A method such as ours, presented in this BSSA, uses expert opinion to chose the data sets (catalog of past earthquakes from historical or archeological records, dated cumulative offsets, trench data), but then lets the Bayesian framework determine the best parameters for given candidate renewal models, and also lets the framework come up with the optimal combination of models in the Bayesian sense. In principle, the model could end up with 0.5, .3, .2 for BPT, Weibull, and lognormal for a given fault, and 0.8, 0.15, 0.05 for another fault of possibly the same mechanism in the same type of tectonic environment. Would that be okay? Or is there something about the data that you would not quite trust that much?

Article A Bayesian Framework to Rank and Combine Candidate Recurrenc...