Validation of the Georgia Tech Regional Tidal Current Resource Assessment Model and GIS-web Tool

Abstract

The performance of Georgia Tech’s Regional Oceans Modeling System (GT-ROMS), developed to predict accurate and defensible estimates of tidal current power density for the U.S. coastline, is evaluated using point measurements of depth-averaged current speed and tidal elevation from thirty-nine tidal current speed monitoring stations and twelve tidal elevation stations for two classes of stations along the U.S. coast. Class 1 corresponds to validation stations independent of those Class 2 stations used to calibrate the model. Correlation statistics, including phase shift, amplitude ratio, and R2 value, are calculated along with model performance metrics that include the root mean square error, Nash-Sutcliffe efficiency coefficient, and mean absolute error. In addition, predicted and measured current speed, depth-averaged current speed frequency and cumulative frequency histograms means, standard deviation and maximum and minimum value are compared. Final results presented herein are for those sites for which the differences in model and measured depths are less than 30%. The results indicate a fair model performance for predicting currents speeds in medium and high density regions with average NSE (Nash-Sutcliffe efficiencies) coefficients of 0.27 and 0.58 for Class 1 and 2 stations respectively. In regards to IEC validation criteria for acceptable differences in predicted and measured max current speeds to be less than 30%, approximately 80% of the Class 1 stations and 89% of Class 2 stations pass this criteria for the high and medium power density regions. Correlation coefficient R2 values range from 0.76 to 0.89 for average current speeds and 0.79 to 0.85 for tidal elevations. The model has an overall average tendency to over predict mean current speeds for the medium and high power density regions. Frequency histograms of predicted current speeds with measured speeds indicate that the model may benefit from additional calibration before using the model for selection of feasible tidal current energy device locations and assessing the U.S. total available tidal current power.