Understanding the Geospace Phenomena Connected to Localized Perturbations in Earth’s Magnetic Field

University Of New Hampshire, Durham NH

Investigators

Abstract

During intervals of increased geomagnetic activity, increased currents in geospace can induce currents in the ground or in long, manmade conductors, such as power lines. These geomagnetically induced currents (GICs) can drive power outages and damage power components while also affecting pipelines and train systems. Developing the ability to predict GICs is important to protecting infrastructure and limiting the impact of geomagnetic storms on public safety and the economy. This project addresses GIC prediction by seeking to understand the connection between localized temporal changes in magnetic field measurements (dB/dt) and the magnetospheric and ionospheric phenomena causing them. This work will support the training of two graduate students that will be prepared to enter the STEM workforce with knowledge of space science, space weather, and machine learning and support the career of a woman PI leading the project. This project is jointly funded by the Magnetospheric Physics program, the Established Program to Stimulate Competitive Research (EPSCoR), and the Aeronomy program. Several studies have shown that peaks in dB/dt can be very localized, on the scales of hundreds of km. Thus, forecasting is needed at a localized level to provide power companies with actionable warnings. The current physics-based models used by the Space Weather Prediction Center lack the resolution needed to include physical phenomena at the spatial scales of the dB/dt peaks. Higher resolution models are being used for scientific studies, but are computationally expensive and take longer to run, making them more challenging to use for timely forecasting. An advantage of machine learning (ML) models is that once trained, the runtime to make predictions is significantly lower than physics based direct modeling. ML networks’ ability to model nonlinear relationships in datasets can allow us to better understand the phenomena that result in localized dB/dt through the use of model explainability techniques. The following science questions will be addressed: (1) What are the spatial characteristics of localized magnetic perturbations? (2) What magnetosphere and ionosphere phenomena correlate with localized magnetic perturbations and why? The first question will be tackled using two methods that take advantage of the NSF-funded SuperMAG database: by characterizing the Region to Specific Difference (a parameter that compares the value at one location to the average values within a defined region) and by interpolation with spherical elementary current systems, and then comparing the results. The second question will utilize explainable machine learning techniques to explore magnetosphere and ionosphere phenomena that correlate with the spatially localized perturbations, incorporating datasets from satellites (e.g., DMSP, AMPERE, TWINS, and THEMIS). The results of this work will improve our understanding of geomagnetically active intervals, magnetosphere-ionosphere coupling, and the causes of localized perturbations in the ground magnetic field. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.