Polarimetric Tomography Applied to Synthetic Multi-Spacecraft White-Light Images: Observing Coronal Mass Ejections in 3D

Abstract

A discrete tomography method has been developed that is able to reconstruct three-dimensional coronal mass ejection (CME) density structure. We test the method by producing synthetic coronagraph imagery for three CME events using the CORona--HELiosphere (CORHEL) model. We combine images from different numbers of observing spacecraft, ranging from three to seven, and we perform the method separately using polarimetric and non-polarimetric reconstructions, as a means to test their relative effectiveness. In all cases, we show that increasing the number of observing spacecraft reduces the mean relative absolute error (MRAE) between the simulated and reconstructed density. Furthermore, the MRAE is generally lower when using polarimetric reconstructions compared to non-polarimetric reconstructions. Methods applied to localise the CME front work well for all spacecraft configurations, and are improved when using polarimetric, over non-polarimetric, reconstructions. The presence of a CME front in the simulations can be identified with an accuracy of $(72\pm9)\%$, $(70\pm8)\%$ and $(52\pm12)\%$ for CME1, CME2 and CME3 via polarimetric reconstructions using only three spacecraft at L1, L4 and L5. The radial position of the CME front can be constrained to a high level of precision when using polarimetric reconstructions using the same three spacecraft; $0.003\pm0.004$\,au, $0.004\pm0.005$\,au and $0.005\pm0.004$\,au for CME1, CME2 and CME3, respectively. We expect that at least four spacecraft are required in order to derive accurate information about 3D CME structure. We find no strong evidence of improvement when including out-of-ecliptic observers, but that their inclusion increases the volume of space within which the inversion can be performed.

Figures (12)

• Figure 1: Sketch illustrating the placement of the synthetic spacecraft placed through the MHD simulation data cubes and two of the configurations considered in this work -- see Table \ref{['tab:orbit_configs']} for a full list. (a) The '4sc' configuration, with observers placed at L1, L4, L5, and a polar orbit P1. (b) The '6sc' configuration, featuring three pairs of spacecraft in a ring formation at 1 au from the Sun.
• Figure 2: Overview of the three CMEs simulated with CORHEL-CME, showing (a) CME1, (b) CME2, and (c) CME3. For each event, the top panel shows selected field lines of the pre-eruptive (or at $t=0$ in the simulation) RBSL flux rope over the source AR on the photosphere, the middle panel displays CME field lines 15 minutes after the eruption, and the bottom panel features a slice of the CME scaled density along the solar equatorial plane at the time when the front reaches approximately 15 $R_{\odot}$ (the plot covers the full COR domain, i.e. up to 30 $R_{\odot}$).
• Figure 3: Synthetic total-brightness coronagraph images of the 29/11/2020 event (CME2) at $t=28$ for each of the nine spacecraft. Each image has been processed to subtract the background and reduced to $128\times128$ pixels with a 4$R_\odot$ occulter applied to obscure pixels near the Sun.
• Figure 4: Processed images and the corresponding solutions and residuals from tomography applied to the 29/11/2020 event (CME 2) at $t=28$ using six observing spacecraft. Each block of six images shows (top left) processed image used as input to tomography, (top middle) image produced from tomographic density solution, (top right) relative absolute error between the simulation and solution images, (bottom left) degree of polarisation in the simulation images, (bottom middle) degree of polarisation from the tomographic density solution, (bottom right) relative absolute error between the simulation and solution degree of polarisation.
• Figure 5: Two-dimensional histograms comparing radiance values in every pixel in the simulated and solution images, for each CME at every time-step. (a) shows results for CME1, (b) for CME2 and (c) for CME3. In each group, the top row represents solutions using tb measurements only, and the bottom row represents solutions using a combination of pb and tb. Each column represents each of the five different spacecraft configurations, labelled above the panels.The MRAE is printed in each case, as is the total number of pixels contributing to each histogram, $n$. The identity line is shown in red. The count number in each group of histograms is scaled logarithmically to emphasise low values.