4

I'm evaluating the use of H3 as a DGGS for a project, and so far I've been unable to find any reference of the area of the largest and smallest hexagons in the grid at each resolution. The "official" resolution table provided on the H3 website lists only the average edge lengths and areas, and I know that the maximum area of a hexagon in a given resolution is approximately 1.6 times the area of the smallest, but I need precise values.

Is anyone aware of a table that lists these, or is this likely something I'll have to do myself by just generating grids at each resolution and finding the maximum and minimum areas?

4

We don't have a table like this, though we do have a ticket to create one FWIW. For max and min, you don't have to iterate over the entire grid.

  • The smallest cells are the pentagons at the vertices of the icosahedron. You can access them using getPentagonIndexes(res). If you're specifically interested in hexagons, the smallest hexagons are the pentagon's neighbors, which you can access with kRing(pentagon, 1).

  • The largest cells are the hexagons at the centers of the icosahedron faces. I don't know if we expose this data, but you can get the cells fairly easily - take the 20 largest cells from getRes0Indexes and then use h3ToCenterChild(base, res) to get the child at a given resolution.

I used this logic to make the table here: https://observablehq.com/@nrabinowitz/h3-area-stats

You can also see a visualization of area variation here: https://observablehq.com/@nrabinowitz/h3-area-variation

8
  • Extremely helpful, thanks! I assume the same idea would hold true in terms of intercell distance, yes? And that the only real-world variation aside from this would be due to the earth not being quite exactly a sphere (so a minimum-size hexagon near the pole would in fact contain a little bit less real-world area than one near the equator, though they'd appear equal in the model)?
    – DanM
    Mar 5 '20 at 19:40
  • If I understand the question, intercell distance is a function of edge length, like cell area, so the variation should be similar. The variation is due to the fact that the cells are gnomically projected onto the icosahedron faces, so they're distorted somewhat due to the projection - other than that, yes, there's probably some distortion based on the spherical model H3 uses. Mar 5 '20 at 22:38
  • @nrabinowitz in the h3-area-stats link you provided, can you explain what the min and max areas are? A res-8 hexagon has an edge of about half a km and in your table its min area is 447,526 km². How is that computation made?
    – user159330
    Mar 16 '20 at 15:20
  • @bsod - Arg, just realized that the library I'm using output sq meters, not sq kilometers, so I had the labels wrong! I've updated the labels in the table, apologies for any confusion. Mar 17 '20 at 17:48
  • Hi, perhaps your h3-area-stats will be more useful (to show H3 behavior or grid-area profile by resolutions) when replacing MAX and MIN by percentiles, something like 90th and 10th percentiles. May 2 '21 at 1:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.