Investigate TurfJS for geometry repair/simplification

[prashtx]

https://github.com/Turfjs/turf-simplify
https://github.com/Turfjs/turf-kinks

We need to confirm that the libraries behave well with problematic parcels that we've encountered in the past.

[prashtx]

It looks like turf-kinks doesn't detect overlaps with no crossover, as in

{
  "type": "Polygon",
  "coordinates": [[ [5,5],[5,6],[6,6],[7,6],[6,6],[6,5],[5,5] ]]
}

It also doesn't detect repeated points, although those are easier for us to deal with separately:

{
  "type": "Polygon",
  "coordinates": [[ [5,5],[5,6],[6,6],[6,6],[6,5],[5,5] ]]
}

[hampelm]

Should we file a bug? Is a self-overlap without crossover technically a
kink?

On Wed, Dec 31, 2014 at 3:09 PM, Prashant Singh [email protected]
wrote:

It looks like turf-kinks doesn't detect overlaps with no crossover, as in

{
"type": "Polygon",
"coordinates": [[ [5,5],[5,6],[6,6],[7,6],[6,6],[6,5],[5,5] ]]
}

[image: image]
https://cloud.githubusercontent.com/assets/1307783/5590209/631f1f26-90e4-11e4-942f-4251212a61eb.png

It also doesn't detect repeated points, although those are easier for us
to deal with separately:

{
"type": "Polygon",
"coordinates": [[ [5,5],[5,6],[6,6],[6,6],[6,5],[5,5] ]]
}

—
Reply to this email directly or view it on GitHub
#267 (comment)
.

[prashtx]

We can possibly use turf-envelope to index a feature's bounding box. It's possible to get a degenerate polygon out of turf-envelope, so we'd still need to do some validation. If we commit to simplifying all indexed polygons/multipolygons, then we start with computing the envelope. If the area is 0, we can index the centroid instead. The envelope is always a rectangular polygon, so we won't encounter weird zero-width arms. It should either be valid or have zero area.