NiklasPhabian
commented
9 months ago ok. There seem to be multiple issues going on. You project the points to LambertAziumthalEqual area. Therefore, all the subsequent geometries are in this projection. When you bootstrap the gdf, you'll need to tell it and set the CRS to polar_crs, rather than geod_crs.
geopandas.GeoDataFrame(df_dict, crs=polar_crs, geometry=vert_polys)Next, you cannot hand those projected coordinates to STARE. STARE knows nothing about projections and assumes all x/y coordinates are wgs84 lat/lons. Since STARE doesn't know about projections, neither does pystare. Unfortunately, STAREPandas also does not give a damn about projections as well, even though it could/should.
So before you could do a starepandas.sids_from_geoseries, you have to reproject them to WGS84 coordinates. IMO, the easiest way would be to do
usi_geo_df = usi_geo_df.set_geometry('vert_poly_geo').to_crs(4326)['geometry']
usi_geo_df.make_sids()Or:
usi_geo_df = usi_geo_df.set_geometry('vert_poly_geo').to_crs(4326)
sids = starepandas.sids_from_geoseries(polys_pds, level=opts.mcms_q_level, convex=False, force_ccw=False, n_partitions=opts.n_parts, num_workers=opts.n_cores)Word of caution: While multiple geometry columns are supported in geopandas, the CRS is a setting of the dataframe, not the individual column. I.e. be very very careful of having multiple geom columns with differing CRS
I saved of a single ETC track geopandas dataframe
mcms_merra2_npac_2017_usi_tagrp_niklas.pklfor you here is some relevant code to it's creation.After this, I try to make SIDs etc.
The first curious thing is the column
vert_poly_geoare not shown to be in projection units when I print the mcms_tagrp_gdf.head(), so I am curious why they are projected like thegeometrycolumn.Here I try to find the SIDs
opts.mcms_q_level = 10opts.n_parts = 60opts.n_cores = 6and here is the error.
If I set convex=True I get the same error.