stronger self-purification capacity in the atmospheric environment and a stronger inhibitory effect on fog-haze pollution. Though their effects vary, the atmospheric environment self-purification capacities of both Beijing–Tianjin–Hebei and Northeast China contribute to the overall aggravation of environmental pollution. In the Beijing–Tianjin–Hebei region, there is a positive correlation between atmospheric environment self-purification capacity and environmental pollution, with a coefficient of 0.041. In the Northeast China region, there is a positive correlation between atmospheric environment self-purification capacity and environmental pollution, with a coefficient of 0.769. In the Beijing–Tianjin–Hebei region and Northeast China region, the coefficients prove insignificant. This is because of relatively low rainfall and wind in the Beijing–Tianjin–Hebei region and Northeast China and the limited ability of the atmospheric environment to self-purify to reduce fog-haze pollution.

5. Research Conclusions and Suggestions for Policies

5.1. Research Conclusions

Based on the above empirical analysis, the following conclusions are drawn: Firstly, the main reason for fog-haze pollution is that the weighting of particulate matter data is too high and the impact of particulate matter is very small. The outbreak of haze in China was caused by the implementation of the gravimetric method for detecting haze in 2011 and 2012, resulting in the low-cost emission of a large number of ultrafine particles. Secondly, the impact of atmospheric environment self-purification capacity on fog-haze pollution has a dual threshold effect with green technology innovation: when the level of green technology innovation exceeds a certain level, the inhibition effect of atmospheric environment self-purification capacity on fog-haze pollution is significantly enhanced, and the threshold value of level of green technology innovation can be determined by measuring the reasonableness of the layout of pollution sources and other indicators, and when that is greater than the first threshold, atmospheric environment self-purification capacity can significantly reduce the impact of fog-haze pollution, whereas when this level reaches the second threshold, atmospheric environment self-purification capacity can also significantly reduce the impact of fog-haze pollution. Thirdly, atmospheric environment self-purification capacity has obvious regional differences, as the atmospheric environment self-purification capacity in the prefecture-level city region of Guangdong is significantly higher than that in the prefecture-level city region of Hebei, and it can directly improve local fog-haze pollution; i.e., there is a significant negative correlation between atmospheric self-purification capacity and environmental pollution. Atmospheric self-purification capacity has a certain spatial spillover effect, which can directly improve haze pollution in surrounding areas, or indirectly improve it through green technology innovation. Fourthly, due to the irrational layout of high-pollution industries, the particulate matter emitted by other high-polluting industries exceeds the limit of atmospheric self-purification capacity, and the haze pollution in some parts of China is even more severe.

5.2. Suggestions for Policies

Our policy suggestions are as follows: The first is to formulate differentiated green input policies. Government departments should formulate and implement differentiated green input and atmospheric self-purification capacity enhancement policies according to the levels of atmospheric pollution, self-purification capacity, and green technology innovation in different places, such as for areas with serious atmospheric pollution and a better foundation for green technology innovation to increase their green technology innovation input, develop targeted policy measures to enhance their green technology innovation, and make full use of green technology innovation capacity to inhibit fog-haze pollution through atmospheric environment self-purification capacity. The innovation capability of green technology should be fully utilized, and haze pollution can be suppressed through atmospheric environment self-purification capacity. In areas where atmospheric pollution is not severe and the foundation for green technology innovation is not good, more green funds should be invested in improving atmospheric environment self-purification capacity so as to fully leverage its ability to suppress haze pollution. Secondly, the scientific enhancement of atmospheric environment self-purification capacity, especially in seriously polluted areas, can effectively inhibit regional environmental pollution. These places should make full use of atmospheric environment self-purification capacity to systematically improve the quality of the atmospheric environment. Atmospheric environment self-purification capacity is affected both by natural factors and by human factors. Due to the lack of elasticity in natural factors affecting atmospheric environment self-purification capacity, such as seasonal changes, terrain conditions, the maximum thickness of the mixed layer, and atmospheric stability, it is difficult for these factors to undergo significant changes. Therefore, by strengthening the influence of human factors in these areas, such as scientifically planning the proportion of built-up areas for urban green development, effectively enhancing artificial wetlands, and reasonably arranging industrial development, we can provide effective and targeted policy recommendations to effectively improve the self-purification capacity of the regional atmospheric environment. Thirdly, for the gradient transfer of polluting industries based on environmental self-purification capacity, according to the calculation results for atmospheric environment self-purification capacity combined with the location entropy of the production value of pollution-intensive industries and the PM2.5 emission concentration of each province and city, the transfer area for air pollution-intensive industries should be scientifically selected. Then, we can systematically improve the guarantee system for the gradient transfer of polluted industries based on environmental self-purification capacity based on supply aspects of land, capital, and talents.

The research shortcomings and prospects of the article include the following: due to limitations in research funding and time, the relative complexity of the problem, the causal mechanism behind green technology innovation and atmospheric environment self-purification capacity, the potential endogeneity of green technology innovation and the level of economic development in different regions, the reliability of green patents in measuring the level of “green technology innovation” in different contexts, the reliability of green patents in measuring green technology innovation by selecting only green patents interested in air quality, etc., we plan to conduct related research in subsequent studies.