Quantitative Comparison of Methods for Predicting the Arrival of Coronal Mass Ejections at Earth based on multi-view imaging

Abstract: We investigate the performance of six methods for predicting the CME time of arrival (ToA) and velocity at Earth using a sample of nine Earth-impacting CMEs between May 2010 and June 2011. The CMEs were tracked continuously from the Sun to near Earth in multi-viewpoint imaging data from STEREO SECCHI and SOHO LASCO. We use the Graduate Cylindrical Shell (GCS) model to estimate the three-dimensional direction and height of the CMEs in every image out to $\sim$200 R$\odot$. We fit the derived three-dimensional (deprojected) height and time data with six different methods to extrapolate the CME ToA and velocity at Earth. We compare the fitting results with the in situ data from the WIND spacecraft. We find that a simple linear fit after a height of 50$R\odot$ gives the best ToA with a total error $\pm$13 hours. For seven (78%) of the CMEs, we are able to predict the ToA to within $\pm$6 hours. These results are a full day improvement over past CME arrival time methods that only used SOHO LASCO data. We conclude that heliographic measurements, beyond the coronagraphic field of view, of the CME front made away from the Sun-Earth line are essential for accurate predictions of their time of arrival.