The simulated streamflow needs to calibrate to match the observed flow. That can be possible by getting fitted catchment parameters. Two approaches are available for calibration and validation. One is to use the whole observed at as calibration and second is to split data for validation. Both use almost the same methodology. The objective is to obtain fitted parameters. SWAT CUP does the calibration validation of the whole series. So both scientific concepts are explained theoretically and practically. But the whole data series concept is used by most of the researchers so its explained first, then data split validation concept is explained. So, this is to you which concept you want to use. Writing parameters to the SWAT model is also covered and the calibrated model was run in the course. Mathematics behind the task is a little complex so I suggest downloading a few research papers similar to your study to explain that equations. Such as R Square, NSE, PBIAS, RSR. These all statistics calculated in SWAT Cup.

This is the complete Scientific Course.