Real time physics-based solar wind forecasts for SafeSpace

We present the solar wind forecast module implemented on the Sun - interplanetary space - Earth's magnetosphere chain of the H2020 SafeSpace project. The wind modelling pipeline, developed at the IRAP, performs real-time robust simulations (forward modelling) of the solar wind from the surface of the Sun up to the L1 point, from which magnetospheric indices are determined using a neural network. The pipeline effectively combines in a synergistic way different mature research models: determination of the background coronal magnetic field, computation of many individual solar wind acceleration profiles (1 to 90 solar radii), propagation across the heliosphere and formation of CIRs (up to 1 AU or more), estimation of synthetic diagnostics (white-light and EUV imaging, in-situ time-series) and comparison to observations and spacecraft measurements. Different magnotograms sources (WSO, SOLIS, GONG, ADAPT) can be combined, as well as coronal field reconstruction methods (PFSS, NLFFF), wind models (MULTI-VP), and heliospheric propagation models (CDPP/AMDA 1D MHD, EUHFORIA), magnetospheric indices neural network (ONERA). We tested and validated the wind forecasts against different in-situ observatories using a dynamic time-warping metric. We provide a web-based service that continuously supplies a full set of bulk physical parameters (wind speed, density, temperature, magnetic field, phase speeds) of the solar wind up to 6-7 days in advance, at a time cadence compatible with other space weather applications.