# GIS Quest - Part IV

Part I defined a polygon to limit the dataset and Part II calculated trails completed within that polygon. At the time of writing the completion rate is at 82%, so not too far off done. What comes next? Defining an arbitrary polygon in Part I was a nice start, but perhaps it’s time to swtich to a structure that tiles nicely.

## Squares

If we start with the world and divide it into 32 squares, and then recursively subdivide each of those squares into 32 new squares, until we reach a required accuracy, we’d get something like this:

The above scheme, subdividing the world into recursive segments of 32
squares, and labelling each of those squares using a Z-order curve is
called a Geohash. To
cover the area within the polygon I would need to include something like
`u4ez9t`

, `u4ez9v`

, `u4ez9k`

,
`u4ez9s`

, `u4ez9u`

, `u4ez97`

,
`u4ez9e`

, `u4ez9g`

, `u4ez9d`

, and
`u4ez9f`

, so let’s try something else.

## Hexagons

As we all know, hexagons are the bestagons. Hexagons do tile nicely, but they do not subdivide as nicely as squares and triangles. Still, there do exist geospatial indexing systems that use hexagons, such as H3. We don’t really need the subdivision though, and the polygon we defined in Part I was already pretty close to a hexagon, so let’s leave it at that:

### Bestagon 0

And the quest continues…