Forecasting of Solar Eruptions using Statistical Mechanics, Ensemble, and Bayesian Forecasting Methods

Period of Performance: 06/30/2014 - 03/30/2015


Phase 1 SBIR

Recipient Firm

Northwest Research Associates, Inc.
3380 Mitchell Lane
Boulder, CO 80301
Principal Investigator


The technological systems on which today's war-fighters depend, reflecting today's society, are vulnerable to the effects of sudden high-energy events from our Sun. Because the propagation speed for sudden increases in X-rays from a solar flare is the speed of light, it is necessary to truly forecast these events in order to mitigate their effects; the ionospheric response and influx of damaging radiation is almost as fast, with satellite charging and communications disruptions resulting. NorthWest Research Associates (NWRA) proposes to present an autonomous, flexible, verifiable forecasting system for flares, eruptions and energetic particle events from the Sun. The forecasting system uses presently available multi-wavelength data sources and state-of-the-art research in solar physics: key components include parameterizing observations of solar magnetic fields, surface- and sub-surface plasma flows, prior event history, and chromospheric/coronal pre-event energization. A forecast of solar activity is constructed using Multi-Variate Non-Parametric Discriminant Analysis that incorporates Bayesian statistics to produce a probabilistic forecast. We propose to tailor the results of prior innovative research to the unique USAF requirements, including full autonomy, verifiable performance and high flexibility regarding forecast timing and relative error costs.