Economic Value of Terminal Aerodrome Forecasts at Incheon Airport, South Korea

: Meteorological information at an arrival airport is one of the primary variables used to determine refueling of discretionary fuel. This study evaluated the economic value of terminal aerodrome forecasts (TAF), which has not been previously quantitatively analyzed in Korea. The analysis data included 374,716 international ﬂights that arrived at Incheon airport during 2017–2019. A cost–loss model was used for the analysis, which is a methodology to evaluate forecast value by considering the cost and loss that users can expect, considering the decision-making result based on forecast utilization. The value was divided in terms of improving fuel efﬁciency and reducing CO 2 emissions. The results of the analysis indicate that the annual average TAF value for Incheon Airport was approximately 2.2 M–20.1 M USD under two hypothetical rules of refueling of discretionary fuel. This value is up to 26.2% higher than the total budget of 16.3 M USD set for the production of aviation meteorological forecasts by the Korea Meteorological Administration (KMA). Further, it is up to 10 times greater than the 2 M USD spent on aviation meteorological information fees collected by the KMA in 2018.


Introduction
Weather has a significant effect on aircraft operations management [1]. A flight manager strongly considers the weather condition of the destination airport when loading aviation fuel prior to the aircraft's departure [2]. To reduce fuel costs, accounting for more than 25% of the airline's operating costs [3], meteorological information, which strongly affects the fuel payload, is vital for optimizing airlines' operations and flight route and planning decisions. However, despite the importance of the information, some factors, such as users' perceptions of meteorological information as public goods and lack of standards for appropriate prices, make it difficult to increase or adjust the charge.
When information from public institutions is concentrated on a small number of users, paid use of public information is implemented to realize its benefit principle and to use it as a resource for rapid provision and quality improvement [4]. The Korea Meteorological Administration (KMA) has been providing terminal aerodrome forecasts (TAF) and collecting charges for international flights' domestic airport landings and foreign airlines passing through the Korean airspace since 2005 in accordance with the Weather Act of Korea and recommendations by the International Civil Aviation Organization (ICAO). The charges collected in 2017 from airlines were 1.2 M USD (1 USD = 1165.51 KRW, average exchange rate in 2019). However, at 7.6% of the TAF production cost (16 M USD), this amount was very low. In response, the KMA raised the 2018 TAF usage fee by 84.8% (4.5 USD) per landing and 118.1% (2 USD) per airspace passage [5]. Although the KMA increased the TAF price, the production cost match collection ratio was still 12.1% in 2018. The low payback rate might trigger future attempts by the KMA to realize appropriate information that was not needed because of inaccurate forecasts or to properly load fuel that was inevitably needed because of accurate forecasts. Figure 1 provides a schematic of the decision to load additional fuel using TAF and the landing or deviation occurrence at the arrival airport. Severe weather occurred with additional fuel on board, and the landing rate is p. The landing rate without additional fuel is allocated as q. Costs and losses associated with the combination of forecasts and weather events and the relative frequency of each case are defined in Table 1. In Table 1, C2 is equal to the insurance cost because no event occurs. Insurance costs refer to fuel costs that are unnecessarily burned because of additional fuel loaded to prepare for adverse weather. L is the loss from being unprepared for severe weather. If additional fuel was not loaded, the cost was assumed to be zero when a weather event did not occur. Fnn refers to the relative frequency between severe weather forecasts and actual occurrences. D is the deviation cost.

= + (1 − )
(1) The economic value (EV) is defined as the difference between the expected cost of not using TAF (ECNT) and the expected cost of using TAF (ECT). Each expected cost can be calculated using variables defined in Table 1 as follows: Costs and losses associated with the combination of forecasts and weather events and the relative frequency of each case are defined in Table 1. In Table 1, C 2 is equal to the insurance cost because no event occurs. Insurance costs refer to fuel costs that are unnecessarily burned because of additional fuel loaded to prepare for adverse weather. L is the loss from being unprepared for severe weather. If additional fuel was not loaded, the cost was assumed to be zero when a weather event did not occur. F nn refers to the relative frequency between severe weather forecasts and actual occurrences. D is the deviation cost. The economic value (EV) is defined as the difference between the expected cost of not using TAF (EC NT ) and the expected cost of using TAF (E CT ). Each expected cost can be calculated using variables defined in Table 1 as follows: Because aircraft fuel consumption is affected by flight distance (t) and aircraft size (a), the TAF value can be considered separately for each flight, t, and a. The value used by the airline is defined as follows:

Data Acquisition and Description
Actual aviation data from airlines could not be collected, which is difficult to value using a cost-loss model. Generally, private companies treat internal information related to costs-losses, deviations, and cancellations as management secrets; therefore, they tend to be reluctant to disclose such information to competitors or the public. Thus, this study collected the available data listed in Table 2 and attempted a generalized analysis. Arrival data for international flights were collected during 2017-2019. Of 374,716 flights, 254 deviated during the three-year period, 160 of which returned because of meteorological factors. Aircraft arrived at Incheon Airport from 229 departure airports. Data included arrival time, departure airport, number of passengers, aircraft type, arrival/deviation status, and reasons for deviation, but not departure time. To estimate departure time, we collected the flight time between the departure airport and Incheon Airport using web crawling. Departure time is calculated as flight time subtracted from arrival time. Fuel economy by aircraft type considers fuel-tank capacity (gallons) and maximum flight range (km) as provided by aircraft manufacturers. Table 3 presents flight distance, fuel-tank capacity, and fuel economy calculated using the two variables. Major model types were used because the level of categorized aircraft types varies by source. For example, the value of fuel economy of A220 aircraft in Table 3 is 0.92, which is an average fuel economy of A220-100 and A220-300. This method was applied to all aircraft data. Meteorological data were downloaded from COMIS4. The data collection period was 31 December 2016-31 December 2019, considering that dispatchers use TAF prior to aircraft departure. The total number of TAFs, including original and amended forecasts, is 5195. Annual average fuel prices per gallon for 2017, 2018, and 2019 were 1.5 USD, 1.90 USD, and 1.88 USD, respectively, with a maximum value of 2.18 USD and a minimum value of 1.25 USD [22]. Cost of fuel purchased by the airline was assumed to be the average monthly fuel price at the time of aircraft departure. Energy Information Administration (EIA) (2016) indicated that CO 2 emissions per gallon of aviation fuel are 21.1 lbs. [23]. The average monthly carbon emission price for 2017-2019 was downloaded from the KRX ETS market information platform [24]. Annual average prices per ton were 17.7 USD, 19.5 USD, and 25.3 USD for 2017, 2018, and 2019, respectively.

Methodology
This study defined some conditions because actual airline flight data could not be used. The captain and dispatcher consider the weather at the arrival airport before departure and decide on the payload and whether to load discretionary fuel as preparation for deviations and delays. Interviews with airline staff led us to set their decision-making time to six hours before aircraft departure. Contingency fuel is prescribed under the following condition: "it shall not be less than a 5-min of flight at 450 m (1500 ft) altitude above the destination airport in standard atmosphere" [25]. Additionally, in 2012, ICAO announced a plan to adopt a 5% contingency fuel rule by revising its aircraft fuel requirements; Korean airliners are operating flights with 5% contingency fuel after gaining approval from the Minister of Land, Infrastructure and Transport [26]. There are no current criteria for the minimum discretionary fuel payload. It is assumed to be identical to two contingency fuel rules (hereafter the "5-min rule" and "5% rule"). The amount of fuel consumed during a Sustainability 2021, 13, 6965 6 of 13 five-minute flight was calculated based on the distance traveled at the speed of 425 km/h, regardless of the type of aircraft. The speed of 425 km/h is the conversion of 230 knots indicated in air speed (KIAS) to ground speed, and 230 KIAS is the maximum airspeed at altitudes lower than 14,000 ft, as set by ICAO. The visibility criterion for determining the discretionary fuel payload was set at 400 m, the low-visibility alert standard for Incheon Airport. A 5% rule is established for the amount of 5% of trip fuel.
Decision-makers who disregard TAF information increase the volume of discretionary fuel payloads in consideration of safety because they do not know the weather at the arrival airport. Additionally, aircrafts become heavier proportional to the increase in discretionary fuel. An increase in aircraft weight erodes fuel efficiency and over-burning of fuel in proportion to the decline in fuel economy. Over-burning increases greenhouse gas emissions. The TAF value considered the lower fuel economy and higher greenhouse gas emissions. Consequently, the increase in combustion (b) from additional fuel (β) is calculated by considering each aircraft category (a), the weight (m) of a particular flight (i), and the distance (d) shown, as presented in Table 4. Aircraft in the second column in Table 4 are from Ryerson et al., who modeled fuel consumption for some range of aircraft types using Piano 5-an aircraft performance analysis software [26]. The first column in Table 4 provides simplified aircraft types for analysis. Results showed that if A220 and A320 flights operating at 2000 mil add 10 lbs. of fuel, the additional fuel burn can be calculated as 0.019 × 10 + 4.644 × 10 −5 × 2000 × 10 = 1.12 (lbs.). Aircrafts that were not analyzed by the precedent study were grouped by size, and aircrafts within the same category were assumed to have the same fuel economy [27]. Aircraft sizes refer to the ICAO classification of sizes. Table 5 presents the ICAO airfield reference code, classified by aircraft width [28].
Airline emissions are important to the discussion of environmental problems and operational cost savings because domestic aviation is subject to allocation of emission rights, and a quota of approximately 6 million Korean allowance units (KAUs) was allocated during the second planned period (2018-2020) in Korea. A KAU represents the amount of greenhouse gas emissions allowed to be allocated to individual greenhouse gas emitters within the scope of total greenhouse gas emissions allowed to achieve the national greenhouse gas reduction target. Moreover, ICAO determined the Carbon Offsetting and Reduction Scheme for International Aviation in 2016, which will be implemented in 2021. In this case, airlines exceeding emission standards must purchase and offset emissions rights in the carbon market. Measures such as buying eco-friendly aircraft and reducing loads might be considered for reducing emissions; however, these methods add to an airline's expenses [29]. In other words, greenhouse gas emissions that can be reduced by TAF utilization represent the hidden value of TAF information. Therefore, we calculated the TAF value based on the price of greenhouse gas emissions rights per unit considering the CO 2 emitted by over-burned fuel, which results from the additional fuel weight increase. The equation is as follows: where f amount is the amount of excess fuel (gallon.), 21.1 is CO 2 conversion factor (lbs.) from aviation fuel (gallon.), and P KAU is the monthly price of KAU.

Fuel Efficiency
The TAF value used by national airlines arriving at Incheon Airport was calculated as approximately 6.2 M USD (5-min rule) and 55 M USD (5% rule), respectively, for 2017, 2018, and 2019. The case of Incheon Airport alone confirms that the KMA has provided airlines a value exceeding 13% of their annual budgets for TAF production in the 5% rule case, which is considered to provide an average of 148 USD per flight and 0.8 USD per passenger (67,497,168 passengers over three years). Figure 2 shows the monthly changes in TAF values. The Pearson's correlation between the TAF value and the price of aviation fuel are higher than 0.96, in line with the results in Gruenigen et al.'s (2014) work [21]. The result refers to the linear relationship between the two variables and shows that the TAF value varies considerably depending on fuel prices. Figure 3 shows the average TAF value for the aircraft sizes presented in Table 5. In Figure 3, the TAF value of C-group aircrafts is 5.6 and 18.2 USD and that of F-group aircraft is 56 and 630.7 USD. This indicates that the use of TAF resulted in large aircrafts gaining 34 times more value than small aircrafts. F-group aircrafts are large, such as B747s and A380s, and the amount of discretionary fuel required to be loaded exceeds that of small aircrafts in the absence of TAF. Adding more discretionary fuel means decreased fuel efficiency, which contributed to improving the TAF value (Equation (3)). Figure 4 provides results classifying TAF values by flight time. In Figure 4, the TAF value of short-haul flights under three hours averaged 5.6 and 11.0 USD, and the value for long-haul flights exceeding 10 h was 49.3 and 674.9 USD. Changes in TAF values over flight time are associated with TAF values by aircraft size in Figure 3 because long-haul flights often use relatively large aircrafts, and the amount of discretionary fuel that must be loaded equals the increased weight of the aircraft.  Figure 3 shows the average TAF value for the aircraft sizes presented in Table 5. In Figure 3, the TAF value of C-group aircrafts is 5.6 and 18.2 USD and that of F-group aircraft is 56 and 630.7 USD. This indicates that the use of TAF resulted in large aircrafts gaining 34 times more value than small aircrafts. F-group aircrafts are large, such as B747s and A380s, and the amount of discretionary fuel required to be loaded exceeds that of small aircrafts in the absence of TAF. Adding more discretionary fuel means decreased fuel efficiency, which contributed to improving the TAF value (Equation (3)).    Figure 3 shows the average TAF value for the aircraft sizes presented in Table 5. In Figure 3, the TAF value of C-group aircrafts is 5.6 and 18.2 USD and that of F-group aircraft is 56 and 630.7 USD. This indicates that the use of TAF resulted in large aircrafts gaining 34 times more value than small aircrafts. F-group aircrafts are large, such as B747s and A380s, and the amount of discretionary fuel required to be loaded exceeds that of small aircrafts in the absence of TAF. Adding more discretionary fuel means decreased fuel efficiency, which contributed to improving the TAF value (Equation (3)).     Figure 3 shows the average TAF value for the aircraft sizes presented in Table 5. In Figure 3, the TAF value of C-group aircrafts is 5.6 and 18.2 USD and that of F-group aircraft is 56 and 630.7 USD. This indicates that the use of TAF resulted in large aircrafts gaining 34 times more value than small aircrafts. F-group aircrafts are large, such as B747s and A380s, and the amount of discretionary fuel required to be loaded exceeds that of small aircrafts in the absence of TAF. Adding more discretionary fuel means decreased fuel efficiency, which contributed to improving the TAF value (Equation (3)).   The TAF value depends linearly on the price of aviation fuel, but that is not all. The performance of the TAF itself affects the benefit. We assumed the "pessimistic" forecast to observe the change in the value using the result of the TAF skill. The pessimistic forecast announces a more low-visibility forecast, even at the risk of an increase in false alarms. The virtual pessimistic forecast could be made by increasing F 12 in Table 1. At this time, the maximum value of the pessimistic forecast is the sum of F 12 and F 22 , which means that low visibility is announced every day, even though it does not occur. Table 6 shows how much the TAF value changes relative to the original forecast. Excessive forecasting of low visibility by increasing pessimistic forecasts allows airlines to load more fuel that originally would have been unnecessary, thereby leading to a decline in the TAF value. Excessively optimistic forecasts could also be made as opposed to pessimistic forecasts. Optimistic forecasts help save costs by reducing airlines' additional fuel payloads. However, if low visibility actually occurs, a loss will also occur. The optimistic forecast can be developed by increasing F 21 ; however, the values of F 11 (0.000029) and F 21 (0.005076) were so small that significantly analyzing the overall TAF value changes were not possible. However, Gruenigen et al. (2014) explained that airlines do not use overly optimistic forecasts because of safety regulations, even if the use of optimistic forecasts improves TAF values [21].

CO 2 Reduction
The solid line in Figure 5 shows monthly changes in the value of fuel economy resulting from the use of TAF during 2017-2019. The value of the increased fuel efficiency is 0.7 M USD (5-min rule) and 6.3 M USD (5% rule) during the 2017-2019 period, as much as the reduced discretionary fuel from the utilization of the TAF. This amount represents maximum 2.1 M USD annually, 16 USD per flight and 9 cents per passenger. The value of reduced CO 2 emissions was calculated by multiplying the price of emission rights and the monthly amount of emissions. The additional CO 2 emissions when TAF is not utilized for three years totaled 33,518 tons (5-min rule) and 299,989 tons (5% rule), and the monthly change is indicated by the dashed line in Figure 6.     Additionally, the values in terms of reduced CO2 emissions and the price of emission allowances correlated highly (>0.98). Various benefits were strongly sensitive to "price". The TAF value can be estimated as being heavily influenced by two price factors-fuel or emission right prices-without dramatic changes in number of flights and CO2 emissions. All correlation coefficients between "value" and "variables" are significant at 99% confidence.  Figure 7 shows the results of correlation analysis between the main variables and TAF value. Fuel price shows a high correlation with the values of fuel economy (>0.96). Additionally, the values in terms of reduced CO 2 emissions and the price of emission allowances correlated highly (>0.98). Various benefits were strongly sensitive to "price". The TAF value can be estimated as being heavily influenced by two price factors-fuel or emission right prices-without dramatic changes in number of flights and CO 2 emissions. All correlation coefficients between "value" and "variables" are significant at 99% confidence.   Figure 7 shows the results of correlation analysis between the main variables and TAF value. Fuel price shows a high correlation with the values of fuel economy (>0.96). Additionally, the values in terms of reduced CO2 emissions and the price of emission allowances correlated highly (>0.98). Various benefits were strongly sensitive to "price". The TAF value can be estimated as being heavily influenced by two price factors-fuel or emission right prices-without dramatic changes in number of flights and CO2 emissions. All correlation coefficients between "value" and "variables" are significant at 99% confidence. In this study, the variables were generalized. In reality, each flight would have been equipped with discretionary fuel that took into consideration distance, aircraft type, weight, weather conditions, and airline policies. However, the lack of actual flight (planning) data resulted in a collective calculation of discretionary fuel loads based on the aircraft's fuel economy. The fuel volume required for a five-minute flight was used as basic data for valuations (Table 7).

Conclusions
This study quantitatively assessed the economic value that the captain or dispatcher arriving at Incheon Airport could expect by using TAF to predict weather at the airport before departure. The TAF value of fuel efficiency analyzed using a cost-loss model was shown to be 2 M USD and 18 M USD per year. In the result of the 5% rule, the value of flights arriving at Incheon Airport from using the TAF was assessed somewhat higher than that of the precedent study. Gruenigen et al. (2014) estimated the value of TAF utilization by national airlines at 15 M USD at Zurich Airport [21]. One reason might be that the annual average of arrival flights at Incheon Airport ( 125 K) is more than that for Zurich Airport's arrival flights ( 110 K). However, the difference in TAF values between the two airports could be compounded by various factors, including flight distance, aircraft type and fuel economy, low-visibility criteria, fuel prices, and so on. Therefore, in this study, simple comparisons between the results are not appropriate because of the absence of the airline's actual operational data, which might be considered the most important. The value of the CO 2 emission reduction incurred as a result of overloaded discretionary fuel were assessed at an annual average of 233.2 K USD and 2.1 M USD each. In total, the profits earned from using TAF by all international flights arriving at Incheon Airport between 2017 and 2019 are 2.2 M or 20.1 M USD per year (Table 8). The charge for using TAF for international aircraft arriving at airports across the country was 1.8 M USD in 2018. This is less than the value of the TAF evaluate based on the 5-min rule. In other words, even if assessed conservatively, the KMA is contributing significantly to the reduction of potential cost expenditure by airlines by providing TAFs. Bade et al. [30] surveyed key stakeholders about using weather information, including airline captains, dispatchers, air traffic managers, and airport maintenance managers. Respondents provided the potential benefits of using weather information to reduce costs through optimal flight route selection and effective handling of diversions/resolutions, reduced winter airport management operating costs (employees, snow removal, and others), and increased air traffic management efficiency through airport overspace optimization [30]. Regardless of the value associated with reducing discretionary fuel information derived from this study, the potential benefits that the respondents provided suggest that TAF can be used as critical data for improving the aviation industry's safety and profitability. The TAF value can be significantly greater than the annual average of 20.1 M USD, considering that the airspace and foreign national aircraft are excluded from this study and other domestic airports. The meteorological community should use TAF's economic value information to properly inform the aviation industry of the value of weather information to enhance its understanding of policies, such as fare hikes.
We evaluated TAF values only for Incheon Airport, and a limit exists of not utilizing actual airline flight and flight plan data. An overall value estimation is required by expanding to all domestic airports using airlines' actual flight plan data. Since the total economic value of TAF covers a considerable range of outcomes (from $2.2 M to $20.1 M), it is necessary to show the form of distribution about potential discretionary fuel-mounted by flight to produce more accurate results. It can be expected that the 5-min rule will apply for short-haul flights and the 5% rule for long-haul flights; however, unfortunately, the relevant data are held by the airlines and generally not disclosed, making analysis difficult. If airlines' data could be collected, it would be possible to estimate the value of each airline and airport separately for various weather types of meteorological information (e.g., wind) as well as visibility. Nevertheless, the results of this study are meaningful because we derived TAF values where no quantitative assessments were made in Korea; therefore, the results could be utilized as a source to improve awareness of aviation weather information.