To ensure a safe and sustainable operation of pumped-storage hydropower stations (PSHSs), representative transient field tests and model calibration are normally required. Previous studies have identified the uncertainty associated with the pump-turbine model, i.e. pump turbine characteristic curves (PTCCs), as the main bottleneck restricting the accuracy of transient simulation. This study focuses on adjusting the PTCCs to improve the transient simulation by fusing field data. The reconstruction starts with deforming PTCCs by adjusting some feature points. Two point distribution models (PDMs) were then built to represent the potential deformations of PTCCs. Finally, measured data under transient and steady conditions is integrated with the PDMs to reconstruct the PTCCs. The simulation results using the reconstructed PTCCs demonstrate a significant accuracy improvement compared to those using the original PTCCs in an experimental platform. The reconstructed model's accuracy is enhanced when load rejection experiments for reconstructing PTCC are conducted under low water head and large guide-vane opening conditions. These conditions provide a wider range of training data on the PTCCs. This study can serve as a calibration foundation for digital twin systems, enabling the smart and sustainable operation of PSHSs.