Multi-channel coherency migration (MCM) location algorithm: improving its computational efficiency and location accuracy

Research output: Contribution to conferenceAbstractpeer-review


Multichannel Coherency Migration (MCM) is a waveform based location algorithm that continuously calculates the coherencies between waveforms from various receiver pairs and migrates the source location at the point with maximum coherency without the need of phase picking. However, its computational demand is high.
We incorporate a collapsing grid method to reduce the computational time of the MCM without reducing its accuracy. Instead of finding the source location in a large searching area using a fine grid, we propose the use of iteratively finer grids around the maximum likelihood position of the hypocenter to accelerate the calculation and improve the location resolution. At each iteration the search grid volume around the potential source locations are identified by the previous iteration, and the searching area and grid element size are reduced to achieve a better resolution. We show that this iterative grid search approach facilitates near real time execution of the MCM algorithm. When using synthetic data from at least 9 stations, the collapsing grid method locates an event to the true location with up to three iterations, reducing the run time up to 64 times compared to the original MCM algorithm for the same element size and search volume area.

Further, we improve the accuracy of the collapsing grid approach. The 1st step is to reduce the focused volume, after the first iteration, the grid will contain only potential source locations of higher coherency. At the end of the 1st iteration the algorithm isolates the coherency values that are above the 99.9% quantile and their corresponding potential source locations. Next, it only focuses on the locations that appear most frequently in that list, the origin time is ignored at this stage. The grid collapses around each of those locations separately to build the new grid search volumes for the next iteration. This new approach reduces the search volume in the 2nd and 3rd iteration, every time focusing on the most likely locations resulting in a single best location in the end.

The new approach increases the confidence of the accuracy of the results as we can achieve 100% accuracy locating an event using 6 number of stations instead of 9. Both approaches are minimising the computational time to almost real time while being accurate on the result to help us make fast and confident decisions
Original languageEnglish
Number of pages1
Publication statusPublished - 13 Dec 2021
EventAGU Fall meeting 2021 - New Orleans, United States
Duration: 13 Dec 2021 → …


ConferenceAGU Fall meeting 2021
Country/TerritoryUnited States
CityNew Orleans
Period13/12/21 → …


  • waveform based location algorithm
  • multichannel coherency migration (MCM)
  • collapsing grid method

Cite this