MBE, RMSE of two series of data, correlation.

I have a series of Data generated as output of a mathematical model which I already know the MBE and RMSE (real measured values and simulated ones by the model)

Afterwards I used the same input for the previous model but for another model, but i do not have the real-measured data, that means I can just assume that the calculation made by the first model is very correct/precise, and thus used as reference for the output values of the second model. If I calculate the RMSE and MBE based on this, is that correct? can I at the end sum up the MBE and RMSE of the first model with the measured values...