This project will investigate the formation and evolution of magnetic flux ropes in Earth's magnetotail. It will address four main questions: (1) Are flux ropes formed by simultaneous magnetic reconnection at multiple x-lines and if so what determines the number of reconnection sites? (2) How do flux ropes in the plasma sheet evolve with time? Do smaller ropes coalesce to form larger ones? Or can internal x-lines form inside a large flux rope to generate smaller, secondary ropes? (3) Is there a connection between the onset of the formation of a flux rope and the onset of a magnetic substorm? (4) Does the location of the near-Earth neutral line vary and if so what controls it? A significant part of the data analysis will be done using a new data mining tool. This part of the project will be done as a subaward to the developer of the data mining tool. The data mining tool will be used to automate the detection of flux ropes. The detailed structure and evolution of selected events will be done using a Grad-Shafranov reconstruction -- a 2-D analysis technique that assumes the axis of the flux rope is an axis of symmetry.
Most of the research for this project will be carried out by an early-career woman scientist. She will also be interacting with space scientists at the University of Michigan and at the University of Leicester in the UK. An innovative data mining technique will be used to construct a large database of flux rope events which will be made available to the space science community.
Hot plasma flowing outwards from the surface of the Sun permeates the solar system and is known as the solar wind. The Earth’s magnetic field carves out a cavity in the solar wind known as the magnetosphere, which acts to slow and deflect the solar wind around the planet. The solar wind is able to impart energy and momentum to the Earth’s system via a process known as magnetic reconnection, and the interaction between the two plasma regimes drives large scale circulation of plasma and magnetic field lines through the Earth’s magnetosphere. Magnetic reconnection allows solar wind plasma to enter the magnetosphere, and can lead to bright auroral displays as well as disrupting radio communications and GPS systems, causing increased drag on orbiting satellites, and inducing strong currents in the Earth’s ionosphere. The aim of this project was to apply innovative data-mining algorithms to magnetic field and plasma data taken by orbiting spacecraft in order to automatically identify the signatures of magnetic reconnection. The application of such techniques allows detailed analysis of many decades of data in a completely objective manner, which is not feasible using current techniques requiring ten minute intervals of data to be analysed by eye. Hundreds of examples of these reconnection signatures were used to train the data-mining algorithm to identify the correct signatures. This was an iterative process, requiring careful addition of examples, assessment of accuracy, and further training until sufficient accuracy was achieved. Further testing and data preparation was carried out and the data-mining tool is now able to receive many decades of data and correctly identify reconnection signatures over 95% of the time. The outcomes of this project are twofold. The successful development of the algorithm has provided a universal tool capable of application to a much broader range of data sets, in particular to all four years of MESSENGER data from Mercury, where reconnection signatures are observed every few seconds. Looking into the future, Magnetospheric Multiscale (MMS) is a NASA mission due to launch in 2015 and designed specifically to identify reconnection dynamics at the Earth. The application of this tool to MMS data will be an invaluable asset to the scientific community. The second outcome of this project has been that lists of these reconnection-related signatures are now available (upon request) to the scientific community. Scientific papers using these events to better understand the process of magnetic reconnection have been presented at international conferences by both the Co-PI (Dr. Imber) and a PhD student at the University of Michigan (Nicole Pothier), and a Michigan summer intern (Mojtaba Akhavantafti) employed on the grant gained valuable experience of both data-analysis and machine learning techniques.
