Title:

SITL Ground Loop: Automated Selection of MP Crossings in the SITL Workflow

The Magnetospheric Multiscale Mission (MMS) is a constellation of four NASA spacecraft orbiting the Earth and collecting data about magnetic reconnection, a phenomena that drives space weather such as the appearance of aurora near the Earth's poles. Certain regions and physical phenomena the spacecraft experience throughout their orbit, such as magnetopause crossings (MPs), are rich in scientific data that scientists on the ground would like to analyze. The spacecraft collect two levels of data: survey data that is quick to transmit to Earth but of too low resolution to perform science, and burst data that is slow to transmit but is fit for performing science. However, due to memory and data transmission limitations, the MMS spacecraft cannot feasibly downlink all burst data they collect. Instead, a team of volunteer scientists (scientists-in-the-loop/SITLs) look through survey data for regions of particular scientific interest (such as magnetopause crossings) to download in burst mode. To assist the SITLs in this task of selecting regions of interest, we trained a neural network to automatically classify and select regions of interest in survey data. Our initial model, the Ground Loop System (GLS), was trained to specifically identify magnetopause crossings. The existing automatic selection system (ABS), designed to accomplish the same task as the GLS, has been shown to underselect relevant regions, especially magnetopause crossings. The GLS has been successfully running at NASA’s Science Data Center, the home of the MMS mission and SITL team since October of 2019. The model has selected 78% of all SITL-selected magnetopause crossings, 44% more than the ABS. This indicates both its value to the SITL and its accuracy in selecting magnetopause crossings.

Name

Comment*

Comments are viewable only by submitter

Captcha*