Fig. 11.

WWI and incidence rate lag estimates, in days, with varying sample days for the WWTP of Reims ($n=1000$ subsampling experiments with random sampling of 60% of incidence rate curve). Default corresponds to the WWI as it is routinely processed with every single data point available. Other possibilities are obtained through resampling twice a week on specific weekdays. The Red dotted line indicates the zero offset level. The Blue dotted line is the median level over the 14 medians. As the difference in variance between the set of median time lags from the 7 WWTPs of Fig. 7 and the set of median time lags from the 14 two-days combinations displayed here is not statistically significant, subsampling could be one of the factors explaining the variability in optimal time lags between WWTPs shown in Fig. 7.