Solar wind propagation delay predictions between L1 and Earth based on machine learning

Kurzfassung

GNSS positioning errors, spacecraft operations failures and power outages potentially originate from space weather in general and the solar wind interaction with the geomagnetic field in particular. Depending on the solar wind speed, information from L1 solar wind monitor spacecraft only give a lead time to take safety measures between 20 and 90 minutes. This very short lead time requires end users to have the most reliable warnings when potential impacts will actually occur. In this study a machine learning algorithm is presented that is suitable to predict the solar wind propagation delay between Lagrangian point L1 and the Earth. This work introduces the proposed algorithm and investigates its operational applicability. The propagation delay is measured from interplanetary shocks passing the Advanced Composition Explorer (ACE) first and their sudden commencements within the magnetosphere later, as recorded by ground-based magnetometers. Overall 380 interplanetary shocks with data ranging from 1998 to 2018 builds up the database that is used to train the machine learning model at the time of the interplanetary shocks. The training of the machine learning model uses ACE data which contains all three components of solar wind speed and the position of the spacecraft. The performance assessment of the machine learning model is examined on the basis of a 10-fold cross-validation. The major advantage of the machine learning approach is its simplicity when it comes to its application. After training, values for the six features have to be fed into the algorithms only and the evaluation of the propagation delay can be continuous. The machine learning model is validated against a simple convective solar wind propagation delay model and a vector method that takes the normal vector of the shock front into account. Additionally, this work aims at providing greater interpretability of the machine learning approach to predicting the propagation delay. For that purpose also drop-column feature importance and Shapley values of the trained algorithm are presented and analysed.