*3.3. HEC-RAS Model Performance with the SRC and WA-ANN Techniques*

To sum up the above-elaborated calibration and validation exercises using SRC and WA-ANN-based boundary conditions, their statistical performance was compared and tabulated in Table 7. The statistical results (Table 7) clearly indicated a preferable performance of the model using WA-ANN-based sediment load boundary conditions. As SRC reconstructed the missing sediment load data with R2 and NSE at 0.635 and 0.655, respectively, the model calibration took a long time to adjust transport parameters for attaining stability. However, due to better recondition accuracy using WA-ANN (R<sup>2</sup> = 0.771 and NSE = 0.771), the HEC-RAS model simulated the bed-levels changes with great stability. As the SRC overestimated sediment load, therefore to flush extra sediments, we needed to adjust the transport parameters that might not represent the correct physics of the transport processes in the reservoir. Therefore, more accurate boundary conditions played a vital role in precise modeling of the transport processes by keeping transport parameters within the physical limits.

**Table 7.** Statistical performance of HEC-RAS model with the SRC and WA-ANN techniques during the calibration and validation periods.

