PostGIS not recognizing geometry type after writing to database with geopandas - postgis

Question

I have a PostGIS database set up that is spatially enabled.
engine = create_engine(f'postgresql://user:password#localhost:5432/{dbname}')
I have a valid geopandas dataframe (i.e., if I export as .shp or .gpkg I can load into a GIS app with no errors) with a geometry column named 'geometry'. I write these data to the database with:
gdf.to_postgis('table_name', con=engine, if_exists='append', index_label='id')
If I read all records from this table:
gpd.read_postgis('table_name', con=engine, geom_col='geometry')
I get the following warning:
SAWarning: Did not recognize type 'geometry' of column 'geometry' pandas_sql.meta.reflect(only=[sql])
If I try to execute an postgresql statement such as:
gpd.read_postgis('SELECT DISTINCT(st_cnty_fips) FROM table_name', con=self.engine, geom_col='geometry')
I get an error:
line 65, in _df_to_geodf
raise ValueError("Query missing geometry column '{}'".format(geom_col))
ValueError: Query missing geometry column 'geometry'
Using PGAdmin4 I can see that the column exists. What am I doing wrong?

Answer 1

Geopandas requires a geometry columna after filtering out geometry additionally or adding it should work.
The parameter geom_col='geometry' allows you to specify the name of this column that geopandas needs.
Need to add geometries.
For example:
SELECT your_query_with_out_geom, geometry FROM table_name;
In your case, could:
gpd.read_postgis('SELECT DISTINCT(st_cnty_fips), geometry FROM table_name', con=self.engine, geom_col='geometry')