Re-Evaluating Historical Earthquakes in Australia and Indonesia with Archival and Seismological Data

Abstract

Seismic hazard assessments require a detailed record of earthquake activity. This record needs to be as complete and continuous as possible in space and time to constrain the distribution and rates of seismicity. Relying on instrumentally recorded earthquakes alone fails to account for earthquakes with recurrence times longer than the 100+ year history of instrumental seismology. Therefore, seismic hazard studies must often incorporate information on earthquakes that occurred before the instrumental era, i.e., historical earthquakes.

I re-evaluate historical earthquakes in Australia and Indonesia on an event-based and regional scale by relying on primary archival and seismological materials. For Indonesia, I created the Gempa Nusantara database. It is the largest macroseismic intensity dataset ever compiled for the region, with 7380 intensity data points for 1200 historical earthquakes between 1546 and 1950. This dataset was used to examine the completeness of the historical record by examining colonial influences, comparing with the paleoseismic record and previous studies, and using these intensities to test seismic hazard curves for selected cities in Indonesia. The latter in particular demonstrated that the annual frequency of damaging shaking corresponds well with these curves for many, but not all, of Indonesia's largest cities. I also investigated the pre-digital (pre-1964) instrumental record in Indonesia by correlating instrumental readings from early seismographs with felt earthquakes. This allowed the identification of 25 instrumented earthquakes between 1889 and 1902. I also used amplitudes from Milne seismographs to calculate surface-wave magnitudes for 173 earthquakes between 1898 and 1919. Comparing felt effects with instrumental locations helped to identify earthquakes that were erroneously located by sparse instrumental data in the pre-digital period (1904 to 1964).

For Australia, I reviewed two of its largest historical earthquakes. The first was a M~6 earthquake on 7 June 1918 that was presumed to have had an epicentre off the coast of Bundaberg in Queensland. I was able to prove that this location is incorrect and that the correct location (~24.93 S and ~150.88 E) was inland. My results were guided by a review of macroseismic effects including previously untapped first-hand accounts, seismograms, and felt aftershocks. By examining the original seismograms, I show that previously reported S-P times were incorrect. S-P times I re-picked, now offer independent support for an inland source. Using instrumental amplitudes from seismological stations in Australia, I computed Ms ~5.9 and a revised Mw ~6.0 for this event.

Lastly, I re-visited macroseismic observations and seismological data from the 1954 Adelaide earthquake which is the most damaging earthquake in South Australia to date. Using a similar approach with macroseismic data, I concluded that the 1954 epicentre was further to the east (~35.05S and ~138.86E) than previously assumed. This location is supported by previously unconsidered macroseismic effects at sites within the Adelaide Hills including numerous accounts of areal and subsurface hydrological effects. I computed an instrumental Mw of ~5.4. Inverting macroseismic data gave an intensity magnitude of Mw ~6.0 for this event. The discrepancy between the two magnitudes is hypothesised to result from the event having a higher-than-normal static stress drop and a greater focal depth.

In summary, my research shows how important it is to tap often underutilised data to correct errors, fill gaps in modern earthquake catalogues and quantify uncertainties that are geological, numeric or societal. My findings also underscore the equal importance of integrating historical and scientific datasets with a balanced weighting. These steps are necessary to correctly guide future seismic hazard assessments in Australia and Indonesia, and globally.