Item

Abstract

Determining the sources of solar brightness variations1,2, often referred to as solar noise3, is important because solar noise limits the detection of solar oscillations3, is one of the drivers of the Earth’s climate system4,5 and is a prototype of stellar variability6,7—an important limiting factor for the detection of extrasolar planets. Here, we model the magnetic contribution to solar brightness variability using high-cadence8,9 observations from the Solar Dynamics Observatory (SDO) and the Spectral And Total Irradiance REconstruction (SATIRE)10,11 model. The brightness variations caused by the constantly evolving cellular granulation pattern on the solar surface were computed with the Max Planck Institute for Solar System Research (MPS)/University of Chicago Radiative Magnetohydrodynamics (MURaM)12 code. We found that the surface magnetic field and granulation can together precisely explain solar noise (that is, solar variability excluding oscillations) on timescales from minutes to decades, accounting for all timescales that have so far been resolved or covered by irradiance measurements. We demonstrate that no other sources of variability are required to explain the data. Recent measurements of Sun-like stars by the COnvection ROtation and planetary Transits (CoRoT)13 and Kepler14 missions uncovered brightness variations similar to that of the Sun, but with a much wider variety of patterns15. Our finding that solar brightness variations can be replicated in detail with just two well-known sources will greatly simplify future modelling of existing CoRoT and Kepler as well as anticipated Transiting Exoplanet Survey Satellite16 and PLAnetary Transits and Oscillations of stars (PLATO)17 data.

In this study, we employed and combined the newest observations and modelling techniques to reproduce total solar irradiance (TSI) variability with high precision at all timescales from minutes to decades. We computed the magnetic component of the TSI variability with the Spectral And Total Irradiance REconstruction for the satellite era (SATIRE-S) model10, which is one of the most successful and refined models of magnetically driven solar irradiance variability4,9. The high-cadence solar magnetograms and continuum images recorded by the Helioseismic and Magnetic Imager onboard the Solar Dynamics Observatory (SDO/HMI)18 allowed us to expand SATIRE-S calculations of magnetically driven TSI variability to timescales of as little as 12 min (see Methods). Our calculations of the granulation-driven TSI variability were based on recent three-dimensional simulations19 of convective gas currents both above and below the solar surface with the MPS/University of Chicago Radiative MHD (MURaM)12 code (see Methods). Therewith we are able to cover timescales of hours, critical for exoplanet detection studies but less understood until now20, with much better accuracy than was possible before21 (see Methods). We also stress that, unlike empirical approaches20,21, our calculations of the granulation were purely physics based, leaving us little freedom to change the properties of the granulation to match the observations.

Some statistical models20 of TSI variability account for larger convective structures, such as supergranules and mesogranules. However, there is currently no evidence that these structures have an intrinsic brightness contrast of non-magnetic origin22 so we did not include them in our modelling. We also refrained from including oscillations (which dominate TSI variability for periods of about 5 min and have been extensively used for helioseismology) in our modelling and focused on the TSI variations considered in helioseismology as noise.

While the amplitude of the granulation component of the TSI variability does not depend on time, the magnetic component is linked to the specific configuration of faculae (that is, bright concentrations of magnetic field23) and spots on the visible solar disk, and thus depends on solar magnetic activity. Therefore, we considered four intervals of the TSI record representing different levels of activity (Table 1). The three 1 month intervals at 2-min cadence allowed the study of the high-frequency component of the variability (Fig. 1), whereas the 19 year interval with a daily cadence was used to assess long-term changes.