Verification of high-speed solar wind stream forecasts using operational solar wind models
High-speed solar wind streams emanating from coronal holes are frequently impinging on the Earth's magnetosphere causing recurrent, medium-level geomagnetic storm activity. Modeling high-speed solar wind streams is thus an essential element of successful space weather forecasting. Here we evaluate high-speed stream forecasts made by the empirical solar wind forecast (ESWF) and the semiempirical Wang-Sheeley-Arge (WSA) model based on the in situ plasma measurements from the ACE spacecraft for the years 2011 to 2014. While the ESWF makes use of an empirical relation between the coronal hole area observed in Solar Dynamics Observatory (SDO)/Atmospheric Imaging Assembly (AIA) images and solar wind properties at the near-Earth environment, the WSA model establishes a link between properties of the open magnetic field lines extending from the photosphere to the corona and the background solar wind conditions. We found that both solar wind models are capable of predicting the large-scale features of the observed solar wind speed (root-mean-square error, RMSE $\approx$ 100 km/s) but tend to either overestimate (ESWF) or underestimate (WSA) the number of high-speed solar wind streams (threat score, TS $\approx$ 0.37). The predicted high-speed streams show typical uncertainties in the arrival time of about 1 day and uncertainties in the speed of about 100 km/s. General advantages and disadvantages of the investigated solar wind models are diagnosed and outlined.