[1] We investigate transient fluid flux through the seafloor recorded near the Costa Rica trench during the 2000 Costa Rica Seismogenic Zone Experiment using a 2-D fully coupled poroelastic finite element model. We demonstrate that the observed hydrologic anomalies are consistent with a model of propagating slow slip at the subduction interface between the frontal prism and downgoing plate. There are two sources of volumetric strain that drive fluid flux at the seafloor in response to fault slip at depth: (1) compression and dilation in the vicinity of the tips of a slipping patch and (2) extension and compression due to flexure of the seafloor. The superposition of these two effects results in distinctive spatial and temporal patterns of fluid flow through the seafloor. In a forward modeling approach, time series from shear ruptures with a range of fault length-to-depth ratios in a heterogeneous crust are generated and compared with flow rate observations. Assuming a constant propagation rate and an elliptical profile for the distribution of slip along the decollement, the set of model predictions enables us to infer the probable rupture location, extent, propagation velocity, and duration from a single flow rate time series. The best fit model suggests that the slow slip event initiated within the toe at a depth of less than 4 km and propagated bilaterally at an average rate of 0.5 km dÀ1 . This interpretation implies that stress in the shallow subduction zone is relieved episodically. Furthermore, the Costa Rica data suggest that episodic slow slip events may initiate in the prism toe without being triggered by a seismic event further downdip.

[1] To enable testing of the hypothesis that fluid flow from cold seeps responds to episodic tectonic events, we have developed a flow meter with a temporal resolution on the order of minutes. The Optical Tracer Injection System (OTIS) measures flow rates through the sediment-water interface (SWI) of 0.1 to >500 m/yr ± 10%, samples fluid for postrecovery chemical analysis, and is adaptable for use as a long-term real-time monitoring station. Laboratory and in situ testing demonstrate that the instrument meets temporal accuracy and resolution requirements necessary to detect transient flow events. In a deployment at an active seep site in Monterey Bay, the OTIS measured flow rates with tidal period variability. Time series analysis indicates flow rate and tide height records are in phase. This flow rate response to tides, together with seep fluid temperatures measured at the SWI of greater than 9°C, suggests the seep is connected via a highly permeable fracture or other conduit to an overpressured aquifer at a depth of at least 110 m. We infer the hydrogeology of this well-like system is further complicated by additional fluid sources since 3 He/ 4 He ratios are elevated, and there is a lack of correlation between flow rate and temperature records at frequencies lower than tidal frequencies. This short-duration deployment demonstrates the OTIS's potential to capture flow and chemistry transients associated with earthquakes and creep in the offshore environment.

An array of boreholes, drilled through a regionally continuous hydrologically confining layer of sediments into extrusive igneous basement rocks of the Juan de Fuca Ridge eastern flank, has been instrumented with CORK hydrologic observatories for long‐term monitoring and fluid sampling. Omission of seals between nested casing strings reaching into basement at one site created a low‐resistance connection between basement and the overlying water column, and despite the natural superhydrostatic state of basement water at that location, a “runaway” condition of cold seawater downhole flow into the crust was established, which persisted for more than 4 years. The existence of this condition, along with perturbations generated by it and by initial drilling operations observed at a properly sealed hole 2.4 km away, have been used with analytic and finite element model solutions to constrain formation permeability. The minimum threshold permeability allowing stable downhole flow is roughly 4 × 10−13 m2. A value of permeability similar to this (3–4 × 10−13 m2) is estimated on the basis of the elapsed time for initial perturbations to propagate between the sites (∼2.5 days). The amplitude of the long‐term flow perturbation observed at the sealed site (roughly 1.7 kPa) is smaller than that predicted by modeling (5–10 kPa). Models for flow in an anisotropically permeable layer show that this could be the consequence of low vertical permeability (e.g., arising from massive volcanic or sediment interlayering) or high permeability in the direction of the tectonic fabric generated at the ridge axis. Disagreement between the permeabilities estimated here with previous large‐scale estimates appropriate for the cross‐strike direction (the primary direction between the borehole sites) (10−10–10−9 m2) is difficult to reconcile; it is possible that the holes are poorly connected to zones of high permeability that facilitate the large lateral fluid and heat fluxes previously inferred at this young crustal site.