Planes, Trains, and Automobiles

Although I knew that transportation was a major part of carbon emissions, fully exploring my household data really drove that point home.

However, the high-level data only unearthed more questions that needed exploring. Are plane emissions primarily driven by the fact that we travel further by plane? How far do we travel by plane vs. other forms of transport? How much better is train travel than plane travel, especially for short-haul trips? Can we replace one form of transport with another to reduce emissions? Is this feasible?

To find answers to these questions, understanding the carbon intensity factors for each form of transport is crucial.

The carbon intensity factor, or emission factor, tells you the carbon intensity of a particular mode of transport, measured in grams of carbon dioxide per kilometre (gCO₂/km). Calculating emission factors can be complex. For example, car emission factors depend upon the number of occupants. With two people in a car, fuel emissions may not change significantly, but emissions per person per kilometre will be halved. Additionally, emission factors calculated for burning petrol fail to account for the entire lifecycle of fuel extraction, refining, and transportation, which contributes to overall emissions.

Planes …

To analyse our flight emissions (using data spanning 20 years from 2003-2022), I used a third-party tool tries to account for various aspects of plane emissions¹.

These aspects include flight distance, of course, but also include estimates for the plane type, average occupancy on the route, and corrections for high altitude emissions' increased impact compared to lower altitude emissions.

Firstly, I plotted flight length (in kilometres) against carbon dioxide emissions (in kilograms):

This plot reveals adverse mix of short-haul and long-haul flights. Notably, even the shortest flight was estimated to emit 110 kgCO₂. Considering that the UK's average per capita carbon emissions were 5.2 tonnes in 2021, this means that the shortest flight accounted for over 2%, or a week's worth, of the annual emissions of a typical resident.

The longest flight emitted 1,810 kg, or 1.8 tonnes, of carbon dioxide, amounting to roughly one third of the average annual emissions. This London to Singapore flight, lasting only 13 hours, released carbon dioxide equivalent to a full four months' worth of the average UK resident's emissions.

These figures are truly staggering and underscore the substantial carbon dioxide emissions associated with air travel.

The question remains: is this solely due to the speed and distance covered by planes, or are planes inherently more carbon-intensive? The gradient of the best-fit line in the plot above represents the carbon intensity of planes (gCO₂/km)². In this case, the average carbon intensity across all our plane journeys was 175 gCO₂/km. We'll revisit this value later.

… trains …

Estimating train emissions in detail is more challenging. Unlike planes, trains can run on diesel or electricity, and the carbon emissions depend upon the local electricity generation mix. For example, a train running through France, with its low-carbon nuclear power generation, is greener than train running through a country relying on coal-burning power plants.

To estimate train emissions in Europe, I used excellent calculator provided by EcoPassenger³. This calculator considers factors such as train type, local electricity grid, and average passenger numbers for a given route.

Unfortunately this calculator doesn't cover journeys outside of Europe. For trains taken elsewhere, I relied on various sources, including specific publications by train companies and third-party analyses. In some cases, I made extrapolations based upon the similarity between diesel trains in different parts of the world.

Once again I plotted train journey length (in kilometres) against carbon dioxide emissions (in kilograms):

While a strong linear relationship between journey length and carbon dioxide emissions is not evident here, the general trend for longer journeys emitting more carbon dioxide remains. This plot's higher scatter reflects the greater complexity and variability involved in calculating train emissions.

This plot also highlights that the longest train journey is considerably shorter than the longest plane journey. Even if I ignore stopovers in different cities, my longest train journey was barely over 1,500 kilometres. It's also clear the common dioxide emissions from trains are significantly lower than emissions from planes. The greatest amount of carbon dioxide emitted from a single train journey was just under 30 kg, far less than even the shortest plane flight at 110 kg.

To make an easier comparison between plane and train emissions, I plotted both data sets on the same axes:

Now the disparity is staggeringly clear.

Train journeys are typically shorter than plane journeys, which is not unexpected. What is clear though, is the carbon dioxide emissions from train journeys are substantially lower than those from plane journeys. The average carbon intensity of all our train journeys, represented by the slope of the line, averages just 34 gCO₂/km - only one fifth of the average plane journey's value of 175 gCO₂/km.

… and automobiles

For car journeys I have reliable data for many of my trips because I've always recorded the amount of petrol I fill up with and the distance covered for each tank of fuel.

Unfortunately I don't have precise data on average occupancy for personal car trips i.e. the number of people in each car for each journey. Therefore I estimated occupancy values based upon historical data on how we typically used the car for that year e.g. there were years when I did a lot of single-occupancy commuting. On average, our personal car occupancy was about 1.5, which is close to the English average of 1.6⁴.

To convert petrol consumption into carbon dioxide emissions, I used the whole lifecycle emission factor provided by Carbon Independent⁵. This value, just under 3.2kgCO₂/litre of petrol, includes not only burning of the fuel, but also the energy expended in extraction, refining, and transport of that fuel.

Considering our average occupancy, our personal car travel has a carbon intensity of 117 gCO₂/passenger-kilometre, or a raw carbon intensity of 176 gCO₂/kilometre driven (117 gCO₂/passenger-kilometre x 1.5 passengers/car = 176).

This value of 176 g CO₂/kilometre driven is notably higher than the official values for each of my cars. Over this period, I've driven two cars with official values of 106 g/km and 119 g/km, respectively.

Although these official values don't consider the full lifecycle of fuel extraction, refining, and transportation, there is still a discrepancy with my observed figures.

In total, we drove 170,000 km in the first car (at 106 gCO₂/km) and 35,300 km in the second car (at 119 gCO₂/km), up to the end of 2022. Combining these should give me a theoretical distance-weighted average of 108 gCO₂/km. However, the real value of 176 gCO₂/km is 1.63x the theoretical value.

This 1.63x discrepancy cannot be fully explained by the full lifecycle carbon emission value. The whole lifecycle estimates from Carbon Independent start with a base emission of 2.4 kgCO₂/L and add 1.32 kgCO₂/L to account for fuel processing etc. This gives a final value of 3.17 kgCO₂/L. Carbon Independent's whole lifecycle value is therefore 1.32x higher than the base fuel emissions, not as much as my measured 1.63x discrepancy.

This discrepancy means that the official emissions estimate for my cars is too low. This will translate into higher fuel consumption than official estimates and is something that many people are familiar with – you never seem to get as good economy as the manufacturer claims you will. In my case, my cars are burning 20% more fuel than the manufacturers claim they do.

As well as personal car use, we've also hired cars on holidays. In these cases the average occupancy is always 2.0 because we travelled together. However, accurate petrol consumption data is not always available. Therefore, I've estimated consumption based upon places we drove to and the make and model of car we hired, alongside published economy figures for those cars. It's likely that these are underestimates, as we discovered in our personal car, but there's nothing I can do without more accurate data.

When calculating emissions for hired cars, the carbon intensity was 125 gCO₂/passenger-kilometre, or a raw carbon intensity of 250 gCO₂/kilometre driven (125 gCO₂/passenger-kilometre x 2.0 passengers/car = 250).

It's not too surprising that these values are higher than those for our personal car journeys as many of our hire car usage happened in the USA, where larger and les fuel-efficient cars are commonplace. Notably, I never hired a car with an engine smaller than 2.7 L in the USA, while in the UK I have never had one with an engine larger than 1.2 L.

What does it all look like together?

I've summarised all these emission factors into a single plot that shows carbon emissions at our average occupancy rate:

I've also included coach travel. This is a value I estimated for our journeys, but wasn't a large contributor to overall emissions because we've travelled a relatively short distance by coach (<3,000 km).

When showing just the raw carbon emissions, assuming a single person traveling in each car, the car emission rates significantly increase, equalling or even exceeding those of planes:

The final summary of cumulative distance travelled and carbon dioxide emissions spans 20 years, from 2003 to 2022, and looks like this:

Although cars are slower than planes, we spend much more time in cars meaning that the overall distance is higher. It's also clear that trains and coaches are almost rounding errors in our carbon dioxide emissions. An increase in travel by train and a decrease in travel by plane will be a key area to address as part of our personal net zero journey.

Bonus graph

If you've made it this far, here's a nice bonus graph for you! This graph came about because of my exploring when and where we've taken flights. It shows the cumulative time that's elapsed since either of us last got onto a plane, plotted against time.

The graph climbs, at one day per day, until we get on a plane. At that point it drops back to zero, giving a characteristic saw-tooth pattern.

Our record was 1,354 dayS (just over 3.5 years) between flights. If we are to break that record again it will have to be 2027 before either of us uses a plane again. This chart is live and updates daily, allowing you to track our progress.