On the issue of trend and noise in the estimation of extreme rainfall properties

In this paper the issue of potential trend and noise in the estimation is investigated for 6 rain series ranging from 19 to 55 years in duration. Different to recent research that predicts a – climate induced–increase in heavy precipitation no clear indication for such trend was found in the investigated historical rain series. Another important aspect is the length of the rain series that is required for the estimation of extreme rainfall properties and the associated uncertainty. The analysis indicates that at least a period of 10 years should be used for the estimation. But even so the possible deviation expressed in terms of the 90 percentile is in the order of 5 to 10% of the ‘true value', defined as the value derived when the whole given series is used for the estimation.