Chronological split selection was based on the sensitivity analysis in Table 4. The 70-1515 split was selected because it gave the lowest test MAE (2.961) and lowest test RMSE (4.533), while maintaining a positive KGE (0.198). The 60-25-15 split achieved the highest KGE (0.385), but with slightly higher test MAE (2.976) and RMSE (4.60). The 65-20-15 split showed similar test errors but weaker generalisation, and the 75-15-10 split performed worst overall, with the highest test errors and a negative KGE (-0.040). Therefore, 70-15-15 was adopted for subsequent zone-specific evaluation. Table 4 – Chronological split sensitivity analysis: mean MAE, RMSE and KGE across zones MAE RSME KGE Training Validation Test Training Validation Test 70-1515 1.602 1.416 2.961 2.207 1.942 4.533 0.198 75-1510 1.184 1.064 3.014 1.645 1.422 4.840 -0.040 60-2515 1.206 1.217 2.976 1.666 1.654 4.601 0.385 65-2015 0.775 0.683 2.965 1.095 0.989 4.580 0.235 3.4.2 Zone-wise test performance Zone-wise test performance is summarised in Table 5 for models trained with and without meteorological predictors. This comparison was used to assess whether external weather improved one-step deformation forecasts. Overall, the effect of meteorological data was not consistent across sectors. The Toe showed the clearest improvement when meteorological variables were included, with RMSE decreasing from 3.80 to 3.46 mm and MAE from 2.54 to 2.36 mm. The East changed only marginally, whereas performance worsened in the West and slightly in the Crest. KGE was not used for zone-wise interpretation because values remained low, with a maximum of 0.356 in the Crest. This indicates that, although absolute errors were acceptable, the models did not reproduce the temporal variability of the increment series sufficiently well for KGE to provide a robust basis for interpretation, increasing the risk of overfitting. Therefore, performance was assessed primarily using MAE and RMSE. Table 5 – Zone test performance for one-step-ahead forecasting of Δ Meteorological data Metric Crest East Toe West No RSME 4.687 4.445 3.802 4.924 MAE 3.287 2.712 2.539 3.206 Yes RSME 4.712 4.422 3.646 5.352 MAE 3.132 2.980 2.360 3.372
RkJQdWJsaXNoZXIy MTM0Mzk2