ABSTRACT
The rainfall source used to run one-year simulation was the National Weather Service Multisensor precipitation estimates (MPE) radar-rainfall products. This source has a mean systematic error correction for Puerto Rico and in some places cannot remove the local bias, correctly, principally for small areas. Some systematic errors in the MPE rainfall product were revealed in the simulation period, where the MPE sensed larger amounts of rainfall than actually occurred within the study Mayaguez Bay Drainage Basin area. In these cases the observed discharges were lower than the simulated for Anasco and Rosario rivers. The stream flow illustrate cases in which the up-scaled model could not reproduce the observed flow because the rainfall could not be quantified accurately using the MPE product. Forcing the model to produce maximum and minimum peak flows by judiciously parameterizing the model showed that the predictability limits of the model were well the magnitude of the observed flow.
