%0 Journal Article
%T Numerical simulation of wind waves on the R¨ªo de la Plata: evaluation of four global atmospheric databases
%A Paula Martin
%A Walter Dragani
%A Bibiana Cerne
%A Guadalupe Alonso
%J Brazilian Journal of Oceanography
%D 2012
%I Universidade de S?o Paulo
%X The performance of NCEP/NCAR I, NCEP/DOE II, JRA-25 and ERA-Interim global databases, implemented as atmospheric forcings of the SWAN model in the R¨ªo de la Plata region, was quantitatively tested by calculating the bias, the mean square root error, the determination coefficient and the slope of the line fitted between observed and simulated wave parameters (significant wave height, mean period and direction). Even though statistical estimators showed no evident differences for wave periods and directions some noticeable differences were observed for simulated significant wave heights depending on the forcing used. The lowest bias (0.22 m) was obtained when the SWAN model was forced by ERA-Interim. With regard to the mean square root errors, the lowest values were obtained when NCEP/NCAR I (0.16 m) and NCEP/DOE II (0.19 m) were used as forcing. In addition, the best slope for simulated heights (0.79) was obtained using NCEP/DOE II. Computed determination coefficients for heights, periods and directions were very similar (0.89-0.93) for all the simulations carried out in this study. Energetic and severe wave events were given special consideration. The most energetic wave episode recorded in the R¨ªo de la Plata mouth (24 August, 2005) was analyzed and discussed in particular. It was concluded that during energetic atmospheric conditions the best agreement is achieved by implementing NCEP/DOE II as forcing. In the light of these results it is concluded that NCEP/DOE II is the most suitable atmospheric forcing to simulate wave heights with the SWAN model in the R¨ªo de la Plata region. Na regi o do Rio de la Plata, o desempenho das rean¨¢lises globais do NCEP/NCAR I, NCEP/DOE II, JRA-25 e ERAInterim implementadas como for antes atmosf¨¦ricas do modelo SWAN foram quantitativamente acessados atrav¨¦s do vi¨¦s, erro quadr¨¢tico m¨¦dio, coeficiente de determina  o e inclina  o da reta. Estes ¨ªndices foram obtidos dos parametros de ondas observados e simulados (alturas significativas de ondas, per¨ªodo principal e dire  o). Embora as estimativas estat¨ªsticas n o mostrem diferen as evidentes para per¨ªodos e dire  es, algumas diferen as not¨¢veis foram obtidas para altura de ondas simuladas, dependendo do vento utilizado. O menor vi¨¦s para altura significativa (0.22 m) foi obtido quando o SWAM foi for ado com a ERAInterim, enquanto o NCEP/NCAR I (0.16 m) e NCEP/DOE II (0.19 m) forneceram menor erro quadr¨¢tico m¨¦dio. A melhor inclina  o da reta entre simula  o e observa  o de altura significativa (0.79) foi obtida usando NCEP/DOE II. No per¨ªodo de estudo, o maior epis¨®dio
%K Reanalises de NCEP
%K Reanalises japon¨ºs de 25 anos
%K Modelo Swan
%K Rio da Prata
%K NCEP/NCAR Reanalysis I
%K NCEP/DOE reanalysis II
%K JRA-25 reanalysis
%K ERA-Interim
%K SWAN model
%K R¨ªo de la Plata
%U http://www.scielo.br/scielo.php?script=sci_arttext&pid=S1679-87592012000400008