A stochastic derivation of the areal reduction factor (ARF) of the sto
rm intensity is presented: it is based on the analysis of the crossing
properties of the rainfall process aggregated in space and time, As a
working hypothesis, the number of crossings of high rainfall intensit
y levels is assumed to be Poisson-distributed and a hyperbolic tail of
the probability of exceedances of rainfall intensity has been adopted
. These hypotheses are supported by the analysis of radar maps during
an intense storm event which occurred in Northern Italy. The reduction
factor derived from this analysis shows a power-law decay with respec
t to the area of integration and the duration of the storm. The areal
reduction results as a function of the storm duration and of its frequ
ency. A weak, but significant decrease of the areal reduction factor w
ith respect to the return period is shown by the functions derived, an
d this result is consistent with that of some recent studies on this t
opic. The results derived, although preliminary, may find useful appli
cations for the definition of the design storm in urban catchments of
a size greater than some square kilometres and with duration of some h
ours.