Characterizing the Relationship of Tremor and Slip during Recent ETS Events in Northern Cascadia using Strainmeters, GPS, and Tremor Observations

We investigate the relationship between slip and tremor during multiple recent slow slip events in northern Cascadia. While the relationship of geodetically detectable slow slip and nonvolcanic tremor appears to be broadly coincident, the exact spatial and temporal characteristics remain unclear at a finer scale. Typical GPS derived slip distributions tend to be spatially and temporally smoothed and offset slightly updip of tremor distributions. These discrepancies may be real, or they may be a consequence of the resolution of GPS data or an artifact of the inversion methodology. Borehole strainmeters provide additional independent geodetic constraints for characterizing slip, provide greater temporal resolution, and greater precision than GPS. However, various non-tectonic artifacts and other sources of error have limited the number of usable stations and made deriving reliable information from strainmeters during slip events difficult.

We utilize strainmeters with low levels of noise and minimal observable artifacts to constrain forward models and to provide additional independent observations in joint geodetic inversions with GPS data. A series of slip distributions are derived by inverting strainmeter and GPS data using the Kalman-filter-based Extended Network Inversion Filter. To compare the tremor distributions to the geodetically derived slip we also construct slip distributions using tremor occurrences as a proxy for localized slip on the plate interface. The magnitude of slip per tremor occurrence is then scaled to best match the observed surface displacements. Separate slip distributions informed by GPS and tremor are then used to predict strain time series. The comparisons between strain predictions and observations produce mixed results. This may indicate that that tremor and slip are not always coincident. This is particularly evident during the Aug. 2010 event, where the peak GPS-derived slip is located in a region with decreased tremor activity. Preliminary results also show that shifting the peak tremor-derived slip updip a few kilometers decreases the overall misfit of the displacements, supporting the conclusion that the peak tremor activity occurs slightly downdip of the peak slip.