The model climate in the AMIP experiment was very different from reality and that of most models compared: the west subtropical winds were strongly underestimated; the field of sea-level pressure was very distinct from reality, and so on. Now the reason for this is understood. The version of the model for the AMIP experiment had an error in the Matsuno scheme: for advection, horizontal velocities from timestep N were with the vertical velocity from N-1 timestep. The equations for temperature and humidity were written in the divergence form, but the velocities did not satisfy the continuity equation. In advection of these values there was false forcing, and this is the reason for a false cascade of energy from low to high frequency and its dissipation there. After removing this error and making some other changes (for example, including of horizontal diffusion of the fourth order that effectively vanishes very high space frequencies) the AMIP experiment was repeated. The new results indicate that there isn't much difference between our model's climate and real data.