The high pore pressure gradients inherent to accretionary complexes affect the force balance of the wedge via seepage force, which acts in the direction of flow and is proportional to the pressure (head) gradient. If sufficiently large, this seepage force can offset gravity and friction and lead to failure. At the toe of the wedge sediments are weak, slopes are over-steepened by folding and faulting, and fluid pressure gradients can be high; these conditions are conducive to seepage-induced failure. For the 14--16¿ slope at the toe of the southern Cascadia wedge, the pore pressure gradient necessary to initiate failure is λ=0.74--0.86. The gradient necessary to cause failure is sensitive to surface slope and sediment strength, but is insensitive to porosity. Reasonable estimates of sediment strength for most accretionary wedges require pore pressure gradients ranging from 10 to 60% of lithostatic to cause failure. These values are within the range of modeled and measured pore pressures in accretionary complexes, suggesting that seepage-induced slope failure should be an expected feature in this environment. If these failure features are observed, then their presence can be used to constrain the pore pressure gradient within the wedge, independent of any assumptions regarding fluid discharge or permeability. If seepage failure repeats and is localized in the same region, then it can lead to channel, gully, and canyon formation.

Two convergent margins, southern Cascadia and northern Hispaniola, show many regularly spaced headless canyons that cannot be attributed to downslope erosive flow. We suggest that these canyons are forming from internally driven seepage-induced failure. Both the Oregon and Hispaniola accretionary wedges also contain evidence for non-uniform fluid flow based on the observed and inferred presence of vents. Using Darcy's Law, the pore pressure constraint from the slope failure analysis and an estimate of the total fluid discharge, we examine the relationship between wedge and vent permeabilities, the areal extent of focused fluid venting, and the percent of the total fluid discharge that flows out of vents. Given reasonable estimates of the total fluid discharge out of the southern Cascadia wedge, we find that the wedge must be less permeable than 2¿10-17 m2 in order for focused fluid venting to occur at all. If the permeability of the vents is much higher than the wedge permeability, then the vents will occur over a very small percentage of the wedge; these vents, however, could accommodate much of the fluid flowing out of the wedge. Using permeability measurements from samples collected at the toe of the Oregon margin , we estimate that vents at the toe of the southern Cascadia accretionary complex comprise less than 0.2% of the wedge area, but that these vents can accommodate up to 60% of the total fluid discharge. ¿ American Geophysical Union 1992