Adjustment of extreme wind speed in regional climate downscaling over southwestern South Atlantic

• Published in: International Journal of Climatology
• Main Authors: da Silva, Natália Pillar, Crespo, Natália Machado, Kaufmann, Clarisse Lacerda Gomes, Lima, José Antônio Moreira, Andrioni, Marcelo, de Camargo, Ricardo, da Rocha, Rosmeri Porfírio
• Other Authors: Petrobras
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2022
• Subjects: Atmospheric Science; Argentina; Uruguay; South Atlantic Ocean
• Online Access: http://dx.doi.org/10.1002/joc.7876; https://onlinelibrary.wiley.com/doi/pdf/10.1002/joc.7876; https://onlinelibrary.wiley.com/doi/full-xml/10.1002/joc.7876; https://rmets.onlinelibrary.wiley.com/doi/pdf/10.1002/joc.7876

Abstract In this study, we estimated near‐surface wind speed 100‐year return values over the southwestern South Atlantic Ocean, using present‐day (1979–2018) simulations from the regional climate models (RCMs) WRF and RegCM4. Extreme wind events, associated with hazardous conditions over coastal and oceanic areas, must be well represented in numerical models for risk assessment, and few studies focused on the added value offered by RCMs to wind extremes. Events were selected with the peaks‐over‐threshold method and extremes were calculated by fitting peaks to a generalized Pareto distribution. For the assessment of model performance, we used the satellite‐based dataset from the Cross‐Calibrated Multi Platform (CCMP), which has great agreement with in situ observations. While modern reanalysis underestimated higher wind speed quantiles, the CCMP was able to represent these quantiles. Dynamical downscaling with the WRF (RegCM4) indicates an underestimation (overestimation) of wind speed for upper quantiles. To mitigate the effects of these differences in the extreme value estimate, we applied a linear adjustment in the simulated wind speed using the CCMP as reference. This application reduced the bias for higher wind speeds in simulations for all regions, except over the coastal area near Argentina and Uruguay, where downscaling already realistically represents extreme events. The spatial distribution of the simulated extremes is compatible with previous results based on reanalysis and satellite data, although with finer‐scale structure, especially over the southern South Atlantic, a region frequently affected by cyclone occurrence and extreme near‐surface winds. The extreme wind speed maps estimated for 100 years reflect these conditions with values reaching around 30 m·s −1 in the area, even after the wind adjustment. Besides, the fine‐scale features aggregate important new information related to extreme winds in the region, which is a relevant added value in the study of return values estimates.