Landslides mapped in 14 watershed analyses in Oregon and Washington pr
ovide a regional test of a model for shallow landsliding. A total of 3
224 landslides were mapped in watersheds covering 2993 km(2) and under
lain by a variety of lithologies, including Tertiary sedimentary rocks
of the Coast Ranges, volcanic rocks of the Cascade Range and Quaterna
ry glacial sediments in the Puget Lowlands. GIS (geographical informat
ion system) techniques were used to register each mapped landslide to
critical rainfall values predicted from a theoretical model for the to
pographic control on shallow landsliding using 30 m DEMs (digital elev
ation models). A single set of parameter values appropriate for simula
ting slide hazards after forest clearing was used for all watersheds t
o assess the regional influence of topographic controls on shallow lan
dsliding. Model performance varied widely between watersheds, with the
best performance generally in steep watersheds underlain by shallow b
edrock and the worst performance in generally low gradient watersheds
underlain by thick glacial deposits. Landslide frequency (slides/km(2)
) varied between physiographic provinces but yielded consistent patter
ns of higher slide frequency in areas with lower critical rainfall val
ues. Simulations with variable effective cohesion predicted that high
root strength effectively limits shallow landsliding to topographic ho
llows with deep soils and locations that experience excess pore pressu
res, but that low root strength leads to higher probabilities of failu
re across a greater proportion of the landscape. (C) 1998 John Wiley &
Sons, Ltd.