It is not unclear how their method, they explain it quite clearly and in detail [0]. In general, you are correct that road and (especially) rail require substantially more infrastructure than air travel. For combustion vehicles, however, such infrastructure is less than 10% of the total impact of such travel [1]. As energy source get cleaner, the relative impact of road construction and maintenance will increase. However, we run into difficult questions of allocation of burden, as essentially all damage to roads is done by heavy transport (damage goes by something like axle load to the fourth power), and by weather (irrespective of how much the road is used).

It is rather difficult to determine the actual impact of any one person's decision to fly or not fly. The state of the art considers the aircraft, the average load factor of the operator (or even load factor over a specific route), and the demand for urgent freight on that route (i.e. an empty seat might be compensated by more freight). Moreover, the marginal impact of one person is not the same as the marginal impact of e.g. a new travel policy for a large institution. Do to low data availability, we often take the average impact instead of the marginal.

For those actually interested in helping making open source and open data tools to help others make informed decisions, please check out BONSAI [2,3] (I am affiliated with the BONSAI NGO).

[0] http://ecopassenger.hafas.de/bin/help.exe/en?L=vs_uic&tpl=me... [1] https://pubs.acs.org/doi/abs/10.1021/acs.est.8b00261 [2] https://github.com/BONSAMURAIS/bonsai [3] https://bonsai.uno/