Forecasting the Solar Wind with Sequential Monte Carlo Assimilation of Satellite Data

Space weather affects life on Earth and in outer space. Human technologies are affected by coronal mass ejections and similar outbursts of solar activities. Accurate prediction of the solar wind and its polarity can help understand the Sun and its dynamic environment. The Wang-Sheeley-Arge (WSA) phenomenological model of the coronal magnetic field can estimate solar-wind speed and interplanetary magnetic field polarity in the inner heliosphere by using photospheric magnetic field maps. WSA has historically used two parameters, the source surface and interface radii, to tune its predictions. In this talk, I describe how our team used sequential Monte Carlo, also called particle filtering, in the assimilation of satellite data to adjust the values of these radii. Adaptive optimization, applied to week-long timescales across several months of historical data, yielded approximately double predictive performance. In addition to improved forecasts, this statistical study highlights challenges in parameter estimation for the nearest and most-observed solar-mass object: the Sun.

*Approved for Los Alamos Unlimited Release: LA-UR-21-21918.