Modelling the delivery of landslide-generated sediment to channel
networks is challenging due to uncertainty in the magnitude-frequency
distribution of failures connected to the channel network. Here, we
investigate a simplified treatment of hydrological connectivity as a
means for improving identification of coarse sediment delivery to
upland rivers. Sediment generation from hillslopes and channel banks
and its delivery to the channel network are modelled based on a
modified form of SHALSTAB coupled to a network index version of
TOPMODEL. The network index treatment has two important hydrological
effects: (a) it only allows saturated areas to connect to the
hydrological network when there is full saturation along the associated
flow path; and (2) overland flow associated with unconnected but
saturated zones is assumed to remain within the catchment and to
contribute to a reduction in the catchment-averaged saturation deficit.
We use this hydrological treatment to restrict sediment delivery to
situations where there is surface hydrological connection (i.e.
saturation) along the complete flow path that connects failure areas to
the drainage network. This represents an extreme restriction on the
possibility of connected failure as it does not allow for failed
material to connect if failures are associated with partial saturation
or where delivery involves runout across areas where hydrological
connection is not maintained. The impact of this restriction is
assessed by comparing model predictions with field mapping of connected
failures and data from continuously recording coarse sediment sensors,
for two storm events. The hydrological connection requirement
restricted connected failures to zones closer to the drainage network
and resulted in a better level of agreement with the field mapped
failures. Simulations suggested that in the study catchment the
majority of sediment inputs occur from hydrologically-connected areas
close to the channel network during moderate sized rainstorms that
occur relatively frequently. (c) 2007 Elsevier B.V. All rights reserved.