Modeling of Greenhouse Gases Emissions from Hong Kong’s Air Transport Industry: 2011 to 2030

Abstract

The air transport industry has played a crucial role in Hong Kong’s economic growth. However, aircraft operations produce a considerable volume of greenhouse gases emissions. By analyzing aviation kerosene consumption data from the first quarter of 2011 to the fourth quarter of 2018, this study developed a seasonal autoregressive integrated moving average (ARIMA) model—ARIMA(1,1,0)(0,1,1)₄—that accurately reflects the actual consumption patterns. This model was then utilized to forecast aviation kerosene consumption from the first quarter of 2019 to the fourth quarter of 2024, a period marked by Hong Kong’s social unrest, followed by the pandemic and post-pandemic effects of COVID-19. As COVID-19 transitioned to an endemic stage, the number of aircraft movements has steadily risen over the past three years, resulting in increased aviation kerosene consumption. This study assessed the reduction in aviation kerosene consumption and the corresponding greenhouse gases emissions during the first quarter of 2020 to the fourth quarter of 2024, primarily attributed to the impacts of the COVID-19 pandemic. It was determined that the reduction reached a peak of 15,973 kT of CO₂ in 2022, subsequently falling to 7020 kT of CO₂ in 2024. Utilizing both actual and forecasted consumption data, this study estimated greenhouse gases emissions from the Hong Kong air transport industry for the years 2011 to 2030.

1. Introduction

Air transport plays an important role in the global economy, as it supports international trade, tourism, and cultural exchange. It allows for the swift transportation of goods and individuals over long distances, significantly enhancing economic development and globalization. The International Air Transport Association (IATA) reported that airlines worldwide facilitated 4.5 billion passenger journeys and carried goods valued at over USD 6.7 trillion in 2019 [1]. The air transport industry directly and indirectly supported approximately 87.7 million jobs. However, the industry faced severe challenges due to the COVID-19 pandemic in 2020 [1]. The International Civil Aviation Organization (ICAO) indicated that global air passenger traffic dropped by around 60% in 2020 [2]. The downturn began in China, where the domestic aviation market experienced a significant decline early in 2020 [3,4]. As the pandemic spread, its effect on air transport became evident worldwide, reaching a peak in April 2020, when air passenger traffic fell by 94% compared to April 2019 [1,4]. The majority of passenger aircraft were grounded, as countries implemented travel restrictions, quarantines, and social distancing measures. This led to the collapse of air passenger connections between cities worldwide. While freight services continued, the disruption was significant, since approximately half of air cargo is transported on passenger flights [5]. The near halt of passenger operations severely affected global supply chains that modern economies depend on. Additionally, the demand for medically related products caused an increase in air cargo volume, prompting the conversion of some passenger aircraft into cargo carriers. From April 2020 to December 2022, air passenger traffic began to recover along various trajectories in different regions [4]. The U.S. air transport industry showed a steady recovery, aided by a relatively high tolerance for COVID-19 among the government and the public. By December 2022, U.S. air passenger traffic reached 80.2 million, reflecting only a 10.5% decrease from December 2019, the final month prior to the pandemic [4]. In Europe, the air transport industry also experienced significant recovery from June 2020 to December 2022, with passenger traffic totaling 57.5 million in December 2022, which was about 24.2% lower than the pre-COVID-19 level [4]. In China, air passenger traffic stood at 18.7 million in December 2022, representing a decline of over 64.5% from 52.8 million in December 2019. Fortunately, following the end of strict COVID-19 measures in late December 2022, China’s air passenger traffic began to recover rapidly, reaching 50.6 million in December 2023 and 56.98 million in December 2024.

The strategic importance of air transport is particularly evident in regions that serve as major international hubs, with Hong Kong being a prime example. Hong Kong stands out as a prominent center for both international and regional aviation. Its status as an international aviation hub is embedded in its constitutional framework, as the Basic Law requires the preservation of this status [6]. The city’s sophisticated infrastructure and free-market economic policies have historically supported its position as a crucial hub for passengers and cargo alike [6,7]. Furthermore, Hong Kong’s geographical location makes it an ideal gateway to mainland China and the wider Asia–Pacific region, enhancing its connectivity and attractiveness as a stopover destination [8,9]. The Guangdong–Hong Kong–Macao Greater Bay Area (GBA) initiative further underscores Hong Kong’s significance in the aviation industry. This initiative aims to integrate the region’s economies and create new opportunities for the city to reinforce its status as a hub for international and regional aviation [6]. Moreover, the city’s collaboration with nearby airports, such as through the Hong Kong–Zhuhai–Macao Bridge, enhances its connectivity and supports the Belt and Road economic corridor, further solidifying its strategic position in global air transport [10]. However, the COVID-19 pandemic has had a profound adverse effect on Hong Kong’s air transport. Air passenger traffic plummeted by 86.2% while air cargo was found to drop modestly by 6% in 2020, compared to 2019 levels [11].

Air transport operations require a huge investment and a considerable amount of energy [12]. Furthermore, most aircrafts must be refueled before departing from an airport. The use of aviation kerosene is essential for providing the necessary thrust to take off, ensuring a smooth flight, and allowing for a safe landing at the destination. However, the combustion of aviation kerosene results in the emission of various greenhouse gases (GHGs), including carbon dioxide (CO₂) and nitrogen oxides (NO_x), as well as water vapor, sulfur oxides, and particulate matter (PM). This study specifically aims to develop a seasonal autoregressive integrated moving average (ARIMA) model to analyze aviation kerosene consumption in Hong Kong, utilizing data from the first quarter of 2010 to the fourth quarter of 2018, prior to the adverse effects of Hong Kong’s social unrest and the significant impact of the COVID-19 pandemic, and to utilize the then developed model to assess the reduction in GHG emissions due to the pandemic.

More specifically, this study will commence with determining the relationship between the number of aircraft taking off from Hong Kong and aviation kerosene consumption. Subsequently, it will identify the optimal seasonal ARIMA model to characterize aviation kerosene consumption using quarterly data from 2010 to 2018. The model will be employed to forecast quarterly aviation kerosene consumption from 2019 to 2030. By utilizing emission and conversion factors from the Intergovernmental Panel on Climate Change (IPCC), this study will estimate GHG emissions linked to aviation kerosene consumption in Hong Kong from 2010 to 2030 [13] and evaluate the reduction in GHG emissions resulting from the COVID-19 pandemic. This study addresses the existing gap regarding the insufficient modeling of air transport following the COVID-19 pandemic [11], as well as the scarcity of studies dedicated to modeling air transport and its GHG emissions throughout the past, present, and future of air transport operations. Consequently, the findings of this study offer valuable insights into GHG emissions from the Hong Kong air transport industry for the years 2011 to 2030. Furthermore, this study uncovers important connections between aircraft movements, aviation kerosene consumption, and critical incidents, such as the COVID-19 pandemic. This indicates that either aircraft movements or aviation kerosene consumption can serve as an early warning system for potential economic disruptions. With China’s commitment to reaching peak carbon emission by 2030 and achieving carbon neutrality by 2060 [14], Hong Kong, as a key transport hub within the country, faces a considerable challenge if its air authority, airlines, and aviation fuel suppliers continue with business as usual. This research explores the significant challenge present by the increase in GHG emissions. It assesses the shortcomings of existing mitigation efforts and highlights the pressing need for innovation solutions. Potential strategies for reducing GHG emissions include improving operational efficiency, embracing sustainable aviation fuels, and electrifying specific operations. This encompasses the deployment of electric unmanned aerial vehicles for transporting goods and passengers between Hong Kong and its nearby cities in mainland China.

The rest of this paper is structured as follows: Section 2 provides a literature review on air transport modeling, focusing on predictive modeling using time series data, followed by Materials and Methods, in Section 3. Section 4 presents the Results. Section 5 covers the discussion and conclusions, including the implications and limitations of this study.

2. Literature Review on Air Transport Modeling

Air transport modeling is a multifaceted field that encompasses various approaches and methodologies to address the complexities of air transportation systems. This field covers the modeling of airport networks [15,16], airline operations and their optimization [17,18], aircraft performance and simulation models [19], as well as predictive modeling of air traffic [20,21]. Additionally, it includes the modeling of air passenger and cargo movements, airline flights, and demand forecasting [4,11,22,23,24,25,26,27,28]. Notably, the modeling and forecasting of air passenger and cargo traffic has been a focal point of research for several decades [4,11,22,23,24,25,26,28].

2.1. Modeling of Air Passenger and Cargo Traffic and Demand Forecasting

Accurate modeling in air transport is crucial for anticipating demand and managing resources effectively. ARIMA models, along with their seasonal variants, have been prominent in forecasting air transport demand. For instance, a study compared ARIMA models with regression models, concluding that ARIMA models are superior in simulating weekly air transport passenger numbers by route, particularly in the context of the U.K. Civil Aviation Authority’s forecasting approach [28]. Conversely, the seasonal ARIMA models are more versatile in addressing the variations in air passenger and cargo movements on a monthly basis [11]. This approach can be traced back to the foundational work of Box and Jenkins (1970) in their seminal book, Time Series Analysis: Forecasting and Control [22], where they applied the seasonal ARIMA model to analyze total monthly international airline passenger data in the U.S. from 1949 to 1960. Their model, referred to as “ARIMA(0,1,1)(0,1,1)₁₂,” has since been recognized as the Airline model. Over the past decades, this model has been effectively utilized to characterize air passenger movements in the U.S. [4,22], in Europe [4], and in China [4], including Hong Kong [11]. It has also been used to characterize monthly air cargo traffic in Hong Kong appropriately [11]. Specifically, Chen et al. [23] reported that seasonal ARIMA models demonstrated superior accuracy in predicting air passenger traffic, outperforming other models, such as the Holt–Winters and grey forecasting model for inbound air passenger traffic to Taiwan. Furthermore, Coshell [29] indicated that seasonal ARIMA models outperformed the Holt–Winters and Naïve models regarding both the goodness of fit and forecasting precision for outbound air passenger flows from the U.K. to 20 different destinations.

2.2. GHG Emission from Aircraft

Commercial aviation is a significant and growing contributor to GHG and other pollutants, which play a role in climate change and alterations in the atmosphere [30]. The engines of aircraft emit a variety of pollutants, such as CO₂, water vapor, NOx, and black carbon (BC), all of which contribute to the greenhouse effect and climate change [30]. The impact of aircraft emissions is particularly pronounced at high altitudes, where they can lead to increased ozone formation and cloudiness, thereby intensifying the greenhouse effect [31]. The release of NO_x at cruising altitudes of 10–12 km is particularly concerning due to its longer atmospheric lifespan of 5 to 10 days, in contrast to the mere hours that similar emissions last at ground level [32,33]. Consequently, NO_x emissions from commercial aircraft contribute to a net increase in ozone levels, especially in the upper troposphere [32]. Vouitsis et al. [34] examined airborne nanoparticles related to transportation and the influence of aircraft emissions on the formation of secondary organic aerosols. They reported that while regulations have significantly reduced emissions from road transport, air and maritime transport continue to contribute to urban nanoparticle pollution in numerous locations.

3. Materials and Methods

Hong Kong has long been recognized as a major air transport hub, playing a crucial role in global aviation networks. Its strategic geographic location and advanced infrastructure have enabled it to serve as a key transit point for both travelers and air cargo [5]. Notably, prior to the relocation of its airport to Chek Lap Kok, the former Kai Tak Airport was already the third-busiest airport globally for international travelers and the largest for freight in the 1990s [35]. Anticipating the constraints of an airport situated in the city center, the Hong Kong Government embarked on the construction of a new airport on Chek Lap Kok, a man-made island adjacent to Lantau Island, in 1989. The Hong Kong International Airport was inaugurated with a single runway in July 1998, with an estimated construction cost of around USD 60 billion. The project marked a significant achievement in Hong Kong’s infrastructure development, underscoring its commitment to preserving its role as a leading international aviation hub [36]. Within a year of its opening, the airport opened a second runway in May 1999. The construction of a third runway commenced in 2016. The third runway was officially opened in November 2022 for USD 18 billion. Between 2011 and 2018, Hong Kong International Airport ranked as the third busiest airport for international passenger traffic and held the title of the busiest airport for cargo [37].

3.1. Data Sources and Data

Aircraft operations and the consumption of aviation kerosene are intricately related. The Hong Kong Civil Aviation Department (HKCAD) oversees air traffic control services for all aircraft within the Hong Kong Flight Information Region and maintains air traffic statistics. Monthly data on aircraft departures from 2011 to 2024 was sourced from the HKCAD website at https://www.cad.gov.hk/english/statistics.html (accessed on 1 May 2025) [38]. For information on aviation kerosene consumption, the Hong Kong Census and Statistics Department (HKCSD) publishes annual and quarterly reports titled “Hong Kong Energy Statistics,” which include data on aviation kerosene usage categorized by “sales of oil products by type of users,” accessible at https://www.censtatd.gov.hk/en/scode90.html (accessed on 1 May 2025) [39]. This study obtained aviation kerosene consumption data for Hong Kong from the first quarter of 2011 to the fourth quarter of 2024, which was the most recent data available at the time of this writing. Given that the aviation kerosene consumption data is reported quarterly, the monthly aircraft departure figures from Hong Kong for the years 2011 to 2024 were converted to quarterly data.

3.2. Seasonal ARIMA Model

In time series, seasonality refers to a consistent pattern of fluctuations that recurs every S time intervals, where S indicates the number of intervals before the pattern reappears. For data collected quarterly, S is typically set to 4 intervals per year. In a seasonal ARIMA model, the seasonal autoregressive (AR) and moving average (MA) components forecast the value of x_t by utilizing data points and errors from previous periods that are multiples of S. Indeed, the seasonal ARIMA model effectively integrates both seasonal and non-seasonal elements within a comprehensive framework. The model is denoted as ARIMA(p, d, q)(P, D, Q)_S, where p represents the order of the non-seasonal AR, d signifies the degree of non-seasonal differencing, q indicates the order of the non-seasonal MA, P denotes the order of the seasonal AR, D represents the degree of seasonal differencing, Q indicates the order of the seasonal MA, and S specifies the duration of the recurring seasonal pattern. The model can be expressed mathematically as

$$\Phi \left(B^S\right)\varphi \left(B\right){∆}_S^D{∆}^d\left(x_t\right)=\mathrm{\Theta }\left(B^S\right)\theta \left(B\right){\epsilon }_t$$

(1)

where x_t is the nonstationary time series, and ε_t is the Gaussian white noise process. Additionally, the non-seasonal components include the AR and MA parts:

$$\varphi \left(B\right)=1-{\varphi }_{1}B-\cdots -{\varphi }_p{B}^p$$

(2)

$$\theta \left(B\right)=1+{\theta }_{1}B+\dots +{\theta }_q{B}^q$$

(3)

The seasonal components also consist of the AR and MA parts:

$$\Phi \left(B^S\right)=1-{\varphi }_1{B}^S-\cdots -{\varphi }_P{B}^{PS}$$

(4)

$$\left(B\right)=1+{\Theta }_1{B}^S+\cdots +{\Theta }_Q{B}^{QS}$$

(5)

The differencing operators are

$${\Delta }^d={\left(1-B\right)}^d$$

(6)

$${\Delta }_S^D={\left(1-B^S\right)}^D$$

(7)

The backshift operator B is defined as

$$B^k{x}_t=x_{t-k}$$

(8)

In this research, a seasonal ARIMA model was utilized to analyze quarterly aviation kerosene consumption data in Hong Kong, covering the period from the first quarter of 2011 to the fourth quarter of 2018. The data were sourced from the Hong Kong Census and Statistics Department [39] and imported into IBM SPSS 29.0 for time series modeling using the Forecasting module. Following the methodology proposed by Box and Jenkins [22], the development of a seasonal ARIMA model for a specific time series should involve three key steps: model identification, model fitting, and model verification.

During the initial phase, the numbers of differencing, i.e., d and D, which render the time series stationary, were determined through plots of the differenced time series. The autocorrelation and partial autocorrelation functions were employed to ascertain the degrees of AR and seasonal AR components, represented as p and P, as well as the degrees of MA and seasonal MA components, indicated as q and Q, following the differencing process.

In the second phase, the key parameters ϕ_i, Φ_i, θ_i, and Θ_i were estimated using maximum likelihood estimation techniques. The final phase involved assessing the model’s accuracy by comparing the predicted aviation kerosene consumption data with the actual figures. Various metrics, including mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE), coefficient of determination (R²), and Bayesian information criterion (BIC), were utilized to evaluate forecasting precision. Subsequently, the model was employed to project quarterly aviation kerosene consumption from the first quarter of 2019 to the fourth quarter of 2030. Additionally, with the availability of quarterly data from the first quarter of 2020 to the fourth quarter of 2024, the effects of COVID-19 on aviation kerosene consumption were analyzed. This study utilized the forecasting module provided by IBM SPSS 29.0 software for model identification, fitting, and verification.

3.3. GHG Emission from Aviation Kerosene

Aviation kerosene is composed of hydrocarbons such as paraffins (alkanes such as n-dodecane), olefins, naphthene, and aromatics, along with antioxidants and metal deactivators. Its major ingredient is often represented by n-dodecane (C₁₂H₂₆) or similar long-chain alkanes for combustion studies. The complete combustion of n-dodecane in air can be represented as C₁₂H₂₆ + 18.5 O₂ → 12 CO₂ + 13 H₂O. Therefore, the products of this combustion reaction are carbon dioxide and water vapor. According to the IPCC [13], emissions from aircraft engines are primarily made up of approximately 70 percent CO₂, nearly 30 percent water vapor, and less than 1 percent each of NO_x, carbon monoxide (CO), sulfur oxides (SO_x), non-methane volatile organic compounds (NMVOC), PM, and other trace components. The IPCC [13] also notes that modern engines emit minimal to no nitrous oxide (N₂O) and methane (CH₄). Furthermore, the IPCC (2006) estimates that around 10 percent of aircraft emissions, excluding CO and hydrocarbons, occur during ground operations at the airport and during the takeoff and landing phases. The majority of emissions, about 90 percent, are released at cruising altitudes. For CO and hydrocarbons, the distribution is approximately 30 percent local emissions and 70 percent at higher altitudes. The IPCC provides a default CO₂ emission factor for aviation kerosene of 71,500 kg/TJ, with a lower and upper limit of 69,800 kg/TJ and 74,400 kg/TJ, respectively. The emission factors for CH₄, N₂O, and NO_x are reported as 0.5 kg/TJ (0.215 to 1 kg/TJ), 2 kg/TJ (0.6 to 5 kg/TJ), and 250 kg/TJ (187.5 to 312.5 kg/TJ), respectively [13]. According to the IPCC Sixth Assessment Report, the global warming potential (GWP) values for CO₂, CH₄, and N₂O over a 100-year period are 1, 29.8, and 273, respectively [40]. Given that the calorific value of aviation kerosene is 0.0358 PJ/ML [41], the quarterly energy consumption from aviation kerosene is calculated for the period spanning the first quarter of 2011 through the fourth quarter of 2024. GHG emissions from aviation kerosene are determined by multiplying the GHG emission factors by the quarterly energy consumption, factoring in the GWP values for the various GHG.

4. Results

4.1. Association Between the Number of Aircraft Taking off and the Consumption of Aviation Fuel

Figure 1 illustrates the number of aircraft departures alongside the consumption of aviation kerosene measured in kiloliters from the first quarter of 2011 through the fourth quarter of 2024. The data indicates a rise in aircraft departures from 39,940 in the first quarter of 2011 to 54,525 by the fourth quarter of 2018. However, this upward trend stalled in the third and fourth quarters of 2019 due to social unrest in Hong Kong, which disrupted airport operations. In the first three quarters of 2020, numerous passenger flights were canceled as a result of the COVID-19 pandemic, leading to a sharp decline in aircraft departures to 31,629 in the first quarter of 2020, 15,389 in the second quarter of 2020, and a low of 15,247 in the third quarter. Throughout 2021 and 2022, the number of aircraft departures fluctuated significantly. Fortunately, following the announcement at the end of 2022 regarding the cessation of China’s zero COVID-19 policy, aircraft departures began to rise steadily, increasing from 21,011 in the fourth quarter of 2022 to 48,139 in the fourth quarter of 2024. Likewise, Figure 1 indicates that the consumption of aviation kerosene followed a very similar trend from the first quarter of 2011 to the fourth quarter of 2024.

Figure 2 illustrates the relationship between the number of aircraft departures and the consumption of aviation kerosene measured in kiloliters. The data indicates a strong, positive, and statistically significant correlation between these two variables (r = 0.987, p < 0.001). The slope of the trend line suggests that for every thousand aircraft that take off, the consumption of aviation kerosene is expected to rise by 31.77 million liters (MLs).

4.2. Modeling of Aviation Kerosene Consumption

Data on aviation kerosene consumption from the first quarter of 2011 to the fourth quarter of 2018 were input into an IBM SPSS 29.0 data file. The forecasting module selected was expert modeling utilizing ARIMA models, which determined that the appropriate model for the dataset was a seasonal ARIMA(1,1,0)(0,1,1)₄. The MAE value was 33,137, the MAPE value stood at 1.814, and the RMSE value was 49,107. The R-squared value was found to be 0.910, with a normalized BIC of 21.844. When this model was evaluated against alternative models, as presented in

Table 1. MAE, MAPE, RMSE, R², and BIC for different seasonal ARIMA models.

Model	MAE	MAPE	RMSE	R2	BIC
(0,0,0)(0,0,0)4	131,563	7.023	158,117	0.000	24.050
(0,1,0)(0,0,0)4	62,270	3.387	78,222	0.749	22.645
(0,1,0)(0,1,0)4	43,700	2.430	61,552	0.849	22.177
(1,1,0)(0,1,0)4	42,760	2.336	57.416	0.874	22.160
(1,1,1)(0,1,0)4	42,856	2.338	58,552	0.874	22.322
(1,1,0)(0,1,1)4	33,137	1.841	49,107	0.910	21.844
(1,1,1)(0,1,1)4	33,487	1.824	50,897	0.909	22.163
(1,1,1)(1,1,1)4	33,557	2.832	51,664	0.910	22.315

, the seasonal ARIMA(1,1,0)(0,1,1)₄ model demonstrated the best fit based on MAE, MAPE, RMSE, R², and BIC values. Furthermore, the residual autocorrelation function (ACF) and partial autocorrelation function (PACF) plots exhibited no significant spikes, further validating the suitability of the identified seasonal ARIMA model, as illustrated in Figure 3.

Figure 4 illustrates the actual consumption of aviation kerosene alongside the predicted consumption derived from the seasonal ARIMA (1,1,0)(0,1,1)₄ model. It also includes the upper and lower confidence levels of the predicted values at a 95% confidence interval. The results demonstrate that the model effectively captures the aviation kerosene consumption data from the first quarter of 2011 to the fourth quarter of 2018. This model has been employed to forecast aviation kerosene consumption from the first quarter of 2019 through the fourth quarter of 2030. The projection assumes a steady increase in aviation kerosene consumption, disregarding any potential external disruptions.

Figure 5 illustrates the anticipated consumption of aviation kerosene, derived from the established seasonal ARIMA model, covering the timeframe from the first quarter of 2019 to the fourth quarter of 2030. The projections indicate that by the end of the fourth quarter of 2025, the consumption of aviation kerosene is expected to attain 2650 ML. Furthermore, by the fourth quarter of 2030, the aviation kerosene consumption is projected to increase to 2980 ML. Figure 5 also depicts the actual consumption of aviation kerosene from the first quarter of 2019 to the fourth quarter of 2024, represented by the blue line. It is expected that in the latter half of 2025, the actual consumption of aviation kerosene, following a robust recovery trend, may align with the forecasted consumption, as indicated by the red line. Subsequently, it is likely to experience steady growth similar to that projected by the seasonal ARIMA model.

Figure 6 depicts the decline in aviation kerosene consumption from the first quarter of 2019 to the fourth quarter of 2024, a timeframe that encompasses the COVID-19 pandemic. The decline was calculated using the discrepancies between projected and actual consumption figures, as given in Figure 5. The decline was initially modest in the first quarter of 2019, at approximately 39.2 ML, peaked at 1584.9 ML in the second quarter of 2022, and subsequently decreased to 600.5 ML by the fourth quarter of 2024. Analyzing the annual consumption of aviation kerosene reveals a reduction of 490.2 ML in 2019, attributed to the adverse impacts of social unrest. The unrest involved individuals blockading the departures area in August, during the third quarter, and disrupting transport to and from the airport in September, which falls in the fourth quarter of 2019. The unrest eventually subsided following the enactment of the Hong Kong National Security Law in June 2020. However, this decline intensified significantly to 4779.6 ML in 2020 due to the COVID-19 pandemic, followed by 5436.7 ML in 2021 and 6191.9 ML in 2022. In the post-pandemic period, the reductions were recorded at 4060.7 ML in 2023 and 2721.5 ML in 2024, respectively.

4.3. Estimation of GHG Reduction Due to the COVID-19 Pandemic

The net calorific value for aviation kerosene is 0.0358 TJ/kL, while the default emission factors established by the IPCC for aviation kerosene are 71,500 kg/TJ for CO₂, 0.5 kg/TJ for CH₄, and 2 kg/TJ for N₂O. Consequently, 1000 ML of aviation kerosene yields a heating value of 35,798 TJ, resulting in an emission of 2559.5 kilotons (kT) CO₂, 17.9 tons of CH₄, and 71.6 tons of N₂O. Utilizing the GWP values of 1, 29.8, and 273 for CO₂, CH₄, and N₂O, respectively, the total emissions from 1000 ML of aviation kerosene amount to 2579.6 KT of CO₂-eq. Figure 7 illustrates the estimated GHG reduction from the first quarter of 2020 through the fourth quarter of 2024, which covers the impact of the COVID-19 pandemic. The GHG reduction was estimated at 1971 kT of CO₂-eq in the first quarter of 2020, peaked at 4088 kT of CO₂-eq in the second quarter of 2022, and subsequently decreased to 1549 kT of CO₂-eq by the fourth quarter of 2024. Annually, the GHG reductions were estimated to be 12,330 kT of CO₂-eq in 2020, followed by 14,025 kT in 2021, 15,973 kT in 2022, 10,475 kT in 2023, and 7020 kT in 2024.

4.4. The Projected GHG Emissions Based on the Seasonal ARIMA Model from 2025Q1 to 2030Q4

Figure 8a,b depict the projected consumption of aviation kerosene, as modeled by the seasonal ARIMA approach, along with the corresponding GHG emissions for the timeframe spanning the first quarter of 2025 through the fourth quarter of 2030. The data indicates that aviation kerosene consumption could reach 2549 ML in the first quarter of 2026, escalating to 2980 ML by the fourth quarter of 2030. On an annual basis, it is forecasted that the consumption of aviation kerosene will progressively rise from 10,505 ML in 2026 to 11,591 ML in 2030. Regarding GHG emissions, the consumption of aviation kerosene is expected to result in the release of 27,177 kT of CO₂-eq in 2026, increasing to 29,899 kT of CO₂-eq by 2030.

4.5. Actual and Projected GHG Emissions with Sensitivity Analysis Using Monte Carlo Simulations

The IPCC (2006) suggested that the 95 percent confidence intervals for CO₂ emission for aviation kerosene was between 69,800 and 74,400 kg/TJ [13]. Additionally, the 95 percent confidence intervals for CH₄ and N₂O emissions from aviation kerosene were reported as 0.215 to 1 kg/TJ for CH₄ and 0.6 to 5 kg/TJ for N₂O. The emissions of CO₂, CH₄, and N₂O are characterized by lognormal distributions [13]. To assess the confidence intervals of GHG emissions, Monte Carlo simulations were executed 1000 times, addressing the uncertainties related to CO₂, CH₄, and N₂O emissions resulting from aviation kerosene consumption in a probabilistic framework. The findings revealed that total GHG emissions in 2018 amounted to 21,786 kT of CO₂-eq, with a 95 percent confidence interval ranging from 21,203 to 22,685 kT. This figure decreased to 8482 kT of CO₂-eq by 2022, with a 95 percent confidence interval of [8268, 8828] kT. Furthermore, it is anticipated that total GHG emissions will reach 29,899 of CO₂-eq by 2030, with a 95 percent confidence interval of [29,156, 31,187] kT. Figure 9 illustrates both the actual and predicted GHG emissions, along with their respective 95 percent confidence intervals.

5. Discussion

The results of the ARIMA model presented in Table 1 indicate a notable improvement through differencing: changing from the first (0,0,0) to (0,1,0) results in a dramatic reduction in all error metrics (MAE, MAPE, RMSE), along with a significant increase in R-squared. This highlights the non-stationarity of the original time series and the necessity of differencing to make it stationary. Comparable trends are observed in the additional improvements, with more complex models down the list. However, the diminishing improvement in performance suggests that incorporating more AR and MA terms can enhance the model’s ability to capture the underlying patterns; yet, beyond a certain point, specifically up to the seventh model, there is no increase in predictive power, and it actually results in higher MAE, RMSE, and BIC values compared to the sixth model. Notably, Model (0,1,0)(0,1,0)₄, which incorporates a seasonal difference, shows significant improvement when compared to Model (0,1,0)(0,0,0)₄, clearly indicating the presence of a seasonal differencing component. Similarly, the integration of AR, differencing, seasonal differencing, and seasonal MA components seems to capture the data’s dynamics effectively; however, additional complexity provides limited benefits. Considering the BIC and overall performance, the sixth Model (1,1,0)(0,1,1)₄ appears to be the best choice. It provides a good balance between model fit and complexity.

5.1. Implications

Figure 6 illustrates a responsive relationship between reductions in aviation kerosene consumption and critical incidents, indicating that such consumption may serve as a timely proxy for broader economic activity. Consistent with prior research, which indicates the efficacy of mobility data in anticipating economic disruptions [42], this study corroborates those findings, utilizing a more economically grounded and reliable indicator: fuel consumption. The observed relationship is particularly salient given the potential for increased prioritization of national security and domestic production in light of the current geopolitical conditions, such as the overarching trend of the Great Decoupling. This shift may contribute to monitoring the fragmentation of the global aviation network. Consequently, aviation kerosene consumption data offers valuable insights for assessing the evolving structure of global trade and connectivity.

Despite the disruptions brought about by the COVID-19 pandemic, this study forecasts that the development of aviation activities and associated GHG emissions will revert to their pre-pandemic trajectories. Based on the forecast, the climate crisis seems to be inevitable; there should be a complex interplay of factors. Considering the demand signals, our current technological advancements in aircraft efficiency and alternative fuels only offer limited mitigation potential. As reported by the World Economic Forum, the production of sustainable aviation fuel is significantly lagging behind expectations [43]. Even if sustainable aviation fuel refineries manage to secure sufficient investment, they will only be able to produce around 17 million tons annually by 2030, which could satisfy approximately 4 to 5% of total aviation fuel consumption [43]. Furthermore, the price of sustainable aviation fuel is roughly two and a half times higher than that of conventional jet fuel [44]. A novel implication of the aforementioned points is the necessity for a shift in policy focus. While traditional carbon pricing mechanisms are crucial, they are inadequate to tackle the systemic challenges confronting the aviation industry. There is an urgent need for new strategies. However, this research can only highlight the issues with current practices; even with drastic measures and seemingly forceful policy measures, the number of flights, aviation fuel consumption, and the associated GHG emissions continue to rebound. A shift towards more efficient and intelligent aviation systems is essential, including the electrification of short-haul operations, such as the adoption of electric unmanned aerial vehicles for transporting goods and passengers between Hong Kong and its nearby cities in mainland China. Moreover, alternative transportation methods and policies that alter travel behaviors and consumption priorities are crucial to reducing the demand for air travel and the associated GHG emissions. Additionally, the central government of China has recognized Hong Kong as a key hub in its recent expansion of the Chinese high-speed rail network, which increases the number of mainland cities with direct rail links to and from Hong Kong, consequently reducing the need for flights from Hong Kong to mainland Chinese cities in the short to near term [45].

5.2. Limitations of This Study and Future Research

This study acknowledges several limitations. Firstly, although this study utilized quarterly data from official sources, spanning the first quarter of 2019 through the fourth quarter of 2018 for ARIMA modeling, data uncertainties and the limited baseline for ARIMA forecasts may lead to increased errors, especially in long-term predictions and volatile environments, thereby limiting the model’s reliability beyond 2030 and the precision of decision-making. Secondly, ARIMA modeling assumes parameter constancy over the estimation window. However, structural breaks—such as regime shifts caused by changes in fundamentals—can violate this assumption. This can cause parameter instability and diminishing forecast accuracy for the pro-2024 and pro-2030 periods. Other models that are less sensitive to structural changes may perform better, such as long short-term memory (LSTM), Prophet, XGBoost, and ARIMA/machine learning hybrid models [25,26,27,46,47]. However, considering pragmatic constraints and explainability, the ARIMA approach offers valid and useful approximations. Given that the business-as-usual approach remains a prominent strategy within the Hong Kong air transport industry, the ARIMA model is still one of the most preferred choices due to its practical value in serving the purpose of this study. Thirdly, the accessibility of data and the scope of this study prevented a thorough analysis of all potentially relevant factors, such as differences in fuel demand between long-haul and short-haul flights, different aircraft types, and load factors. This could lead to an over-homogenization of consumption patterns. Future research could consider estimates such as global fuel consumption averages for different flights and route distributions. Fourthly, the assumption of continuous, globally interconnected travel may not hold true in all scenarios. The discussion considers minimal effects from escalating Sustainable Aviation Fuel costs and the limited scalability of electric or hydrogen propulsion for long-haul flights, as well as other emerging air travel technologies. Furthermore, the impact of current mitigation strategies may materialize with a longer lag than anticipated, extending beyond the scope of our data.

Future research should explore the potential of using aviation kerosene consumption patterns as a leading indicator for assessing economic activity, especially at the local level, due to their greater availability and reliability. Given the limited effectiveness of current mitigation strategies, we recommend prioritizing research into the design of more immediate or short-term economic incentives to encourage broader adoption of sustainable practices and policies. Additionally, future research should investigate the likely trajectory of emissions following the adoption of sustainable aviation fuel and the implementation of regional carbon pricing in Hong Kong.

Finally, it is important to note that this study’s findings should not be limited to the Hong Kong air transport industry. The methodology we have adopted is likely applicable to other hub airports, especially those in similarly dynamic and densely populated areas, such as Tokyo, Singapore, Dubai, London, and Atlanta. Future research could explore the association between aviation kerosene consumption and flight data in these hubs, as well as forecast the trajectories of GHG emissions from major airport hubs.

6. Conclusions

This study pioneered the quantification of GHG emission reductions in Hong Kong’s air transport industry resulting from the COVID-19 pandemic. Furthermore, it developed a seasonal ARIMA model utilizing actual quarterly aviation kerosene consumption data from Hong Kong spanning 2011 to 2018. The model was subsequently employed to forecast quarterly aviation kerosene consumption from 2019 to 2030. It was found that the decrease in GHG emissions peaked at 15,973 kT of CO₂ in 2022 due to the COVID-19 pandemic. Additionally, the findings indicated that if a business-as-usual approach is maintained, the annual GHG emissions from the air transport industry in Hong Kong could escalate to 29,899 kT of CO₂-eq by 2030, representing an increase of over 37% from 21,786 KT of CO₂-eq in 2018. Therefore, it is essential for Hong Kong to shift towards a more efficient and intelligent aviation system, which includes the electrification of short-haul operations, such as the implementation of electric unmanned aerial vehicles for transporting goods and passengers between Hong Kong and its nearby cities in mainland China. Moreover, the adoption of Sustainable Aviation Fuel should be considered. Lastly, alternative transportation methods and policies that modify travel behaviors and consumption priorities are vital for diminishing the demand for air travel and the corresponding GHG emissions.