EPR was used in a similar manner to previously to attempt to derive an expression for a daily regeneration rate based on all flushing actions and their repeats. An extensive data set was available with 624 individual flushing actions available in theory (before issues of missing data had to be dealt with) with accumulation rates calculable for each successive pair of flushing operations at each site. All the variables in Figure 10 were used in the EPR, i.e., eight candidate inputs in total. In this case study, the amount of material in the second flush was used as calculated by the integrative method (recall that in the UK national case study 1 the average turbidity had not been the dominant predictive factor, however it had proved to be so for the Dutch local scale study). Results are presented here for standard polynomial structured equations produced by EPR. The MOGA process ran for 4,320 generations. Table 4 provides three models produced along with the CoD when utilising all data. Note the relatively high prediction from just using the Flush2 amount, but that the addition of soil temperature further improves the forecast.
Top three Pareto optimal daily regeneration rate estimation models identified by EPR
| Model structure | CoD |
|---|---|
| 0.897 | |
| 0.947 | |
| 0.954 |
| Model structure | CoD |
|---|---|
| 0.897 | |
| 0.947 | |
| 0.954 |