search

opengeospatial / geotiff

18 stars 9 forks source link

Refunding handling geodetic parameters in GeoTIFF on basis of AS Topic 2 #95

Open EmDevys opened 3 years ago

EmDevys commented 3 years ago

From Roger Lott 9 Dec.: The provisions in GeoTIFF 1.1 (vs GeoTIFF 1.0) are not (fully) adequate for 21st century needs. In particular there is no proper capability for dealing with heights, and spatial 3D CRSs are incomplete. If you have a Lidar survey with sensor platform coordinates in geographic 3D and the georeferencing is made in this CRS (with today's technology, the standard practice) then you cannot express this in GeoTIFF v1.1. Nor can you express a compound CRS, or its vertical component. The possible values for GTModelTypeGeoKey need to be extended and the 'temporary' provisions of Annex D completely replaced with normative requirements for the additional GTModelTypeGeoKey values. Also the need to associate a coordinate epoch with a dynamic CRS is required. This was all suggested during the v1.1 work but deemed to be beyond the scope of the minimum change strategy.

EmDevys commented 3 years ago

In GeoTIFF 1.1, Requirement 14.4 and guidance in Annex D.3 allow to use ellipsoidal height, though what is proposed may not be compliant with the way to model this in AS Topic 2. So Lidar surveys with ellipsoidal heights (or other similar usages) may use GeoTIFF 1.1. GeoTIFF 1.1 is more problematic with projected cRS and ellipsoidal height, though Annex D.3 explains how to handle this with GeoTIFF 1.1.

joanma747 commented 2 years ago

Do we what to extent GeoTIFF to all CRS. TIFF is a 2D grid so we should limit to what is possible in TIFF. The issue is focus in the vertical CRS. We could review this. Can we review other kinds of CRSs supported in WKT (or use WKT directly; see https://github.com/opengeospatial/geotiff/issues/56)

EmDevys commented 2 years ago

GeoTIFF (as is) is aimed at georeferenced or geocoded raster (imagery such as orthoimagery or 2D-grid of values providing information at a location on the ground - at a 2 D location). This information may be classified or thematic imagery / raster or elevation (in this last case, this is sometimes called 2D1/2 such as Terrain Elevation Model), and there is then a need to define the Vertical CS ). That is what GeoTIFF 1.0 aimed to provide / encode, and what has been clarified in OGC GeoTIFF 1.1.

Though the 2.0 version allows to broaden the scope, I guess the TIFF 2D limitation (as you mention it) - or 2D1/2 as reminded here) and the fact that GeoTIFF is designed to represent geospatial 2D raster (therefore connected to a location on the earth or another planet) should still be the core of GeoTIFF 2.0. If there are use cases or requirements for addressing other dimensions (and CRSs), no objection, but should probably be an extension.