Require PROJJSON instead of WKT2:2019 #97
Conversation
format-specs/geoparquet.md
Outdated
| [OGC Topic 2: Referencing by coordinates abstract specification / ISO-19111:2019](http://docs.opengeospatial.org/as/18-005r4/18-005r4.html). | ||
| Apart from the difference of encodings, the semantics is intended to be exactly | ||
| the same as WKT2:2019, and PROJJSON can be morphed losslessly from/into | ||
| WKT2:2019. |
There was a problem hiding this comment.
It looks like PROJJSON mentions a few deviations from WKT2:2019. Maybe "exactly" and "losslessly" are too strong here.
How about "Apart from the difference of encodings, the semantics of PROJJSON are intended to match WKT2:2019, and a CRS in one encoding can generally be represented in the other"?
There was a problem hiding this comment.
Regarding this, it could be wise, for greater interoperability, to discourage the use of BoundCRS objects, softly for top-level (as this maps to standard WKT2), and strongly for BoundCRS embedded in CompoundCRS (as this is a PROJ extension), for the PROJJSON CRS of GeoParquet.
BoundCRS objects are mostly the port of PROJ.4 +nadgrids/+towgs84/+geoidgrids constructs that are less necessary nowadays since PROJ can use its database to infer transformations between datums.
So the list of top-level objects suggested for use in PROJJSON would be:
- GeographicCRS 2D (or 3D if 3D data),
- ProjectedCRS 2D (note: ProjectedCRS 3D with ellipsoidal height is supported by PROJ, but still a bit of esoteric)
- CompoundCRS with horizontal part being a GeographicCRS 2D or a ProjectedCRS 2D and vertical part being a VerticalCRS
- and perhaps, GeodeticCRS with cartesian geocentric coordinate system (X,Y,Z) (probably of marginal use)
| BBOX[-90,-180,90,180]], | ||
| ID["OGC","CRS84"]] | ||
| ``` | ||
| The CRS is likely equivalent to OGC:CRS84 for a GeoParquet file if the `id` |
There was a problem hiding this comment.
Thanks for putting this together. I have some thoughts on it, but seemed to be a bigger scope than just a line in a PR, getting into if we might be able to tweak projjson. See #98
|
Still seing this open/ I'm confused: I thought that PROJJSON was already part of GeoParquet v0.4 ! And existing GDAL versions (latest 3.5.x and 3.6.0) that write GeoParquet v0.4 do write PROJJSON ! |
|
@rouault sorry, my mistake for not closing this. It was superseded by #96, which did indeed bring in PROJJSON support. The distinction here was about whether or not |
Resolves #95 based on associated discussions in #90 and #52.
Supersedes #94 by incorporating a modification of that text here.
Supersedes #96 by requiring
crsinstead of allowing it to be optional, with associated text changes to express this. PROJJSON for OGC:CRS84 and rules for inferring it are equivalent to #96.This includes a minor formatting change of the table including crs but only the crs row changed content and position within the table.