Changes in the intensity and frequency of precipitations are likely to affect the hydrologic
cycle much more than changes in the mean climate. In this work, the hypothesis that long
term series of short duration maximum annual rainfall are stationary has been evaluated.
Values of annual maximum of 1, 3, 6, 12, 24 hours precipitations from 21 weather stations in
Friuli Venezia Giulia region were collected for the period 1922‐2009; trend analysis on raw
data series was performed using linear regression and Mann‐Kendall test. For each duration
and location series, EV‐I distribution was used to calculate the critical values (quantiles) of
precipitation according to return periods of 2, 20, and 100 years on moving windows of 20
years including at least 15 useful values; on these data, trend analysis was performed using
linear regression and break points were searched using the Pettitt test.
Linear regression and Mann‐Kendall test performed on raw data suggested that in most
cases the maximum value of short‐duration rainfall increased along time; however, the trend
was only seldom statistically significant.
On the contrary, regression of 1, 3, 6, and 12 hours precipitations quantiles computed for all
return periods showed a significant increase with time, while the opposite was observed for
24h quantiles. Break points were detected in most series, with differences between periods
up to 50%.
The results indicate that trends of intense rainfall are not univocally defined, therefore
suggesting that data for hydrological design must be carefully selected.
