Abstract
The 50-year return value of 10 m wind was calculated at 12 locations in Japan with upper air soundings. Comparison was made between observations and historical simulations from the Database for Policy Decision-making for Future climate changes (d4PDF), between the Block-Maxima and Peaks-Over-Threshold methods, and between empirical and fitted values. The occurrence of tropical cyclones (TCs) and vertical wind profiles at different locations were examined to account for the model performance. Model results were adjusted producing improvements over 9 of 12 locations. Then, 4 K warmer simulations with six possible warming patterns were compared with historical simulations. Over all locations, TC occurrence reduced for all six patterns, but TC influence on extreme wind increased for five patterns. However, no widespread changes (locations or patterns) were seen in the return values. Changes were associated with TCs more than extratropical cyclones, and the primary source of uncertainty seemed to be the warming pattern.
