Authors: Syed Raza (The University of Alabama in Huntsville), Talwinder Singh (Georgia State University), Nikolai Pogorelov (The University of Alabama in Huntsville)

Coronal Mass Ejections (CMEs) are among the most violent events in our solar system. Earth-directed CMEs can cause geomagnetic storms, provided that they carry a negative out-of-ecliptic magnetic field component. Many models of varying levels of complexity have been developed to predict CME Time of Arrival (TOA) at Earth. However, despite these efforts, CME TOA errors remain over 10 hours. In this study, we use a hybrid approach that combines ensemble magnetohydrodynamic (MHD) simulations of flux-rope-based CMEs propagating through a data-driven Solar Wind (SW) background with Machine Learning (ML) algorithms to improve CME TOAs. To simulate CMEs, we use the Multi-Scale Fluid-Kinetic Simulation Suite (MS-FLUKSS, Pogorelov et al., 2014). Graduated Cylindrical Shell (GCS) is a popular model used to determine the CME properties through multiple viewpoint coronagraph images. We utilize the GCS model to perform ensemble simulations of 15 CME events, for which the initial CME speed is varied to create 21 ensemble members. To improve TOA for each CME, ML methods are used to compare the results of ensemble modeling with the data  from the Heliospheric Imagers (HI) aboard STEREO-A and STEREO-B.