Notebook 2.1 - Do the DWD Stations Actually Cover the Districts?

Published 3 April 2026

agri-weather-yield-drivers notebooks dwd station-coverage nuts3

Before building weather features, I wanted to see where the DWD stations fall inside the NUTS-3 districts. Notebook 2.1 maps those stations and counts how much support each district has.

A district with few stations needs different handling from one with dense coverage. Checking that now gives me a way to carry the gap into later analysis as an explicit limitation.

Technical lane: Data Evaluation Business lane: Product & Delivery

Validation intent.

I check coverage before feature engineering so I can identify districts with weak station support and explain how that affects the assumptions.

Data source
DWD CDC station metadata
Spatial join
Point-in-polygon assignment to NUTS-3 districts
Output class
District coverage quality labels and counts
2.1

Evaluate DWD Stations at NUTS-3

Spatially join station points to NUTS-3 polygons and quantify districts with zero or low station coverage.

DWD CDCPoint-in-polygonCoverage audit
Core notebook sequence completed: 33%

Key output

The notebook joins DWD CDC station points to district polygons and produces coverage counts and quality labels. You can use that table to find districts with zero or low station coverage before working with their weather features.

I keep weak station coverage in the data quality metadata so it remains visible beside the later model outputs.

- Quality gate principle

What I am watching

I want to keep those labels attached when I aggregate weather features and interpret the risk scores. A ranking is easier to judge when you can also see how much station data supports it.

Open notebook source

More evaluation posts