Understanding the capability of a simulation to reproduce observed features is a requirement for its use in operational space weather forecasting. We compare statistically ionospheric seasonal variations in the Grand Unified Magnetosphere-Ionosphere Coupling Simulation (GUMICS-4) global magnetohydrodynamic (MHD) model with measurements. The GUMICS-4 data consists of a set of runs that was fed with real solar wind measurements and covers the period of one year. Ionospheric convection measurements arefrom the SuperDARN radars, and electric currents are derived from the magnetic field measured by the CHAMP satellite. Auroral electrojet indices are used to examine the disturbance magnetic field on ground. The signatures of electrodynamic coupling between the magnetosphere and ionosphere extend to lower latitudes in GUMICS-4 than in observations, and key features of the auroral ovals — the Region 2 field-aligned currents, electrojets, Harang discontinuity, and ring of enhanced conductivity— are not properly reproduced. The ground magnetic field is even at best about 5 times weaker than measurements, which can be a problem for forecasting geomagnetically induced currents. According to the measurements, the ionospheric electrostatic potential does not change significantly from winter to summer but field-aligned currents enhance, whereas in GUMICS-4, the electrostatic potential weakens from winter to summer but field-aligned currents do not change. This could be a consequence of the missing Region 2 currents: the Region 1 current has to close with itself across the polar cap, which makes it sensitive to solar UV conductivity. Precipitation energy and conductance peak amplitudes in GUMICS-4 agree with observations.