There’s an interesting email thread going on about tiling in a DataCube/QB context. Rather than hijack that thread I thought I’d post a related matter here, in case it’s of interest. We have a tiling spec for multidimensional arrays in CoverageJSON, described here:
And there is a concrete example in the playground (https://covjson.org/playground/) - click on Examples -> Grid (Tiled).
As you can see, the range of a Coverage (i.e. the multidimensional array of actual data values) can be split into nD tiles (in “tileSets”). Each tile is an nD array (encoded in JSON of course). There can be multiple tileSets encoding the same range, to support different use cases (e.g. “flat and wide” tiles for users who are interested in maps and “long and thin” tiles for users who are interested in timeseries).
Jon Blower | CTO, Institute for Environmental Analytics
Philip Lyle Building, University of Reading, Whiteknights Campus, Reading RG6 6BX
T: +44 (0)118 378 5213 M: +44 (0)7919 112687