Skip to main content

Training tool for SUVI thematic map machine learning

Project description

suvi-trainer

CodeFactor PyPI version

The Solar Ultraviolet Imager (SUVI) aboard the National Oceanic and Atmospheric Administration's Geostationary Operational Environmental R-Series satellites is used in machine learning applications to create thematic maps, images showing where different features are on the Sun. This tool allows domain experts to load images, manipulate them, and create labeled maps, which are used in training solar image classifiers.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. At the moment, it is advised to run this code in a fresh virtual environment with Python 3.5. Installation should only require running the setup.py script.

Installing

pip3 install suvitrainer

You will need to download and edit the configuration file to include your name and the upload password. Please contact me for the password and further information.

Please note that the extra scripts may require packages that are not automatically installed. They are auxiliary and not fundamental for the annotation tool to run. Everything needed for run_suvitrainer.py and the main annotation tool should be automatically installed. If you want to use an auxiliary script download the repo.

Running

The run_suvitrainer.py script should provide all needed functionality for the average user. It takes a couple optional arguments: verbosity and dates. The verbosity, -v or --verbose argument will print helpful status information while running. The date option allows three methods of specifying which date to run on: simply a date string (2018-08-05T17:52), a path to a local file that contains a list of date strings where each one is on a different line, or a url to an online list of dates. The default is to pull using the url for dates.txt stored in this repository. This is preferred to create a large curated data-set with some repeats for validation.

The list of options for input images:

  • halpha (from GONG)
  • any of the AIA EUV channels listed as aia-[WAVELENGTH] with the wavelength in angstroms, e.g. aia-131
  • any of the SUVI EUV L1B channels listed as products, e.g. suvi-l1b-fe131 (see the FTP site for a list)
  • any locally saved SUVI l2 composite images (only available internally to NOAA employees at this time), e.g. suvi-l2-ci094

For more detailed usage, see the user guide.

Data

The output of training is saved as a FITS file which is later converted into a labeled png. Labeled data will be available shortly. A couple examples are available here. This labeled data is then used in machine learning classification of solar images.

Contributing

Please contact me (email or just through pull/issue requests) for updates. If you're interested in labeling images or would like access to the labeled database, please contact me through email.

Authors

  • J. Marcus Hughes - Initial work

License

This project is licensed under the MIT License - see the LICENSE file for details

Acknowledgments

  • Guidance from Dan Seaton, Jon Darnel, and Vicki Hsu

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
suvitrainer-1.1.9-py3-none-any.whl (18.0 MB) Copy SHA256 hash SHA256 Wheel py3
suvitrainer-1.1.9.tar.gz (18.0 MB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page