weathermart: the weather data market

Documentation: https://weathermart.readthedocs.io/en/latest/

weathermart provides a common provider/retriever interface over several weather data sources, including EUMETSAT satellite products, MeteoFrance OPERA radar products, GRIB-based forecast data, and terrain datasets.

Installation

Run the setup script from the repository root:

chmod +x install.sh
./install.sh
source .venv/bin/activate

install.sh installs uv, creates the local .venv, installs the package with all extras, and sets up the CODA and HARP dependencies needed for EUMETSAT products such as METOP and MTG.

Credentials

Some retrievers require external accounts and API credentials.

EUMETSAT

Create an account in the EUMETSAT portal and generate API credentials:

• Portal: https://api.eumetsat.int/api-key/

Store the credentials in .eumdac_credentials.json in the repository root:

{
  "consumer_key": "YOUR_EUMETSAT_CONSUMER_KEY",
  "consumer_secret": "YOUR_EUMETSAT_CONSUMER_SECRET"
}

The satellite examples in src/weathermart/examples/ use this file by default.

MeteoFrance OPERA Radar

Create an account at the MeteoFrance API portal and generate an API token:

• Portal: https://portail-api.meteofrance.fr

Store the token in .meteofranceapi_token.json:

{
  "OPERA_API_TOKEN": "YOUR_METEOFRANCE_API_TOKEN"
}

Alternatively, you can export the token directly:

export OPERA_API_TOKEN=YOUR_METEOFRANCE_API_TOKEN

Implemented Retrievers

The package can retrieve data from:

• NWP forecasts via GRIB files
• Radar data via the MeteoFrance OPERA API
• Satellite data via the EUMETSAT API
• DEM raw products via URL-based downloads

Retrievers can be used directly or through the provider, which checks the cache first and only retrieves missing data.

Provider Example

import numpy as np
import pandas as pd

from weathermart.default_provider import default_provider

provider = default_provider()
config = {
    "ICON-CH1-EPS": ["CLCT", "TOT_PREC", "U_10M", "V_10M", "QV_2M", "T_2M", "P", "SP"],
    "dates": pd.date_range("2023-08-01", "2024-09-09"),
}

provider.provide_from_config(
    config, datatype="forecast", ensemble_members=0, step_hours=np.arange(1, 13)
)

Examples

Example scripts are available in src/weathermart/examples/.

For example:

python src/weathermart/examples/example_msg.py
python src/weathermart/examples/example_mtg.py
python src/weathermart/examples/example_metop.py