Python Geopandas – How to Intersect Polygons Within a Single GeoDataFrame in Python

Question

I am trying to find a way to intersect a polygons layer with itself in python to identify each newly produced intersection polygon using geopandas, but am not sure how to actually perform this with a single geodataframe, as opposed to intersecting one geodataframe with another geodataframe.

I have this map of NYC boroughs. Each borough has a "borocode"

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import geopandas as gpd
from shapely.geometry import Polygon, LineString, Point

nybb_path = gpd.datasets.get_path('nybb')
boros = gpd.read_file(nybb_path)
boros.set_index('BoroCode', inplace=True)
boros.sort_index(inplace=True)
boros.plot()

We get this output geodataframe:

and see this:

I then draw buffers around them via:

boros['geometry'] = boros.geometry.buffer(8000)
boros.plot(cmap='Greens', edgecolor='black', alpha=0.5)

which produces this output dataframe, which looks the same, but the geometries are different:

and I then see this:

This leads to overlapping regions, as we can see above. There are 8 of these "overlapping" regions from what I can make out here.

What I want to do is identify the regions where these newly buffer polygons "overlap", or I suppose this would be called an intersection. I want to assign them a unique ID, and create a new column that shows which original polygons "participate" in creating each intersection zone.

And so, my goal is to create this geodataframe:

ID                   Overlaps     Number_of_Overlaps                                 geometry
----------------------------------------------------------------------------------------------
0                   Manhattan                      1     POLYGON ((XX.XX XX.XX, XX.XX XX...))
1                       Bronx                      1     POLYGON ((XX.XX XX.XX, XX.XX XX...))
2                    Brooklyn                      1     POLYGON ((XX.XX XX.XX, XX.XX XX...))
3                      Queens                      1     POLYGON ((XX.XX XX.XX, XX.XX XX...))
4               Staten Island                      1     POLYGON ((XX.XX XX.XX, XX.XX XX...))
5            Manhattan, Bronx                      2     POLYGON ((XX.XX XX.XX, XX.XX XX...))
6    Manhattan, Bronx, Queens                      3     POLYGON ((XX.XX XX.XX, XX.XX XX...))
7    Manhattan, Staten Island                      3     POLYGON ((XX.XX XX.XX, XX.XX XX...))
8           Manhattan, Queens                      2     POLYGON ((XX.XX XX.XX, XX.XX XX...))
9               Bronx, Queens                      2     POLYGON ((XX.XX XX.XX, XX.XX XX...))
10   Manhattan, Bronx, Queens                      3     POLYGON ((XX.XX XX.XX, XX.XX XX...))
11              Bronx, Queens                      2     POLYGON ((XX.XX XX.XX, XX.XX XX...))
12        Manhattan, Brooklyn                      2     POLYGON ((XX.XX XX.XX, XX.XX XX...))
...

The documentation on this in geopandas only includes examples where I am overlaying one "layer" on top of another "layer", whereas with my example I am trying to find these intersection zones within the same layer itself.

My inclination would be to go with this code using the .overlay() function within geopandas, going with:

NYC_intersections = boros.overlay(boros, how='intersection')

However, this function actually requires 2 geodataframes to be intersected, so I would need to intersect 2 different geodataframes, such as with:

res_intersection = boros1.overlay(boros2, how='intersection')

However, I only have my single geodataframe to work with, so I am not sure how to actually use this code.

My question ultimately here is, how can I take my buffered polygons layer and intersect it with itself to then create a geodataframe where each polygon, including the intersection zone polygons, is identified, listing the polygons that make up each "zone" and the number of polygons making up each zone?

Answer 1

You should be using union:

import geopandas as gpd

df = gpd.read_file(r'/home/bera/Desktop/GIStest/countries.shp')
df = df[["NAME","geometry"]]

dfbuff = gpd.GeoDataFrame(df, geometry=df.buffer(40000))

u = dfbuff.overlay(dfbuff, how="union")
u = u.loc[u["NAME_1"]!=u["NAME_2"]] #Drop the self unions

#The union will create duplicates, like polygon1-polygon2 and polygon2-polygon1
u["countries"] = u.apply(lambda x: '-'.join(sorted(x[["NAME_1", "NAME_2"]])), axis=1)
u = u.drop_duplicates(subset="countries") #Drop them by name