Flood effectiveness observations imply that two families of processes
describe the formation of debris flow volume. One is related to the
rainfall?erosion relationship, and can be seen as a gradual process,
and one is related to additional geological/geotechnical events,
those named hereafter extraordinary events. In order to discuss the
hypothesis of coexistence of two modes of volume formation, some
methodologies are applied. Firstly, classical approaches consisting
in relating volume to catchments characteristics are considered.
These approaches raise questions about the quality of the data rather
than providing answers concerning the controlling processes. Secondly,
we consider statistical approaches (cumulative number of events distribution
and cluster analysis) and these suggest the possibility of having
two distinct families of processes. However the quantitative evaluation
of the threshold differs from the one that could be obtained from
the first approach, but they all agree in the sense of the coexistence
of two families of events. Thirdly, a conceptual model is built exploring
how and why debris flow volume in alpine catchments changes with
time. Depending on the initial condition (sediment production), the
model shows that large debris flows (i.e. with important volume)
are observed in the beginning period, before a steady-state is reached.
During this second period debris flow volume such as is observed
in the beginning period is not observed again. Integrating the results
of the three approaches, two case studies are presented showing:
(1) the possibility to observe in a catchment large volumes that
will never happen again due to a drastic decrease in the sediment
availability, supporting its difference from gradual erosion processes;
(2) that following a rejuvenation of the sediment storage (by a rock
avalanche) the magnitude?frequency relationship of a torrent can
be differentiated into two phases, the beginning one with large and
frequent debris flow and a later one with debris flow less intense
and frequent, supporting the results of the conceptual model. Although
the results obtained cannot identify a clear threshold between the
two families of processes, they show that some debris flows can be
seen as pulse of sediment differing from that expected from gradual
erosion.