Built this to explore THETIS-MRV, the EU's public ship emissions database. Around 12,000 vessels per year, 2018-2024. Each dot is a vessel positioned at its flag state.
The interesting pattern: the biggest clusters aren't where the shipping companies are. They're in Panama, Liberia, and the Marshall Islands, the world's largest open registries.
2024 is the first year ships had to pay for their carbon emissions under EU law. You can switch between CO₂ total, EU ETS cost, and ship type views.
> The interesting pattern: the biggest clusters aren't where the shipping companies are. They're in Panama, Liberia, and the Marshall Islands, the world's largest open registries.
Genuine question, how is this interesting? Surely it just renders the data useless for assessing anything other than "what are the most popular open ship registries"?
Fair point. The dataset only has annual aggregates per vessel, no voyage routes or port calls, so flag state is the only geographic dimension available. The more useful part is the emissions data: CO₂ per vessel, ETS costs per company, and how the fleet changed from 2018 to 2024. The globe is the interface for exploring that.
There is nothing in the dataset that would require the use of a globe to visualize anything. You could have drawn this as bar charts and it would give us the same information (with the added advantage of not being limited to a few countries at a time). Or even a 2D earth map.
It just turns on my CPU fans and gives me no insights.
The navigation is broken on this for me. If I zoom in, the mouse panning is still at the zoomed out level. I mean I have to move the mouse increasingly accurately and in small increments as I zoom in, as though I'm still panning the zoomed-out map. And the zoomed-out map itself is not exactly fluid.
Built this to explore THETIS-MRV, the EU's public ship emissions database. Around 12,000 vessels per year, 2018-2024. Each dot is a vessel positioned at its flag state.
The interesting pattern: the biggest clusters aren't where the shipping companies are. They're in Panama, Liberia, and the Marshall Islands, the world's largest open registries.
2024 is the first year ships had to pay for their carbon emissions under EU law. You can switch between CO₂ total, EU ETS cost, and ship type views.
Wrote about the data and the rendering challenges here: https://www.marcohaber.dev/blog/seafloor
Repo: https://github.com/marcoshaber99/seafloor
> The interesting pattern: the biggest clusters aren't where the shipping companies are. They're in Panama, Liberia, and the Marshall Islands, the world's largest open registries.
Genuine question, how is this interesting? Surely it just renders the data useless for assessing anything other than "what are the most popular open ship registries"?
Fair point. The dataset only has annual aggregates per vessel, no voyage routes or port calls, so flag state is the only geographic dimension available. The more useful part is the emissions data: CO₂ per vessel, ETS costs per company, and how the fleet changed from 2018 to 2024. The globe is the interface for exploring that.
There is nothing in the dataset that would require the use of a globe to visualize anything. You could have drawn this as bar charts and it would give us the same information (with the added advantage of not being limited to a few countries at a time). Or even a 2D earth map.
It just turns on my CPU fans and gives me no insights.
That's fair. Thanks for checking it out.
The navigation is broken on this for me. If I zoom in, the mouse panning is still at the zoomed out level. I mean I have to move the mouse increasingly accurately and in small increments as I zoom in, as though I'm still panning the zoomed-out map. And the zoomed-out map itself is not exactly fluid.
Good catch, thanks. Just pushed a fix, rotation speed now scales with zoom level. Should feel much better when zoomed in.