Examination of systematic mislocation of South Sandwich Islands earthquakes using station pairs: Implications for inner core rotation

One of the most robust observations for the inner core rotation is that the differential PKP BC‐DF traveltimes from South Sandwich Islands (SSI) earthquakes to College, Alaska (and some other stations in Alaska), have increased systematically over the past 50 years. The time shift in the differential time residuals is some 0.3 s over 30 years. This temporal change is thought to result from a shift of the inner core structure from a superrotation of the inner core. However, the observation has been hotly debated and has been suggested to be an artifact of systematic earthquake mislocation. Here we examine this issue using three mutually independent approaches, with a goal to quantify the amount of the systematic mislocation, if any. All three approaches involve differencing data between north stations at a similar azimuth as the SSI‐Alaska azimuth and south stations at nearly the opposite azimuth. Mislocation along this azimuth has greatest impact on the PKP differential times. In approach 1, we examine how the double difference of P traveltimes between a north station and a south station changes with time. Because the differential apparent slowness between the P traveltimes of the pairs is 8 to 16 times that of the differential BC‐DF times, mislocation would cause a time shift in the double difference an order of magnitude larger than that of BC‐DF times. In approach 2, we compare directly difference in P arrival times between two SSI events that are close by and between a north and a south stations, which avoids using earthquake locations and Earth models entirely. In approach 3, we compare PKP traveltime residuals between a group of north and south stations. Because PKP traveltimes are generally not used for location in earthquake bulletins, they provide an independent data set to check earthquake location errors. The results from approaches 1 and 2, both of which use P data, are generally consistent. The mislocation is up to 3.6 ± 4.2 km (one standard deviation), explaining less than 10% of the observed COL temporal change. The results from approach 3 using PKP data show a larger mislocation but with a larger error (14.4 ± 10.8 km over 30 years). It can explain about 30% ± 22% of the observed COL temporal change. Therefore inner‐core superrotation is still the best explanation for the observed temporal change in the differential traveltimes.