add DimensionalData extension #178
Conversation
ext/GeometryOpsDimensionalDataExt/GeometryOpsDimensionalDataExt.jl
Outdated
Show resolved
Hide resolved
| import GeometryOps as GO | ||
| import GeoInterface as GI | ||
|
|
||
| function GO.polygonize(A::DD.AbstractDimArray; dims=(DD.X(), DD.Y()), crs=GI.crs(A), kw...) |
There was a problem hiding this comment.
What happens with the dims that aren't given (time etc)?
There was a problem hiding this comment.
Yeah I haven't done that yet, this should be WIP.
But I image we make a feature collection with a feature for each slice, and put the lookup values as properties?
Otherwise return a DimArray of geoms of the remaining dimensions ?
Probably a single feature collection or table is most useful?
There was a problem hiding this comment.
Actually this is a case for R style vector data cubes.
But we can just broadcast over the other dimensions and get a DimArray of X/Y geometries with time/etc dimensions.
We should be able to write that directly as a FeatureCollection with GeoJSON/ArchGDAL/Shapefile without any more work as long as we name the array :geometry.
There was a problem hiding this comment.
Do we then want to add an assert here that the dims can only have length 2?
|
Does that last commit set minimum version to Julia 1.10 instead of 1.9? |
|
Oops wrong commit message, we can fix it in the squash |
|
Is this just waiting on the datacube stuff/implementation? |
|
Waiting for me to have time 😭 |
| bounds_vecs = if DD.isintervals(lookups) | ||
| map(DD.intervalbounds, lookups) | ||
| else | ||
| @warn "`polygonsize` is not possible for `Points` sampling, as polygons cover space by definition. Treating as `Intervals`, but this may not be appropriate" |
There was a problem hiding this comment.
Doesn't this just mean that the ring vertices are located at the points? That seems pretty exact...
There was a problem hiding this comment.
Hmm not sure I understand you. It's just saying that points don't cover any space so polygonizing them doesn't make sense
There was a problem hiding this comment.
I was thinking of it more in a sense that the polygonizing operation is the concave hull of the set of points that satisfy f...but that is also probably a different algorithm!
There was a problem hiding this comment.
Yeah, somehing else! polygonize is like a groupby/combine for lots of little square polygons
This adds a DD extension for
polygonizeso that any DD array can be polygonized with X/Y (or custom withdimskeyword) lookup values.