A Multiple Linear Regression-Based Approach for Storm Surge Prediction Along South Brazil

Knowledge of the characteristics of storm surges is of paramount importance for an effective coastal hazard risk assessment. The vast Majority of storm surges in southern Brazil are generated by extratropical cyclones. In this context, this study focuses on spatial and temporal analysis of storm surges caused by the cyclones within a 30-year period. The wind and pressure from CFSv2 data were analyzed from 1979 to 2010 time span, with a time lag of 9 h identified for pressure and 26 h for wind stress. A multiple linear regression model was then employed to estimate the expected intensity of the storm surge. The results suggest that the model represented 82% of the variations in the storm surge levels. The accuracy of the surge predictions was validated by the recorded storm surges using tide gauge data for Itajaí port. The agreement between the modelled and the observed values were considered good by explaining 74% of the variations in the storm surge. Finally, Analysis of future scenarios of 1-, 10-, 50- and 100-return period revealed that positive storm surge events may be higher, increasing coastal risks in the coastal zone of southern Brazil.