Landbruget.dk

Making Danish agricultural data transparent and universally accessible.

Landbruget.dk organizes data from 18+ Danish government sources into a single, queryable platform. We collect, clean, and publish agricultural, environmental, and regulatory data so that journalists, researchers, and citizens can hold the industry accountable.

Project Structure

landbruget.dk/
├── frontend/               # Next.js 16 — interactive map + data visualization
├── frontend-pesticide/     # Next.js 16 — PFAS/pesticide exposure maps
├── data-explorer/          # Next.js 16 — browser-based SQL queries on Parquet files
├── backend/                # Python data pipelines (medallion architecture)
│   ├── pipelines/          # 12 data pipelines (CHR, unified, climate, etc.)
│   └── common/             # Shared utilities (DuckDB, R2, CRS)
├── supabase/               # PostgreSQL migrations + Edge Functions
├── schema/                 # Data catalog (183 datasets) + relationship docs
├── docs/                   # Pipeline index, data lineage, troubleshooting
├── scripts/                # Utility scripts (worktree setup)
└── .github/                # CI/CD workflows (30+ GitHub Actions)

Tech Stack

Layer	Technology	Purpose
Frontend	Next.js 16, React 19, TypeScript, Tailwind CSS 4	Web applications
Maps	MapLibre GL JS, PMTiles	Geospatial visualization
Backend	Python 3.11+, DuckDB, ibis-framework	Data pipelines
Database	Supabase (PostgreSQL 15 + PostGIS)	Storage + API
Data Storage	Cloudflare R2	Raw + processed data
CI/CD	GitHub Actions	Pipeline orchestration
Deployment	Vercel	Frontend hosting
Linting	oxlint (frontend), ruff (backend)	Code quality
Testing	Playwright (frontend), pytest (backend)	Quality assurance

Quick Start

Prerequisites

• Node.js 18+
• Python 3.11
• uv
• Supabase CLI

Workspace Setup

./scripts/setup-worktree.sh

This installs frontend dependencies, Playwright browsers, and the shared Python workspace via uv, pinned to Python 3.11 via .python-version, then verifies npm test, npm run lint, and uv run pytest can resolve their local tooling.

Frontend

cd frontend
cp .env.example .env.local    # Configure Supabase credentials
npm ci
npm run dev                   # http://localhost:3000

Data Explorer

cd data-explorer
cp .env.local.example .env.local  # Add Google API key for Gemini
npm install
npm run dev                       # http://localhost:3000

Backend Pipelines

uv sync --python 3.11 --all-packages --group dev

# Run a specific pipeline
cd backend/pipelines/unified_pipeline
uv run python -m unified_pipeline bronze --source cadastral

Data Architecture

Medallion Architecture

All data flows through three layers:

• Bronze — Raw data preserved exactly as received from sources. No transformations.
• Silver — Cleaned, validated, and standardized. Type coercion, deduplication, format normalization.
• Gold — Analysis-ready. Joins across datasets on CVR/CHR/BFE identifiers, derived metrics.

Coordinate Reference System

All geospatial processing uses EPSG:25832 (UTM 32N, meters). Data is transformed to EPSG:4326 (WGS84) only at the final Supabase upload step.

Data Identifiers

All datasets join on one or more of:

Identifier	Name	Format	Purpose
CVR	Company Registration	8 digits	Links to companies
CHR	Central Herd Register	6 digits	Links to livestock herds
BFE	Cadastral ID	Variable	Links to land parcels

Data Sources

We collect from 18+ official Danish government sources including:

• Landbrugsstyrelsen — Field boundaries, crop data, agricultural subsidies
• Fødevarestyrelsen (FVM) — Livestock registry (CHR), veterinary data, pig movements
• Miljøstyrelsen — Pesticide database (BMD), environmental company registry (DMA)
• Geodatastyrelsen — Cadastral data, administrative boundaries
• Danmarks Statistik — Agricultural statistics
• DMI — Weather and climate data
• Arbejdstilsynet — Workplace safety inspections
• Datafordeleren — Property ownership data
• GEUS — Borehole pesticide data (Dataverse)

See docs/PIPELINE_INDEX.md for the full pipeline documentation.

Pipelines

Pipeline	Purpose	Schedule
unified_pipeline	18+ government data sources	Weekly (Mon 2 AM UTC)
chr_pipeline	Livestock registry + veterinary data	Weekly
svineflytning_pipeline	Pig movement tracking	Weekly (Wed 2 AM UTC)
climate	Farm-level CO2e emissions	On demand
bmd_scraper	Pesticide database	Monthly
dma_scraper	Environmental company registry	Monthly
drive_data_pipeline	Google Drive regulatory docs	On demand
bbr_buildings	Building registry	Monthly
arbejdstilsynet_inspections

Development

Running Tests

# Frontend
cd frontend && npm test         # Playwright E2E
cd frontend && npm run lint     # oxlint

# Backend
uv run --all-packages --group dev pytest   # pytest
uv run --all-packages --group dev ruff check backend
uv run --all-packages --group dev ruff format backend

Branch Naming

Format: <type>/<short-description> — e.g. feat/map-view, fix/chr-data-load

Types: feat, fix, docs, refactor, test, chore, ci, perf, build

Commit Messages

<type>(<scope>): <subject>

Examples: feat(frontend): add interactive map view, fix(pipeline): correct CHR transformation

Contributing

1. Create a branch from main following the naming convention above
2. Make your changes
3. Run all tests (npm test + uv run --all-packages --group dev pytest)
4. Run linters (npm run lint + uv run --all-packages --group dev ruff check backend)
5. Open a pull request — all PRs require review before merge

License

Code

The source code in this repository is licensed under the MIT License.

Data

The MIT license does not cover the data. This includes both:

• Upstream data ingested from Danish public-sector sources (Landbrugsstyrelsen, CHR Registry, Geodatastyrelsen, Miljøstyrelsen, Danmarks Statistik, DMI, and others — see docs/PIPELINE_INDEX.md), and
• Derived datasets we publish to the Cloudflare R2 CDN (JSON, Parquet, PMTiles).

Each dataset retains the licensing terms of its original source — typically Danish government open-data terms or, where applicable, a Creative Commons license. Where the source is not openly licensed, the original copyright and conditions of the issuing public authority apply. Reusing data from this project requires complying with the upstream source's terms.

See /om-os for the project's overall data policy and /kilder for per-source attribution.

The data is provided "as is and as available" — no warranty is given as to completeness, accuracy, or timeliness.