A distributed wind downscaling technique for wave climate modeling under future scenarios

Abstract

The aim of this study is to develop a Weibull-based distributed downscaling technique for wind field as forcing for the wave models to investigate the wave climate under future scenarios. For this purpose, the statistical downscaling approach modifies Weibull distribution parameters of the global circulation model wind speeds based on the corresponding features of wind data of ECMWF (European Center for Medium-Range Weather Forecasts). The proposed technique has the advantage of modifying the wind components in each grid point based on the corresponding values in the same grid point of ECMWF wind field. Hence, it is superior to other existing models due to considering the spatial variation. The previous models using inverse distance weighting suffer from heterogeneity and ignoring spatial variation in areas with high gradient of wind speed. Moreover, the Weibull-based technique outperforms the existing statistical downscaling techniques in terms of accuracy. Prior to investigate future distribution of wave characteristics, performance of the selected GCM was evaluated and compared against the corresponding models obtained from the available regional climate models. Future projections of wind fields (RCP4.5, RCP8.5) were downscaled for the period of 2081 to 2100 with the proposed model as driving force for wave modeling in the Persian Gulf. To investigate the impacts of climate change on wave characteristics, results of the wave simulations from a third generation wave model (SWAN) for future scenarios are compared with those of the historical period (1981–2000) in monthly, seasonal, and annual scales. Generally, for RCP8.5, the results indicate a decrease in future significant wave height and peak wave period about 15% and 5%, respectively. However, the change of wave direction is marginal. Moreover, wave models forced with RCP4.5 wind data provide slightly higher average values in terms of wave height and peak wave period compared to those of RCP8.5.