Periodic Reporting for period 1 - SYCOS (Solar cYcle prediCtion tool using sOlar internal oscillations (SYCOS))
Reporting period: 2023-02-01 to 2024-10-31
The overall objective of the SYCOS project was to advance space weather forecasting by using an innovative new method based on the analysis of solar oscillations. We proposed to use a special class of these oscillations, namely the surface gravity (or (f)undamental) mode, captured in the upper part of the convection zone, to reveal the geometry and strength of the sub-surface magnetic fields. Changes in these sub-surface magnetic fields potentially enable us to predict the formation of solar active regions (ARs) - the source regions of eruptive events - on scales of the order of a day before their emergence. This possibility would greatly enhance our capabilities of mitigating the imminent threats.
The project was based on two pillars: the theoretical investigation using world-leading global magnetoconvection simulations of the solar dynamo, developed in the ERC CoG “UniSDyn” (Käpylä et al, 2012, 2016), and the observational evidence outlined in Singh et al. (2016). The simulation efforts have revealed that the surface gravity mode becomes affected by the subsurface magnetic field, resembling an AR-like, bipolar, magnetic field configuration, by an enhancement of its energy at high spatial wavenumbers (Singh et al., 2021). The observational efforts required an upgrade of the semi-automated high-throughput pipeline to extract the f-mode energy for different types of regions, described in Korpi-Lagg et al. (2022).
Major challenges in the project were the demanding theoretical implementation in the magnetoconvection simulations and the calibration of the f-mode data, both setting very high demands on computer resources. Especially the latter, the calibration of the f-mode data, turned out to be more complicated than originally estimated, and a completely new calibration concept needed to be developed. The new concept was successfully applied to several AR emergences, and hindcasts the potential to become a powerful tool not only for AR emergence forecasting, but for a broader analysis of the f-mode (and p-mode) oscillations beyond the simple determination of its strength, revealing details about the fundamental processes in AR formation.
The project was based on two pillars: the theoretical investigation using world-leading global magnetoconvection simulations of the solar dynamo, developed in the ERC CoG “UniSDyn” (Käpylä et al, 2012, 2016), and the observational evidence outlined in Singh et al. (2016). The simulation efforts have revealed that the surface gravity mode becomes affected by the subsurface magnetic field, resembling an AR-like, bipolar, magnetic field configuration, by an enhancement of its energy at high spatial wavenumbers (Singh et al., 2021). The observational efforts required an upgrade of the semi-automated high-throughput pipeline to extract the f-mode energy for different types of regions, described in Korpi-Lagg et al. (2022).
Major challenges in the project were the demanding theoretical implementation in the magnetoconvection simulations and the calibration of the f-mode data, both setting very high demands on computer resources. Especially the latter, the calibration of the f-mode data, turned out to be more complicated than originally estimated, and a completely new calibration concept needed to be developed. The new concept was successfully applied to several AR emergences, and hindcasts the potential to become a powerful tool not only for AR emergence forecasting, but for a broader analysis of the f-mode (and p-mode) oscillations beyond the simple determination of its strength, revealing details about the fundamental processes in AR formation.
The observational aspect of Task 1, the finalization of the quiet-Sun pipeline, could be completed according to schedule. It revealed an interesting directional dependence of the f-mode power in the quiet Sun over the solar cycle, presented in several meetings and conferences (see figures “cycle-long latitudinal dependence of the poloidal/toroidal f-mode power”). However, the extension of this quiet-sun data product to produce heat maps of the f-mode power for the whole Sun, irrespective of its activity level, turned out to be very sensitive to an accurate calibration procedure. The fitting methods described in Singh et al. (2016) and Waidele et al. (2022) lacked the required accuracy, and a new method based on the computation of averaged ring-diagrams had to be developed. This method relies on the calculation of so-called “flat fields”, i.e. average ring-diagrams over the period of one to two Carrington rotations, requiring an accurate reprojection of the HMI data to a fixed grid irrespective of the changing orbital geometry. This step was implemented successfully, the source code was made publicly available on our f-mode gitlab repository. Although this additional complication slowed down the process of producing standard heat maps of the Sun (see example in Figure “F-mode heatmap”), it opens a new diagnostic window by allowing not only the f-mode strength to be analyzed in detail, but also changes in the spatial and temporal frequency domain.
The numerical work related to the project concerned with developing more realistic magnetoconvection simulation setups, where the magnetic field would more closely resemble an emerging active region, and turbulence and dynamo action would be driven by convection. All these improvements were achieved, and also the core computational engine of the numerical method was replaced with an efficient GPU-accelerated library Astaroth (Pekkilä et al., 2022). This enabled tenfold speed-up in the computational work, and allowed us to study f and p modes in better accuracy and improved physics included. The theoretical expectations were confirmed in these setups, and furthermore we were able to study the effect of various field geometries on the oscillation modes (Warnecke et al., in preparation). Combining these results with observations, it is now possible to deduce information of the strength and orientation of the magnetic field in the subsurface layers.
The numerical work related to the project concerned with developing more realistic magnetoconvection simulation setups, where the magnetic field would more closely resemble an emerging active region, and turbulence and dynamo action would be driven by convection. All these improvements were achieved, and also the core computational engine of the numerical method was replaced with an efficient GPU-accelerated library Astaroth (Pekkilä et al., 2022). This enabled tenfold speed-up in the computational work, and allowed us to study f and p modes in better accuracy and improved physics included. The theoretical expectations were confirmed in these setups, and furthermore we were able to study the effect of various field geometries on the oscillation modes (Warnecke et al., in preparation). Combining these results with observations, it is now possible to deduce information of the strength and orientation of the magnetic field in the subsurface layers.
Although the initial goal of the project, the production of daily heat maps of the Sun, could not be fully achieved, we could produce sample heat maps demonstrating the power of the developed method. The encountered difficulties resulted in a broadening of the applicability of our research not only for the forecasting, but also for the deepening of the understanding on how ARs are formed already in subsurface layers before their appearance on the solar surface, with potential new insights on the fundamental open questions in solar physics related to the dynamo mechanism(s) acting in the solar interior.
The magnetoconvection simulations produced during this project are unprecedented in resolution and physics included to study magnetic effects on solar oscillations. The simulation data forms the basis of training neural networks for predicting the emergence of active regions.
The magnetoconvection simulations produced during this project are unprecedented in resolution and physics included to study magnetic effects on solar oscillations. The simulation data forms the basis of training neural networks for predicting the emergence of active regions.