TEMPORAL CLUSTERING OF EARTHQUAKES AND TIME-DEPENDENT SEISMIC HAZARD ANALYSES: EXAMPLES FROM THE WASATCH AND LOST RIVER FAULTS

Many faults throughout the Basin and Range Province show evidence for temporal clustering of earthquakes. These variations in rates of activity significantly affect probabilistic seismic hazard analyses (PSHA), but paleoseismic records are typically too short to rigorously incorporate these variabilities. Two exceptions include the Wasatch fault in Utah and the Lost River fault in Idaho, where deep trenches on selected segments revealed long paleoseismic records. We develop time-dependent models for PSHAs along these similar faults with very different event chronologies. Data from previous studies on the Brigham City (BC) and Salt Lake City (SLC) segments of the Wasatch shows an apparent clustering of 6 events since 9 ka and and only 1 event between 9 and 15 ka. In contrast, the Lost River fault shows three separate clusters of events during the past 160 ka, with 2 events occurring around 20 ka, 3 events between 45 and 70 ka, and 2 events between 130 and 150 ka. We applied the approach of the 1999 Working Group on California Earthquake Probabilities to calculate conditional probabilites and equivalent Poisson (time-dependent) recurrence intervals (TDRIs) for surface-faulting earthquakes. We assumed a lognormal renewal model and coefficients of variation (COV) from 0.3-0.7 to calculate conditional probabilities for the next 50 yrs. Resulting TDRIs for the Wasatch fault vary by more than an order of magnitude, depending on: 1) the COV; 2) whether the short-term (<6-9 ka) or long-term (<15 ka) paleoseismic record is used (ie., how temporal clustering is addressed); and 3) the elapsed time since the youngest surface-faulting event. For the SLC segment, elapsed time is less than mean recurrence, and TDRIs are similar to Poisson hazard models. Thus, using the time-dependent model has little effect on the hazard, but results are still signficantly affected by the length of the paleoseismic record used. In contrast, elapsed time exceeds mean recurrence for the BC segment, resulting in consistently much shorter TDRIs than for Poisson models,which significantly increases the hazard regardless of the length of paleoseismic record used. Results for the Lost River fault are forthcoming.