Evaluating Solar Wind Forecast Using Magnetic Maps That Include Helioseismic Far-Side Information

TL;DR

The paper addresses the challenge of limited full-Sun magnetic-field observations by incorporating helioseismic far-side information into global magnetic maps via FARM. Using the WSA solar wind model and EUHFORIA, the study demonstrates that including far-side data can meaningfully improve ecliptic forecasts (up to ~50% in correlation) and reveals substantial 3D heliospheric differences driven by far-side active regions. The improvements are typically localized and depend on data currency and solar-cycle geometry, with ADAPT maps often performing best at ACE, while FARM yields notable gains at Earth and STEREO-A. The work highlights the potential and limitations of far-side information for space-weather forecasting and outlines concrete steps—such as tuning surface flux transport parameters and integrating more robust far-side detections—to further enhance predictive capabilities.

Abstract

To model the structure and dynamics of the heliosphere well enough for high-quality forecasting, it is essential to accurately estimate the global solar magnetic field used as inner boundary condition in solar wind models. However, our understanding of the photospheric magnetic field topology is inherently constrained by the limitation of systematically observing the Sun from only one vantage point, Earth. To address this challenge, we introduce global magnetic field maps that assimilate far-side active regions derived from helioseismology into solar wind modeling. Through a comparative analysis between the combined surface flux transport and helioseismic Far-side Active Region Model (FARM) magnetic maps and the base surface flux transport model without far-side active regions (SFTM), we assess the feasibility and efficacy of incorporating helioseismic far-side information in space weather forecasting. We are employing the Wang-Sheeley-Arge Solar Wind (WSA) model for statistical evaluation and leveraging the EUropean Heliospheric FOrecasting Information Asset (EUHFORIA), a three-dimensional heliospheric MHD model, to analyze a case study. Using the WSA model, we show that including far-side magnetic data improves solar wind forecasts for 2013-2014 by up to 50% in correlation and 3% in root mean square error and mean absolute error, especially near Earth and Solar TErrestrial RElations Observatory - Ahead (STEREO-A). Additionally, our 3D modeling shows significant localized differences in heliospheric structure that can be attributed to the presence or absence of active regions in the magnetic maps used as input boundaries. This highlights the importance of including far-side information to more accurately model and predict space weather effects caused by solar wind, solar transients, and geomagnetic disturbances.

Figures (6)

• Figure 1: Sample magnetic maps from 2 October 2013 that show the SFTM (before inclusion of far-side active regions), FARM (with far-side active regions included), and ADAPT (reference global magnetic maps without far-side sources). The position of Earth’s longitude is marked by a blue 'x'.
• Figure 2: Modeled and observed in situ solar wind velocities are shown for the following time periods: 16 May to 16 July 2013 for STEREO-A, 11 November 2013 to 11 January 2014 for STEREO-B, and 1 March to 1 May 2013 for ACE. The observed in situ velocity, is shown in black. Model predictions based on FARM and SFTM maps are represented in blue and green, respectively. The red shaded region indicates the range of solutions from the ADAPT ensemble. Gray shaded areas denote time intervals identified as CMEs in the Richardson and Cane ICME catalog (ACE) or L. Jian’s ICME lists (STEREO-A, -B). Red arrows highlight periods where FARM results visibly outperform SFTM results.
• Figure 3: Statistical evaluation of two years of WSA results from FARM, SFTM, and ADAPT maps for the location Earth (ACE), STEREO-A (STA) and STEREO-B (STB) respectively. Panel a and c show the RMSE and MAE between model results and in situ measurements, panel b depicts the Pearson correlation coefficient, and panel d the normalized DTW distance. The bars and error bars represent the $80\%$ and $95\%$ confidence intervals, respectively, while the horizontal bar indicates the median value.
• Figure 4: EUHFORIA model results using FARM (right column) and SFTM magnetic maps (left column) from 8 June 2011. From top to bottom the panels depict the photospheric magnetic field that was used as input to the coronal model, the solar wind at 0.1 AU which is the input to the heliospheric model and the solar wind results at 1 AU. The simulation results correspond to 15 June 2011 23:59 UT The blue, red, and green dots denote the projected locations of Earth, STEREO-A, and STEREO-B, respectively.
• Figure 5: EUHFORIA model results of the same run as Figure \ref{['fig:mhdruns']} but showing the equatorial plane (column a) and the meridional planes at the position of Earth (column b), at STEREO-A (column c) and STEREO-B (column d). The simulation results correspond to 15 June 2011 23:59 UT. FARM results are shown on the top row and SFTM results on the bottom row. The blue, red, and green dots denote the projected locations of Earth, STEREO-A, and STEREO-B, respectively. Note that the longitudes are given in Stonyhurst coordinates for clarity.