River kilometres per country, measured on the ellipsoid

PythonexpertSpatial analysis

A transboundary-water NGO is ranking countries by the length of major river mapped inside their borders, as the first cut of a shared-basin index. The previous analyst projected everything to Web Mercator, which made Russia's rivers half again as long as Brazil's for no hydrological reason. The brief is to cut every river at the borders and measure each piece on the ellipsoid, so the number means the same thing in Norway and in Nigeria.

Write a script that produces river kilometres per country.

The datasets are mounted where the script runs:

/data/countries.geojson — country and dependency polygons
/data/rivers.geojson — 461 river and lake centrelines, MultiLineStrings

geopandas, shapely, pyproj, pandas and numpy are available. Assign the finished table to a variable named `result`: a GeoDataFrame with one row per country, carrying the country geometry and

iso_a3, name
river_km — total length of centreline inside the country, km, measured geodesically

Cut each river at the country boundaries and measure the pieces with a geodesic length (pyproj.Geod.geometry_length on WGS 84). A river that crosses a border is credited to each side for the part on that side. Do not write a file — the runner reads result.

The input files

/data/countries.geojson — EPSG:4326; the geometry is in geometry (Polygon / MultiPolygon).

column	type	meaning
`country_id`	string	CO- plus the ISO 3166-1 alpha-3 code
`name`	string	Short name
`name_long`	string	Long name
`iso_a3`	string	ISO 3166-1 alpha-3; -99 where Natural Earth assigns none
`iso_a2`	string	ISO 3166-1 alpha-2; -99 where Natural Earth assigns none
`continent`	string	Continent
`subregion`	string	UN subregion
`pop_est`	number	Population estimate (persons)
`pop_year`	integer	Year of the estimate
`gdp_md`	number	GDP, millions of US dollars (USD m)
`gdp_year`	integer	Year of the GDP figure
`economy`	string	Natural Earth economy class
`income_group`	string	World Bank income group

/data/rivers.geojson — EPSG:4326; the geometry is in geometry (MultiLineString).

column	type	meaning
`river_id`	string	River identifier, `RV-` and four digits
`name`	string	Name as labelled
`name_en`	string	English name
`featurecla`	string	River \| Lake Centerline
`scalerank`	integer	Prominence, lower is more prominent

Run executes the script in your browser and shows you its output and a preview. Submit executes it again on the server and grades what result holds.

What to submit

Submit: A Python script that assigns its output to result.
Input files: /data/countries.geojson · /data/rivers.geojson
Required properties: iso_a3, name, river_km

Objectives

Find every river–country pair that intersects
Cut each river to the country polygon
Measure the pieces geodesically and sum per country
Assign the finished GeoDataFrame, one row per country, to result

Constraints

These are where the teaching is. Read them twice.

Length is geodesic on WGS 84 — not planar in any projection, not degrees
A river crossing a border contributes to each side its own part
Every country appears, with zero where no centreline is mapped
result carries the country geometry with iso_a3, name and river_km

Graded requirements

Cuts rivers at the country boundary
Measures length geodesically
One row per country
Carries iso_a3, name and river_km
Total river kilometres inside all countries
The country with the most river kilometres

Cutting a global line layer at boundaries and measuring on the ellipsoid is the method under river, road and pipeline statistics in every international dataset, and the projection shortcut is the most common way they go wrong.

Python

Loading editor

Files: /data/countries.geojson, /data/rivers.geojson · assign result

Loading map

What will be checked

6 scored, 0 informational

Cuts rivers at the country boundary
17%
Measures length geodesically
17%
One row per country
17%
Carries iso_a3, name and river_km
8%
Total river kilometres inside all countries
25%
The country with the most river kilometres
17%