Skip to content

ODRE.md

Latest commit

 

History

History
59 lines (36 loc) · 3.39 KB

ODRE.md

File metadata and controls

59 lines (36 loc) · 3.39 KB
Raw

Production Assets Pipeline (ODRÉ)

Back to README

This document details the extraction and spatial mapping algorithms applied to the French National Registry of Electricity Production Assets. The pipeline implements a geocoding process to assign precise coordinates to power plants, enabling the spatial correlation of installed capacity with local meteorological conditions.

Part 1: Collection Module (collect_production_assets.py)

This module manages the acquisition of the national registry and its enrichment via an auxiliary georeferencing dataset. It solves the issue of missing coordinates in regulatory datasets by merging data based on INSEE codes (the standard French administrative municipality identification system).

Data Sources

To achieve complete coverage, the pipeline aggregates two distinct open data streams:

1. Asset Registry (Primary Source)

  • Provider: Open Data Réseaux Énergies (ODRÉ)
  • Dataset: National Registry of Electricity Production and Storage (Registre national des installations de production) {cite:p}ODRE:Registre.
  • Granularity: Per installation (exact locations are anonymized for small installations < 36 kVA, or kilovolt-amperes).
  • Format: CSV Export via API v2.1.
  • License: Etalab Open License.

2. Geographic Reference (Auxiliary Source)

  • Provider: OpenDataSoft (Public)
  • Dataset: Georef France Commune {cite:p}OpenDataSoft:Georef.
  • Role: Provides the geographic centroid (latitude, longitude) for every municipality in France to use as a fallback.

Data Schema Output

Column Name Type Description
decommissioning_date DATE Date the asset was decommissioned (NULL if active).

Note: The persisted installations table also retains auxiliary source columns used internally for joining and disambiguation: insee_code_fallback, region_code, energy_source_code, technology_code, asset_name, and the reference-centroid columns ref_lat / ref_lon.

Part 2: Spatial Transformation Logic

While this module is primarily a collector, it performs spatial transformations to prepare the data for the "multi-barycenter" aggregation used downstream in transform_meteo.py.

Geocoding Logic

The aggregation of meteorological variables requires defining the center of gravity for each energy source. This module provides the atomic data (points) needed to calculate those regional centers.

  • Input: Administrative Codes (INSEE).

  • Logic: Prioritize specific native coordinates when available; fallback to the administrative municipality centroid when they are missing or anonymized.

    $$Coordinate_{final} = \text{COALESCE}(Coordinate_{native}, Centroid_{municipality})$$

  • Output: A dataset where every MW of installed capacity has a valid spatial coordinate (latitude, longitude).

Note on Privacy & Resolution: For small installations (<36 kVA, typically residential solar panels), the exact location is anonymized by the provider for privacy. The centroid imputation method is statistically sufficient for regional-scale weather correlation. The spatial error margin introduced by using a municipality centroid (~2 to 5 km) is significantly smaller than or roughly equal to the grid resolution of standard numerical weather models used by Météo-France (e.g., AROME is ~1.3 km, ARPEGE is ~5 km).

Back to README