Statistical Evaluation of Monthly Marine Surface Winds of CMIP6 GCMs in the western North Pacific

Abstract

 For over years, fundamental component and dataset in climate projection had been covered by general circulation models (GCMs) output mainly from the Coupled Model Inter-comparison Project (CMIP). Marine surface winds are an important output of GCMs and they provide input to marine forecasts and warning systems. Their accuracy have direct implications for marine safety, air-sea fluxes, wave and ocean models. Western North Pacific (WNP) is known as a highly vulnerable region to oceanic and atmospheric hazards, such as storm surges, waves and typhoons. Therefore, it is necessary to examine the quality of surface winds from CMIP6 GCMs in comparison with obserations or reference data. The ERA-Interim reanalysis is considered as the reference data. Monthly surface winds in WNP and its four sub-regions over 30 years (1979-2008) are retrieved from 32 ensembles of CMIP6 and the ERA-Interim reanalysis. The surface winds are evaluated by using five statistics, bias, correlation coefficient (R), root mean square deviation (RMSD), standard deviation (σ) and Index of Agreement (IoA). In general, overestimation over sea and underestimation in land are found with approximately 1 m/s variance in wind speed. RMSD and standard deviation on average is relatively similar for each GCM and ensemble. Distinctive seasonal characteristics of surface winds are found in each sub-region. Based on the findings, it can be concluded that the use of individual ensemble result from CMIP6 GCMs could result in serious bias. Therefore, it is recommended to use multi model ensemble-averaged values for the application of CMIP6 surface winds such as wave climate projections.