This is the objective function surface of an unsteady 2D HEC-RAS model that minimizes the volume error and maximizes the NSE by trying 10,000 different combinations of the curve number and Manning’s n. That best model (still with a 5.75% volume error) is very hard to find on a relatively flat valley. Calibrating the RAS parameters for an arid region like southern New Mexico is not easy at all without much flow data.