Sensitivity of rip current forecasts to errors in remotely-sensed bathymetry

Of all rip currents generated on an in-situ bathymetry, 55% were reproduced on the remotely-sensed bathymetry, showing that models predicting nearshore currents on remotely-sensed bathymetry have predictive value. Positive rip current predictions were promoted significantly by accurate reproduction of the pattern and amplitude of nearshore bars at length scales between 200 and 400 m. In contrast to the length-scale aware validation technique applied here, commonly used domain-wide bulk error metrics lack important information about spatial variations in the quality of remotely-sensed bathymetry.