The travel-related carbon dioxide emissions of atmospheric researchers

Most atmospheric scientists agree that greenhouse gas emissions have already caused signiﬁcant changes to the global climate system and that these changes will accelerate in the near future. At the same time, atmospheric scientists who – like other scientists – rely on international collaboration and information exchange travel 5 a lot and, thereby, cause substantial emissions of carbon dioxide (CO 2 ). In this paper, the CO 2 emissions of the employees working at an atmospheric research institute (the Norwegian Institute for Air Research, NILU) caused by all types of business travel (conference visits, workshops, ﬁeld campaigns, instrument maintainance, etc.) were calculated for the years 2005–2007. It is estimated that more than 90% of the emis- 10 sions were caused by air travel, 3% by ground travel and 5% by hotel usage. The travel-related annual emissions were between 1.9 and 2.4 t CO 2 per employee or between 3.9 and 5.5 t CO 2 per scientist. For comparison, the total annual per capita CO 2 emissions are 4.5 t worldwide, 1.2 t for India, 3.8 t for China, 5.9 t for Sweden and 19.1 t for Norway. The travel-related CO 2 emissions of a NILU scientist, occurring in 24 days 15 of a year on average, exceed the global average annual per capita emission. Norway’s per-capita CO 2 emissions are among the highest in the world, mostly because of the emissions from the oil industry. If the emissions per NILU scientist derived in this paper are taken as representative for the average Norwegian researcher, travel by Norwegian scientists would nevertheless account for a substantial 0.2% of Norway’s 20 total CO 2 emissions. Since most of the travel-related emissions are due to air travel, water vapor emissions, ozone production and contrail formation further increase the relative importance of NILU’s travel in terms of radiative forcing.

a lot and, thereby, cause substantial emissions of carbon dioxide (CO 2 ). In this paper, the CO 2 emissions of the employees working at an atmospheric research institute (the Norwegian Institute for Air Research, NILU) caused by all types of business travel (conference visits, workshops, field campaigns, instrument maintainance, etc.) were calculated for the years [2005][2006][2007]. It is estimated that more than 90% of the emis-10 sions were caused by air travel, 3% by ground travel and 5% by hotel usage. The travel-related annual emissions were between 1.9 and 2.4 t CO 2 per employee or between 3.9 and 5.5 t CO 2 per scientist. For comparison, the total annual per capita CO 2 emissions are 4.5 t worldwide, 1.2 t for India, 3.8 t for China, 5.9 t for Sweden and 19.1 t for Norway. The travel-related CO 2 emissions of a NILU scientist, occurring in 24 days 15 of a year on average, exceed the global average annual per capita emission. Norway's per-capita CO 2 emissions are among the highest in the world, mostly because of the emissions from the oil industry. If the emissions per NILU scientist derived in this paper are taken as representative for the average Norwegian researcher, travel by Norwegian scientists would nevertheless account for a substantial 0.2% of Norway's

Introduction
The global atmospheric concentration of carbon dioxide (CO 2 ) has increased from a 25 pre-industrial value of about 280 ppm to 379 ppm in 2005 (IPCC, 2007). The primary 7374 source of this increased atmospheric concentration results from fossil fuel use, with land-use change providing another significant but smaller contribution. The enhanced atmospheric CO 2 concentrations produced a radiative forcing of 1.66 W m −2 in the year 2005 (IPCC, 2007), which is the largest contribution of all forcing agents to the total radiative forcing. The transport sector is one of the largest sources, contributing 5 about 19% of all CO 2 emissions according to the EDGAR 3.2 fast track inventory for the year 2000 . Emissions of CO 2 by aircraft were 0.51 Gt CO 2 /year in 1992, about 2% of total anthropogenic carbon dioxide emissions or about 13% of carbon dioxide emissions from all transportation sources (Penner et al., 1999). Aircraft emissions have been growing at a fast rate and in 2005 have reached about 0.65 Gt 10 CO 2 /year (Kim et al., 2007).
Most atmospheric scientists are concerned that greenhouse gas emissions have already caused significant changes to the global climate system and that these changes will accelerate in the near future. At the same time, atmospheric scientists -as all other scientists -rely on the international exchange of information and on international 15 collaboration. Often, this requires traveling to conferences, workshops or project meetings. Atmospheric researchers also perform field campaigns or maintain monitoring sites in sometimes remote locations. The CO 2 emissions from the resulting travel, if counted on a per-capita basis, are likely to be substantial. However, to my knowledge this has never been properly quantified and published in the open literature. 20 In this paper, I make a first case study by estimating the CO 2 emissions caused by all business trips undertaken by the employees of my institute, the Norwegian Institute for Air Research (NILU), during the years 2005-2007. NILU's staff has grown from 134 to 160 persons during this period, 60 to 79 of whom were scientists ( ing for city and other location names in the Wikipedia on-line encyclopedia (see http://en.wikipedia.org/wiki/Main Page). Most of the coordinates are not those of an airport but of the traveller's final destination. Figure 1 shows all destinations and great circle routes from the home airport, and Table 2 lists the number of trips undertaken every year (381-557). It is seen that while European destinations generally dominate, 15 a substantial number of trips were made to destinations outside Europe. Travel within Norway, except for travel between the offices in Kjeller and Tromsø (also an almost 2-hour-long flight), is relatively unimportant. Trips in the institute's immediate vicinities were ignored completely while for the remaining trips within Norway it was assumed that air travel was used. Errors resulting from these assumptions would impact the 20 annual CO 2 budget by less than 1%. From the available information, the total flight distance covered by a trip is calculated to be where D GC is the great-circle distance between the Oslo (or Tromsø) airport and the 25 destination location. The factor 2 accounts for the return trip, F 1 =1. average deviation of actual flight lengths from great circle distance (about 10 percent according to Kim et al., 2007), and F 2 is an empirical correction factor. The factor F 2 is applied here because only the final destination of a trip is known but the actual routing may have involved more than one flight which increases the total distance traveled and also reduces the fuel use efficiency as shorter flights have a lower efficiency (see 5 below). We take F 2 to be 1.0 and 1.2 for D GC less than 800 km and more than 800 km, respectively, assuming that short distances are always covered by direct flights while long distances often require two or more flights. The CO 2 emission per flight kilometer is calculated according to where C is the emission factor for CO 2 , 3.15 kg CO 2 /kg jet fuel; E U is the specific energy usage per available seat kilometer (ASK), taken to be 1.2 MJ/ASK and 2.0 MJ/ASK for flights of more than 1000 km and less than 1000 km, respectively (Babikian et al., 2002). The higher fuel consumption for the shorter flights is due to the larger fraction of time spent taxiing at airports and in the fuel-intensive ascent phase; H is the lower 15 heating value of kerosine, 43.1 MJ/kg; L is the average passenger load factor, assumed to be 0.75 (Morrell, 2007). Combining equations 1 and 2, the CO 2 emissions of a trip can be calculated, which are then summed over all trips of a year. Overall, I estimate that annual CO 2 emissions calculated with the above procedure are accurate to within about 30%. The largest 20 uncertainties are due to E U varying by about 20-30% between different aircraft types and depending on flight length (Babikian et al., 2002), the empirical factor F 2 , and the passenger load factor, which can be quite different for different airlines. Morrell (2007) give L values between 0.51 and 0.91 for different airlines (the higher values are for a charter company and are not representative for the business travel considered here).

25
The various parameters in equations 1 and 2 were chosen such that the CO 2 emission estimates are thought to be conservative -thus, underestimation is more likely than overestimation. For every trip undertaken by a NILU employee, I have added 160 km of ground travel, assuming that about 80 km are required for the return trip between employees' homes and the airport (for instance, Gardermoen airport is about 50 km outside Oslo), 60 km are required for the return trip between the destination airport and the final destination (assuming this to be a typical distance of an airport from a city center), and 20 km are 5 covered at the destination site. These numbers are thought to be very conservative. I have assumed that 0.03 kg/km gasoline are consumed for ground transportation on average, again a conservative estimate given that taxis or private cars are often used. In total, ground transportation contributed less than 3% to total emissions, such that a more refined procedure was not deemed necessary.

10
CO 2 emissions occur also during the stay in a hotel, for instance due to heating or hot water preparation. For Switzerland, the average CO 2 emission per visitor night in a hotel is estimated at 11 kg CO 2 (Schegg and Amstutz, 2004). For New Zealand, the corresponding estimate is 8 kg CO 2 per visitor night (Becken and Patterson, 2006), whereas the German Environmental Protection Agency uses a value of 18 kg CO 2 15 (Schächtele and Hertle, 2007). Here, I use an average of these values, 12 kg CO 2 per visitor night. A detailed analysis of the travel information for the year 2007 revealed that a NILU employee spent on average 3.4 nights away from home per trip. This number was also used for the other years to calculate the corresponding CO 2 emission from the number of trips undertaken.  Table 2 summarizes the distances traveled and corresponding CO 2 emissions caused by NILU's staff. In total, NILU employees traveled between 2.2 and 3 million kilometers per year. This corresponds to 15 000-19 000 km per employee. Since most trips, especially the longer ones, were undertaken by scientists, it makes sense to count 25 the emissions per scientist and use this as the main yardstick for later comparisons. The average NILU scientist traveled 30 000-43 000 kilometers, roughly once around the Earth, every year. The resulting CO 2 emissions ranged from 269-368 t/year, which corresponds to 1.8-2.3 t CO 2 per employee or 3.7-5.2 t CO 2 per scientist. More than 97% of these emissions were caused by air transportation, less than 3% by ground transportation. Hotel usage (Table 3) adds another 5% and, thus, is a relatively minor contributor to the CO 2 emissions.

5
The total travel-related CO 2 emissions (i.e., sum of transportation emissions and emissions from hotel usage) for 2005-2007 are listed in Table 4 and the per-capita emissions for the year 2007 are compared with total national per-capita emissions in various countries in Table 5. The per-capita emissions (2.4 and 5.0 t CO 2 per employee and per scientist, respectively) appear relatively small compared to the Norwegian average per-capita emission (19.1 t CO 2 ). However, Norway's per-capita CO 2 emissions are among the highest in the world, which is due mainly to oil production which accounts for almost 60% of the national emissions. Such a comparison is, thus, misleading. NILU's travel-related per-scientists emissions are much higher than the Norwegian emissions from road transport. NILU's per-scientist emission are higher than the total 15 per-capita emissions in many Asian countries, including China, and they also exceed the global average per-capita CO 2 emissions. They are furthermore comparable to the total per-capita emissions in some highly industrialized nations, for instance in France, Sweden or Switzerland. Given that these emissions occur in just 24 days of the year (the average time a NILU scientist spent abroad), this result is quite remarkable. 20 In this study I have made emission estimates only for CO 2 and more than 90% of these emissions resulted from aircraft use. However, aircraft impact climate not only through their CO 2 emissions. They emit nitrogen oxides which produce ozone, a particularly effective greenhouse gas at cruising altitudes (Fabian and Kärcher, 1997). Aircraft also emit water vapor, which at these high altitudes also adds a small posi-25 tive radiative forcing. Furthermore, aircraft produce contrails and possibly also lead to enhanced cirrus cloud cover, both causing a positive radiative forcing (Marquart et al., 2003). The radiative forcing due to these effects is more uncertain than the radiative forcing from CO 2 but in total may effectively more than double the CO 2 radiative  (Penner et al., 1999). Consequently, the per-capita climate impact of NILU's business trips is relatively larger than the per-capita CO 2 emissions alone would suggest. In other words, if radiative forcing values instead of CO 2 emissions would have been compared, the relative importance of NILU's travel would increase. An important question remaining from this study is how representative the travel be-5 havior of NILU's scientists is for the whole (atmospheric) science community. There are several factors which may produce larger travel activity of a NILU scientist compared to other scientists. Most importantly, Norway is on the northern end of Europe such that almost all trips by NILU scientists involve relatively long distances. Furthermore, a substantial fraction of NILU's revenue comes from projects sponsored by the European 10 Union, the United Nations and other international sources. NILU also maintains small branch offices in various countries (e.g., United Arab Emirates, Poland) and operates monitoring stations both in the Arctic and in Antarctica which need to be visited regularly. Finally, NILU employees also travel occasionally between the Kjeller and Tromsø offices. However, there are also factors that discourage NILU scientists from traveling, 15 for instance the fact that few intercontinental flights are departing from Oslo, such that nearly always multiple flights are needed to reach a destination on another continents. This is even worse for scientists working in Tromsø who must travel to Oslo first, before embarking on an international flight. The long duration of such trips is a strong incentive for reducing the number of trips.

20
A larger number of studies such as this one will be needed to fully evaluate the climate impact of scientists. Atmospheric researchers, most of whom are concerned about the climate impact of anthropogenic activities, may actually travel less than other scientists, making a more complete assessment all the more important. Still, it is clear from this study that travel-related CO 2 emissions by scientists are substantial and need 25 to be taken into account. Awaiting a more complete assessment and assuming in the meantime that the per-capita emissions of the NILU scientists are typical at least for researchers in Norway, the total CO 2 emissions caused by the travel of all Norwegian scientists can be calculated. According to RCN (2008), 54000 persons were involved in research and development activities in Norway in the year 2005. As this number includes part-time personnel, the number of person years spent in research and development, 30500, may be a more appropriate scaling factor. Assuming that Norwegian researchers emit 5 t CO 2 per person year on travel, the total number for Norway would be 152.5 kt CO 2 , which is almost 0.2% of Norway's CO 2 emissions.

4 Conclusions
In this paper, the CO 2 emissions of the employees working at an atmospheric research institute (NILU) caused by all types of business travel (conference visits, workshops, field campaigns, instrument maintainance, etc.) were calculated for the years 2005-2007. More than 90% of the emissions were caused by air travel, 3% by ground travel 10 and about 5% by hotel usage. The travel-related annual emissions were between 1.9 and 2.4 t CO 2 per employee or between 3.9 and 5.5 t CO 2 per scientist. For comparison, the total annual per capita CO 2 emissions are 4.5 t worldwide, 1.2 t for India, 3.8 t for China, 5.9 t for Sweden and 19.1 t for Norway. Norway's per capita CO 2 emissions are among the highest in the world, due to large emissions from the oil industry, and 15 are probably not suitable for comparison. However, compared to the total per capita emissions of other nations, NILU's CO 2 emissions from business travel are quite high, given that the emissions occur during only 24 days of the year. On a per-scientist basis, they even exceed the global average per capita emission. Assuming that the per capita emissions derived in this study are typical at least for Norwegian researchers, travel by 20 the Norwegian research and development personnel would account for almost 0.2% of the national total CO 2 emissions. The importance of the travel-related emissions increases further when considering that most of these emissions are due to air travel, whose total radiative forcing effect (including high-altitude ozone formation, water vapor emissions and contrail formation) may be more than twice the radiative forcing due to the CO 2 emissions alone.
Given the substantial climate impact of transportation, scientists should re-think their Interactive Discussion current travel behaviour. Also funding agencies should re-evaluate their funding criteria. Currently, international collaboration is highly encouraged both by national as well as international funding bodies. Consequently, research proposals involving a lot of travel tend to be more successful than proposals with less travel, even though the scientific benefit from the enhanced travel may not always be clear. While continued