rcpann 0.0.6

Ring Current Proton Artifician Neural Network Model

Modeling ring current proton distribution using artificial neural network (multi-layer perceptron).

This model is based on RBSP/RBSPICE measurements from 2013-2018. The RBSPICE measures proton flux at 14 energy channels from 45 keV to 598 keV, and this model provides proton spin-averaged flux in those 14 energy channels.

The 'lws_training_display_v1.0.ipynb' demonstrates the training of the model, as well as the illustration of comparision between test dataset and predictions over long-term. The processed data are stored in 'rcpann_model_files.zip'. Two additional subprograms are needed: 'time_double.string.py' which converts time formats and 'test_correlation.py' which plot comparison between test data and predictions.

To use this model, the users basically use 'rcpann_model_v1.0.ipynb', which load the model stored in 'rcpann_model_files.zip', and output the results (the log10 of flux in units of keV^-1c m^-2 s^-1). We also shows 4 examples of how to use this model. To simulate proton flux in a specific event, the uses also need the geomagnetic data stored in 'geomag_2012_2018.csv'.

The data for the training and modeling are available at https://doi.org/10.5281/zenodo.7651736.

Please contact [e-mail] for support.