Aviation and the Global Atmosphere

	Other reports in this collection

9.3.4. Comparisons of Present-Day and 2015 Forecast Emissions Inventories (NASA, ANCAT/EC2, and DLR)

Table 9-4 lists the totals for calculated fuel burned and emissions from the NASA, ANCAT, and DLR inventories for 1976, 1984, 1992, and 2015. Because these inventories consisted of 3-D data sets, the differences in spatial distributions as well as totals are compared. The NASA inventories also included emissions of CO and HC, which are summarized in Table 9-5.

The NASA inventories include piston-powered aircraft in the general aviation fleet. This category of aircraft is excluded from the ANCAT and DLR inventories, but the contribution to total fuel burned from these aircraft is small (2.6% of fuel burned in 1992). Piston-powered aircraft are large contributors to CO and HC emissions relative to the amount of fuel they burn (39% of CO and 13% of HC emissions in 1992). This large relative contribution is reflected in the emissions indices of these two pollutants in the general aviation category.

A comparison of calculated global total values for fuel burned and NOx emissions from the NASA, ANCAT, and DLR inventories for 1992 and 2015 is shown in Figure 9-9. All three inventories for 1992 have approximately the same calculated values for total fuel burned in the civil air fleet; the difference in total fuel (7% maximum) arises almost entirely from different calculated contributions for military aviation operations, for which the ANCAT inventory calculates 33% lower fuel burned. Because military fuel is estimated to be between 13 and 18% of total fuel in 1992, the effect of this large difference in estimates between military sectors on the total is small. Use of the NASA inventories as a base is arbitrary and does not imply that differences from the NASA results are errors. Exclusion of turboprop operations from the ANCAT inventory results in about a 2% underestimate (if data from the NASA inventory are used).Calculated values for total NOx emissions from the three inventories for 1992 are within 9% of each other. The ANCAT and DLR values are higher than those from NASA-a result of a combination of differing fleet mixes, a different method of calculating NOx emissions, and the offsetting effects of civil and military calculations. This variation is also reflected in the calculated EI(NOx) for the fleet components: The ANCAT and DLR inventories have a total fleet emission index that is 15% higher than that of the NASA inventory.

Differences between inventory totals widen for the 2015 case, although total fuel burned is still within 8%. Total NOx emissions in the NASA 2015 forecast are almost 15% greater than those in the ANCAT forecast, a result of different assumptions about the direction of NOx reduction technology (the NASA assumptions result in an increase in NOx emissions index in the civil sector, whereas the ANCAT forecasts assume a reduction). Other differences between the NASA, ANCAT, and DLR inventories relate to the distribution of calculated fuel burned and emissions, geographically (latitude and longitude) and with altitude. Although all three inventories place more than 90% of global fuel burned and emissions in the Northern Hemisphere, there are differences between inventories in the details of the distribution. Figure 9-10 shows the distribution of fuel burned as calculated in 1 month (May) of the 1992 NASA inventory. The most heavily trafficked areas are clearly visible (United States, Europe, North Atlantic, North Asia).

For geographical comparison purposes, data in the files of the NASA and ANCAT 1992 inventories were divided into 36 regions, defined by 60� spans of longitude and 20� spans of latitude. Figure 9-11 shows the differences between the ANCAT and NASA 1992 inventories with regard to geographical distribution. The major differences between the NASA and ANCAT inventories (on a geographical basis) lie in the estimate of fuel burned and NOx emissions in the regions covering North America and Europe. The ANCAT inventory places 32% of total fuel burned and 30% of total NOx over North America, whereas the NASA inventory places 27% of fuel burned and 27% of NOx over that region. ANCAT places 16% of the fuel and 15% of the NOx over Europe, whereas the NASA inventory places 21% of the fuel and 19% of total NOx over that region. Part of this difference may be explained by the 10% scaling of U.S. traffic assumed by ANCAT as a method of approximating the U.S. charter market.

NASA and ANCAT fuel and NOx emissions projections for 2015 are similar to the respective 1992 inventories in that no new city pairs were used in the 2015 traffic projections. Growth rates from 1992 to 2015 vary with region, so the geographical distribution of emissions changes over time. The altitudinal distributions of fuel burned in the present-day NASA, ANCAT, and DLR inventories are shown for civil aviation in Figure 9-12 and for military aviation in Figure 9-13. The civil aviation distributions are similar, with the NASA inventory showing more fuel burned at higher altitudes. The military distributions are quite different, with fuel burned in the NASA inventory concentrated at the higher altitudes and fuel burned in the ANCAT inventory at lower altitudes. This difference may be because of a higher proportion of transport operations in the NASA inventory. The altitudinal distribution of NOx emissions follows closely that of fuel burned. The three inventories show that more than 60% of the fuel burned and NOx emissions occur above 8 km, whereas a major fraction of CO and HC are emitted near the ground.

Although the three inventories show comparably low variations for total global monthly figures over the year, the seasonal dependency can be quite large for some regions (Figure 9-14). Operations in the North Atlantic and North Pacific show a clear yearly cycle, with a maximum in the northern summer and a minimum during winter. In contrast, Southern Hemisphere operations show little seasonal variation overall, with small peaks in February and November. DLR has also examined longer trends in fuel burned and emissions for air traffic (Schmitt and Brunner, 1997). 3-D gridded inventories of fuel burned and emissions were calculated for 1982 through 1992 using ICAO statistics on annual values for international scheduled air traffic and ABC time table data of all scheduled air traffic for the same week of September in 1986, 1989, and 1992. Emissions inventories were produced for each of these data sets using the same methods as in the 1992 DLR inventory described above. These inventories concentrate on scheduled services because reasonably accurate calculations are possible for this segment of aviation. Because these data do not include nonscheduled flights, military traffic, general aviation, or former Soviet Union/China traffic, they are of limited use in global modeling studies. However, they do provide a consistent set of data to track the growth of the international and domestic scheduled sector. Table 9-6 gives the totals for the yearly inventories.

9.3.5. Error Analysis and Assessment of Inventories

Simplifying assumptions used in creating all of the 3-D emissions inventories have introduced systematic errors in the calculations. An analysis of the effects of the simplifying assumptions on fuel burned used in the 1992 NASA inventory has been performed by Baughcum et al. (1996b). All of the assumptions have the effect of biasing the calculation toward an underestimate of fuel burned and emissions produced, as detailed in Table 9-7. The effects of the assumptions on the ANCAT and DLR inventories may be expected to be similar, because most of the simplifying assumptions used in those inventory calculations were similar to those in the NASA inventory.

The assumption of great-circle flight paths results in an underestimate of distance flown, although the practice of routing to take advantage of winds may result in lower fuel consumption than a great-circle path for a given flight. A study of international and domestic flights from German airports showed an average increase in flight distance of 10% for medium- and long-haul flights above 700 km, with larger deviations from great-circle routes for shorter flights (Schmitt and Brunner, 1997). Ground delays and in-flight holding at relatively low altitudes caused by congestion in the air traffic control system also adds to fuel consumption. Aircraft in service are subject to factors that may increase fuel consumption by up to 3% (e.g., engine deterioration, added weight from added systems, and increased surface roughness). Factors that cause underestimates of fuel burned do not necessarily operate at the same time, so they are not additive. Sutkus et al. (1999) compared fuel burned for certain carriers and certain specific aircraft types reported to DOT by U.S. air carriers, with the value for fuel burned calculated for these carriers and aircraft types in the 1992 NASA inventory. The comparison shows that a combination of factors outlined above results in systematic underestimation of total fleet fuel burned by 15-20% for domestic operations. The assumptions in the foregoing analysis apply to the civil aviation fleet. An error analysis of the calculation of fuel burned and emissions from military operations is not possible, given the nature of the estimates used in the calculations.

The present-day inventories described above have reported global fuel consumption values for 1992 ranging from 129 to 139 Tg. However, reported aviation fuel production was somewhat larger, at 177 Tg (OECD, 1998a,b). Calculated fuel consumption therefore accounts for 73-80% of total fuel reported produced in 1992. Simplifying assumptions used in calculating the inventories probably account for most of the difference. Reported fuel production values are not an ideal reference, however, because they do not necessarily represent fuel delivered to airports for use in aircraft. Jet fuel, in particular, is a fungible product; it can be reclassified and sold as kerosene or mixed with fuel oils or diesel fuel, depending on market requirements (e.g., when low freezing point fuel oil is needed in winter). Other distillate fuels from refineries may satisfy jet fuel requirements and could be purchased and used as jet fuel. As a consequence, reported jet fuel production data do not provide a rigorous upper or lower limit to jet fuel use. Fuel production data represent a compilation of reports of varying accuracy from many (not all) countries, whose overall accuracy has not been evaluated (Baughcum et al., 1996b; Friedl, 1997).

OECD data on aviation fuel production from 1971 (the first year the data includes the former Soviet Union) to 1996 are shown in Figure 9-15. These data shown are the sum of OECD and non-OECD country production data. Reported data include production of aviation gasoline, naphtha-type jet fuel (mostly JP-4, used for military aircraft), and kerosene-type jet fuel (Jet A, the most common transport aircraft jet fuel). Also shown are calculated values of aviation fuel burned from the NASA, ANCAT, and DLR present-day inventories. (NASA values have been increased by 15% as a rough estimate of systemic underestimate of civil fuel burned.)

Aviation gasoline has declined as a percentage of total aviation fuel-from 4% of production in 1971, to just over 1% in 1995. Production of naphtha-type jet fuel reached just over 10% of total fuel in 1983, but has since declined to less than 1% as military aviation has phased out its use in favor of kerosene-type fuels. Prior to 1978, production of naphtha-type fuel was not reported as a separate item in the OECD database; it was included in the kerosene-type production data.

The three inventories are in good agreement; given the different approaches and data sources used, the inventory results (particularly for the present day) are remarkably consistent. Assumptions regarding the state of NOx reduction technology in 2015 cause the biggest difference in the results of the three forecasts. The 1992 and 2015 inventories of NASA, ANCAT, and DLR are all suitable for calculating the effects of aircraft emissions on the atmosphere, taking account of differences in the details of the inventories and systematic underestimates examined above. To correct for the systematic underestimation of fuel burned in the inventories when calculating the effects of aviation CO2 emissions, fuel burned values for 1992 should be increased by 15% and those for 2015 should be increased by 5%, based on the assumption that inefficiencies in the air traffic control system responsible for extra fuel burned will be much reduced by 2015. A summary of the results from these inventories is given in Table 9-8. The DLR "trend" inventories (1982-92) include only a portion of total aviation operations (scheduled international service for all years and total scheduled service in 1986, 1989, and 1992); as such, they are valuable for historical growth analysis and for comparisons with the NASA and ANCAT/EC2 scheduled traffic segments.