Time dependence of PKP(BC)-PKP(DF) times: Could this be an artifact of systematic earthquake mislocations?

Differential inner core rotation was first detected from time-dependent observations of differential travel times between PKP(DF) waves (which go through the inner core) and PKP(BC) waves (which turn at the bottom of the outer core) (Song and Richards. 1996). Criticisms concerning the reported detection have been raised, the most significant one being the possible bias from potential systematic earthquake mislocations. In this study. I examine this issue by comparing both differential AB-BC and BC-DF times between pairs of events with similar locations and waveforms. The procedure is very similar to the "doublet" technique proposed recently by Poupinet et al. (2000), who concluded that the time dependence of the BC-DF times from the earthquakes in South Sandwich Islands recorded at College, Alaska observed by Song and Richards (1996) is an artifact of earthquake mislocations. Bacause the difference in ray parameters between AB (another branch of PKP, which turns at the mid-outer core) and BC is more than twice the difference between BC and DF, systematic mislocations have much larger effect on differential AB-BC times than on differential BC-DF times. My analysis of similar event pairs for the South Sandwich Islands to College, Alaska path shows a clear time-dependence of increasing BC-DF times over time. The AB-BC times of these same event pairs do not show any increasing trend, rather, they show a decreasing trend. The results rule out the possibility that the observed time dependence of the BC-DF times is caused by earthquake mislocation, providing strong support for the differential inner core rotation.