iLoc

Motivation for iLoc

• Error ellipses are often unreliable due to the violation of assumptions of independent, Gaussian errors. Similar ray paths crossing unmodeled 3D structures introduce correlated travel time prediction errors.
• Ignoring correlated errors results in underestimated error ellipses and increasing location bias with increasing number of stations.

A brief history of iLoc

• Original algorithm was developed under an AFRL contract (Bondár and McLaughlin, 2009a).
• ISC location algorithm operational since 2010 (Bondár and Storchak, 2011). The ISC has relocated its entire bulletin (Storchak et al., 2017, 2020).
• iLoc has additional features, and open source since 2014.
• CsLoc at EMSC since 2019, (Steed et al., 2019; Bondár et al., 2020).
• iLoc plugin in SeisComp since 2020.

iLoc in a nutshell

• Can be used in a routine operational environment (fast).
• Assumes that an event is already formed, and the phases associated to the event.
• Accounts for correlated travel-time prediction errors.
• Initial hypocenter guess from Neighbourhood Algorithm search (Sambridge, 1999; Sambridge and Kennett, 2001).
• Linearised inversion using a priori estimate of the full data covariance matrix (Bondár and McLaughlin, 2009a).
• Attempts for free-depth solution only if there is depth resolution, otherwise sets default depth from a global grid of reliable free depth events from historical seismicity.
• Robust magnitude estimates with uncertainties.
• Gutenberg-Richter, Veith-Clawson and Murphy-Barker depth-distance corrections for mb and mB.
• Uses seismic, hydroacoustic and infrasound observations.
• Arrival time, slowness and azimuth measurements are used in the location.
• Identifies seismic phases w.r.t the initial guess, then the best hypocenter estimate from the NA grid search.
• Uses all valid ak135 (Kennett et al., 1995) phases in location.
• Elevation and ellipticity corrections (Dziewonski and Gilbert, 1976; Kennett and Gudmundsson, 1996).
• Depth-phase bounce point corrections (Engdahl et al., 1998).
• Uses RSTT travel-time predictions for Pn/Sn and Pg/Lg (Myers et al., 2010).
• RSTT provides its own uncertainty estimates (Begnaud et al., 2020, 2021).
• Optional use of a local velocity model.
• Predictions for local phases are calculated up to 3 degrees, beyond that iLoc switches to RSTT/ak135 predictions.
• Local phase travel time predictions are calculated for Pg/Sg, Pb/Sb, Pn/Sn.
• Supports IMS1.0/ISF1.0 and ISF2.0, ISF2.1 bulletin formats.
• Supports IDC oracle, NDC-in-a-Box PostgreSQL, ISC PostgreSQL and SeisComp MySQL database schemas.
• Optional Google Earth kmz output files.
• Performs Bondár and McLaughlin, (2009b) ground truth candidate event selection test for relocated events.
• Highly parameterizable (config files, command line arguments).