boost distance query from point to polygon

Several ways have been developed to boost distance queries from point to point and point to line (see here). However, I have not found a solution to fast distance query from point to polygon using R-Tree or cKDTree. Here, Geoff Boeing provides a way to boost spatial query, but looks like (to me) can only detect whether points are falling into the polygon using .intersect(). My question is how to calculate the nearest distance from point gdf to polygon gdf quickly using R-Tree, spatial index, etc. I did some tests on small dataset.

# Load libraries    
%matplotlib inline
import numpy as np
import pandas as pd
import geopandas as gpd
import matplotlib.pyplot as plt
from geopandas import GeoSeries
from shapely.geometry import Point

# New York City
nyc = gpd.read_file(gpd.datasets.get_path('nybb'))

# Random points within the bounding box
np.random.seed(123)

xmin, xmax, ymin, ymax = 900000, 1080000, 120000, 280000
xc = (xmax - xmin) * np.random.random(20) + xmin
yc = (ymax - ymin) * np.random.random(20) + ymin
geopts = gpd.GeoDataFrame(GeoSeries([Point(x, y) for x, y in zip(xc, yc)]))
geopts.rename(columns={0:'geometry'}, inplace=True)

# Plot NYC and points
ax = nyc.boundary.plot(edgecolor='black')
geopts.plot(ax=ax, color='red')

# Distance from point to polygon
geopts['dist2poly'] = geopts.geometry.apply(lambda x: nyc.distance(x).min())
geopts

However, with big datasets, this approach can be time-consuming. Any suggestions to incorporate spatial index into this distance query? Thanks.

[shuai-zhou]

@martinfleis Thanks, this worked. Forgive me if I bothered you too much, but I spent the weekend on some examples and still could not work out a solution. To make it simple, I was wondering how to get the same results just like the straightway way geopts['dist2poly'] = geopts.geometry.apply(lambda x: nyc.distance(x).min()) did. Thanks, much appreciated.

[adriangb]

Glad you found a solution! Just to quickly elaborate on the example @martinfleis have above, it is important to note that if you set num_results to a small number, you may not get accurate results. If you use too large of a number, things will be very slow. There is a much more detailed discussion of this in #1271.