Abstract
Output from the global forecast system (GFS) meteorological model and measurements derived from the QuikSCAT satellite are investigated, comparing them with in situ data obtained from moored buoys. In order to undertake these comparisons, one year (April 2003–March 2004) of wind data from the GFS model and from the QuikSCAT satellite were co-located with the in situ data; retaining spatial differences of less than 0.4° longitude−0.5° latitude and temporal differences of <30 min. The buoy data were converted from 3 m to 10 m height above the sea surface by means of the logarithmic profile method. To investigate the agreement between the data pairs, correlation coefficients were computed; likewise linear regression models were fitted to the different data sets. The accuracy and precision of the QuikSCAT and GFS model were estimated through the computation of the bias and the RMS differences (RMSD). The statistical results reflect a better agreement between the QuickSCAT and the buoy winds (r > 0.68 for the wind speed, r > 0.86 for the direction), rather than between the GFS and the buoy winds (r > 0.61 for the wind speed, r > 0.74 for the direction). Similarly, there is higher accuracy and precision in the QuikSCAT measurements (RMSD < 2.23 m s−1 for the wind speed, RMSD < 48.3° for the direction). Nonetheless, the GFS still represents a good agreement with the winds obtained from the buoys. In both cases, there is better agreement and higher accuracy and precision in the open ocean waters (0.82 < r < 0.98), than near to the coast (0.61 < r < 0.91), both for wind speed and direction.
Acknowledgements
The authors would like to thank the JPL SeaWinds Project and CERSAT, at IFREMER, Plouzané (France), for the production and distribution, respectively, of ‘SeaWinds on QuikSCAT Level 2B Ocean Wind Vectors, in 25 km Swath Grid’. We would like to thank also the SeaWiFS Project (Code 970.2) and the Distributed Active Archive Center (Code 902) at the Goddard Space Flight Center, Greenbelt, MD 20771, for the production and distribution of the GFS model data. Finally, thanks to Puertos del Estado and the Met Office for distributing the moored buoys wind data. Leire Ibaibarriaga (AZTI-Tecnalia) is acknowledged for her advice on the statistical methods. Professor Michael Collins (University of Southampton, UK, and AZTI-Tecnalia) is acknowledged for his contribution to the manuscript.