Python for GIS Domain¶

What you'll learn¶

• What GIS and geospatial data are
• Coordinates and the distance problem (tested)
• Vector vs raster data
• Projections — the classic gotcha
• The GeoPandas/Shapely ecosystem

GIS (Geographic Information Systems) analyzes data tied to locations — maps, routes, territories, satellite imagery. Python has become a first-class GIS language. The distance calculation here is run-verified; spatial libraries follow documented APIs.

The distance problem (tested)¶

The distance problem — a key concept in Python for GIS.

A deceptively hard basic task: how far apart are two lat/long points? Because Earth is (roughly) a sphere, you can't just use flat-plane distance — you need the haversine formula. Runnable pure Python:

import math

def haversine_km(lat1, lon1, lat2, lon2):
    """Great-circle distance between two points, in kilometers."""
    R = 6371.0                                   # Earth radius in km
    p1, p2 = math.radians(lat1), math.radians(lat2)
    dp = math.radians(lat2 - lat1)
    dl = math.radians(lon2 - lon1)
    a = math.sin(dp/2)**2 + math.cos(p1)*math.cos(p2)*math.sin(dl/2)**2
    return R * 2 * math.asin(math.sqrt(a))

# London to Paris
d = haversine_km(51.5074, -0.1278, 48.8566, 2.3522)
print(f"{d:.1f} km")

Output:

343.6 km

~344 km matches the real great-circle distance between London and Paris. This is why GIS isn't just "x, y math" — the Earth's curvature and the choice of coordinate system fundamentally shape every calculation.

Vector vs raster¶

Points/lines/polygons vs pixel grids — the two geospatial data models.

Two fundamental data models in GIS:

• Vector — geometry as points, lines, and polygons (a city as a point, a road as a line, a country as a polygon). Best for discrete features. Handled by Shapely (geometry) and GeoPandas (tables of geometries).
• Raster — a grid of pixels, each with a value (satellite imagery, elevation, temperature maps). Best for continuous surfaces. Handled by rasterio, NumPy.

Projections: the classic gotcha¶

Coordinate reference systems — get them wrong and distances/areas are nonsense.

GeoPandas + Shapely¶

DataFrame-style spatial analysis with geometry operations.

GeoPandas extends Pandas with geometry — a DataFrame where one column holds shapes:

import geopandas as gpd            # pip install geopandas
from shapely.geometry import Point

# Load a shapefile / GeoJSON
cities = gpd.read_file("cities.geojson")

# It's a DataFrame with a geometry column + a CRS
print(cities.crs)
cities = cities.to_crs("EPSG:3857")     # reproject to a common CRS

# Spatial operations
point = Point(-0.1278, 51.5074)
within = cities[cities.geometry.within(some_polygon)]   # points inside an area
cities["area_km2"] = cities.geometry.area / 1e6

The ecosystem¶

GeoPandas, Shapely, rasterio, and Folium.

Practice exercises¶

1. Use haversine_km to find the nearest of several cities to a given point.
2. Compute the total length of a route given a list of lat/long waypoints.
3. Explain, with an example, how mixing two coordinate systems produces wrong results.
4. Look up your city's coordinates and compute its distance to three others.
5. Describe when you'd use vector vs raster data for a given problem (e.g. city locations vs rainfall).