Author:

Abstract:

We developed COolfluid COroNa UnsTructured (COCONUT) - a novel global coronal model based on the COOLFluiD numerical code. The steady-state solution is determined by the inner boundary defined by magnetogram data in combination with fixed hydrodynamic values representative of the typical corona, while inside the numerical domain of our simulation, the corona is described by the set of ideal-MHD equations with gravity. The latter is solved with use of an implicit solver on an unstructured grid. Our code has passed a set of benchmark tests and proved its accuracy for simple dipole / quadrupole solutions as well as for a wide range of magnetograms, both during solar minima and solar maxima. With various numerical optimization techniques and an adaptive CFL step, we decreased the computation time while maintaining the high robustness and reliability. Finally, we coupled the obtained results with the heliospheric wind model of EUHFORIA 2.0 space weather forecast to show its forecast abilities. All this leads to an accurate MHD solution obtained within only a few hours of computation, which is crucial for space weather forecast systems. Here we present some of the numerically obtained results for selected magnetograms chosen to represent a variety of stages of the solar activity, from minimum to maximum, with each of them corresponding to a particular solar eclipse, to allow us the direct comparison of simulations with the observed coronal structures. Following the commonly used procedure, the input raw / original MDI and HMI magnetograms are pre-processed by projection on spherical harmonics and a selection of a maximum frequency for the reconstruction. This is equivalent to a smoothing of the map and results in removal of the small, intense magnetic structures on the solar surface. These are in fact numerically more challenging to capture, while their contribution to the overall structure of the solar wind at 0.1 AU and the largescale coronal magnetic fields has not been thoroughly investigated yet. With several maps and several levels of accuracy of reconstruction we address this problem and show the map resolution and pre-processing impact on accuracy of the numerical results. This is especially important for computationally challenging maximum-activity magnetograms which require significant pre-processing for stability and satisfactory computational speed. To verify our numerical results, we use a validation scheme proposed by Wagner et al. 2022 (from less to more sophisticated methods, i.e., visual classification, feature matching, streamer direction and width, brute force matching, topology classification). We investigate the distribution of magnetic structures as predicted by our simulations and compare them to the obtained coronal magnetic field topology. The detailed comparison with observations reveals that our model recreates relevant features such as the position and the shape of the streamers (by comparison with white-light images), the coronal holes (by comparison with EUV images) and the current sheet (by comparison with WSA model at 0.1 AU). We conclude that an unprecedented combination of accuracy, computation speed and robustness is accomplished at this stage, with possible improvements in the foreseeable future, such as inclusion of more physics or implementation of the adaptive mesh refinement technique. Our results also show that the final solution is still very sensitive to the map chosen and its pre-processing, especially for solar maximum-activity cases