Developing a Data Analysis Tool to Predict Radar Data from Magnetometer Data during the Farley Buneman Instability

During the Farley Buneman instability, large amounts of current are flowing in the ionosphere. Despite being mainly observed through radar campaigns, these currents generate large magnetic fields which should be noticeable on surface magnetometers. By developing a data analysis program using machine learning, we can measure the relationship between radar and magnetometer data and expand the overall dataset of observed accounts of the Farley Buneman instability.
The architecture of this model will incorporate concepts from atmospheric and space sciences as well as machine learning methods like polynomial regressors and support vector machines. This multi-stage model will be used to accurately map incoming magnetometer data to the corresponding radar data. Additionally, the model will predict particle velocities and temperatures to be able to expand the dataset of radar stations using readily available magnetometer data. Additionally, to achieve higher accuracies, hyperparameters will be tuned and set using relevant ionospheric plasma physics concepts.
This experiment can open new research possibilities by being able to better utilize existing magnetometer data to find historical accounts of the FB instability and through predicting radar data from magnetometer data we can expand the working dataset we have for magnetometer stations around the world.