Number density as a function of pseudo R and z (see text) in the selected data set. Gaia data N_D (top left) and our final model N_S (bottom left). Difference between data and model relative to the Poisson noise (N_D–N_S)/sqrt(N_D) (top right). Relative difference between counts (N_D − N_S)/N_D (bottom right). The relative difference shows that the model accuracy is generally better than 10%, apart from the regions showing the vertical wave (see text).