A new paper has been published in Geophysical Research Letters, with contribution from current G&G postgraduate research student Thomas Hall, investigating Segmentation of the Sumatran fault. 

In this paper the authors use a clustering algorithm to partition seismicity into discrete clusters. From a seismology point of view cluster are representatives of source zones and depending on the spatial scale clusters can represent different geological features. In this instance they selected seismicity attributed to the great Sumatran fault and assume that defined clusters indicate segments along the Sumatran fault. From this study they show that there are 16 segments that range from 30-190km in length (Figure 1).

Link: http://onlinelibrary.wiley.com/doi/10.1002/2014GL060242/abstract 

Figure 1: Earthquake clusters along the Sumatran fault. Each individual cluster is coloured differently. Inset B explores K, the number of clusters fitted to Sumatran fault seismicity. Competing optimum K are identified by large KL indices [Krzanowski and Lai, 1988] in the range K = 2-25.