CORAL MICROATOLL PALEOGEODESY ON SIMEULUE ISLAND, SUMATRA, REVEALS EARTHQUAKE CLUSTERS AND PERSISTENT RUPTURE SEGMENTATION

Coral microatoll paleogeodesy, with its ability to provide century-long or multi-century continuous records of high precision paleo-elevation data with remarkably precise ages, allows for unprecedented resolution in the reconstruction of the timing and extent of past megathrust ruptures. The main limitations of the method are that (1) data are commonly missing immediately following large uplift events; (2) sudden subsidence events cannot easily be distinguished from slow subsidence events; and (3) it would be difficult to distinguish paleo-events separated by less than a year, because of erosion of centuries-old “fossil” microatolls.

Simeulue island, off the west coast of northern Sumatra, straddles the boundary of the 2004 (M_W 9.1) and 2005 (M_W 8.6) Sunda megathrust ruptures. The 2004 and 2005 earthquake ruptures nucleated north and southeast of Simeulue, respectively, and each propagated bilaterally toward the 100-km-long island.

We focus on the mid-14th to mid-15th century, for which corals provide a continuous history of relative sea level at both the southern end of the 2004 patch and the northern end of the 2005 patch. U-Th dates of fossil coral microatolls suggest that northern Simeulue, which rose in 2004, experienced moderate to large uplifts around AD 1394 and 1450 and minor uplift around 1430. At a site in southern central Simeulue that rose in 2005, the 1394 and 1450 uplifts are very small or entirely absent, whereas the 1430 uplift there is as large as in 2005.

Coral paleogeodesy and historical intensity data also show that the spatial extent of an earthquake in AD 1861 was similar to that in 2005.

Collectively, the spatiotemporal distribution of uplifts around Simeulue reveals numerous ruptures that occurred either northwest or southeast of central Simeulue, but in every case for which we have data, the earthquake does not appear to have ruptured across the central Simeulue region. Thus, central Simeulue appears to be a persistent barrier to rupture.

In addition to the earthquake cluster that occurred over a 60-year period in the 14th-15th centuries at the southern end of the 2004 patch, uplift in the final event of the sequence may have been greater locally than that in 2004. This implies that additional uplift during another earthquake may occur under northern Simeulue in the coming decades.