Satellite Consumer

Download and convert satellite data for use in ML pipelines

Satellite data is a valuable resource for training machine learning models. Forecasting renewable generation requires knowledge of the weather conditions, and those weather conditions can be inferred and enriched using satellite data.

EUMETSAT provide a range of satellite data products, which are easily available in NAT image format. In order to improve its accessibility for training models, this consumer processes downloaded data into the Zarr format.

Note

This repo is in early development and so will undergo rapid changes. Breaking changes may occur in the CLI and the API without warning.

Installation

Install using the container image:

$ docker pull ghcr.io/openclimatefix/satellite-consumer

or, if you prefer a CLI:

$ pip install git+https://github.com/openclimatefix/satellite-consumer.git

This will put the sat-consumer-cli command in your virtual environments bin directory.

Example usage

$ docker run \
    -e SATCONS_COMMAND=consume \
    -e SATCONS_SATELLITE=rss \
    -e EUMETSAT_CONSUMER_KEY=<your-key> \
    -e EUMETSAT_CONSUMER_SECRET=<your-secret> \
    -v $(pwd)/work:/work \
    ghcr.io/openclimatefix/satellite-consumer

This will download the latest available data for the rss satellite and store it in the /work directory. For a description of all the possible configuration options, see Documentation.

Documentation

The satellite consumer provides a number of commands for different logical processing of raw data. These commands (and their options) can be seen when using the cli entrypoint:

$ sat-consumer-cli --help

When running the satellite consumer using the environment entrypoint (as in the docker container)

$ sat-consumer

the command is chosen via an environment variable. There are also a number of common configuration options that are shared between all commands:

Variable	Default	Description
SATCONS_COMMAND		The command to run (consume/merge).
SATCONS_SATELLITE		The satellite to consume data from.
SATCONS_WORKDIR	/mnt/disks/sat	The working directory. In the container, this is set to /work for easy mounting.
SATCONS_HRV	false	Whether to download the HRV channel.
EUMETSAT_CONSUMER_KEY		The EUMETSAT consumer key.
EUMETSAT_CONSUMER_SECRET		The EUMETSAT consumer secret.

Each command then has its own set of configuration options:

Consume:

Downloads scans for a given time and window into a zarr store in the given working directory.

Variable	Default	Description
SATCONS_TIME		The time to consume data for (when using the consume command). Leave unset to download latest available.
SATCONS_WINDOW_MINS	0	The time window to consume data for in minutes (defaults to a single scan).
SATCONS_WINDOW_MONTHS	0	The number of months to consume data for (takes precedence over SATCONS_WINDOW_MINS).
SATCONS_VALIDATE	false	Whether to validate the downloaded data.
SATCONS_RESCALE	false	Whether to rescale the downloaded data to the unit interval.
SATCONS_NUM_WORKERS	1	The number of workers to use for processing.

Merge:

Merges consumed stores for a given time window into a single store in the working directory.

Variable	Default	Description
SATCONS_SATELLITE		The satellite to consume data from.
SATCONS_WINDOW_MINS	210	The time window to merge data for.
SATCONS_CONSUME_MISSING	false	Whether to consume missing data.

FAQ

How do I add a new satellite to the consumer?

Currently the consumer is built to the specific data requirements of Open Climate Fix. However, adding a new satellite in the from EUMETSAT shouldn't be too hard, provided it uses the same seviri_l1b_native format and sensor channels - just update the available satellites in config.py.

Development

OCF recommends using uv for managing your virtual environments.

$ git clone [email protected]:openclimatefix/satellite-consumer.git
$ cd satellite-consumer
$ uv sync

Running the CLI

The python package contains a CLI entrypoint for ease of use when developing, which is available to your shell via the sat-consumer-cli command, assuming you have built the project in a virtual environment, and activated it.

Linting and static type checking

This project uses MyPy for static type checking and Ruff for linting. Installing the development dependencies makes them available in your virtual environment.

Use them via:

$ python -m mypy .
$ python -m ruff check .

Be sure to do this periodically while developing to catch any errors early and prevent headaches with the CI pipeline. It may seem like a hassle at first, but it prevents accidental creation of a whole suite of bugs.

Running the test suite

There are some additional dependencies to be installed for running the tests, be sure to pass --extra=dev to the pip install -e . command when creating your virtualenv (uv sync includes the development dependencies by default, so uv users can ignore this!).

Run the unit tests with:

$ python -m unittest discover -s src/satellite_consumer -p "test_*.py"

Note

If you have created your virtual environment using uv, the above can be run via the Makefile, using make typecheck, make lint, and make test respectively.

Further reading

On the directory structure:

• The official PyPA discussion on "source" and "flat" layouts.

---

Contributing and community

• PR's are welcome! See the Organisation Profile for details on contributing
• Find out about our other projects in the here
• Check out the OCF blog for updates
• Follow OCF on LinkedIn

Contributors

---

Part of the Open Climate Fix community.