On the effect of the sampling frequency of sea level measurements on return period estimate of extremes—Southern European examples