small and it indicated that the simulated values fit the observed
data well. And the simulated values estimated with mean k fit the
observed data the best. Thus, pollutant load estimation and according
water quality management could be carried out based on
the model.
The normality assumption of the model residuals used in the
Bayesian framework was assessed via QQ plots (Fig. 5).
Although some outlier values are evident in the tails of the
distribution, these figures demonstrate that the model residuals are
modeled appropriately via a Gaussian distribution.