Open bates-j opened 5 months ago
My current take is that maybe this has more to do with however USACE/HEC came up with the projection string than anything else. If you search for the Muncie projection "NAD_1983_StatePlane_Indiana_East_FIPS_1301_Feet" there's a result for a deprecated ESRI projection: https://epsg.io/102673 -- maybe it's that? Also seems like when geopandas
writes out to JSON it uses a projection short code rather than the full WKT, so if pyproj
can't recognize the WKT then I guess things get weird.
Would it be worth checking/verifying/pulling projection data from spatialreference.org to handle this? I've found ras handles the format they provide even more so than ESRI. They also provide links for what to use in place of deprecated spatial references (e.g. the example Thomas provided https://spatialreference.org/ref/esri/102673/).
The "Muncie" HEC-RAS model downloaded from the HEC website has a projection (EPSG:2965) with well-known text that deviates slightly from the accepted versions (OGC and ESRI). Because of this,
pyproj
was unable to parse and recognize the correct projection, and consequently there was trouble gettinggeopandas
to correctly write out the correct OGC URN within JSON format. Further investigation needs to be performed to ensure the way projections are currently handled will meet all of the long-term needs of the codebase.