[gdal-dev] Multidimensional raster support in GDAL

[Edzer Pebesma]

On 10/26/2017 09:39 AM, Ari Jolma wrote:
> Edzer Pebesma kirjoitti 26.10.2017 klo 10:26:
>> More in general, we have multidimensional arrays where horizontal space
>> is not an x y image (GDAL raster band) but for instance a sequence of
>> feature geometries (e.g., in situ sensor data, or demographic data by
>> administrative region, time, and age class). We may want to keep those
>> use cases in mind when working on this from the GDAL perspective, since
>> pretty soon you will want to cut out a point or region, or aggregate
>> grid values over a polygon area while keeping all other dimensions.
> 
> With vector data the case is different, yes. You may have 'a thing'
> (say, a province), whose spatial representation changes significantly in
> time along with its characteristics (say, population). Or you may have a
> moving sensor.

For me, the static vs dynamic space discretization choice does not align
with the vector vs. raster choice, but for the case of representing
spatiotemporal phenomena with dense arrays, which is what we discuss
here, we'd better stick to static space discretization.

> 
> But with raster data, I'd stick very strongly to the notion that the
> spatial representation is immutable. The only thing that may be

Yes.

> important(? I may disregard something here) is that when you do
> reprojections, then the area that the cell represents changes. Is this
> too strong statement?

That's a different discussion, really. But an interesting one!

> 
> Ari
> 
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/gdal-dev

-- 
Edzer Pebesma
Institute for Geoinformatics
Heisenbergstrasse 2, 48151 Muenster, Germany
Phone: +49 251 8333081