Pygeoapi provider [Completed ]

[chinedu117]

@isedwards @david-i-berry

Please review the changes notably :

• the Geography column definition SRID 4326
• the storage of Point object as binary (WKB)
• the derivation of lat lon from Point object instead of writing join queries (note: location is part of observation in CDM and OGC OMS)
• etc Please let me know your thoughts as i write tests for these.

[david-i-berry]

In PostGIS we can have mixed geography types in the same column so at this stage I would leave it as a generic geography type. As way of example see a quick test below:

[david-i-berry]

The geographies in the column need to have the same SRID but we may want to allows SRIDs other than 4326. We can take this question to the TT-CDM but my preference is to be flexible as long as the CRS / SRID is reported in the data model.

[chinedu117]

The geographies in the column need to have the same SRID but we may want to allows SRIDs other than 4326. We can take this question to the TT-CDM but my preference is to be flexible as long as the CRS / SRID is reported in the data model.

Thanks for the insight. Could we instead enforce SRID 4326 ? We can always convert from one SRID to another using postgis or geoalchemy's ST_Transform method. So we would only expect the srid to be in one format for all our calculations.

[david-i-berry]

The geographies in the column need to have the same SRID but we may want to allows SRIDs other than 4326. We can take this question to the TT-CDM but my preference is to be flexible as long as the CRS / SRID is reported in the data model.

Thanks for the insight. Could we instead enforce SRID 4326 ? We can always convert from one SRID to another using postgis or geoalchemy's ST_Transform method. So we would only expect the srid to be in one format for all our calculations.

That's a good point, in terms of implementation we can insist on 4326 but then provide the mechanism for transformation to other SRIDs.

[chinedu117]

[chinedu117]