CC BY 4.0 · generated 2026-10-09

Take the data

The complete 0xCarbon dataset is free to download and reuse under CC BY 4.0: 2,723 destinations across 46 countries, each with monthly temperature, sunshine and rainfall, return CO₂e and journey time by rail and air from 32 departure cities across 16 countries, and 4608 precomputed frontiers.

This dataset did not exist before. If it is useful to you — a newsroom, a paper, a tool, an argument with a colleague — use it. Attribution is the only condition.

2,723destinations across 46 countries, each with twelve months of temperature, sunshine and rainfall — drawn from a pool of 2,723 candidates

4,608precomputed frontiers: 3 questions × 32 cities × 12 months × 4 time budgets

85,770origin–destination journeys costed by rail and by air

What is not in it

Every one of the 2,723 settlements that clear the size and distance filters carries twelve months of climate data, so none is excluded for want of it. Nothing in this dataset is interpolated, imputed or filled from a neighbouring town: a destination either carries its own twelve months from the source or is absent. The climate source is ERA5 reanalysis — observations blended with a weather model on a grid — not a thermometer in the town square; the methodology says what that means.

Download

File	What's in it	Format
frontiers.json	Every frontier, self-describing: licence, model parameters and emission factors travel with the data	JSON

How to cite

0xCarbon (2026). European travel carbon and climate dataset. Derived from UK Government GHG Conversion Factors 2026 (OGL v3.0), ERA5 via Open-Meteo (CC BY 4.0) and GeoNames (CC BY 4.0). Available at https://0xcarbon.com/data/

Shape of the data

Frontiers are keyed question:origin:month:hours — so warmest:london:1:15 is the warmest January options from London within a fifteen-hour one-way journey, and sunniest:london:11:10the sunniest November options within ten hours. Each entry carries the destination's name, country, return kg CO₂e, mean temp in °C, the mode (rail or fly), modelled hours, the quality it was ranked on as q, its temp, sunand rain for that month, and a URL for the destination page.

Read the methodology before using these numbers in anything that matters — particularly the section on what the method gets wrong. Rail journey times are modelled, not timetabled.