geopandas-distance-calculation

name	geopandas-distance-calculation
description	How to calculate distance between points and other geometries (like lines or polygons) in metric units using GeoPandas.

Distance Calculation with GeoPandas

This skill demonstrates how to calculate the distance between geographic features (like points and lines) accurately. The crucial step is projecting geographic coordinates (degrees) into a projected coordinate system (meters).

Prerequisites

pip install geopandas shapely

Basic Usage

When calculating distances on the Earth's surface, you must not calculate distance directly in EPSG:4326 (which uses degrees). Instead, you project the data into a metric coordinate system like EPSG:4087 (World Equidistant Cylindrical) or EPSG:3857 (Web Mercator, though it distorts distance). EPSG:4087 or EPSG:6933 (Cylindrical Equal Area) are common for global calculations, or a local UTM zone. EPSG:4087 provides distances in meters.

import geopandas as gpd
from shapely.geometry import Point

# 1. Load data
points_gdf = gpd.read_file('points.geojson')
lines_gdf = gpd.read_file('lines.geojson')

# 2. Define a metric CRS (e.g., EPSG:4087 for World Equidistant Cylindrical)
METRIC_CRS = "EPSG:4087"

# 3. Project both datasets to the metric CRS
points_proj = points_gdf.to_crs(METRIC_CRS)
lines_proj = lines_gdf.to_crs(METRIC_CRS)

# 4. Optional: If you want distance to any part of the lines network, combine them
network_geometry = lines_proj.geometry.unary_union

# 5. Calculate distance (the result will be in meters because of the CRS)
points_gdf['distance_m'] = points_proj.geometry.distance(network_geometry)

# 6. Convert to kilometers
points_gdf['distance_km'] = points_gdf['distance_m'] / 1000.0

# 7. Find the point furthest away
furthest_point = points_gdf.nlargest(1, 'distance_km').iloc[0]

print(f"Furthest point distance: {furthest_point['distance_km']:.2f} km")

Tips

• Always verify your metric CRS when doing distance calculations. EPSG:4326 will return distances in degrees.
• Combining features with .unary_union before calling .distance() is significantly faster than calculating distances to each feature individually.