Skip to content

DWD Pipeline

Overview

The DWD pipeline processes numerical weather prediction data from the Deutscher Wetterdienst (DWD), specifically the ICON-D2 high-resolution model.

It is designed for near-real-time weather forecasting and risk assessment use cases.

Data sources

  • ICON-D2 forecast model (open data)

Data is downloaded from: https://opendata.dwd.de/

Output datasets

  • RAW
  • Full ICON-D2 GRIB forecasts
  • STAGING
  • Parsed meteorological variables
  • SILVER
  • Enriched wind and gust datasets
  • GOLD
  • Shareable weather indicators (e.g. gust alerts)

Licensing follows dl-de/by-2-0.

Execution & Docker image

Docker image:

ghcr.io/celine-eu/pipeline-dwd

Run locally:

task pipeline:dwd:run

Configuration & overrides

Customizable options: - Geographic bounding boxes - Forecast steps & variables - Retention and cleanup policies

See: - flows/config.yaml - dbt cleanup operations

Contributing

Contributions may include: - additional ICON products - new derived indicators - improved cleanup or validation logic

Ensure: - licensing remains compatible - derived datasets are documented in governance