A methodology is presented for estimating the probability of a flood occurr
ing on a particular catchment from heavy convective rain, at a lead-time of
up to seven days. The procedure is based upon the physical characteristics
of the land surface, which are remotely measured at a resolution of the or
der of 1 km. The sensible heat flux derived from these measurements is used
to initialise a simple convective scheme and the vertical velocities of as
cending parcels of air at the top of the atmospheric boundary layer estimat
ed. The precipitable water content of the atmosphere is parameterised by a
probability density function. The combination of this with the vertical vel
ocity of the air leads to the derivation of the probability of occurrence o
f a flood-producing convective rainfall event.
The system has been tested for a limited period over the River Irwell catch
ment in the Northwest of England and the results show some correlation betw
een the predicted probabilities and actual convective rainfall observed. Ho
wever, as the result of the procedure is probabilistic, a much longer trial
is required to assess its reliability. This paper describes the physical b
asis for the proposed scheme and the limitations and weaknesses of the proc
edure are outlined. The case study is presented and followed by a discussio
n of the steps required to develop the model further and prepare it for pos
sible operational situations. (C) 2000 Elsevier Science B.V. All rights res
erved.