Solar Energetic Particle (SEP) events forecasting in the framework of SPRINTS
Abstract
We present the Space Radiation Intelligence System (SPRINTS) and its current capabilities to forecast solar energetic particles (SEP) using streamlined data-driven and machine-learning processes. SPRINTS is designed as a community ecosystem to ensure scientific transparency and can be deployed to any infrastructure environment. Its current data is comprised of GOES X-ray and proton data from 1986-present and ACE/DSCOVR from 1997-present in a Timescale database with APIs. The time-series database is supported by flare, SEP, and CME event catalogs as well as event associated catalogs such as flares associated with SEPs, CMEs and radio bursts. Within the framework of SPRINTS, these catalogs can be improved through scientific crowd-sourcing methods (e.g., versioning) thereby allowing critical alignment of both underlying data and event relationships. This forms a ML-ready dataset process for the community interested in establishing consistent train, test and validation and verification processes when building models to predict flares, SEPs, and CMEs. SPRINTS is coupled to the MagPy (i.e., MAG4) forecasting capability whereby it takes probabilistic forecast parameters of interest based on free-energy proxies for flare parameters including flare flux, fluence and peak ratio of the long and short X-ray channels required by the post-eruptive machine-learned models developed. This gives a continuous pre- and post-eruptive forecasting capability as new information (e.g., flare eruptions and CME kinematic) becomes available to the system. We will present initial models results for the Air Force Research Laboratory SEP forecast requirements (10 MeV @ 10 pfu, 10 MeV @40 pfu, 30 MeV @10 pfu, 50 MeV @10 pfu, 100 MeV @ 1 pfu) at 12 hour and 24 hour cadences. SPRINTS is currently providing forecasts in real-time through a REST API and has supporting dashboards near-real time forecasts, historical analysis, and event relationship analysis.
- Publication:
-
44th COSPAR Scientific Assembly. Held 16-24 July
- Pub Date:
- July 2022
- Bibcode:
- 2022cosp...44.1147E