GeoPandas Set CRS on Points

Dance Party2 picture Dance Party2 · Aug 15, 2016 · Viewed 30.3k times · Source

Given the following GeoDataFrame:

h=pd.DataFrame({'zip':[19152,19047],
               'Lat':[40.058841,40.202162],
               'Lon':[-75.042164,-74.924594]})
crs='none'
geometry = [Point(xy) for xy in zip(h.Lon, h.Lat)]
hg = GeoDataFrame(h, crs=crs, geometry=geometry)
hg

       Lat          Lon     zip     geometry
0   40.058841   -75.042164  19152   POINT (-75.042164 40.058841)
1   40.202162   -74.924594  19047   POINT (-74.924594 40.202162)

I need to set the CRS as I did with another GeoDataFrame (like this):

c=c.to_crs("+init=epsg:3857 +ellps=GRS80 +datum=GGRS87 +units=mi +no_defs")

I've tried this:

crs={'init': 'epsg:3857'}

and this:

hg=hg.to_crs("+init=epsg:3857 +ellps=GRS80 +datum=GGRS87 +units=mi +no_defs")

...but no luck.

Some important notes:

  1. The other GeoDataFrame for which the above .to_crs method worked was from a shape file and the geometry column was for polygons, not points. Its 'geometry' values looked like this after the .to_crs method was applied:

    POLYGON ((-5973.005380655156 3399.646267693398... and when I try the above with the hg GeoDataFrame, they still look like regular lat/long coordinates.

  2. If/when this works out, I'll then concatenate these points with the polygon GeoDataFrame in order to plot both (points on top of polygons).

  3. When I try concatenating the GeoDataFrames first before using the .to_crs method, and then I use the method on both the point and polygon rows at once, I get the following error:

    ValueError: Cannot transform naive geometries. Please set a crs on the object first.

Thanks in advance!

Answer

Ufos picture Ufos · Aug 30, 2018

Geopandas API got cleaned up, and now works without surprises. Make sure to use the lastest stable version and read the docs.

Setting the CRS on a GeoDataFrame using its EPSG code is as simple as

gdf.set_crs(epsg=4326, inplace=True)

where gdf is a geopandas.geodataframe.GeoDataFrame. Watch out for the explicit inplace!

So in the example above it would be:

import pandas as pd
from shapely.geometry import Point
from geopandas import GeoDataFrame

df = pd.DataFrame({'zip':[19152,19047],
               'Lat':[40.058841,40.202162],
               'Lon':[-75.042164,-74.924594]})

geometry = [Point(xy) for xy in zip(df.Lon, df.Lat)]
gdf = GeoDataFrame(df, geometry=geometry)

gdf.set_crs(epsg=4326, inplace=True)
# ^ comment out to get a "Cannot transform naive geometries" error below

# project to merkator
gdf.to_crs(epsg=3395)

     zip        Lat        Lon                          geometry
0  19152  40.058841 -75.042164  POINT (-8353655.485 4846992.030)
1  19047  40.202162 -74.924594  POINT (-8340567.652 4867777.107)