Exploration of long records of extreme rainfall and design rainfall inferences

Long records of annual maximum daily rainfall from Europe and the USA, with lengths exceeding 100 years, are statistically analysed to investigate the adequacy of typical extreme value distributions for extreme rainfall and their effect on design rainfall amounts. Statistical analyses show that the conventionally employed Extreme Value Type I (EV1 or Gumbel) distribution may yield inappropriate extrapolations for the upper tail of distribution function of extreme rainfall, whereas this distribution would seem as an appropriate model if fewer years of measurements were available (i.e., parts of the long records were used). In contrast, the extreme value type II (EV2) distribution appears to be suitable for the examined long series. Thus, the results of the analyses agree with a recently expressed scepticism about the EV1 distribution which tends to underestimate the largest extreme rainfall amounts.