Toward a Next Generation Particle Precipitation Model: Mesoscale Prediction Through Machine Learning (a Case Study and Framework for Progress). Issue 6 (8th June 2021)

• Record Type: Journal Article
• Authors: McGranaghan, Ryan M.; Ziegler, Jack; Bloch, Téo; Hatch, Spencer; Camporeale, Enrico; Lynch, Kristina; Owens, Mathew; Gjerloev, Jesper; Zhang, Binzheng; Skone, Susan
• Is Part Of: Space weather. Volume 19:Issue 6(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-06-08
• Subjects: data science -- evaluation -- machine learning -- magnetosphere‐ionosphere coupling -- particle precipitation -- space weather; Space environment -- Periodicals; 551.509992
• DOI: 10.1029/2020SW002684
• Languages: English
• ISSNs: 1542-7390

Abstract

Abstract: We advance the modeling capability of electron particle precipitation from the magnetosphere to the ionosphere through a new database and use of machine learning (ML) tools to gain utility from those data. We have compiled, curated, analyzed, and made available a new and more capable database of particle precipitation data that includes 51 satellite years of Defense Meteorological Satellite Program (DMSP) observations temporally aligned with solar wind and geomagnetic activity data. The new total electron energy flux particle precipitation nowcast model, a neural network called PrecipNet, takes advantage of increased expressive power afforded by ML approaches to appropriately utilize diverse information from the solar wind and geomagnetic activity and, importantly, their time histories. With a more capable representation of the organizing parameters and the target electron energy flux observations, PrecipNet achieves a >50% reduction in errors from a current state‐of‐the‐art model oval variation, assessment, tracking, intensity, and online nowcasting (OVATION Prime), better captures the dynamic changes of the auroral flux, and provides evidence that it can capably reconstruct mesoscale phenomena. We create and apply a new framework for space weather model evaluation that culminates previous guidance from across the solar‐terrestrial research community. The research approach and results are representative of the "new frontier" of space weather research at the intersection of traditional and data science‐driven discovery and provides a foundation for future efforts. Plain Language Summary: Space weather is the impact of solar energy on society and a key to understanding it is the way that regions of space between the Sun and the Earth's surface are connected. One of the most important and most challenging to model are the way that energy is carried into the upper atmosphere (100–1, 000 km altitude). Particles moving along magnetic field lines "precipitate" into this region, carrying energy and momentum which drive space weather. We have produced a new model, using machine learning (ML), that better captures the dynamics of this precipitation from a large volume of data. Machine learning models, carefully evaluated, are capable of better representing nonlinear relationships than simpler approaches. We reveal our approach to using ML for space weather and provide a new framework to understand these models. Key Points: We utilize a data‐driven organization of input parameters to produce a new total electron energy flux nowcast model (see Figure 4) Extended input features and higher expressive power provided by machine learning (ML) approach yields more capable mesoscale and peak flux specification (see Figure 7) and an overall reduction in specification errors compared with the state‐of‐the‐art (see Table 1) A framework for evaluation of ML (and any model) in geospace is suggested, building on momentum in the community (see Section 6)