How to use GeoPandas buffer function to get buffer zones in kilometers?

Question

I am having trouble with using GeoDataFrame.buffer function. That is probably because of my lack of understanding crs. Here is the case: I have events dataset (points longs & lats) and locations dataset (points longs & lats). I want to check which of the events occurred within 2km radius circle from locations. I know how to perform spatial join in geopandas, but the problem is in buffer(distance, resolution) function. When creating both GeoDataFrames I used crs: 'epsg:4326' and in distance parameter use 0.2 (without any good reason). So my question is:

How to adjust crs and distance parameter to make it in kilometers?

Here is my sample code. The intersection should contain only events with id 1 & 2. However it does contain event 3 which is about 10km far from any of locations. Also it does not contain event with id 4, which is good, because that one is hundreds kilometers from any of locations.

events = pd.DataFrame([
    {'id': 1,
     'longitude': 54.606402,
     'latitude': 18.347709},
    {'id': 2,
     'longitude': 54.604681,
     'latitude': 18.346534},
    {'id': 3,
     'longitude': 54.544322,
     'latitude': 18.293896},
    {'id': 4,
     'longitude': 50.843233,
     'latitude': 11.130688}
])
locations = pd.DataFrame([
    {'id': 'a',
     'longitude': 54.604972,
     'latitude': 18.346815},
    {'id': 'b',
     'longitude': 54.605917,
     'latitude': 18.347249}
])

locations_gpd = gpd.GeoDataFrame(locations,
                                 geometry=gpd.points_from_xy(locations.longitude, locations.latitude).buffer(0.2, 16),
                                 crs={'init': 'epsg:4326'})

events_gpd = gpd.GeoDataFrame(events,
                              geometry=gpd.points_from_xy(events.longitude, events.latitude),
                              crs={'init': 'epsg:4326'})

intersection = gpd.sjoin(locations_gpd, events_gpd, how='left')

Answer 1

Reproject to a projected coordinate system in meters, for example EPSG:32634. Then the buffer distance will be in meters:

import geopandas as gpd
import pandas as pd

events = pd.DataFrame([
    {'id': 1,
     'longitude': 54.606402,
     'latitude': 18.347709},
    {'id': 2,
     'longitude': 54.604681,
     'latitude': 18.346534},
    {'id': 3,
     'longitude': 54.544322,
     'latitude': 18.293896},
    {'id': 4,
     'longitude': 50.843233,
     'latitude': 11.130688}
])
locations = pd.DataFrame([
    {'id': 'a',
     'longitude': 54.604972,
     'latitude': 18.346815},
    {'id': 'b',
     'longitude': 54.605917,
     'latitude': 18.347249}
])

locations_gpd = gpd.GeoDataFrame(locations,
                                 geometry=gpd.points_from_xy(locations.longitude, locations.latitude),
                                 crs='epsg:4326')

events_gpd = gpd.GeoDataFrame(events,
                              geometry=gpd.points_from_xy(events.longitude, events.latitude),
                              crs='epsg:4326')

locations_gpd = locations_gpd.to_crs("EPSG:32634")
locations_gpd.geometry = locations_gpd.geometry.buffer(2000, 6)

events_gpd = events_gpd.to_crs("EPSG:32634")

intersection = gpd.sjoin(locations_gpd, events_gpd, how='left')