Validation Dataset for DE9IM functions in DGGS

[rggibb]

Logically, one could just take any existing DE9IM validation dataset and import it into a DGGS and perform the equivalent tests. However DGGS provides the capability to specify the precison with which to do a test, and the results can be both precision and DGGS dependent. For example:

For the pair of shapes:

Logically we would expect square.overlaps(triangle) to return True

But if we import theses shapes into a DGGS at three different levels, we might get the following:

Further the nuances of the result are both DGGS architecture dependent (hexagons vs squares etc) and DGGS RS dependent even within the same architecture.

So how do we address this in order to get a robust test suite with consistent validation results for all DGGS RSs.

[jerstlouis]

@rggibb This is a very interesting topic.

Even without implementing a DGGS API, the issue of precision came up in our implementation of DE9IM and CQL2 for OGC API - Features, since we import the data in a DGGS data store. The CQL2 ATS includes a series of useful spatial comparison tests.

So I would say that DE9IM functions are automatically dependent on refinement levels in this case.

Perhaps some guidance in this regard is needed for the Abstract Test Suite specifically for the CQL2 Filtering of both Zone Data and Zone Queries, which can include DE9IM spatial operators.

Our deployment of the CQL2 test dataset is available here, and we support DGGS zone data and zone queries using CQL2 for the three DGGRS in Annex B. Some extensive testing / comparisons with other implementations would be useful to make progress on this issue.