Kattenhorn, S.A. (2002)
Fluid redistribution around faults after earthquakes with consequences for aftershock locations.
Eos, Transactions of the American Geophysical Union 83, Fall Meeting Supplement, Abstract T21A-1053
Kattenhorn, S.A. (2002)
Fluid redistribution around faults after earthquakes with consequences for aftershock locations.
Eos, Transactions of the American Geophysical Union 83, Fall Meeting Supplement, Abstract T21A-1053
Earthquake events along faults commonly modify the ambient hydrologic system adjacent to the faults. For example, the 1983 M7.3 Borah Peak, Idaho, earthquake resulted in anomalous hydrologic activity in the vicinity the Lost River fault system. This included the eruption of sand spouts, the creation of temporary lakes, and variations in spring discharge. Such occurrences are ostensibly the result of the fault motions, which result in elastic crustal deformation near to the fault. This deformation is spatially variable, depending on factors such as fault size and shape, slip magnitude and slip sense, and rock elastic parameters.
Perturbations to the pre-earthquake state of stress can be described in terms of changes in the mean normal stress (the isotropic component of the total stress tensor at a point), which may either increase or decrease depending on location with respect to the earthquake rupture. Simple 3-D elastic models of faults can be used to characterize the pattern of these mean normal stress variations. Because there is a 3-D spatial variability to the mean normal stress perturbations, seismic events effectively induce pressure gradients that provide a driving force for fluid redistribution. Fluids will migrate from regions of increased mean normal stress to regions of reduced mean normal stress (i.e., down the pressure gradient), perpendicular to mean normal stress contours. A likely 3-D fluid flow field can thus be predicted based on the results of elastic models.
For the case of normal faults, regions of instantaneous fluid discharge may be expected in increased mean normal stress zones alongside the fault, away from the tips. With increasing strike-slip component of displacement, regions of increased mean normal stress shift towards the tips, becoming anti-symmetrically distributed to either side of opposing fault tips. Instantaneous discharge locations are thus different for pure dip-slip versus pure strike-slip faults. Such predictions can be useful for the creation of hazard maps to indicate regions at risk for liquefaction and surface discharge. In the case of the Borah Peak event, which involved predominantly dip-slip with a small component (17%) of left-lateral motion, regions of increased spring flow correspond well with regions of increased mean normal stress predicted by elastic models.
Unlike the rock elastic response to an earthquake, which is essentially instantaneous, the fluid flow response involves a substantial lag time, governed by the hydraulic conductivity of the rock medium. Fluids are slowly redistributed along the earthquake-induced pressure gradient, resulting in fluids being channeled away from the fault over time, or towards sinks near fault tips or relay zones. For this reason, fluid pressure at any point around a fault changes over time, presumably increasing through time in the down-the-pressure-gradient direction. A corresponding spatial and temporal migration of aftershock locations may result from this fluid flux as effective normal stress reductions occur along suitably oriented fault planes, making them more susceptible to failure. Such temporal changes in aftershock locations cannot be predicted using elastostatic models of Coulomb stress distributions following an earthquake. Aftershock locations for the Borah Peak event show some degree of temporal migration over a period of up to 4 years that correlates reasonably well with the predicted fluid flow field induced by the main earthquake event.
External link: AGU database
