Rudy Magne 1 and Fabrice Ardhuin 1

1. Wave hindcasting in the Mediterranean sea 11th International Workshop on Wave Hindcasting and Forecasting , October 18-23, 2009, Halifax, Canada. Rudy Magne1 and Fabrice Ardhuin1 1 rudy.magne@shom.fr, Service Hydrographique et Océanographique de la Marine, Brest, France Source term parameterizations: Although little used outside of ECMWF, the Bidlot et al. (2005) parameterization was, until recently, the most accurate parameterization for most wave parameters. It provided a welcome improvement on the old WAM Cycle 4 parameterization by reverting to the original definition of the mean wavenumber used in the dissipation term, i.e. a wavenumber corresponding to Tm01 instead of Tm0-1 . However this parameterization does not well represent swell dissipation and is still sensitive (although much less than WAM-Cycle 4) to the presence of swell (Rascle et al. 2008). Otherwise the input and nonlinear interactions are essentially unchanged. The Bidlot et al. (2005) parameterization is the default setting when using switches ST3 and NL1 in WW3. Introduction: Global and regional wave hindcasts and forecasts exhibit much larger errors in coastal areas and enclosed seas than in the open ocean, with typical root mean square error for the significant wave height of the order of 15 to 30% of observed values, compared to 10% or less in the open ocean (Rascle et al. 2008). Recent global-scale improvements in model parameterizations (Ardhuin et al. 2008, 2009) have only marginally improved the situation in such coastal settings and we investigate the causes of this problem in the particular case of the Mediterranean sea, focusing on the Western Mediterranean. The version 3.14 of WAVEWATCH III ™ (WW3) model has been used for this hindcast with a spatial resolution of 0.1°, forced by the 0.5° ECMWF analysis surface winds. Wave hindcast has been done over one year (2007) in the Mediterranean sea, comparing wave model output to in situ stations and altimeter data, using two sets of dissipation source terms, i.e. Bidlot et al., (2005) and Ardhuin et al., (2009). These are activated by the ST3 switch, using different namelist parameters to chose one or the other (Tolman 2009). General results show an underestimation of the wave heights, with strong localized negative bias due to land orography such as in Adriatic Sea, usual when using this kind of wind forcing (Ardhuin et al. 2007). Details results are presented here, highlighting the wind as the main source of error over the Mediterranea sea. In order to correct the defects of Bidlot et al. (2005), the dissipation term was redefined from scratch by Ardhuin et al. (2008, 2009) based on observed swell dissipation rates (see other poster) and an empirical wave breaking term that is loosely based on Phillips’ (1985) saturation ideas, using the exact threshold value measured by Banner et al. (2000), with a partial directional integration of the saturation spectrum consistent with normalization later proposed by Banner et al. (2002). In the T405 version a diagnostic tail is used, and thus there is no need for an explicit cumulative dissipation term, which is otherwise used in T441 (with general better performance at global scales, but slightly worse for short fetches since a reduction of the wind input at high frequency has been introduced to match mean square slope observations). Further work is under way to correct the short fetch bias in T441. Satellite validations: One year of cross-calibrated atlimeters (Queffeulou et al. 2003 , 2006) data have been used to make a wind and wave global validation over the mediterranean sea: ERS2, ENVISAT, Jason-1 and GFO, using the Ardhuin et al., 2009 dissipation source term. Since diffusiometer are assimilated in the ECMWF atmospheric model, they have not been taken into account in this study. Satellites data have been first collocalized in space and time with the model output. Collocalized data have been then assigned to 0.5°x0.5° box, follow by mean statistics computation per box, when more than 100 collocalized values where found (method used by Cavaleri etal., 2006). The upper and lower left hand plots show the NRMSE between model and satellites data respectively for Hs and 10m wind speed. The upper and lower right hand panel show the best fit slope (through the origin) between model and data. Hs and wind speed NRMSE patterns show large regionalized errors ( Adriatic sea, Genoa golf and in the extreme eastern and western part of the Mediterranea sea). An almost perfect geographical correlation exists between wave height and wind speed error, even if hs error is found to be larger. Hs best fit slope correlates also well with the NRMSE parttern, showing a strong understimation where errors are large. Such a correlation is not as clear with the wind speed. It might be due to the faster fluctuation in time of the wind wich leads to a noisy signal. Modelled wave heigth and wind speed are ordered by class. Nrmse and best fit slope are compted against altimeter data for each class. Errors are found to first decrease with wind speed and then increase for high wind speed. Same trends are found for the wave height. Best fit slope curve clearly indicates that low winds are underestimated and high winds are overestimated, which impacts the wave height in the same way. In-situ validations: 22 in situ stations from Meteo France, Puerto del estado (Spain), XIOM (Catalunya, Spain) and ISPRA/Idromare (Italy) have been used for wave model validation over the year 2007. The following plot shows the NRMSE for the significant wave height, using the Ardhuin et al. 2009 dissipation source term. Overall buoys statistics have been computed for both Ardhuin et al. 2009 and Bidlot et al., 2005 parametrizations. Results are summarized on lower tables. Detail statistics have been also provided for the 61001 and 61002 buoys. Overall buoys statistics (year 2007)‏ Ardhuin et al. 2009, T405 ZWND=6m Bidlot et al. 2005, ZWND=6m Wind and wave calibration? Downscaling ? Wind errors are a real problem for wave hindcast or forecast in enclosed sea such the Mediterranean sea. Several tests have been realized in order to pre-calibrate the wind field using altimeter data ( Hs best fit slope correction - wind sector coefficient calibration ) without reel succes. Maybe a weight per box depending on the wind intensity could been tested if we trust the model/altimater best fit slope curves showing an almost linear trend. A post wave calibration is possible using the best fit slope map. .Nevertheless, this is not the aim of this study, keeping in mind to correct the wind field before wave computation. The next step will be a wind downscaling approach using a mixture of SAR and Altimeter data applied to pre-identified meteorological class (Ben Ticha, 2007)… however, any other ideas to improve wind fields will be greatly aprreciated  Buoy 61001 Ardhuin et al. 2009, T405 ZWND=6m Références: Ardhuin F., F. Collard, B. Chapron, P. Queffeulou, J.-F. Filipot and M. Hamon, Spectral wave dissipation based on observations: a global validation, Proceedings of the Chinse-German Joint Symposium on Hydraulics and Ocean Engineering, Darmstadt, Germany, pp. 391-400, 2008. Ardhuin F., L. Marié, N. Rascle, P. Forget and A. Roland, Observation and estimation of Lagrangian, Stokes and Eulerian currents induced by wind and waves at the sea surface, Journal of Physical Oceanography, 2009 , in press. Ardhuin F., E. Rogers, A. Babanin, J-F. Filipot, R. Magne, A. Roland, A-V-D Westhuysen, P. Queffeulou, J-M Lefevre, L. Aouf and F. Collard , Semi-empirical dissipation source functions for wind-wave models: part I, definition, calibration and validations at regional to global scales, Journal of Physical Oceanography, 2009b , under revision. Ardhuin F., L. Bertotti ,J-R. Bidlot,L. Cavaleri,V. Filipetto, J.-M. Lefevre and P. Wittmann, Comparison of wind and wave measurements and models in the Western Mediterranean Sea, Ocean Engineering vol.34, p526-541, 2007 Banner M.K., V. Babanin, I.R. Young, Breaking probability for dominant waves on the sea surface, Journal of Physical Oceanography, vol 30, 3145, 2000 Banner M.K., J.R. Gemmrich, D.M. Farmer, Multiscale measurement of ocean wave breaking probability, Journal of Physical Oceanography, vol 32, 3364, 2002 Bentamy A., H-L. Ayina, P. Queffeulou, D. Croize-Fillon and V. Kerbaol, Improved near realtime surface wind resolution over the Mediterranean Sea, Ocean Science, 2007. Ben Ticha M. B., Fusion de données satellitaires pour la cartographie du potentiel éolien offshore, PhD. Thesis, Ecole des Mines de Paris, 2007 Bidlot J., S. Abdalla, and P. Janssen, A revised formulation for ocean wave dissipation in CY25R1, Tech. Rep. Memorandum R60.9/JB/0516, Research Department, ECMWF, Reading, U. K., 2005. Cavaleri L. and L. Bertotti, Accuracy of the modelled wind and wave fields in enclosed seas, Tellus, 56A, p167-175, 2004 Cavaleri L. and M. Scalvo, The calibration of wind and wave model in the Mediterranean Sea, Coastal Engineering, vol 53, p613-627, 2006 Filipot J-F, F. Ardhuin, A. Babanin, Aunified spectral wave breaking dissipation formulation. Part1. Breaking probability. Journal of Geophysical research, 2009, in revision Phillips O.M., Spectral and statistical properties of the equilibirum range in wind-generated gravity waves, Journal of Fluid Mechanics, vol 156, 505-531, 1985 Queffeulou P., Long term validationof of wave height and wind speed measurements from satellite altimeters. Proceedings of the ISOPE conference, Honolulu, Hawaii, USA, May 25-30, 2003 Queffeulou P., Altimeter wave height validation – an update, OSTST meeting, Venice, Italy, March 16-18, 2006 Rascle N., F. Ardhuin, P. Queffeulou, and D. Croize-Fillon, A global wave parameter database for geophysical applications. part 1: wave- current-turbulence interaction parameters for the open ocean based on traditional parameterizations, Ocean Modelling, vol. 25, pp. 154– 171, 2008. doi:10.1016/j.ocemod.2008.07.006. Buoy 61002 Ardhuin et al. 2009, T405 ZWND=6m