Fig. 2.

Model fit for Munich, Germany. Model simulation for the sampled parameter vector with the highest posterior probability compared with the observed data. In (A) only the case numbers reported by the Robert-Koch Institute and hospital usage for Munich are used for fitting, while in (B) seroprevalence data is also employed. The error bands show the range of plausible values for the observation, confirming that the noise models used are appropriate. In the bottom-right panels of (A, B), where the seroprevalence predicted by the model is plotted, the error bars are only shown at the observation times since the variance of each observation is linked to the number of total antibody tests carried out in each sub-batch.