1. iLoc, best practices
István Bondár
Research Centre for Astronomy and Earth Sciences,
Institute for Geological and Geochemical Research,
Budapest, Hungary
Data Analysis Tools and Methods in Seismology
Scientific Institute, Mohammed V University, Rabat, Morocco,
14 – 16 November 2022
2. Locating events with iLoc
• I usually run events in batch mode first, then review the log file
• Important: when using ISF file as input the very first instruction set applies to all
events!
echo ”ISFInputFile=in.isf ISFOutputFile=out.isf" | iLoc isf > my.log
• Individual event review:
echo ”ISFInputFile=evid.isf ISFOutputFile=evidout.isf" | iLoc isf
• When a database interface is used, instructions can be given to each individual
events in an instruction file
iLoc_SC3db sc3db < instructions.txt > myiloc.log
• Individual event review:
echo ”bud2015colr UpdateDB=0 FixDepth=0" | iLoc_SC3db sc3db
3. Reviewing events with iLoc
• If you are satisfied with the automatic location you don’t need to
relocate it
• OK if it has free-depth solution, small secondary azimuthal gap (sgap)
and most phases were used and they have small residuals
• I always review and relocate events that
• Have not converged
• Do not have a free-depth solution (esp. when it is fixed to 35 km)
• Moved far away from the initial hypocentre after the grid search (more
than a 100 km in distance and more than 30 km in depth)
• You know where they should be (explosions, events that caused surface
rupture)
4. What may cause problems
• Station codes or coordinates are wrong
• iLoc uses the ISC list of registered stations, you always have to check your
unregistered stations if they were in conflict with a registered station code
• Initial guess is wrong
• Reported hypocentres are incompatible with each other
• Initial depth is wrong (e.g. depth is fixed to zero for a deep event)
• Merged events
• Arrival times
• phases do not belong to the event (wrong associations)
• typos, huge picking errors, clock errors
• Basically flat misfit surface, grid search is not able to find a good
initial solution
• Can occur with poor network geometries (sgap > 300)
5. Events to look at
Look in the log file for lines like
• CAUTION: No solution found due to phase loss
• Something is very wrong! Could be bad origin time or depth
• CAUTION: No solution found due to insufficient number of phases
• Could be very few stations or something is wrong
• CAUTION: No solution found due to insufficient number of independent
phases
• Could be very few stations, very bad station geometry or something is wrong
• CAUTION: No solution found due to slow convergence
• Could be incompatible local and regional phases, bad depth, etc.
• airquake, depth fixed to surface
• depth fixed to default depth grid depth
• depth fixed to GRN-dependent depth
• Run iLoc with various initial depths, locations or origin times, if there is no
resolution try various fixed depths.
6. • Automatic solutions may get stuck in a local minima, produced airquakes or the default
depth value was not the best choice for depth. A manual review may also detect
split/joined or fake events. No automatic procedure can deal with these issues.
• However, it is unrealistic to perform a full manual review on a very large dataset. We
devised a strategy for the manual review that prioritizes the events by the potential
severity of location problems:
1. Events that iLoc failed to locate or had secondary azimuthal gap > 354°
2. Problematic solutions, abnormal exits from the iteration loop
3. Airquakes, discarded depth values due to large depth errors
4. Slow convergence
5. Events that moved by a large distance (> 100 km) from the initial guess
6. Large error ellipse (semi-major axis > 100 km)
7. Large secondary azimuthal gap (> 320°) or RMS residual (> 4s)
8. Events fixed to the default depth
9. The rest of the automatic locations are accepted without manual review.
Hierarchical manual review
7. Ground truth candidates
• In the log file look for the lines like this:
GT5cand=1 (nsta=19 ndef=30 sgap= 80.6 dU=0.278 numStaWithin10km=2)
• It means that the event could be promoted to GT5 level!
• These are your precious events, save them in a separate directory.
• Later you can use these events to measure the performance of the
various velocity models