Identifying Solar Wind Time Intervals at Mars: Comparing a physics-based algorithm with a machine learning approach

Authors: Kyle Webster (UCLA), Yingjuan Ma (UCLA), Hao Cao (UCLA)

There is no dedicated solar wind monitor at Mars like there is at Earth. However, Mars Atmosphere and Volatile Evolution (MAVEN) has been collecting solar wind data since its arrival in September 2014. MAVEN has an elliptical orbit trajectory and during many orbits it observes the solar wind at its apoapsis. In order to investigate how the Martian system responds to solar wind conditions, it is important to isolate the time intervals where MAVEN is measuring the upstream solar wind conditions. In this study, we compare the solar wind time intervals from two different methods. One method is from Halekas+ 2017 and it uses a physics-based algorithm to pick threshold values based on expected solar wind conditions. The other method is from the SHARP shock database and it uses a machine learning technique to automatically identify bow shock crossings (from Lalti+ 2022). Our goal is to compare these two methods and provide insights on the strengths and weaknesses of using a physics-based algorithm and a machine learning technique for selecting pristine solar wind intervals.