USE OF THE CYCLIC STRESS METHOD FOR PALEOSEISMIC ANALYSIS

Numerous investigators have used the cyclic stress method to back-calculate seismic parameters (e.g., M and a_max) at paleoliquefaction sites. However, in using the method to back-calculate a_max, a number of uncertainties must be considered. These include the influence of: (1) strength and thickness of a cap layer; (2) the depth of the watertable on surface manifestations of liquefaction; (3) aging and/or cementation; and (4) dike height and extent of liquefaction along the base of a cap. A judiciously designed field investigation is required to assess the influence of these and any other unknown factors.

Most cases used to develop the cyclic stress method boundary curves involve liquefaction at sites with a cap layer thickness between 1 and 3 meters. No consideration was given to the strength or condition (weathering, fissuring, variability, etc.) of the cap layer. As such, the influence of the cap, especially relatively thin or thick caps, is not well-documented. Furthermore, watertable depth and thickness and location of potential source sands at the time of the causative earthquake can profoundly affect the liquefaction manifestation and can be critical to paleoseismic analysis. Because of these unknowns, the best site for back-calculating a_max in a localized vicinity is where the cap is highly variable, especially in thickness, and thick potential source sands are exposed beneath the cap base.

Improper accounting of post-liquefaction aging or cementation can cause back-calculated accelerations to be too high or too low, depending on site-specific conditions [Olson et al. 2001, SRL, 72(1)]. The effects of aging and cementation on various field tests may lead to different back-calculated a_max values. In addition, the proper interpretation of liquefaction severity (as reflected by dike height and extent of liquefaction along the base of a cap) is central to back-analysis. The field investigation must be tailored to assess whether the site conditions will yield the maximum, minimum, or probable a_max value.

Lastly, a regional evaluation of a_max is required to estimate M – preferably by means of using many sites beyond the meizoseismal zone of the earthquake. Only by observing the regional pattern of a_max can one be reasonably confident that the interpretations of M are consistent.