Maximum-probability or Monte-Carlo classification?


As explained in the accompanying paper, we always compare each actual shallow earthquake to all of the plate boundary steps on Earth, one by one, and compute the relative probability that the earthquake was actually caused by that step.  These relative probabilities can be used in two different ways, resulting in subtly different sets of seismic subcatalogs:


Maximum-probability subcatalog set.  The earthquake is assigned to the plate boundary step with the greatest relative probability.  This results in a single "best-estimate" set of subcatalogs.  Note that log-cumulative-number vs. magnitude plots are included along this branch.


Monte-Carlo subcatalog sets.  The relative probabilities for all the plate-boundary steps are summed for each of the 7 plate-boundary types.  Then, a random number generator is used to assign the earthquake to one of these plate boundary types, with relative probability according to the proportions of the sums.  The result is non-unique because it depends on the random number generator and its initialization.  We include 5 alternative realizations, as 5 sets of Monte-Carlo subcatalogs.


(Fortunately, we found that most results of our study are independent of this choice of classification algorithm.  Details are presented in parallel tables in the accompanying paper.)