Why doesn’t model improvement result in better forecast: the 3D drag-based model for CME propagation example

CMEs interact with the solar wind and heliospheric magnetic field which influence their propagation, expansion as well as internal magnetic structure. We understand these processes on a global level, however we are still lacking a detailed qualitative and quantitative understanding of the CME evolution on a level that could result in a reliable forecast. Our limitations are influenced by uncertainties in measurements as well as uncertainties in associating remote-to-insitu events and observation-to-model comparison. These uncertainties affect not only inputs to our CME propagation models, but also evaluation of the outputs. As an example, we present the newly developed adaptation of the widely used drag-based model (DBM, Vrsnak et al., 2013) for 3D geometry, which should in theory provide more accurate forecast. However, we show that for an arbitrary evaluation sample it does not provide significantly different results from its 2D counterpart.