Open m-mohr opened 2 years ago
Hi @m-mohr. Indeed, the temporal dimension is a common aspect of the 2 extensions. I must admit I overlooked the datacube one since it is very specific to the multidimensional arrays of values.
In our use case, we simply need to describe a temporal dimension. The affected datasets may be simple CSV files or complex displacement rasters that are combined with other extension such as insar
.
The relationship with the OGC coverage concept is still very embryonic and would require a deeper work to make something more consistent.
I would be glad to find a way to reuse the work done in datacube. I am just a bit confused to use datacube concept for simple time series datasets. What's your view on that?
It could probably be used to expose a 1D "data cube", but it doesn't feel natural, indeed. I also did not want to enforce the data cube extension to be used, this was more a heads up. Let's discuss in the STAC meeting or so :-)
This looks very similar of what you can already do with the Data Cube extension. Here you'd likely have at least a temporal dimension in cube:dimensions with values (ts:dates) or extent and step (ts:period). ts:parameters sounds like cube:variables. What's the relation between them? Has it been considered and/or aligned? Reading WCS and Coverages at the bottom sounds like there might be a connection.