Closed fwfichtner closed 3 years ago
We will break a lot of backwards compatibility for ukis_psyat.data
.
That is true, but it will make ukis-pysat future proof concerning all aspects related to metadata. In my opinion the benefits overweight the problems related to backwards compatibility by far. Compatibility, moreover, will potentially be affected almost exclusively by the way the metadata are returned (in particular the keys). Method calls should (almost) not be affected as far as I can see for now, since STAC follows a very similar approach to our current custom implementation. Even the returned metadata will be in the same file format (JSON) contain the same subset of attributes that we anyway follow already. Accessing metadata will change to some degree, but we gain the benefit of more flexibility, standardization and interoperability plus the options related to distributed catalogue queries etc.
We should also dicsuss how we want to handly datetime, here's the best practice.
Roadmap to move ukis_pysat.data
to STAC
Reopening since discussion around this topic are still ongoing.
Major point that we are currently still missing is the conversion of metadata to STAC best practises for specific sensors. This should also specifically include the image band list as is described in detail in #23
As soon as we are happy we should consider adding this to stacindex @MWieland.
Let's create new issues, specific for each topic.
Make our metadata a SpatioTemporal Asset Catalogs to be interoperable with other tools. We should also be able to extent it if we need more. @MWieland what do you think about this?
pystac might help with that.
And when we're on it we could make the creation of Cloud Optimized GeoTIFFs a default.