| Summary: | Segmentation of the Sumatran fault is discerned using an analytical approach in which a k-means algorithm partitions earthquakes into clusters of seismicity along the fault. Clusters are tessellated into segment zones from which segment lengths and maximum credible magnitude are estimated. Decreasing the depth of seismicity sampled from 70 to 60 to 50?km reduces interaction with deeper seismicity, and results from the k-means algorithm initially suggest that the fault has K?=?14, 16, and 16 clusters, respectively. After inspection, it becomes clear that the optimum number of clusters is 16. The 16 cluster model developed into zones generates segment lengths ranging from 22 to 196?km and maximum earthquake potentials in the range of Mw 6.5-7.8. The Sumatran fault is dominated by eight great central segments distributed approximately symmetrically about Lake Maninjau. These central fault segments dominate the hazard, which is less in the far north because segments are shorter.
|
|---|
