A unified framework for relative source localization using correlograms

We study the problem of determining an unknown event location relative to previously located events using a single monitoring array in a monitoring well. We show that using the available information about the previously located events for locating new events is advantageous to localizing each event independently. By analyzing confidence regions, we compare the performance of two previously proposed localization methods, doubledifference and interferometry, in varying signal noise and velocity uncertainty. We show that the double-difference method combats the signal noise much better due to the averaging over a larger number of travel time measurements. The interferometric method is superior where the main source of error is the velocity uncertainty between the event locations and the monitoring array. We propose a hybrid method that automatically balances these two approaches and produces a location estimator that is superior to either.