Study on Green Airport Construction and Aviation Pollution Control: A Case Study of Four International Airports

Abstract

In the era of globalization and information technology, the aviation industry has experienced rapid growth. However, the increase in flight numbers has exacerbated environmental issues such as exhaust emissions and noise pollution, raising significant concerns across society. This paper aims to explore the current state of environmental pollution within the aviation industry and propose solutions to promote the development of green airports and effective pollution control measures. This study primarily employs a literature analysis. Initially, a preliminary evaluation index system was established to represent various aspects of aviation pollution. The system was then refined and optimized using the entropy weight method. Subsequently, kernel density estimation and Moran index methods are applied to analyze the temporal and spatial trends in the evaluation indicators. An empirical study is conducted to investigate the degree of endogenous correlation and lag effects among the indices. The results are as follows: (1) Regional neutrality in pollution indicators. The spatial autocorrelation test reveals a lack of significant spatial correlation among the studied aviation environmental pollution indicators, indicating that these variables maintain a degree of regional neutrality. (2) Cargo throughput affects aviation environmental pollution. The PVAR model analysis highlights that cargo throughput has a significant self-impact on aviation environmental pollution, indicating that monitoring and managing cargo operations could be crucial in predicting and mitigating future pollution levels.

1. Introduction

In the contemporary era of globalization and information technology, the aviation sector has become an integral component of modern society. Air transportation significantly bolsters international trade, tourism, and cultural exchange, while simultaneously playing a vital role in global economic integration [1]. Nonetheless, the rapid expansion of the aviation sector has resulted in prominent environmental concerns, capturing the attention of governments and the public internationally. Although the aviation industry has reaped significant economic benefits [2], its environmental footprint has become a matter of grave concern. The convenience and expediency offered by air transport have revolutionized global supply chains and logistics cycles by enabling the swift transit of goods and people across continents and oceans [3]. However, this aviation-dependent economy introduces severe environmental challenges [4].

Studies suggest that the environmental impacts of aviation predominantly emerge in three areas: greenhouse gas emissions, noise pollution, and energy consumption [5]. Aircraft significantly contribute to greenhouse gas emissions, particularly carbon dioxide (CO₂) and nitrogen oxides (NOx), directly affecting air quality and indirectly contributing to global climate change through the greenhouse effect and ozone layer depletion [6]. The International Civil Aviation Organization (ICAO,2024) has set emission standards for aircraft engines to limit pollutants like nitrogen oxides (NOx), hydrocarbons (HC), carbon monoxide (CO), and particulate matter, thereby reducing air pollution. Additionally, ICAO has established noise standards for aircraft to mitigate noise pollution around airports. Many countries and airlines have committed to achieving carbon neutrality in the coming decades, supported by ICAO’s CORSIA initiative, which offsets international flight emissions through carbon offset projects. Governments and airlines are also promoting the use of Sustainable Aviation Fuel (SAF), made from renewable resources, to reduce environmental impact. Efforts to cut ground emissions at airports include the adoption of electric ground support equipment (eGSE) and other low-emission technologies. Enhancements in flight route optimization and air traffic management aim to reduce flight times and fuel consumption, lowering carbon emissions. Regional organizations like the European Union have implemented regulations such as the EU Emissions Trading System (EU ETS), requiring airlines to purchase carbon emission allowances. These combined measures help reduce environmental pollution and promote sustainable development in the aviation industry [7]. Besides gas emissions, noise pollution is another significant concern in the aviation sector [8]. Constant exposure to high-intensity noise can result in hearing loss, sleep disturbances, and cardiovascular diseases [9]. Despite technological advancements in aircraft design aimed at noise reduction, the increasing number of flights continues to worsen noise pollution [10].

The aviation sector’s considerable energy consumption impacts not only aircraft operations, but also airports and the wider aviation supply chain [11]. This has serious environmental consequences, prompting the industry to explore eco-friendly technologies and sustainable development [12]. Green airports, which focus on efficient energy use, effective waste management, noise control, and ecological protection, have emerged as a key solution to these challenges [13]. This review examines the core components and innovations in green airport construction, including renewable energy usage, flight path optimization, soundproof facilities, and advancements in aircraft technology [14,15]. Energy consumption within the aviation sector significantly contributes to its environmental footprint. Researchers such as Kacar et al. [16] emphasize the adoption of energy-efficient technologies in airport infrastructure, including advanced HVAC systems, LED lighting, and automated energy management systems. Baxter’s [17] study found that the photovoltaic (PV) solar systems at Brisbane and Melbourne airports will enable these airports to reduce their CO₂ emissions by around 8000 tonnes per year. The exploration of low-carbon technologies is crucial for minimizing flight emissions. According to Kousoulidou et al. [18], sustainable aviation fuels (SAF) present a viable alternative to conventional jet fuels, offering significant reductions in carbon emissions. Additionally, the development of electric and hybrid aircraft, as discussed by Uysal [19], represents a promising advancement toward more sustainable aviation practices. Effective waste management is an essential component of green airport operations [20]. Sarbassov et al. [21] describe comprehensive waste management programs that include recycling and composting, which significantly mitigate the environmental impact of airport operations. Concurrently, water conservation strategies, such as low-flow fixtures and rainwater harvesting, are critical [22]. Bieliatynskyi et al. [23] concluded that the implementation of the given algorithm on the basis of computer technology will increase the efficiency of wastewater purification and significantly reduce fresh water consumption. Noise pollution is another environmental concern addressed by green airport initiatives [24]. Innovations in constructing soundproof facilities and optimizing flight paths can notably reduce noise impact on surrounding communities [25]. Wu et al. [26] discuss how implementing soundproofing technologies and strategic operational changes can create a more harmonious coexistence between airports and local residents. Furthermore, ecological protection measures, such as preserving natural habitats and implementing green spaces, contribute to the overall environmental sustainability of airports. The current research underscores the multifaceted approach required for green airport construction, integrating energy efficiency, renewable energy, waste management, water conservation, noise control, and ecological protection. While significant strides have been made, ongoing research and development in sustainable aviation fuels and electric aircraft technologies are crucial for further reducing the environmental impacts of aviation. Collaboration among industry stakeholders, policymakers, and researchers will be essential to drive innovation and implement comprehensive green airport strategies.

International cooperation is pivotal in the environmental protection of the aviation industry [27]. Governments and international organizations are implementing stringent environmental standards and policies to guide the aviation sector towards a sustainable future [28]. However, most existing research tends to concentrate on the individual aspects of environmental impacts and suggests solutions accordingly [29]. This study builds upon previous research and critical data, utilizing methodologies such as Life Cycle Assessment (LCA) and Data Envelopment Analysis (DEA) [30] to quantify the environmental performance of airports. This study utilizes a preliminary evaluation index system based on the existing literature [31], which is further refined and optimized through the entropy weight method. It also conducts temporal and spatial trend analyses using Kernel Density Estimation (KDE) and Moran’s I method [32,33]. This study investigates how, despite pandemic disruptions, aviation pollution indicators are stable and increasing, with regional differences decreasing. Negative Moran’s I index shows no spatial correlation, indicating regional neutrality.

2. Materials and Methods

2.1. Evaluation Indicators and Data Collection

This study draws on recent research related to aviation emissions and pollution control based on the “Guiding Principle for Accelerating Energy Conservation and Emission Reduction Efforts” and the widely recognized objective of fostering sustainable development within the aviation industry. To develop the evaluation indicator system, this study references several key documents, including the Green Airport Evaluation Guidelines issued by the Civil Aviation Administration of China (CAAC), the 14th Five-Year Plan for Green Aviation Development, and research conducted by Wei [34]. The system was designed with a focus on scientific rigor, comprehensiveness, and the availability of data. The structure and details of the aviation environmental pollution assessment indicator system are presented in

Table 1. Aviation environmental pollution evaluation index.

Primary Indicators	Secondary Indicators
Airport operation efficiency	Number of takeoffs and landings Passenger throughput Cargo throughput
Pollution prevention and environmental management	Energy consumption
	Waste gas investment
Economic efficiency and sustainable development	Aviation revenue
	Total assets
	Operating cost

.

The panel data utilized in this study covers the period from 2013 to 2022 and includes information from four major airports, Shanghai Pudong International Airport and Hongqiao Airport in China, as well as Los Angeles International Airport and San Francisco International Airport in the United States. The primary sources of these data include official websites such as those of the National Bureau of Statistics of China, Huajing Intelligence, as well as the annual reports and corporate social responsibility (CSR) reports for the years 2013 to 2022 from Shanghai Airport Group, Los Angeles International Airport, and San Francisco International Airport. Additional data were obtained from Statista, the National Environmental Monitoring Center of China, the official website of the U.S. Census Bureau, and the California Climate Investments website. In cases where data were missing, interpolation techniques were applied to estimate the missing values. Furthermore, the variables for the four airports in the evaluation indicator system were averaged to ensure consistency in the analysis. The data sample of this study includes the annual flight takeoffs and landings, passenger throughput, cargo and mail throughput, energy consumption, waste gas and environmental protection investment, aviation revenue, total asset value, and operating cost data of Shanghai Pudong International Airport and Hongqiao International Airport in China, Los Angeles International Airport, and San Francisco International Airport in the United States from 2011 to 2022.

2.2. Entropy Weight Method for Key Indicator Selection

In this study, the entropy value method was used to select important indicators, incorporating the time variable in the process of determining the entropy values. The specific steps are as follows:

Step 1: The standardization of indicator data using the extreme value method is essential to ensure accurate and reliable results. This step helps in identifying and removing any outliers that may skew the data, allowing for a more consistent and meaningful analysis.

$$\begin{gathered} {X^{\prime }}_{\theta ij}={\displaystyle \frac{X_{\theta ij}-X_{jmin}}{X_{jmax}-X_{jmin}}} \text{(For positive indicators)} \\ {X^{\prime }}_{\theta ij}={\displaystyle \frac{X_{jmax}-X_{\theta ij}}{X_{jmax}-X_{jmin}}} \text{(For negative indicators)} \end{gathered}$$

(1)

where $\theta =\mathrm{2013,2014},...,2019$; $i=\mathrm{1,2},\mathrm{3,4}$ represent the four airports: Pudong International Airport, Hongqiao Airport, Los Angeles International Airport, and San Francisco International Airport; and $j=\mathrm{1,2},...,8$ represent the eight secondary evaluation indicators: flight movements, passenger throughput, cargo throughput, energy consumption, environmental protection investment in emissions reduction, aviation revenue, total assets, and operating costs. Therefore, $X_{\theta ij}$ represents the value of the j-th evaluation indicator for the i-th airport in the θ-th year. $X_{jmin}$ and $X_{jmax}$ are the minimum and maximum values of the j-th evaluation indicator, respectively.

Step 2: Non-negative transformation:

$${X^{″}}_{\theta ij}={X^{\prime }}_{\theta ij}+M$$

(2)

A constant M is added to reduce bias. Based on the previous literature [35], where $M=0.00001$.

Step 3: Determination of characteristic proportions for each indicator:

$$P_{\theta ij}={X^{″}}_{\theta ij}÷\sum _{\theta }\sum _i{X^{″}}_{\theta ij}$$

(3)

Step 4: Calculation of entropy value for each indicator:

$$e_j=-k\sum _{\theta }\sum _i{P}_{\theta ij}\mathit{ln}\left(P_{\theta ij}\right)$$

(4)

where $k>0$ is a constant and the entropy value must fall within the range of $0\le e_j\le 1$. Therefore, we choose $k={\displaystyle \frac{1}{99}}$.

Step 5: Calculation of indicator weights:

$$w_j=(1-e_j)/\sum _j(1-e_j)$$

(5)

Step 6: Calculation of comprehensive scores:

$$U_{\theta ij}=\sum _j{w}_j{X^{″}}_{\theta ij}$$

(6)

By applying the entropy value distribution to the aviation environmental pollution evaluation indicator system, we obtained the entropy value weights for the specific indicators, as shown in

Table 2. Aviation environmental pollution evaluation index weight.

Level	Measure Index	Utility	Weight
Airport operation efficiency	Number of takeoffs and landings	Positive indicators	28.60%
	Passenger throughput	Positive indicators	4.12%
	Cargo throughput	Positive indicators	25.49%
Pollution prevention and environmental management	Energy consumption	Negative indicators	22.74%
	Waste gas investment	Positive indicators	0.18%
Economic efficiency and sustainable development	Aviation revenue	Positive indicators	1.16%
	Total assets	Positive indicators	4.12%
	Operating cost	Negative indicators	13.59%

.

As shown in Table 2, four measurement indicators—flight movements, cargo throughput, energy consumption, and operating costs—are assigned relatively high weights, with values of 0.286, 0.2549, 0.2274, and 0.1359, respectively. In contrast, the weights for other indicators such as passenger throughput, environmental protection investment for emissions reduction, aviation revenue, and total assets are much lower, with values of 0.0412, 0.0018, 0.0116, and 0.0412, respectively. The entropy weight method is used to determine the relative importance of each indicator in the evaluation system. Based on this, this study eliminates indicators with weights below 0.1, which includes passenger throughput, environmental protection investment in emissions reduction, aviation revenue, and total assets. Consequently, for the subsequent empirical analysis, this study will focus on four key variables: flight movements (TAL), cargo throughput (CT), energy consumption (EC), and operating costs (OC).

This study incorporates both a time trend analysis and a spatial autocorrelation test of panel data to examine the evolution patterns, characteristics, and spatial structures of the variables. The aim is to identify underlying patterns and relationships within the data, providing a more robust basis for the subsequent Panel Vector Autoregression (PVAR) model, and to explore the dynamic interactions between the variables. The original data used in the analysis are presented in Supplementary Materials Table S1. To ensure consistency in the dataset, especially considering the need for international unit conversions, this study standardizes the units for the key variables. Specifically, flight movements (TAL) are expressed as the number of flights, cargo throughput (CT) is converted into metric tons, energy consumption (EC) is standardized to the equivalent tons of coal needed to produce the same amount of heat, and operating costs (OC) are adjusted to U.S. Dollars based on the current exchange rate. This standardization allows for a consistent and comparable analysis across the different airports and time periods.

2.3. Analysis of Time Change Trend in Aviation Environmental Pollution

Next, we analyzed the spatio-temporal variation trend in the evaluation index series data using the kernel density estimation method and Moran index. We empirically tested the endogenous correlation and hysteresis effect of each index.

In this study, kernel density estimation method is used, and the formula is as follows:

$${\widehat{f}}_h\left(x\right)={\displaystyle \frac{1}{n}}\sum _{i=1}^n{K}_h\left(x-x_i\right)={\displaystyle \frac{1}{nh}}\sum _{i=1}^{n}K\left({\displaystyle \frac{x-x_i}{h}}\right)$$

(7)

Here, n represents the total number of indicators, with n = 4 in this case. $x_i$ represents the specific indicators selected, which include TAL, CT, EC, and OC. h is the bandwidth, and, generally, the value of the bandwidth is determined by the size of n. The relationship between h and n is expressed as $h=c·n^{-{\displaystyle \frac{1}{5}}}$, where c is 1.05 times the standard deviation of the data sequence. K is the chosen kernel function, as determined by the previous literature [36]. This study selects the Gaussian kernel function, which is expressed by the following formula:

$$K\left(x\right)={\displaystyle \frac{1}{\sqrt{2\pi }}}{e}^{-{\displaystyle \frac{1}{2}}{x}^2}$$

(8)

In this study, the above kernel density estimation function is extended from one-dimensional to multidimensional, and the formula is as follows:

$${\widehat{f}}_h\left(x\right)={\displaystyle \frac{1}{n{h}^d}}\sum _{i=1}^N{K}_{\left({\displaystyle \frac{x-x_i}{h}}\right)}$$

(9)

Here, d represents the dimension of x, with d = 3 in this case, corresponding to the time dimension, indicator dimension, and Kernel density dimension. K is the Kernel density function. In this study, data samples were screened by finding outliers in the three-dimensional nuclear density state distribution. As shown in Figure 1, due to the influence of COVID-19, abnormal information is displayed in the nuclear density state distribution during 2020–2022. Therefore, sample data from 2020 to 2022 are deleted in the subsequent model calculation.

2.4. Dynamic Correlation Analysis Based on the PVAR Model

After determining that aviation pollution does not exhibit regional clustering similarity and has a dispersed spatial distribution, we further analyzed the spatial dynamic relationships and interaction mechanisms of various variables. To this end, we employed a PVAR model to analyze the panel data of four representative airports from 2011 to 2019. The mathematical model is as follows:

$$y_{it}={\alpha }_0+\sum _{j=1}^p{\alpha }_{ij}{y}_{i,t-j}+{\gamma }_i+{\theta }_t+{\epsilon }_{it}$$

(10)

where $i$ represents the four airports, $t$ denotes the year, and $p$ indicates the lag order, which is optimally selected based on sample stationarity and subsequent data processing. $y_{it}$ represents the column vector of endogenous variables, ${\mathsf{\alpha }}_{ij}$ denotes the matrix to be estimated, ${\gamma }_i$ is the fixed effects matrix, ${\theta }_t$ represents the time effects, and ${\epsilon }_{it}$ is the regression residual of the model.

3. Results

The observation period is from 2011 to 2022, as shown in Figure 1. From the perspective of spatial distribution, Figure 1a shows that the central value of flight movements gradually shifts to the right, reflecting the steady improvement in the global aviation industry. Additionally, the rapid evolution in this trend is likely associated with global economic growth and population increase. In contrast, the changes in Figure 1b–d show less variation. In terms of distribution mode, in addition to cargo throughput, the data on the flight takeoff and landing, energy consumption, and operating costs of the four major airports show that the concentration has increased significantly around 2020, which is significantly abnormal compared to the previous gaffe. Therefore, we will perform outlier processing of the data during the 2020–2022 epidemic period, and these data will no longer be considered in the subsequent model calculation. From the perspective of the expansion of the main peak width of the Kernel density function, the actual differences among the four airports are gradually decreasing, indirectly reflecting the convergence of strategies adopted by international leading companies in response to the promotion of green aviation practices. In terms of distribution extension, cargo throughput indicates increasingly balanced spatial development, suggesting that the scale of cargo operations across the companies remains at a world-class level. The right tails of the Kernel density function for energy consumption and operating costs have become flatter, indicating a gradual reduction in spatial disparities among airlines in their efforts toward environmental management in aviation.

We applied Moran I index to analyze the spatial correlation of the four key indicators for the sample airports in Figure 1. The results, as shown in

Table 3. Moran I test of four important indicators.

Year	TAL			CT			EC			OC
	Moran I Index	Normal Statistics	p	Moran I Index	Normal Statistics	p	Moran I Index	Normal Statistics	p	Moran I Index	Normal Statistics	p
2011	−0.294	0.135	0.446	−0.497	−0.456	0.324	−0.461	−0.392	0.348	−0.647	−1.171	0.121
2012	−0.318	0.054	0.479	−0.522	−0.516	0.303	−0.475	−0.401	0.344	−0.62	−1.08	0.14
2013	−0.325	0.028	0.489	−0.508	−0.48	0.315	−0.423	−0.259	0.398	−0.603	−1.022	0.153
2014	−0.433	−0.368	0.356	−0.526	−0.529	0.298	−0.122	0.806	0.21	−0.571	−0.917	0.18
2015	−0.562	−0.858	0.195	−0.557	−0.608	0.272	−0.277	0.203	0.419	−0.532	−0.776	0.219
2016	−0.588	−0.96	0.169	−0.578	−0.666	0.253	−0.369	−0.119	0.452	−0.52	−0.724	0.235
2017	−0.607	−1.035	0.15	−0.629	−0.808	0.209	−0.537	−0.702	0.241	−0.444	−0.449	0.327
2018	−0.55	−0.8	0.212	−0.628	−0.816	0.207	−0.865	−0.333	0.024	−0.487	−0.615	0.269
2019	−0.645	−1.168	0.121	−0.648	−0.866	0.193	−0.822	−1.584	0.057	−0.446	−0.458	0.323

, indicate that all Moran’s I indices are negative, meaning that the spatial correlation analysis is not significant. This indicates that the data show little to no spatial correlation. In terms of aviation-related environmental pollution, this suggests that pollution is not concentrated in specific spatial areas. The negative spatial autocorrelation for airport operating costs and energy consumption may reflect regional differences in economic and policy environments, which could impact airport operational efficiency and energy consumption patterns, thus affecting their sustainable development strategies.

Before conducting PVAR modeling, it is important to ensure that the panel data are stationary to avoid spurious regression. This can impact the validity of the model estimation and subsequent impulse response analysis. Therefore, this study conducted stationarity tests on the panel data by region for variables such as the number of flight movements (TAL), cargo throughput (CT), energy consumption (EC), and operating costs (OC). This step is crucial for ensuring the reliability of the results. We utilized several common testing methods, such as the LLC test, IPS test, HT test, and PP-Fisher test, to examine the stationarity of the panel data. Generally, a variable is considered stationary if it passes two or more of these tests. Some variables in this study are still considered stationary, even though they did not pass all of the tests, because they passed at least two tests. For variables identified as non-stationary, we took the logarithm and performed multiple-order differencing until they passed the stationarity tests. After differencing, TAL, CT, EC, and OC are, respectively, renamed as D_TAL, D_CT, D_EC, and D_OC. The results show that, at confidence levels of 10%, 5%, and 1%, the variables are found to be significantly stationary and reject the null hypothesis of non-stationarity. The T-values, which represent the results, are displayed in

Table 4. Unit root test for variables.

Variables	LLC Test	IPS Test	HT Test	ADF Test	PP Test	Conclusion
DTAL	−1.0141 *	−2.9801 **	−0.2641 ***	17.4398	16.9475 **	Pass
DCT	−4.1956 ***	−2.7705 ***	−0.6126 ***	30.5903 ***	98.1368 ***	Pass
DEC	−3.4836 ***	−2.2976 ***	−0.5356 ***	26.7863 ***	19.6487 ***	Pass
DOC	−5.4566 ***	−2.9802 **	−0.3574 *	19.7239 ***	96.2156 ***	Pass

Where *, **, and *** indicate significance at the 10%, 5%, and 1% confidence levels, respectively, the results show that all four variables passed the tests. Additionally, this study performed cointegration analysis using the Kao Test, Pedroni Test, and Westerlund Test on the basis of passing the unit root tests. This further integration check of the data that passed the unit root tests helps to determine whether each series is stationary, which is crucial for interpreting the stability of the constructed model.

.

To capture the underlying patterns and trends in the data more effectively, it is crucial to select an appropriate lag order. Determining the correct lag order is essential for enhancing the accuracy and stability of the model’s predictions. Before proceeding with the Panel Vector Autoregression (PVAR) modeling, it is necessary to determine the optimal lag length. The methods commonly used for selecting the optimal lag order include the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Hannan—Quinn Information Criterion (HQIC). In this study, we employ all three of these criteria—AIC, BIC, and HQIC—to determine the optimal lag order for the model, as presented in

Table 5. The order of lag is determined.

Lag	BIC	AIC	HQIC
1	5.5722 *	3.9687 *	4.7621 *
2	6.24785	4.02943	4.98766
3	42.9875	39.6886	40.8762
4	48.6630	45.4397	47.4982
5	63.4796	60.3792	59.8768

Where * indicates the optimal lag order according to the results from the BIC, AIC, and HQIC criteria.

. The asterisk (*) indicates the optimal lag order according to the results from the BIC, AIC, and HQIC criteria. The model selection process involves identifying the lag order that minimizes these information criteria, ensuring the best-fitting model for the data.

Using the Generalized Method of Moments (GMM) estimation technique, we developed a Panel Vector Autoregression (PVAR) model with D_TAL, D_CT, D_EC, and D_OC as endogenous variables and applied GMM estimation to obtain the first-order regression results for the PVAR model. To address potential fixed effects in the panel data, we further transformed the data using the Helmert transformation. This transformation technique is applied to all subsequent data used in the analysis, as shown in the results presented in

Table 6. GMM panel regression estimation results.

Variables	DTAL(h)	DCT(h)	DEC(h)	DOC(h)
L1. DTAL(h)	0.2107	−0.0384	−0.6923	0.912 ***
	0.9299	0.4785	0.5356	0.3259
L1. DCT(h)	−0.2896	0.0996	0.4909	−0.0820
	0.6321	0.5179	0.4743	0.1443
L1. DEC(h)	0.1609	−0.0633	0.726 **	−0.317 ***
	0.2460	0.1816	0.1868	0.0841
L1. DOC(h)	−0.6165	−0.1744	−0.3778	−0.574 ***
	0.6196	0.1650	0.3155	0.1121

Where **, and *** indicate statistical significance at the 5%, and 1% confidence levels, respectively.

. The table reports the estimated coefficients

For each variable, the symbols *, **, and *** indicate statistical significance at the 10%, 5%, and 1% confidence levels, respectively. The notation “L1.”denotes that the data have undergone first-order regression, and “(h)” indicates that the Helmert transformation has been applied to the data.

After completing the previously described steps, we further tested the robustness and explanatory power of the constructed PVAR model by assessing its long-term stability. This was performed by calculating the unit root eigenvalues, as shown in Figure 2. The results indicate that the values of all variables lie within the unit circle (with a radius of 1), which means that the unit roots of the variables are all less than 1. This confirms the stability of the PVAR model, suggesting that the fluctuations among the variables are stable and exhibit relatively smooth behavior over time.

To explore the causal relationships between the variables, this study conducts a Granger causality test, and the results are presented in

Table 7. Granger causality test results.

Variables	Null Hypothesis	chi2	Degree of Freedom	p
DTAL(h)	DCT(h) is not the null hypothesis of DTAL(h)	0.0931	1	0.711
	DEC(h) is not the null hypothesis of DTAL(h)	0.2967	1	0.578
	DOC(h) is not the null hypothesis of DTAL(h)	1.5196	1	0.207
	None of the variables are null assumptions of DTAL(h)	3.6185	3	0.498
DCT(h)	DTAL(h) is not the null hypothesis of DCT(h)	0.0031	1	0.874
	DEC(h) is not the null hypothesis of DCT(h)	6.7398	1	0.041
	DOC(h) is not the null hypothesis of DCT(h)	0.0426	1	0.688
	None of the variables are null assumptions of DTC(h)	36.2581	3	0.000
DEC(h)	DTAL(h) is not the null hypothesis of DEC(h)	5.6120	1	0.045
	DCT(h) is not the null hypothesis of DEC(h)	0.0015	1	0.875
	DOC(h) is not the null hypothesis of DEC(h)	2.2257	1	0.322
	None of the variables are null assumptions of DEC(h)	17.1718	3	0.024
DOC(h)	DTAL(h) is not the null hypothesis of DOC(h)	4.8996	1	0.089
	DCT(h) is not the null hypothesis of DOC(h)	4.2749	1	0.172
	DEC(h) is not the null hypothesis of DOC(h)	10.2214	1	0.021
	None of the variables are null assumptions of DOC(h)	40.1313	3	0.000

. In this context, if the p-value exceeds the significance thresholds of 10%, 5%, or 1% (i.e., greater than 0.1, 0.05, or 0.01), it suggests that the null hypothesis cannot be rejected. This indicates that there is no statistically significant relationship between the explanatory variable and the dependent variable. As shown in Table 7, the results reveal that the other three variables do not significantly affect flight movements (TAL), which suggests that TAL operates relatively independently. This finding implies that the number of flight movements can be analyzed separately in studies that focus on aviation environmental pollution, with minimal influence from the other related factors.

The results indicate that cargo throughput (CT) displays a certain level of endogenous correlation, with other variables exerting a significant influence on its interaction. The p-value is extremely close to 0, indicating that, apart from CT being a cause on its own, the other variables collectively serve as Granger causes for CT. Similarly, flight movements (TAL) are found to be a Granger cause for energy consumption (EC), indicating that airports’ energy consumption is strongly dependent on the number of flight movements. Additionally, the other three variables also have a significant effect on EC, with a p-value of 0.012. This suggests that interactions between these variables have a notable impact on aviation’s environmental pollution.

Finally, for operating costs (OC), the null hypothesis can be significantly rejected in all cases, indicating that OC is strongly related to the other variables in the context of aviation’s environmental pollution. This implies that decisions regarding aviation pollution are closely tied to cost considerations and that effective environmental policies can often lead to higher economic benefits, as they influence cost structures within the industry.

This study examines the dynamic relationships between variables by analyzing the impact of a one-standard-deviation shock on the current and future values of the variables within the system. This approach provides an intuitive understanding of how random disturbances affect the endogenous variables in the PVAR model. Specifically, the impulse response results for each variable are obtained by simulating 200 iterations using the Monte Carlo method, with the shock period set to 10 periods, as illustrated in Figure 3. The horizontal axis represents the number of lags corresponding to the shock, while the vertical axis shows the response magnitude of each variable to the shock. The red line represents the impulse response function, while the green and blue lines provide a confidence interval for the responses over time. The green line indicates the range within two standard deviations for positive deviations, and the blue line indicates the range for negative deviations.

Based on the Granger causality analysis, we observed that not all variables exhibited mutual relationships. Therefore, our analysis of impulse response functions focuses on the variables that passed the test: D_EC(h) to D_CT(h), D_TAL(h) to D_EC(h), D_TAL(h) to D_OC(h), D_CT(h) to D_OC(h), and D_EC(h) to D_OC(h). These relationships underscore the intricate interactions between internal factors in aviation and environmental pollution. As depicted in Figure 3, when a one-standard-deviation shock is applied to energy consumption, the response of cargo throughput is initially negative. Around period 3, there is a slight positive response, but it gradually decreases and approaches zero over time. This suggests that any temporary positive reaction is minimal, and the overall negative feedback prevails. This indicates that energy consumption exerts a persistent negative effect on cargo throughput, emphasizing the need for airports to consider energy usage in managing cargo throughput, as it significantly contributes to aviation pollution. When a one standard deviation shock is applied to flight movements, the change in energy consumption is slightly positive and approaches zero. This suggests that flight movements have a limited impact on energy consumption, which becomes less significant over time. It implies that emissions from energy consumption are not the main contributors to pollution, possibly because of the role of electricity consumption by airport facilities and the surrounding infrastructure, which reduces the importance of the positive feedback. The diminishing positive response suggests that airports have made efforts to manage carbon emissions, but there is still room for improvement in energy efficiency for infrastructure. When a one-standard-deviation shock is applied to flight movements, the response of operating costs is initially negative, then increases around period 1, followed by positive feedback in period 2, before gradually declining. This indicates that, during the early operational phase, flight movements do not yield significant profits for airports. However, as operations continue, profits rise, primarily due to the high initial investment in manpower, resources, and equipment. After overcoming the transition period, short-term profits can be realized. Nevertheless, airports should consider environmental protection measures, as neglecting them may undermine the positive impact, leading to diminishing returns over time. The response variable in all cases is operating costs, and the patterns for cargo throughput, flight movements, and energy consumption are consistent. Therefore, the analysis follows the same reasoning. Furthermore, these three variables are endogenous factors in aviation environmental pollution. Therefore, airports should focus more on fundamental environmental factors when evaluating cost–benefit relationships.

4. Discussion

This study initially applies the Kernel density estimation method to investigate the temporal evolution trends in the reference data for four variables influencing aviation environmental pollution from 2011 to 2022 from a three-dimensional perspective. This method facilitates a more intuitive comprehension of the development direction and heterogeneity of the research subjects. The visual representation shows that, aside from the unusual disruptions caused by the pandemic in different regions, the historical trends in the indicators are mostly stable and increasing, with the regional differences gradually decreasing. The abnormal sample data from 2020 to 2022 are excluded according to the nuclear density state diagram.

A subsequent spatial autocorrelation test is performed to determine the existence of regional variations in aviation environmental pollution indicators. The results indicate an almost uniformly negative Moran’s I index, signifying an absence of spatial correlation among the studied factors. This implies that the variables under study maintain regional neutrality in the context of aviation environmental pollution research.

Subsequently, a PVAR model is established. Through panel data analysis and the Granger causality test, it is revealed that flight movements exhibit a certain degree of autonomy in relation to other variables, providing a focused viewpoint for addressing aviation emission issues in environmental pollution analysis. The impulse response analysis indicates that cargo throughput exerts the most significant self-impact over the forecast period. Additionally, operating costs also show a significant impact. However, other variables do not demonstrate significant future impact correlations. Therefore, the issue of cargo throughput can be utilized to predict the future impacts of the current state of aviation environmental pollution.

5. Conclusions

This study evaluates aviation environmental pollution by analyzing key operational indicators from four major airports—Shanghai Pudong, Shanghai Hongqiao, Los Angeles International, and San Francisco International—covering the period from 2011 to 2022. It uses the entropy weight method to prioritize variables such as flight movements, cargo throughput, energy consumption, and operating costs, all of which affect pollution levels. The findings highlight steady growth in flight movements and cargo throughput, with decreasing regional differences, suggesting a shift towards standardized green aviation practices. Excluding pandemic-related anomalies ensures that this analysis focused on stable pre-pandemic trends. The Moran’s I indices indicate decentralized pollution sources across airports. This study’s use of the PVAR model reveals that flight movements have independent effects on other variables, while cargo throughput shows complex relationships with energy consumption, flight movements, and operating costs. Energy consumption initially increases with flight movements, but diminishes over time as energy efficiency improves. This study emphasizes the need for the continuous monitoring of sustainability efforts in aviation, recommending a balance between sector growth and environmental sustainability.

In conclusion, this study offers essential insights for both policymakers and industry stakeholders aiming to reduce aviation-related environmental pollution. It demonstrates that the aviation industry is moving toward more standardized, sustainable practices globally, although regional differences still exist. As the industry continues to grow, it is crucial that environmental sustainability remains a key consideration in operational strategies. The ongoing convergence of green practices among major airports signals a positive trend towards achieving a more sustainable aviation future.