Modelling non-radially propagating coronal mass ejections and forecasting the time of their arrival at Earth

We present the study of two solar eruptive events observed on December 7 2020 and October 28 2021. Both events were associated with full halo coronal mass ejections (CMEs) and flares. These events were chosen because they show a strong non-radial direction of propagation in the low corona and their main propagation direction observed in the inner heliosphere is not fully aligned with the Sun-Earth line. This characteristic makes them suitable for our study, which aims to inspect how the non-radial direction of propagation in the low corona affects the time of CMEs' arrival at Earth. We reconstructed the CMEs using SOHO/LASCO and STEREO/COR observations and modelled them with the 3D MHD model EUHFORIA and the cone model for CMEs. In order to compare the accuracy of forecasting the CME and the CME-driven shock arrival time at Earth obtained from different methods, we also used so-called type II bursts, radio signatures of associated shocks, to find the velocities of the CME-driven shocks and forecast the time of their arrival at Earth. Additionally, we estimated the CME arrival time using the 2D CME velocity obtained from the white light images. Our results show that the lowest accuracy of estimated CME Earth arrival times is found when the 2D CME velocity is used (time difference between observed and modelled arrival time, Delta t approximate to -29 h and -39 h, for the two studied events, respectively). The velocity of the type II radio bursts provides somewhat better - but still not very accurate - results (Delta t approximate to +21 h and -29 h, for the two studied events, respectively). Employing, as an input to EUHFORIA, the CME parameters obtained from the graduated cylindrical shell (GCS) fittings at consequently increasing heights, results in a strongly improved accuracy of the modelled CME and shock arrival time; Delta t changes from 20 h to 10 min in the case of the first event, and from 12 h to 30 min in the case of the second one. This improvement shows that when we increased the heights of the GCS reconstruction we accounted for the change in the propagation direction of the studied CMEs, which allowed us to accurately model the CME flank encounter at Earth. Our results show the great importance of the change in the direction of propagation of the CME in the low corona when modelling CMEs and estimating the time of their arrival at Earth.