Elements: Comprehensive Time Series Data?Analytics for the Prediction of Solar Flares and Eruptions

点击翻译按钮获取中文摘要

Solar flares are some of the largest explosive events in our solar system. Together with the accompanying eruptions, solar flares have the potential to disrupt the technology we rely on, such as GPS, radars, high-frequency radio communications between aircraft and air traffic control, communication technology that relies on satellites (such as cell phones and Internet), and electricity grid distribution networks. This project aims to, first, improve scientists' understanding of the time-dependent physical and statistical behavior of solar active regions to the point that we can exploit their observations to predict whether and when they will flare, or erupt, in general, and, second, to enable scientists worldwide to perform comparative, reproducible, and data-driven studies on the prediction of solar explosive events. This project, together with its advanced solar flare prediction software infrastructure, will strengthen our nation's efforts to mitigate the potentially catastrophic impact of solar eruptions. Moreover, by helping achieve reliable forecasts of the timing, location, and magnitude of solar flares (already considered as natural disasters), this project will also contribute to mitigating the impact on other types of infrastructure, such as satellites (GPS, Internet, satellite communications), that are critical to not only our national defense but also a broad range of sectors, such as enterprise operations (e.g., oil-drilling) and communications. This award will also invest in the development of a highly educated, diverse, globally competitive STEM workforce trained in an interdisciplinary environment through the educational and research efforts at Georgia State University. The goal of this project is to improve the scientists' understanding of eruptive solar events by re-shaping the state-of-the-art on data mining techniques. This project will build cyberinfrastructure for data-driven scientific research on complex multivariate time-series data sets. This includes public releases of comprehensive benchmark data sets ideal for data mining research on multivariate time series classification, regression, and clustering, as well as open-source software for these applications (i.e., the public release of pre-trained models). While the domain area is focused on solar physics, this software and data sets can benefit other domains that involve event tracking and mining of spatiotemporal trajectories (e.g., security and healthcare applications of monitoring movement, traffic and weather data analyses, and business predictions). The project will also advance research in Space Weather forecasting through the delivery of reproducible data-driven solar flare predictions, inspired by new research directions (most notably, time series analysis at an unprecedented level) and spearheaded by the data-driven, physically interpretable machine learning models. In this direction, the projects will produce publicly available flare forecasting big data benchmark data sets aligning with recommendations by the National Science and Technology Council and being useful to both Data Science and Space Weather communities. Finally, this project will advance Computer Science research in data mining and information retrieval areas through the above cutting-edge work on time series data analysis.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.