I am trying to model the uncertainty associated with earthquake location parameters (latitude, longitude, and depth) for a hypothetical seismic network.
Several times I've used Monte-Carlo modelling. I've modelled two kinds of location errors: due to inaccuracy of onset picking and due to travel time model uncertainty. For a set of testing points (grid) I did the following: compute arrival times from a supposed event occured in a point, add random values to the times. Then locate the event back by these arrivals but using a different travel time model. After several tests one can get reliable picture showing a systematic error due to travel time model + random scattering due to the travel time picking errors.
This program uses a probabilistic, non-Linear, global-search using the Metropolis-Gibbs method. The program takes into account pick uncertainties and model uncertainty.
You can calculate synthetic arrival times for given locations. Then, you can re-locate your hypothetical catalogue and estimate the of uncertainties for your hypocentral parameters.