Global Catalog of Calibrated Earthquake Locations (GCCEL)
This section describes a repository of calibrated earthquake locations (referred to as the Global Catalog of Calibrated Earthquake Locations, GCCEL or “the Catalog”) that have been determined using the Hypocentroidal Decomposition algorithm for multiple event relocation (Jordan and Sverdrup, 1981), as implemented and developed in the program mloc by Eric Bergman since 1989. The source code, supporting data files and documentation for mloc are posted elsewhere on this website.
Since the late 1990s, mloc’s development has emphasized its use for obtaining calibrated locations, meaning hypocenters with minimal location bias from unknown Earth structure and with reliable estimates of uncertainty. Calibrated hypocenters are determined for clusters of earthquakes (up to about 200 events), typically in source regions up to about 150 km in size in order to minimize the biasing effects of lateral heterogeneity. For GCCEL emphasis is given to events that have been well-recorded at far-regional and teleseismic distances and for which focal depth can be constrained.
The Catalog contains the hypocenters and associated phase data for a globally-distributed set of calibrated earthquake clusters, as well as the regional velocity model, station coordinates, summary plots of the results and a commentary for each cluster. The official website also carries documentation on the standard approach used in these studies and the data file format. The Catalog does not carry the input files used for the relocation analysis with mloc, or most of the standard output files from a calibrated relocation analysis, but those files are all available on this website (see below). A large inventory of previously-analyzed clusters has been reprocessed for inclusion in the GCCEL, and new clusters will be added to the Catalog as they become available.
GCCEL is permanently archived by the U.S. Geological Survey’s National Earthquake Information Center, as an official USGS data product in a collaborative research environment called ScienceBase. The URL for GCCEL is:
The GCCEL data release on ScienceBase has been assigned a DOI of 10.5066/P95R8K8G. Harley Benz manages the publication of GCCEL on ScienceBase. At this time the ScienceBase-hosted website is the official website for GCCEL, from which all currently-available data can be downloaded.
On October 27, 2022 a paper summarizing the GCCEL project and data set has been published by Seismological Research Letters:
Bergman, E. A., H. M. Benz, W. L. Yeck, E. Karasözen, E. R. Engdahl, A. Ghods, G. P. Hayes, and P. S. Earle (2022). A Global Catalog of Calibrated Earthquake Locations, Seismol. Res. Lett. XX, 1–11, doi: 10.1785/0220220217
As of November 5, 2022, the GCCEL database contains 297 calibrated clusters. A compressed KML file of the hypocentral information in the database on this date can be downloaded from this link: earthquakelist.kml.zip (1.5 MB). The same file (uncompressed, possibly more recent) can be downloaded from the official GCCEL website (linked above), but the link for the KML file is not very obvious and the page can sometimes be quite slow to load. It can be found by clicking on the location map in the upper right corner of the main page, to go to a new page displaying the map in a resizable window. Below the window is the link [“Download as Google Earth file format (KML)”]. The uncompressed file is ~50 MB in size currently.
MLOC data for GCCEL
The folders containing all files used for the calibrated relocation analysis (with the program mloc) of every cluster in GCCEL are available for download. Researchers who have installed mloc can use these files to investigate the calibrated relocation analysis of any GCCEL cluster, or to modify the analysis to suit their own needs.
There is one caveat: Each run of mloc uses certain output data from the previous run to set starting location parameters and certain statistical parameters that change slightly from run to run. Therefore, in order to attempt to reproduce the exact output of a cluster posted in GCCEL, it would be necessary to have access to some output files of the previous run. Because of the way the files are stored it would have been a substantial amount of extra work to include those files in the posted archives. A run of mloc started from the data files of the run posted in GCCEL will not, in general, be identical to the posted output. The differences will be minor (relative to uncertainties) in most cases, but could occasionally be significant. If you wish to reproduce any GCCEL results exactly, contact Eric Bergman.