ISCLOC: The Location Algorithm Used by the ISC

István Bondár
International Seismological Centre
Pipers Lane, Thatcham, RG19 4NS, Berkshire,
United Kingdom
www.isc.ac.uk

Version 2.2.6 May, 2015

Bondár, I., and D. Storchak, 2011, Improved location procedures at the International Seismological
Centre, Geophys. J. Int., doi: 10.1111/j.1365-246X.2011.05107.x.

1
Release Notes

Dependencies  
The LAPACK and LBLAS libraries are required. They are part of the MacOS X 10.6 and later
Accelerate framework. On linux/unix machines install the corresponding rpm or download the
libraries from www.netlib.org.
The PostgreSQL client is required if the database interface is to be used. On MacOS X 10.6 – 10.8
it was part of the OS. Unfortunately, in 10.9.0 (Mavericks) Apple decided to remove the client.
However, the postgresql 9.2 libraries are still there. The necessary include files (libpq-fe.h and
postgres_ext.h) are provided in the postgresql_include directory. On linux/unix machines download
the postgresql development (to get the include files) package.
The RSTT package is required for the RSTT-enabled ISC locator. Download it from
www.sandia.gov/rstt. On MacOS X 10.6 and later no compilation is required, on linux/unix
machines compile the package by issuing
make clean_objs
make geotess
make cc
make c
The ISC locator uses Apple’s Grand Central Dispatch for optional multicore processing; obviously
it is part of the MacOS X. On linux/unix libdispatch and its dependencies have to be manually
installed (see http://mark.heily.com/project/libblocksruntime for further information).

RSTT  support  
The ISC locator can be enabled to take travel-time predictions for crustal and mantle phases from a
3D velocity model compliant with the RSTT parameterization (Myers et al., 2010). To do this,
compile the code with the Makefile in the rstt_src directory. See also the README_RSTT and
INSTALL files.

Changes  in  2.2.6  

Added support for ISF2.1 I/O files. When ISF output is required, it is produced in ISF2. Added a
new instruction, fix_depth_median that allows to fix the depth to the median of the reported depths.
Very minor bug fixes in main.c, rstt_main.c, magnitude.c, locator_parallel, phaseids_parallel.c and
travel_times_parallel.c.
Changes  in  2.2.5  
Very minor bug fixes in phaseids_parallel.c and depth_phases.c.

Changes  in  2.2.4  

Minor bug fixes in isf.c and read_textfiles.c. Extra spaces are removed from the instruction line
before interpreting it. Removed dependency on time zone in write_db.c. Added support for the
Murphy-Barker (2003) mb attenuation curves. Introduced a new RSTT-specific configuration
parameter, rstt_model, to specify the pathname of the RSTT model. The rstt_model and
use_RSTT_PgLg parameters can be specified in both the config.txt file or in the instruction line.
The RSTT-specific parameters only affect the behaviour of the RSTT-enabled version of the ISC
locator.

2
Changes  in  2.2.3  and  before  
The performance of the iscloc is boosted by using LAPACK routines to calculate the eigenvalue
decomposition of the symmetric, positive semi-definite data covariance matrix. For large events
with thousands of phases this represents a significant increase in speed. It is assumed that the
LAPACK libraries are installed in unix/linux systems; for Mac OS users LAPACK is already
included in the Accelerate framework.
We also added support for multicore processing via Apple’s Grand Central Dispatch. Several
aspects of the locator beg for parallel processing, such as phase identification, calculation of time
residuals, the block-by-block inversion of the data covariance matrix, and most trivially the
evaluation of the trial hypocentres in the NA search. Grand Central Dispatch allowed us to exploit
multicore CPUs without going through a major redesign of the location software. The multicore
version of the locator runs three times faster on a MacBookPro i7 quadcore than just using a single
core. Grand Central Dispatch obviously works on Mac OS 10.6 and later; for implementing GCD
on linux boxes, please consult the README and INSTALL files, as well as the Makefile.
We have slightly changed the logic of how the Neighbourhood Algorithm (NA) is applied. If
enabled (do_gridsearch=1), the NA search is performed in four dimensions around the initial
hypocentre. The phases are reidentified with respect to each trial hypocentre, and the misfit of the
solution is calculated. The misfit (the Lp-norm of time-defining residuals) is penalized against
using too few phases. This allows for a more thorough exploration of the search space. In order to
speed up the process, accounting for correlated error structure is disabled for events with a
reasonable network geometry and with more than 30 seismic (P or S-type) phases when calculating
the misfit of a trial hypocentre. Finally, we identify the phases according to the best solution from
the NA search, and then test for depth resolution. If there is depth resolution we continue with a
free-depth solution; otherwise we fix the depth to a region-dependent default depth. If the
discrepancy between the default depth and the depth from the NA search is larger than 20 km, we
repeat the NA search, with the depth fixed.
We introduced a new configuration parameter, use_RSTT_PgLg for the RSTT-enabled version of
the locator. If set (use_RSTT_PgLg =1) RSTT Pg/Lg predictions are used for first-arriving crustal P
(Pg or Pb) and S (Sg or Sb) phases. By default this option is turned off.
This release also includes minor bug fixes, as well as improvements in the phase identification
procedures, in the handling of readings (phase and amplitude data reported for an event from a
single station by a single agency), and duplicates (phases reported from the same station by several
agencies), as well as fine tuning of the parameters in the config.txt and ak135_model.txt files.
Note that MS values are not calculated when the input is read from ISF files. It is due to the
shortcomings of the ISF/IMS format that does not provide information on which channel the
surface wave amplitude was measured. Channel information is needed for MS computation.

3
1 Getting Started
1.1 Usage
iscloc [isf|isf2|yyyymmv|-]
• If the argument is 'isf' then iscloc will expect input from an ISF text file. Instructions will be
taken from stdin and the log will be written to the file specified in the config.txt file. The
isf_infile parameter is expected among the instructions, as well as the isf_stafile parameter,
either specified in the config.txt file or given as an instruction. Only the first instruction line
is read, and all events in the input ISF text file are processed according to these instructions.
• If the argument is 'isf2' then iscloc will expect input from an ISF2 text file. Instructions will
be taken from stdin and the log will be written to the file specified in the config.txt file. The
isf_infile parameter is expected among the instructions, as well as the isf_stafile parameter,
either specified in the config.txt file or given as an instruction. Only the first instruction line
is read, and all events in the input ISF text file are processed according to these instructions.
• If the argument is '-' then instructions will be taken from stdin and the log will be written to
the file specified in the config.txt file.
• Otherwise the argument will be used to create the filenames for the instruction file and the
log file. This reflects the ISC-specific operational batch mode.
Examples:
echo "update_db=0 isf_infile=isf.dat isf_outfile=isf.out" | iscloc isf
iscloc - < instruction_file > logfile
echo "11526867" | iscloc -
echo "11526867 indb=edit do_gridsearch=0 depth=NEIC" | iscloc -
echo "11526867 isf_outfile=11526867.isf.out depth=NEIC" | iscloc -
iscloc 200905b

1.2 Environment variables

QETC - Directory pathname for iscloc parameter data files. If it does not exist, it defaults to current
working directory.
QEDIT - Directory pathname for the instruction file (ISC specific). If it does not exist, it reads
instructions from stdin.
QREVISE - Directory pathname for the log file (ISC specific).
PGHOSTADDR, PGPORT, PGDATABASE, PGUSER, PGPASSWORD – PostgreSQL specific
environment variables.
SLBMROOT – Directory pathname for the RSTT root directory. Required if compiled with the
Makefile in rstt_src.

1.3 Instructions
Instructions can either be given on the command line or read from an instruction file. An
instruction line is expected for every event in the form:

4
 isc_evid [par=value [par=value [par [par=value...]]]
where isc_evid is the event identifier (required), and the par=value pairs denote the optional
instruction name and value pairs.
Note:
For ISF input only the first line of the instruction file is interpreted, and the instructions are applied
to all events in the input ISF file. For ISF input the isc_evid parameter is not required.
Tables 1 and 2 provide the list of instructions accepted by iscloc.
Table 1. Instructions
Name Type Description
indb string database account to read from.
outdb string database account to write to.
verbose integer Level of verbosity (0..5).
isf_infile string Pathname to the ISF input text file.
isf_outfile string Pathname to the ISF output text file.
An agency code to fix the hypocentre to. If no
code is given, the hypocentre will be fixed to the
fix_hypo string
median reported hypocentre (lat, lon, depth,
origin time) parameters.
A date-time value or an agency code to fix the
YYYY-MM-DD_HH:MI:SS.SSS origin time to. If no value or code is given, the
fix_time
or string origin time will be fixed to the median reported
origin time.
An agency code to fix the epicentre to. If no
fix_location string code is given, the epicentre will be fixed to the
median reported latitude, longitude values.
A depth value or an agency code to fix the
fix_depth double or string depth to. If no value or code is given, the depth
will be fixed to the median reported depth.
Fix the depth to the region-dependent default
fix_depth_default integer depth (from default depth grid or from Flinn-
Engdahl region number).
Fix the depth to the median of the reported
fix_depth_median integer
depths.
A latitude value or an agency code to start the
locator from. If no value or code is given, the
lat double or string
initial latitude will be set to the median reported
latitude.
A longitude value or an agency code to start the
locator from. If no value or code is given, the
lon double or string
initial longitude will be set to the median
reported longitude.
A depth value or an agency code to start the
locator from. If no value or code is given, the
depth double or string
initial depth will be set to the median reported
depth.
A date-time value or an agency code to start the
YYYY-MM-DD_HH:MI:SS.SSS locator from. If no value or code is given, the
time
or string initial origin time will be set to the median
reported origin time.
Write Neighbourhood Algorithm results to file
write_gridsearch_results integer
[0/1].

5
The instructions below can temporarily (for one event only) override the parameter values set in the
config.txt file.
Table 2. Instructions that can temporarily override configuration parameters
Name Type Description
update_db integer Write results to database [0/1].
isf_stafile string Pathname to the comma-separated station file.
Minimum number of agencies reporting depth phases or PcP/ScS for
ndepagency integer
depth resolution.
Minimum number of time-defining first arriving P and depth phase pairs
mindepthpha integer
for depth-phase depth resolution.
Minimum number of time-defining first arriving P and core phase (PcP,
mincorepha integer
ScS) pairs for depth resolution.
localdist double Maximum epicentral distance for local stations (degrees).
Minimum number of time-defining first arriving P within localdist for
minlocalsta integer
depth resolution.
spdist double Maximum epicentral distance for near-regional stations (degrees).
Minimum number of time-defining first arriving P and S pairs for depth
min_s_p integer
resolution.
do_correlated_errors integer Account for correlated travel-time prediction errors [0/1].
do_gridsearch integer Perform Neighbourhood Algorithm search [0/1].
iseed integer Random number seed.
na_radius double Search radius around starting epicentre (degrees).
na_deptol double Search radius around starting depth (km).
na_ottol double Search radius around starting origin time (s).
na_nsamplei integer Number of initial samples in NA search.
na_nsample integer Number of subsequent samples in NA search.
na_ncells integer Number of cells to be resampled in NA search.
na_itermax integer Maximum number of iterations in Neighbourhood Algorithm search.
Period tolerance around MSZ when searching for horizontal MS
pertol double
amplitudes (s)
Use RSTT crustal phase predictions [0/1]. Valid only for the RSTT-
use_RSTT_PgLg integer
enabled version of the ISC locator.
rstt_model string Full pathname for the RSTT model.

1.4 Input/output
The locator can read bulletin data either from an SQL database, or from ISF text files. With the ‘isf’
and ‘isf2’ command line argument, iscloc will read the input bulletin from the isf_infile file (in this
case the isf_stafile parameter must also be specified). With the command line arguments ‘-‘ or
‘yyyymmv’ iscloc will read the data from the database. The results can be written to an ISF output
file (described by the isf_outfile instruction), to the database, or both. The log and error files are by
default written to the stdout and stderr, respectively. We recommend that the users download the
file of station parameters (isc_stalist and IR2_stalist) from the ISC website (www.isc.ac.uk), where
it is regularly updated. These files are required for ISF and ISF2 input, respectively.
For the ISC database schema description see Appendix A or www.isc.ac.uk. The locator also reads
the data files below (see also Appendix B).

6
Table 3. Data files
Pathname Description
$QETC/iscloc/config.txt Configuration parameters.
Velocity model and phase specific
$QETC/iscloc/<ttime_table>_model.txt parameters. Currently the ak135 and jb travel-
time tables are supported.
Travel-time tables. The ak135 travel-time
tables were generated by libtau; the Jeffreys-
$QETC/<ttime_table>/*.tab
Bullen travel-time tables were generated using
the old ISC locator code.
$QETC/ak135/ELCOR.dat Ellipticity correction coefficients for ak135.
Topography/bathymetry data to calculate
bounce point corrections. The ETOPO1
$QETC/topo/etopo5_bed_g_i2.bin bedrock file was resampled to 5'x5' resolution
(binary file of etoponlat * etoponlon 2-byte
integers).
$QETC/FlinnEngdahl/FE.dat Flinn-Engdahl regionalization.
Default depth grid defined on a 0.5 x 0.5
$QETC/FlinnEngdahl/default.depth0.5.grid
degree grid.
Default depths for Flinn-Engdahl regions for
$QETC/FlinnEngdahl/grn_default_depth.<ttime_table>.dat locations where no default depth grid point
exists.
Generic variogram model to calculate a priori
$QETC/variogram/variogram.model
data covariance matrix.
Magnitude attenuation Q(d,h) curves.
Currently the GR (Gutenberg-Richter) , VC
$QETC/magnitude/<mbQ_table>mbQ.dat
(Veith-Clawson) and MB (Murphy-Barker)
curves are supported.
$QETC/Stations/isc_stalist Station parameters. Required for ISF input.
$QETC/Stations/IR2_stalist Station parameters. Required for ISF2 input.

1.5 Database interface

By default, iscloc reads event data from and writes the results to an SQL database, described by the
ISC database schema (see Appendix A and www.isc.ac.uk). The database platform implemented at
the ISC is PostgreSQL 8.4.
In PostgreSQL the database account is identified by the environment variables $PGUSER and
$PGPASSWORD, the database instance by $PGDATABASE and the database server host by
$PGHOSTADDR and $PGPORT. iscloc connects to the database using these environment
variables.
By default, it reads and writes the account identified by $PGUSER. This can be overridden by the
optional instruction line parameters indb (the database account to read from) and outdb (the
database account to write to) parameters. The user must have read permission to indb and
read/write permission to outdb. New unique identifiers (hypid, magid, stamagid, rdmagid, mszhid
and ampmagid) are obtained from a central database account, nextid_db, if necessary. Writing the
results to the database can be disabled by setting the update_db parameter to 0 in the config.txt file,
or temporarily by setting it in an event instruction line.
Note that if the database platform is anything other than PostgreSQL, the source code (iscloc.h,
main.c, pgsql_funcs.c, read_db.c and write_db.c) should be modified accordingly and the
corresponding SQL library interface should be linked to the code.

7
1.6 Configuration parameters
General iscloc configuration parameters are read from the config.txt file.
Table 4. Configuration parameters
Name Type Description Default
logfile string Pathname for the iscloc logfile. stdout
errfile string Pathname for the iscloc error file. stderr
isf_stafile string Pathname for the station file. isc_stalist
update_db integer Write results to DB [0/1]. 1
nextid_db string Get new unique ids from this account. isc
Reporter id for new hypocentres and
repid integer 100
associations. ISC-specific.
out_agency string Author for new hypocentres and associations. ISC
in_agency string Author for input associations. ISC
ttime_table string TT table name [ak135|jb]. ak135
default_depth double Used if seed hypocentre depth is null. 0.
Maximum depth error for crustal events to
maxdeperror_shallow double 30.
accept a free-depth solution.
Maximum depth error for deep events to accept
maxdeperror_deep double 60.
a free-depth solution.
etopofile string Filename for ETOPO file. etopo5_bed_g_i2.bin
etoponlon integer Number of longitude samples in ETOPO. 4321
etoponlat integer Number of latitude samples in ETOPO. 2161
etopores double Cellsize in ETOPO. 0.0833333
Minimum number of agencies reporting depth
ndepagency integer 2
phases or PcP/ScS for depth resolution.
Minimum number of time-defining first arriving P
mindepthpha integer and depth phase pairs for depth-phase depth 5
resolution.
Minimum number of time-defining first arriving P
mincorepha integer and core phase (PcP, ScS) pairs for depth 5
resolution.
Maximum epicentral distance for local stations
localdist double 0.2
(degrees).
Minimum number of time-defining first arriving P
minlocalsta integer 1
within localdist for depth resolution.
Maximum epicentral distance for near-regional
spdist double 5
stations (degrees).
Minimum number of time-defining first arriving P
min_s_p integer 5
and S pairs for depth resolution.
min_iter integer Minimum number of iterations. 4
max_iter integer Maximum number of iterations. 20
min_phases integer Minimum number of defining phases. 4
allow_damping integer Allow damping of solution [0/1]. 1
Residuals beyond sigmathres * prior
sigmathres double 4.
measurement error will be made defining.
Confidence level for formal uncertainties
confidence double 90.
[90|95|98].
do_correlated_errors integer Account for correlated travel-time prediction 1

8
 errors [0/1].
Perform Neighbourhood Algorithm (NA) search
do_gridsearch integer 1
[0/1].
iseed integer Random number seed. 5590
Search radius around starting epicentre
na_radius double 5.
(degrees).
na_deptol double Search radius around starting depth (km). 300.
na_ottol double Search radius around starting origin time (s). 30.
na_itermax integer Maximum number of iterations in NA search. 5
p-value for norm to compute misfit in NA search
na_lpnorm double 1
[1 ≤ p ≤ 2].
na_nsamplei integer Number of initial samples in NA search. 700
na_nsample integer Number of subsequent samples in NA search. 100
na_ncells integer Number of cells to be resampled in NA search. 25
Comma separated list of agencies not to be UNK,NIED,HFS,HFS1,
no_hypo_agencies string
used in setting the initial hypocentre. HFS2,NAO
body_mag_min_dist double Minimum mb distance (degrees). 21.
body_mag_max_dist double Maximum mb distance (degrees). 100.
surf_mag_min_dist double Minimum MS distance (degrees). 20.
body_mag_max_dist double Maximum MS distance (degrees). 160.
body_mag_min_per double Minimum mb period (s). 0.
body_mag_max_per double Maximum mb period (s). 3.
surf_mag_min_per double Minimum MS period (s). 10.
surf_mag_max_per double Maximum MS period (s). 60.
surf_mag_max_depth double Maximum MS depth (km). 60.
Period tolerance around MSZ when searching
pertol double 5.
for horizontal MS amplitudes (s)
Generate warning if station magnitude range
mag_range_warn_thres double 2.2
exceeds this.
Magnitude attenuation correction table for mb
mbQ_table string GR
[GR|VC|MB|none].
Use RSTT crustal phase predictions [0/1]. Valid
use_RSTT_PgLg integer only for the RSTT-enabled version of the ISC 0
locator.
rstt_model string Full pathname for the RSTT model.

Configuration parameters specific to phase names and/or velocity model are read from the
<ttime_table>_model.txt file.
Table 5. Configuration parameters specific to phases and/or velocity models
Name Type Description
moho double Depth of Moho discontinuity (km).
conrad double Depth of Conrad discontinuity (km).
moho_radius double Distance of Moho discontinuity from centre of the Earth (km).
max_depth_km double Maximum allowable event depth (km).
ssurfvel double Sg velocity for elevation corrections (km/s).
psurfvel double Pg velocity for elevation corrections (km/s).
Table of reported and corresponding ISC phase names to map reported
phase_map table
phases to IASPEI standard phase names.

9
 allowable_phases table List of phase names to which reported phases can be renamed.
List of phase names to which reported first-arriving P phases can be
allowable_first_P table
renamed.
Additional list of phase names to which reported first-arriving P phases
optional_first_P table
can be renamed.
List of phase names to which reported first-arriving S phases can be
allowable_first_S table
renamed.
Additional list of phase names to which reported first-arriving S phases
optional_first_S table
can be renamed.
List of phase names for which residuals will not be calculated (and not
no_resid_phase table
used in the location, e.g. amplitude phases).
List of a priori measurement errors within specified delta ranges for
phase_weight table IASPEI phases.
(phase, mindelta, maxdelta, measurement error values)
mb_phase table List of phase names that contribute to mb calculation.
ms_phase table List of phase names that contribute to MS calculation.

2 Event location
2.1 Initializations
The initialization step sets the starting hypocentre according to the instructions given for the event.
Since the reported hypocentres may exhibit a large scatter, and it is not guaranteed that any of them
is close to the global optimum, the initial hypocentre parameters (latitude, longitude, origin time,
depth) are set to the median of the corresponding reported hypocentre parameters. However, if the
author of the prime (preferred) hypocentre is ISC, EHB, or IASPEI, the initial hypocentre will be
set to the prime hypocentre. Furthermore, if instructions (Table 1) are given to fix or provide
starting values for any of the hypocentre parameters, they will be set according to the instructions.
The event type is read from the event table (or taken from the prime hypocentre if the input is ISF).
If the event type indicates an anthropogenic event, the depth will be fixed to the surface, unless an
instruction is given to fix the depth to some other specific value.

2.2 Phase identification

Reported phases are mapped to IASPEI standard phase names (Storchak et al., 2003) using the map
provided in phase_map (Table 5). The phase_map table lists all possible variations of reported
phase names and the corresponding IASPEI phase name. Note that if the phase_fixed flag is set in
the association table, the phase will not be renamed at all. Unrecognised phase names will be
mapped to a null value.
Once the reported phase names are mapped to the IASPEI standards, they are reidentified with
respect to the initial hypocentre. If the Neighbourhood Algorithm (NA) search is turned on, phases
are identified with respect to each trial hypocentre. Once NA is finished, the phases will be
identified with respect to the refined initial hypocentre guess obtained from the NA search. In order
to avoid oscillating solutions, phase names are kept the same during the iterations of the linearised
least-squares algorithm unless the depth crosses the Moho or Conrad discontinuities between two
iterations. In that case local and regional phases are reidentified.
Phases in a reading (phases reported by a single agency for an event at a single station) are treated
as a group in iscloc. In each reading the phases are identified with respect to the initial hypocentre.

10
Phases that are listed in no_resid_phase (Table 5) are skipped, thus preventing them to be used in
the location.
For a phase in a reading the phase type is determined by the first leg of the ray path indicated by the
first letter in the phase name; for depth phases the phase type is determined by the second letter in
the phase name (e.g. Pn, sP and pwP are P-type phases, Sn, ScP and pS are S-type phases).
Currently only P-type and S-type phases are considered. In other words, Lg, Rg, hydroacoustic or
infrasound phases are not used in the location, even if they had ak135 travel-time predictions.
If the phase_fixed flag is set in the association table for a phase, the time residual is calculated
using the fixed phase name and no further attempt is made to rename the phase. Otherwise iscloc
checks if the phase is in the list of allowable_phases (Table 5). The list of allowable phases was
introduced to prevent the locator renaming phases to unlikely 'exotic' phases, just because a travel-
time prediction fits the observed travel-time better. For instance, we do not want to reidentify a
reported Sn as SgSg, SPn or horribile dictu, sSn. Recall that phases may suffer from picking errors
or the observed travel-times may reflect true 3D structures not modelled by the velocity model.
Introducing the list of allowable phases helps to maintain the sanity of the bulletin and mitigates the
risk of misidentifying phases. However, if a reported phase is not in the list of allowable phases, it
is temporarily added to the list accepting the fact that station operators may confidently pick later
phases. In other words, exotic phase names can appear in the final bulletin only if they were
reported as such.
For each phase in a reading iscloc loops through the (possibly amended) list of allowable phases
and calculates the time residual with respect to the initial hypocentre. Certain rules apply:
• Cannot rename a P-type phase to S-type, and vice versa.
• A phase name can appear only once in a reading.
Further restrictions apply to first-arriving P and S phases. First-arriving P and S phases can be
identified as those in the list of allowable first-arriving P and S phases (Table 5). Occasionally a
station operator may not report the true first-arriving phase due to high noise conditions. To
account for this situation the list of optional first-arriving P and S phases (Table 5) is also checked.
Finally, having observed the rules above, the phase is identified as the phase in the allowable phase
list that has the smallest residual. If no eligible phase has been found, i.e. if the smallest residual is
beyond ±60 seconds, the phase is treated as unidentified.
Once the phases are identified with respect to the initial hypocentre, iscloc sets the time defining
flag and the a priori estimate of the measurement error for each phase. The prior measurement
errors are given in phase_weight (Table 5). The time defining flag is set to true if a valid entry is
found in the phase_weight table. However, the phase is made non-defining if its residual is larger
than a sigmathres times the prior measurement error, or it was explicitly set to non-defining by an
analyst (i.e. the non_def flag is set in the association table). Note that only time-defining phases are
used in the location.
Since phases can be picked and reported by several agencies for the same station, there can be
multiple entries for the same phase arrival. At this stage iscloc considers time-defining phases only.
Arrival picks are considered duplicates if they are reported at the same site for the same event and
if they arrival times is within 0.1 seconds. To account for alternative station codes, the primary
stations codes are used. For duplicates the arrival time is taken as the mean of the arrival times, and
the phase name is forced to be the one with the smallest residual. If accounting for correlated errors
is turned off, duplicates are explicitly down-weighted. However, if correlated errors are accounted
for, down-weighting is not necessary as duplicates are simply projected to the null space.

11
2.3 Travel-time prediction
Currently iscloc supports the JB (Jeffreys and Bullen, 1940) and the ak135 (Kennett et al., 1995)
travel-time tables. One of the major advantages of using the ak135 travel-time predictions is that
they do not suffer from the baseline difference between P, S and PKP phases compared to the
Jeffreys-Bullen tables. Furthermore, ak135 offers an abundance of phases from the IASPEI
standard phase list (Storchak et al., 2003) that can be used in the location, most notably the PKP
branches and depth-sensitive phases.
The ak135 travel-time tables were generated by libtau (Buland and Chapman, 1983); the Jeffreys-
Bullen travel-time tables were generated using the old ISC locator code. Composite tables for first-
arriving P and S phases were also constructed and they are used to get a valid travel-time table
value at local/regional crossover distances without reidentifying the phase during the subsequent
iterations of the location algorithm.
The travel-time table values for a given phase, delta and depth are calculated by bicubic spline
interpolation. Vertical slowness table values are calculated only if vertical partial derivatives are
requested. For depth phases bounce point distances are also calculated.
The P and S velocities in the surface layer to calculate elevation corrections are given by psurfvel
and ssurfvel (Table 5). Ellipticity corrections (Dziewonski and Gilbert, 1976; Kennett and
Gudmundsson, 1996), using the WG84 ellipsoid parameters, are added to the ak135 predictions.
For JB an approximate geoid correction is calculated to account for the ellipticity effects.
Bounce point (elevation correction at the surface reflection point) corrections are calculated by the
EHB algorithm (Engdahl et al., 1998) for depth phases. Water depth corrections for pwP are
calculated for water columns exceeding 1.5 km. We use the ETOPO1 global relief model (Amante
and Eakins, 2009), resampled to 5’ x 5’ resolution, to obtain the elevation or the water depth at the
bounce point. The ETOPO parameters are specified in the config.txt file (Table 4).

2.4 Data covariance matrix

If the observations are assumed to be independent, the data covariance matrix is a diagonal matrix
where the diagonal elements are the a priori estimates of the measurement error variances. The
phase and distance dependent measurement errors are specified in phase_weight (Table 5). The
inverse of the data covariance matrix provides the diagonal weight matrix.
When correlated travel-time prediction errors are present, the data covariance matrix is no longer
diagonal, and the redundancy in the observations reduces the effective number of degrees of
freedom (Bondár and McLaughlin, 2009). Assuming that the similarity between ray paths is well
approximated by the station separation, the covariances between station pairs can be estimated from
a generic P variogram model derived from ground truth residuals. The estimates for the elements of
the data covariance matrix are obtained as
2 2
C D (i, j ) = σ sill − γ (hij ) + δ ijσ phase
2
where σ sill denotes the background variance where the variogram levels off (i.e. where the pairs
become independent), γ ( hij ) is the variogram value for the distance, hij , between the ith and jth
2
stations, δ ij is the Kronecker delta and σ phase is the a priori estimate of the measurement error
covariance for an observed phase. The last term indicates that the measurement error variances add
to the diagonal of the covariance matrix.

12
Because in this representation the covariances depend only on station separations, the covariance
matrix (and its inverse) needs to be calculated only once. We assume that different phases owing to
the different ray paths they travel along, as well as station pairs with a separation larger than 1,000
km are uncorrelated. Hence, the data covariance matrix is a sparse, block-diagonal matrix.
Furthermore, if the stations in each phase block are ordered by their nearest neighbour distance, the
phase blocks themselves become block-diagonal. We determine the nearest neighbour ordering of
the stations by performing a single-linkage hierarchical cluster analysis (de Hoon et al., 2004;
Sibson, 1973) using the distance matrix constructed from the station separations.
To reduce the computational time of inverting large matrices we exploit the inherent block-
diagonal structure by inverting the covariance matrix block-by-block. The singular value
decomposition of the data covariance matrix is written as
C D = U D Λ D VDT
where Λ D is the diagonal matrix of eigenvalues and the columns of UD contain the eigenvectors of
CD. Let C D = BB T , with B = U D Λ1D/ 2 , then the projection matrix

W = B −1 = Λ −D1 / 2 U TD
will orthogonalise the data set and project redundant observations into the null space.

2.5 Neighbourhood Algorithm (NA) search

Linearised inversion algorithms are quite sensitive to the initial guess. However, it is not
guaranteed that any of the reported hypocentres would provide a suitable starting point for the
linearised inversion. In order to find an initial hypocentre guess for the linearised inversion we run
the Neigbourhood Algorithm (Sambridge, 1999; Sambridge and Kennett, 2001; Kennett, 2006)
around the starting hypocentre.
The parameters governing the NA search are read from the config.txt file (Table 4) or can be given
as instructions (Table 2). As with the linearised location algorithm, the forward calculations in NA
use all P and S-type phases and can account for correlated travel-time prediction errors. Note that
for events that have a reasonable network geometry (networks that are not heavily unbalanced) we
temporarily disable the accounting for correlated errors in order to improve performance. During
the NA search, we identify the phases with respect to each trial hypocentre and calculate the misfit
of the trial hypocentre. The misfit is defined as
!
!!! 𝑡𝑟𝑒𝑠! ! 𝑁𝑎𝑠𝑠 − 𝑁𝑑𝑒𝑓
𝑚𝑖𝑠𝑓𝑖𝑡 = +𝛼
𝑁𝑟𝑎𝑛𝑘 − 𝑀 𝑁𝑎𝑠𝑠
where Nass is the total number of P and S-type phases, Ndef is the number of defining phases,
Nrank is the number of independent defining phases, M is the number of degrees of freedom (the
number of model parameters), α (=4.0) is a penalty factor, and p is defined by the parameter
na_lpnorm. Recall that the phase identification process makes a phase non-defining if its residual
exceeds the threshold defined by sigmathres times the prior measurement error. Thus, each trial
hypocentre can have different number of defining phases. The second term in the misfit expression
is introduced to penalize against freakish local minima provided by just a few phases.
The NA parameters, na_nsamplei, na_nsample and na_ncells, are tuned to achieve a reasonable
compromise between speed and the exhaustive exploration of the search space. An initial sample
size (na_nsamplei) of 3000 or larger provides a very thorough initial sampling of the search space,
but for large events with thousands of phases the NA search would be very slow. An initial sample
size somewhere between 500 and 1000 typically provides good results. The parameter na_ncells

13
determines the number of the cells with the best misfits whose neighbourhood is explored in
subsequent NA iterations; na_nsample is the number of samples generated in subsequent iterations.
Thus, for na_ncells=25 and na_nsample=100, each candidate cell is resampled by four new
samples. Keeping the number of cells to be resampled relatively high prevents NA prematurely
falling into a local minima.
Local and near-regional events with lots of stations around the Pg/Pb/Pn crossover distance range
could generate misfit surfaces with quite a number of local minima. In such cases, it is advisable to
run the locator with several parameter settings for NA. On the other hand, for events where the user
may have a good initial guess for the location (especially for anthropogenic events), the NA search
could be safely disabled.
If the instruction write_gridsearch_results is set to 1, the trial hypocentres together with their misfit
and the number of defining phases, as well as the best models after every iteration are written to a
file.

2.6 Depth determination

2.6.1 Depth resolution
Solving for depth is attempted by iscloc only if the data provide sufficient depth resolution. Depth
resolution can be provided by a local network, depth phases, core reflections and to a lesser extent
near-regional secondary phases. We have developed a number of criteria to test whether the
reported data for an event have sufficient depth resolution.
1. Local network: at least minlocalsta stations with time-defining phases within
localdist epicentral distance.
2. Depth phases: at least mindepthpha time-defining first-arriving P and depth phase pairs
reported by at least ndepagency agencies (to reduce the chance of misinterpretation by a
single inexperienced analyst).
3. Core reflections: at least mincorepha time-defining first-arriving P and core reflection pairs
(PcP, ScS) reported by at least ndepagency agencies.
4. Local/near regional S: at least min_s_p time-defining P  and S  pairs within spdist epicentral
distance.  
The parameters governing the depth resolution criteria are read from the config.txt file (Table 4) or
can be given as instructions (Table 2). We attempt a free-depth solution if any of the above criteria
are satisfied; otherwise we fix the depth to a default depth, dependent on the epicentre location.

2.6.2 Default depth for fixed-depth events

If there is insufficient depth resolution provided by the data, or the depth uncertainty for a free-
depth solution exceeds a threshold (Table 4), the hypocentre depth is set to the depth from the
default depth grid if a grid point for the epicentre location exists; otherwise it is set to the median of
reported depths if the initial depth is larger than 100 km; else it is set to a depth (Bolton et al.,
2006) assigned to the corresponding Flinn-Engdahl (Young et al., 1996) geographic region.
The default depth grid (Bondár and Storchak, 2011) is defined on a 0.5º x 0.5º grid and was derived
from the free-depth solutions obtained from the relocation of the entire ISC bulletin using the new
locator as well as the EHB (Engdahl et al., 1998) free-depth solutions, including the fixed-depth
EHB earthquakes that were flagged as having reliable depth estimate (Bob Engdahl, personal
communication, 2010). In total, 815,000 events with reliable depths were used to produce the
default depth grid.

14
The default depth in a grid cell is defined as the median of all depths in the cell, provided that there
were at least five events in the cell, and the 75–25 percent quartile range was less than 100 km. The
latter constraint is imposed to avoid regions with both shallow and deep seismicity.
The default depth grid follows gridline registration, i.e. the nodes are centred on the grid line
intersections and the data points represent the median value in a cell of dimensions (gres x gres)
centred on the nodes. Therefore, a point (lat,lon) falls in a grid cell if the expression below is true.
(|lat - grid_lati| ≤ gres/2 and |lon - grid_lonj| ≤ gres/2)

2.6.3 Depth-phase depth

If the reported depth phases provide sufficient depth resolution, iscloc determines the hypocentre
depth using the depth-phase stacking method (Murphy and Barker, 2006). The depth-phase stack
provides a depth estimate independent from the one obtained by the location algorithm.
The depth-phase stack is always performed if there is depth resolution provided by the depth phases
and the linearised least-squares location algorithm has reached a convergent solution, even if the
depth was fixed either by instructions or by the location algorithm.
The depth-phase stacking method can be described in three steps.
1. The predicted moveout curves (depth phase – first arriving P arrival time) are generated for
each station as a function of depth (they also depend on the epicentral distance).
2. For each observed moveout a depth trace is generated by putting a boxcar at the
corresponding depth. The width of the boxcar is defined by the a priori measurement error
for the depth phase; it is centred on the observed moveout and then projected to the x-axis
(depth).
3. The depth traces are stacked and the depth-phase depth and its uncertainty are defined as the
median and the standard median absolute deviation (SMAD) of the stack, respectively.

2.6.4 Reporting how the hypocentre depth was obtained

The depfix field in the association table describes how the depth was obtained for an event. An
explanation is also given in the iscloc log file as well as in the ISC bulletin when searched on the
ISC website, www.isc.ac.uk.
Table 6. Depth determination
depfix Description
null Free-depth solution.
A Depth fixed by analyst.
B Beyond depth limits; depth fixed to 0/600 km.
D Depth fixed to depth-phase depth.
G Depth fixed to default depth grid value.
H Depth fixed to depth of a reported hypocentre.
M Depth fixed to median of reported hypocentre depths.
R Depth fixed to default Flinn-Engdahl region depth value.
S Anthropogenic event; depth fixed to the surface.

15
2.7 Linearised least-squares location algorithm
2.7.1 Iterative inversion scheme
An iterative linearised least-squares inversion of travel-times (Bondár and McLaughlin, 2009) is
performed by iscloc (Bondár and Storchak, 2011) to obtain a solution for the hypocentre. The
algorithm solves the matrix equation
G W m = WGm = Wd = dW
where G is the (NxM) design matrix containing the partial derivatives of N data by M model
parameters, m is the (Mx1) model adjustment vector [ΔT , Δx, Δy, Δz ]T , d is the (Nx1) vector of
time residuals. If do_correlated_errors (Table 4) is false, W represents the diagonal weight matrix.
If correlated travel-time errors are accounted for, W is the (NxN) projection matrix that
orthogonalises the data set. In other words, we solve the inversion problem in the eigen-coordinate
system in which the transformed observations are independent, that is, dW represents linear
combinations of the observed residuals, the “eigen residuals”. The equation above is solved by
singular value decomposition, which yields the general inverse
G W−1 = VW Λ W−1 U WT
and the model adjustment of
m est = G W−1d W
After the jth iteration, the model vector is adjusted such that mj+1 = mj + mest. Damping can be
applied for large condition numbers.
For free-depth solutions the depth is kept fixed for the first min_iter-1 (Table 4) iterations, to
mitigate the trade-off between depth and epicentre. After each iteration the depth is checked against
negative or excessively large values and fixed to 0 or 600 km. The depth remains fixed for the rest
of the iterations if the number of occurrences of airquakes/deepquakes exceeds 2.
The Gw matrix and the dw vector are recalculated with respect to the new solution. A phase is made
non-defining if its residual exceeds sigmathres times the a priori measurement error (see Tables 4
and 5). Phases are reidentified if the depth crosses the Moho or Conrad discontinuity. Normally the
projection matrix needs to be calculated only once. However, if defining phases were renamed or
defining phases were made non-defining, the projection matrix is recalculated.
The convergence test is based on the Paige-Saunders convergence test value (Paige and Saunders,
1982) and on the history of model and data norms. The convergence test is applied after every
iteration.

2.7.2 Formal uncertainties

Once a convergent solution is obtained, the location uncertainty is defined by the a posteriori
model covariance matrix
T
C M = G −1C D G −1 = VW Λ W−2 VWT
The model covariance matrix yields the four-dimensional error ellipsoid whose projections provide
the two-dimensional error ellipse and one-dimensional errors for depth and origin time. The error
ellipse encompassing the confidence region at a given α percentile level is defined by
(r − rloc ) T C M (r − rloc ) = κ α2

16
where rloc denotes the location vector of the epicentre. We follow Jordan and Sverdrup (1981) to
define κ α2 as

κ α2 = Ms 2 Fα ( M , K + N − M )
where the variance scaling factor s2 is defined as
1
K+ ∑ d w2
s2 = N
K +N −M
and Fα is an F statistic (Zwillinger and Kokoska, 2000) with M and K+N-M degrees of freedom at
the critical level α = confidence% (Table 4) with M = 2 and with N independent observations, that
is, the total number of observations less the number of observations projected to the null space. K is
set to a large value (99999) so that the formal uncertainty estimates approximate “coverage” error
ellipses.

2.7.3 Location quality metrics

Besides calculating the formal uncertainties, iscloc also computes various location quality metrics
(Bondár et al., 2004; Bondár and McLaughlin, 2009) based on the network geometry. The maximal
azimuthal gap, the maximal secondary azimuthal gap and the deviation from a uniformly
distributed station network are calculated for local, near-regional, teleseismic distance ranges as
well as for the entire network.
• Local network: 0 - 150 km
• Near regional: 3 - 10 degrees
• Teleseismic: 28 - 180 degrees
• Entire network: 0 - 180 degrees
The location quality metrics help to assess the accuracy of the hypocentre solution and identify
ground truth candidate events.

2.8 Magnitude calculation

Currently iscloc calculates body and surface wave magnitudes. MS is calculated for shallow events
(depth < surf_mag_max_depth) only. At least three station magnitudes are required for a network
(mb or MS) magnitude. The network magnitude is defined as the median of the station magnitudes,
and its uncertainty is defined as the standard median absolute deviation (SMAD) of the alpha-
trimmed (alpha = 20%) station magnitudes.
The station magnitude is defined as the median of reading magnitudes for a station. The reading
magnitude is defined as the magnitude computed from the maximal log(A/T) in a reading.
Amplitude magnitudes are calculated for each reported amplitude - period pairs.
Note that station, reading, amplitude and MS horizontal and vertical components are always stored
in the database, even if there were insufficient number of station magnitudes to calculate a network
magnitude.

2.8.1 Body-wave magnitudes

Body-wave magnitudes are calculated for each reported amplitude-period pairs, provided that the
phase is in the list of phases that can contribute to mb (mb_phase), the station is between the
epicentral distances body_mag_min_dist and body_mag_max_dist, and the period is between
body_mag_min_per and body_mag_max_per.

17
A reading contains all parametric data reported by a single agency for an event at a station, and it
may have several reported amplitude and periods. For each pair an amplitude mb is calculated.
Amplitude mb = log(A/T) + Q(∆,h)
If no amplitude-period pairs are reported for a reading, the body-wave magnitude is calculated
using the reported logat values.
Amplitude mb = logat + Q(∆,h)
The magnitude attenuation Q(d,h) value is calculated using the Gutenberg-Richter (Gutenberg and
Richter, 1956), or optionally, using the Veith-Clawson (Veith and Clawson, 1972) or the Murphy-
Barker (Murphy and Barker, 2003) tables. For the Gutenberg-Richter tables amplitudes are
measured in micrometers, while for the Veith-Clawson and Murphy-Barker tables amplitudes are
measured in nanometers.
Q(∆,h) = QGR(∆,h) - 3
Q(∆,h) = QVC(∆,h)
Q(∆,h) = QMB(∆,h)
For each reading iscloc finds the reported amplitude-period pair for which A/T is maximal:
Reading mb = log(max(A/T)) + Q(∆,h)
Or, if no amplitude-period pairs were reported for the reading:
Reading mb = max(logat) + Q(∆,h)
Several agencies may report data from the same station. The station magnitude is defined as the
median of the reading magnitudes for a station.
Station mb = median(Reading mb)
Once all station mb values are determined, the station magnitudes are sorted and the lower and
upper α percentiles are made non-defining. The network mb and its uncertainty are then calculated
as the median and the standard median absolute deviation (SMAD) of the alpha-trimmed station
magnitudes, respectively.

2.8.2 Surface-wave magnitudes

Surface-wave magnitudes are calculated for each reported amplitude-period pairs, provided that the
phase is in the list of phases that can contribute to MS (ms_phase), the station is between the
epicentral distances surf_mag_min_dist and surf_mag_max_dist, and the period is between
surf_mag_min_per and surf_mag_max_per.
For each reported amplitude-period pairs MS is calculated using the Prague formula (Vanek et al.,
1962). Amplitude MS is calculated for each component (Z, E, N) separately.
Amplitude MS = log(A/T) + 1.66 * log(∆) + 0.3
To calculate the reading MS, iscloc first finds the reported amplitude-period pair for which A/T is
maximal on the vertical component.
MSZ = log(max(AZ/TZ)) + 1.66 * log(∆) + 0.3
Then it finds the max(A/T) for the E and N components for which the period measured on the
horizontal components is within ±pertol seconds from the period measured on the vertical
component. The horizontal MS is calculated as

18
 2(max  (𝐴! 𝑇! ))!                                                                            𝑖𝑓  𝑀𝑆!  𝑑𝑜𝑒𝑠  𝑛𝑜𝑡  𝑒𝑥𝑖𝑠𝑡

max 𝐴 𝑇 = ! !
! (max  (𝐴! 𝑇! )) + (max  (𝐴! 𝑇! ))      𝑖𝑓  𝑀𝑆!  𝑎𝑛𝑑  𝑀𝑆!  𝑒𝑥𝑖𝑠𝑡

!
2(max  (𝐴! 𝑇! ))                                                                          𝑖𝑓  𝑀𝑆!  𝑑𝑜𝑒𝑠  𝑛𝑜𝑡  𝑒𝑥𝑖𝑠𝑡

MSH = log(max(A/T)H) + 1.66 * log(∆) + 0.3

The reading MS is defined as
⎧( MS Z + MS H ) 2 if MS Z and MS H exist
⎪
Reading MS = ⎨MS H if MS Z does not exist
⎪MS if MS H does not exist
⎩ Z
Several agencies may report data from the same station. The station magnitude is defined as the
median of the reading magnitudes for a station.
Station MS = median(Reading MS)
Once all station MS values are determined, the station magnitudes are sorted and the lower and
upper α percentiles are made non-defining. The network MS and its uncertainty are then calculated
as the median and the standard median absolute deviation (SMAD) of the alpha-trimmed station
magnitudes, respectively.

3 Input/output
3.1 Database
The locator reads the event, hypocenter, hypoc_err, hypoc_acc, association, phase, amplitude,
report and site tables. Note that some of the fields are read from the database only for the sake of
producing an optional ISF output file.
Note that several important assumptions are made:
• There is a prime (preferred) hypocentre identified by the prime_hyp in the event table;
• There is an author identified by in_agency (Table 4) in the event table for every physical
event.
• The etype field in the event table describes the preferred event type (if it is null, ‘ke’ is
taken by default) when the author is in_agency.
• The author of the phases associated to the prime hypocentre is identified by in_agency.
Once the data are read from the database, the reported hypocentres are sorted by their score, the
prime hypocentre being the first. A reported phase name may be replaced if it was reidentified by
NEIC, CSEM, EHB or IASPEI. The phases are sorted by delta, prista, rdid, and time.

Table 7. Data read from database tables

Table Field Description
event prime_hyp hypocentre id of the prime hypocentre
event evid physical event id for the prime hypocentre

19
event etype event type
hypocenter hypid hypocentre id
hypocenter day origin time to the second
hypocenter msec origin time milliseconds
hypocenter lat event latitude
hypocenter lon event longitude
hypocenter depth event depth (depdp if depth is null)
hypocenter depfix fixed depth flag (for ISF output only)
hypocenter epifix fixed epicentre flag (for ISF output only)
hypocenter timfix fixed origin time flag (for ISF output only)
hypocenter mindist closest station distance (for ISF output only)
hypocenter maxdist farthest station distance (for ISF output only)
hypocenter azimgap azimuthal gap (for ISF output only)
hypocenter author author of hypocentre report
hypocenter nsta number of readings (for ISF output only)
hypocenter ndefsta number of defining stations (for ISF output only)
hypocenter nass number of associated phases (for ISF output only)
hypocenter mdef number of defining phases (for ISF output only)
hypoc_err sminax semi-minor axis (for ISF output only)
hypoc_err smajax semi-major axis (for ISF output only)
hypoc_err strike azimuth of semi-major axis (for ISF output only)
hypoc_err stime origin time uncertainty (for ISF output only)
hypoc_err sdepth depth uncertainty (for ISF output only)
hypoc_err sdobs standard error of residuals (for ISF output only)
hypoc_acc score event score (to sort hypocentres)
site prista primary station code (sta if prista is null)
site lat station latitude
site lon station longitude
site elev station elevation
reporter reporter reporting agency (phase data)
association reporter data report id
phase associated to the prime hypocentre, and/or
association phase
phase associated to NEIC, EHB, IASPEI, CSEM hypocentres
association delta epicentral distance
association phase_fixed phase name fixed by analyst
association nondef phase made non-defining by analyst
phase rdid reading id
phase phid phase id
phase sta station code
phase phase reported phase name
phase day reported arrival time to the second
phase msec reported arrival time milliseconds
phase slow reported slowness (for ISF output only)
phase azim reported station-to-event azimuth (for ISF output only)
phase chan reported channel (for ISF output only)
phase sp_fm reported short period first motion (for ISF output only)
phase emergent reported emergent flag (for ISF output only)

20
 phase impulsive reported impulsive flag (for ISF output only)
amplitude ampid amplitude id
amplitude amp reported amplitude
amplitude per reported period
amplitude logat reported log(A/T)
amplitude chan reported channel of amplitude measurement
amplitude amptype reported amplitude type

WARNING: on output, iscloc overwrites the associations that belong to the prime solution!
If a convergent solution is reached, the previous in_agency solution, if any, is removed from the
database. A new hypid is obtained if the previous prime was not in_agency and the event table is
updated accordingly. The hypocenter, hypoc_err, hypoc_acc, network_quality, association,
netmag, stamag, readingmag, ampmag and ms_zh tables are populated with the new solution.
If the locator failed to get a convergent solution, the event, hypocenter and association tables are
reset to the previous prime.

3.2 ISF files

A detailed description of the ISF/ISF2 formats can be found at the ISC website, www.isc.ac.uk.
The ISF2 format is an extension of the ISF format that incorporates the new International
Seismograph Station Registry station naming conventions, agency.network.station.location.
When iscloc reads its input from an ISF/ISF2 file, it is assumed that the prime hypocentre is the last
one among the hypocentres listed in the origin block of an event. The station coordinates are
provided in a separate file, isf_stafile (Table 4), and for the new International Station Registry, in
IR2_stalist. The output is written to an ISF2 format file.

4 References
Amante, C. and B.W. Eakins, 2009, ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data
Sources and Analysis. NOAA Technical Memorandum NESDIS NGDC-24, 19pp, March 2009.
Bolton, M.K., D.A. Storchak, and J. Harris, 2006, Updating default depth in the ISC bulletin, Phys.
Earth Planet. Int., 158, 27-45.
Bondár, I., and D. Storchak, 2011, Improved location procedures at the International Seismological
Centre, Geophys. J. Int., doi: 10.1111/j.1365-246X.2011.05107.x.
Bondár, I., and K. McLaughlin, 2009, Seismic location bias and uncertainty in the presence of
correlated and non-Gaussian travel-time errors, Bull. Seism. Soc. Am., 99, 172-193.
Bondár, I. and K. McLaughlin, 2009, A new ground truth data set for seismic studies, Seism. Res.
Let., 80, 465-472.
Bondár, I., S.C. Myers, E.R. Engdahl and E.A. Bergman, 2004, Epicenter accuracy based on
seismic network criteria, Geophys. J. Int., 156, 483-496, doi: 10.1111/j.1365-246X.2004.02070.x.
Buland, R. and C.H. Chapman, 1983. The computation of seismic travel times, Bull. Seism. Soc.
Am., 73, 1271-1302.
de Hoon, M.J.L., S. Imoto, J. Nolan and S. Miyano, 2004, Open source clustering software,
Bioinformatics, 20, 1453-1454.

21
Sibson, R., 1973, SLINK: An optimally efficient algorithm for the single-link cluster method.
Comp. J., 16, 30-34.
Dziewonski, A.M. and F. Gilbert, 1976, The effect of small, aspherical perturbations on travel
times and a re-examination of the correction for ellipticity, Geophys., J. R. Astr. Soc., 44, 7-17.
Engdahl, E.R., R. van der Hilst, and R. Buland, 1998. Global teleseismic earthquake relocation
with improved travel times and procedures for depth determination, Bull. Seism. Soc. Am., 88, 722-
743.
Gutenberg, B. and C.F. Richter, 1956, Magnitude and energy of earthquakes, Ann. Geof., 9, 1-5.
Jeffreys, H., and K.E. Bullen, 1940. Seismological Tables, Brit. Assoc. Adv. Sci., Gray-Milne
Trust, London.
Jordan T.H. and K.A. Sverdrup, 1981, Teleseismic location techniques and their application to
earthquake clusters in the South-Central Pacific, Bull. Seism. Soc. Am., 71, 1105-1130.
Kennett, B.L.N., 2006. Non-linear methods for event location in a global context, Phys. Earth
Planet. Int., 158, 45-64.
Kennett, B.L.N., E.R. Engdahl, and R. Buland, 1995. Constraints on seismic velocities in the Earth
from traveltimes, Geophys. J. Int., 122, 108-124.
Kennett, B.L.N. and O. Gudmundsson, 1996, Ellipticity corrections for seismic phases, Geophys. J.
Int., 127, 40-48.
Murphy J.R. and B.W. Barker, 2003, Revised B(d,h) correction factors for use in estimation of mb
magnitudes, Bull. Seism. Soc. Am., 93, 1746-1764.
Murphy J.R. and B.W. Barker, 2006, Improved focal-depth determination through automated
identification of the seismic depth phases pP and sP, Bull. Seism. Soc. Am., 96, 1213-1229.
Myers, S.C, M.L. Begnaud, S. Ballard, M.E. Pasyanos, W.S. Phillips, A.L. Ramirez, M.S. Antolik,
K.D. Hutchenson, J. Dwyer, C. A. Rowe, and G. S. Wagner, 2010, A crust and upper mantle model
of Eurasia and North Africa for Pn travel time calculation, Bull. Seism. Soc. Am., 100, 640-656.
Paige, C. and M. Saunders, 1982, LSQR: An Algorithm for Sparse Linear Equations and Sparse
Least Squares, ACM Trans. Math. Soft., 8, 43-71.
Sambridge, M., 1999, Geophysical inversion with a neighbourhood algorithm. I. Searching the
parameter space, Geophys. J. Int., 138, 479-494.
Sambridge, M. and B.L.N. Kennett, 2001, Seismic event location: non-linear inversion using a
neighbourhood algorithm, Pageoph, 158, 241-257.
Storchak, D.A., J. Schweitzer and P. Bormann, 2003. The IASPEI Standard Seismic Phase List,
Seism. Res. Let., 74, 761-772.
Vanek, J., A. Zatopek, V. Karnik, N.V. Kondorskaya, Y.V. Riznichenko, Y.F. Savarensky, S.L.
Solovev and N.V. Shebalin, 1962, Standardization of magnitude scales, Bull. (Izvest.) Acad. Sci.
USSR Geophys. Ser., 2, 108-111.
Veith, K.F. and G.E. Clawson, 1972, Magnitude from short-period P-wave data, Bull. Seism. Soc.
Am., 62, 435-452.
Young, J.B., B.W. Presgrave, H. Aichele, D.A. Wiens and E.A. Flinn, 1996. The Flinn-Engdahl
regionalization scheme: the 1995 revision, Phys. Earth Planet. Int., 96, 223-297.
Zwillinger, D. and S. Kokoska, 2000, CRC Standard Probability and Statistics Tables and
Formulae, Chapman and Hall/CRC.

22
 Appendix A: ISC Database Schema

A detailed description of the entire ISC database schema can be found at the ISC website,
www.isc.ac.uk. This document describes only the database tables the ISC locator reads and writes.

With the introduction of the new ISC locator the ISC database schema has undergone some
changes:
• new tables (ampmag, readingmag, ms_zh, magnitude_type, network_quality) were
added to the schema;
• new fields were added to some tables (hypocenter, hypoc_err, hypoc_acc, netmag,
stamag, association);
• some fields became obsolete (marked as ‘not used’).

In order to maintain the consistency of the ISC database, the unique identifiers assigned to each
measurement or computed entities are governed by the sequences implemented on the main ISC
account. The main account is specified by the nextid_db configuration parameter. The locator
obtains unique identifiers from the sequences below.
• hypid – hypocentre identifier;
• magid – network magnitude identifier;
• stamag_stamagid_seq – station magnitude identifier;
• rdmagid – reading magnitude identifier;
• ampmagid – amplitude magnitude identifier;
• mszhid – MS vertical/horizontal component magnitude identifier;

23
event table
Column Type Description Comment
author varchar(16) author
banished varchar(1) banished [B|null]
evid integer reported event id primary key
lddate timestamp load date
moddate timestamp modification date
prime_hyp integer prime hypocentre id not null
ready varchar(1) ready to publish [R|T|null]
remid integer remark id
reporter integer data report id
etype varchar(4) event type [k|s|d|f][e|h|i|m|n|r|x] [ls] [uk]

Remarks
• There should be an in_agency (Table 4) author for every physical event
• The etype field contains the preferred event type when author = ‘<in_agency>’.
• The etype field is a two letter code where the first letter stands for
k – known
s – suspected
f – felt
d – damaging
and the second letter stands for
e – earthquake
h – chemical explosion
i – induced event, mine collapse
m – mine explosion
n – nuclear explosion
q – quarry blast
r – rockburst, coalbump
x – explosion
exceptions:
uk – unknown
ls – landslide
• The ready field indicates the status of the analyst review process
T – waiting for review
R – reviewed
null – not reviewed by the ISC

24
hypocenter table
Column Type Description Comment
accuracy varchar(10) location accuracy (km)
author varchar(16) author
azimgap numeric(6,3) max azimuthal gap
centroid varchar(1) centroid hypo [C|null]
coll_evid integer reported event id
day timestamp origin time to the second
depdp numeric(8,5) depth-phase depth (km)
depfix varchar(6) fixed-depth flag [null|A|B|D|G|H|M|R|S]
deprecated varchar(1) deprecated flag [D|M|P|null]
depth numeric(8,5) depth (km)
epifix varchar(6) fixed-epicentre flag [F|null]
etype varchar(4) event type
grn integer geographic region number
hypid integer hypocentre id primary key
isc_evid integer physical event id
lat numeric(8,5) latitude (decimal degrees)
lddate timestamp load date
lon numeric(8,5) longitude (decimal degrees)
magid integer magnitude id (not used)
magnitude mumeric(4,2) magnitude (not used)
magtype varchar(2) magnitude type (not used)
maxdist numeric(6,3) max station distance (decimal degrees)
mindist numeric(6,3) min station distance (decimal degrees)
moddate timestamp modification date
msec integer origin time milliseconds
nass integer # of associated phases
ndef integer # of defining phases
ndefsta integer # of defining stations
ndp integer # of depth phases
nsta integer # of readings(!) NOT # of stations
pref_hypid integer preferred hypid for an author
prime varchar(1) prime hypocentre [P|null]
remid integer remark id
reporter integer data report id
srn integer seismic region number
timfix varchar(6) fixed-origin time flag [F|null]
velo_model varchar(6) velocity model
nrank integer # of independent defining phases

Remarks
• nsta is the number of readings (a reading contains all phases reported by a single agency for
an event at a single station), not the number of distinct stations.
• The depfix field can be
null - free-depth solution
A - depth fixed by the analyst
S - anthropogenic event; depth fixed to surface
G - depth fixed to ISC default depth grid
R - depth fixed to ISC default region depth
M - depth fixed to median depth of reported hypocentres
B - beyond depth limits; depth fixed to 0/max_depth_km
H - depth fixed to depth of a reported hypocentre

25
 D - depth fixed to depth-phase depth
• pref_hypid stands for the preferred hypocentre for multiple reports from the same agency
• The deprecated field can be
D – deprecated, M – modified, P – preliminary, or null

hypoc_err table
Column Type Description Comment
author varchar(16) author
confidence varchar(6) confidence level %
deprecated varchar(1) deprecated flag [D|M|P|null]
hypid integer hypocentre id primary key
lddate timestamp load date
moddate timestamp modification date
remid integer remark id
reporter integer data report id
sdepth real depth uncertainty (km)
sdobs numeric(8,4) unweighted RMS (s)
slat numeric(8,5) latitude error (decimal degree)
slon numeric(8,5) longitude error (decimal degree)
smajax real semi-major axis (km)
sminax real semi-minor axis (km)
stime numeric(6,3) origin time uncertainty (s)
strike numeric(4,1) semi-major axis azimuth (decimal degree)
stt numeric(12,4) model covariance matrix
stx numeric(12,4) model covariance matrix
sty numeric(12,4) model covariance matrix
stz numeric(12,4) model covariance matrix
sxx numeric(12,4) model covariance matrix
sxy numeric(12,4) model covariance matrix
syy numeric(12,4) model covariance matrix
syz numeric(12,4) model covariance matrix
szx numeric(12,4) model covariance matrix
szz numeric(12,4) model covariance matrix
sdepdp real depth-phase depth uncertainty (km)

Remarks
• smajax, sminax, stime and sdepth are scaled to the confidence percent level

hypoc_acc table
Column Type Description Comment
hypid integer hypocentre id primary key
reporter integer data report id
score real event score to set prime
nstaloc integer # of defining local stations local: 0-150 km
gtcand integer GT candidate presumed location accuracy
nsta10 integer # of stations within 10 km

26
network_quality table
Column Type Description Comment
hypid integer hypocentre id not null
reporter integer data report id
type varchar(10) distance range not null
du numeric(8,5) deviation from uniformly distributed stations
gap numeric(8,5) largest azimuthal gap
secondary_gap numeric(8,5) largest secondary azimuthal gap
nsta integer # of defining stations
mindist numeric(7,3) min station distance
maxdist numeric(7,3) max station distance

primary key: (hypid, type)

Remarks
• the network quality metrics are given for various distance ranges.
• the type field describes the distance ranges:
local: 0 - 150 km
near: 3 - 10 degrees
tele: 28 - 180 degrees
whole: 0 - 180 degrees

association table
Column Type Description Comment
author varchar(16) author not null
azimdef varchar(1) defining azimuth [A|null]
azimres numeric(6,3) azimuth residual
delta numeric(6,3) epicentral distance (decimal degrees)
deprecated varchar(1) deprecated flag [D|null]
esaz numeric(6,3) event-station azimuth (decimal degrees)
hypid integer hypocentre id not null
lddate timestamp load date
moddate timestamp modification date
net varchar(6) network code
phase varchar(8) IASPEI phase code
phase_fixed varchar(1) fixed phase code {F|null]; phase cannot be renamed
phid integer phase id not null
remid integer remark id
reporter integer data report id
seaz numeric(6,3) station-event azimuth (decimal degrees)
slowdef varchar(1) defining slowness [S|null]
slowres numeric(6,3) slowness residual (s / degree)
sta varchar(6) station code
timedef varchar(1) defining time [T|null]
timeres numeric(9,3) time residual (s)
weight numeric(4,3) weight by locator 1/prior measurement error
nondef varchar(1) do not use in location [U|null]

primary key: (phid, author)

27
phase table
Column Type Description Comment
ampid integer amplitude id
author varchar(16) author
azim numeric(5,2) azimuth (decimal degrees)
chan varchar(3) channel
day timestamp arrival time to the second
delazim numeric(5,2) azimuth picking error (decimal degrees)
delslow numeric(5,2) slowness picking error (s/degree)
deltime numeric(5,2) time picking error (s)
deprecated varchar(1) deprecated flag [D|M|null]
emergent varchar(1) emergent signal [e|E|null]
impulsive varchar(1) impulsive signal [i|I|null]
init varchar(1) first phase in a reading
lddate timestamp load date
lp_fm varchar(1) long period first motion [c|C|+|d|D|-|null]
moddate timestamp modification date
msec integer arrival time milliseconds (ms)
net varchar(6) network code
phase varchar(8) reported phase code
phid integer phase id primary key
pref_rd integer preferred reading id deals with duplicates
rdid integer reading id
remid integer remark id
reporter integer data report id
slow numeric(9,2) slowness (s/degree)
sp_fm varchar(1) short period first motion [c|C|+|d|D|-|null]
sta varchar(6) station code

amplitude table
Column Type Description Comment
amp numeric(12,1) amplitude (nm)
ampid integer amplitude id primary key
amptype varchar(8) amplitude type [0-to-p|p-to-p]
author varchar(16) author
chan varchar(3) channel
day timestamp amplitude time to the second
delamp numeric(10,1) amplitude picking error (nm)
delper integer period picking error (s)
deprecated varchar(1) deprecated flag [D|M|null]
factor integer exponent for moment
lddate timestamp load date
logat numeric(9,2) log(A/T)
moddate timestamp modification date
moment numeric(8,5) scalar moment mantissa
msec integer amplitude time milliseconds (ms)
per numeric(6,2) period (s)
phid integer phase id not null
reporter integer data report id

28
netmag table
Column Type Description Comment
author varchar(16) author
deprecated varchar(1) deprecated flag [D|M|P|null]
hypid integer hypocentre id not null
lddate timestamp load date
magid integer magnitude id primary key
magnitude numeric(4,2) magnitude value
magtype varchar(6) magnitude type
moddate timestamp modification date
nsta integer # of stations
remid integer remark id
reporter integer data report id
uncertainty numeric(4,2) magnitude error
nagency integer # number of agencies

stamag table
Column Type Description Comment
ampid integer amplitude id (not used)
author varchar(16) author
deprecated varchar(1) deprecated flag [D|M|null]
hypid integer hypocentre id not null
lddate timestamp load date
magid integer magnitude id
magnitude numeric(4,2) magnitude value
magtype varchar(6) magnitude type
moddate timestamp modification date
phase varchar(8) phase code (not used)
phid integer phase id (not used)
remid integer remark id
reporter integer data report id
uncertainty numeric(4,2) magnitude error (not used)
stamagid integer station magnitude id primary key
magdef varchar(1) defining magnitude [D|null]
rdid integer reading id
sta varchar(6) station code

Remarks
• stamag stores the station magnitudes calculated from the reading magnitudes.
• the magdef field indicates whether the station magnitude contributed to the calculation of
network magnitude uncertainty. It is set to 1 if the magnitude is between the 20th and 80th
percentile ranges, otherwise to 0.
• stamag is populated even if there are an insufficient number of stations to calculate a
network magnitude. In that case magid is null.

magnitude_type table
Column Type Description Comment
mtypeid integer magnitude type id primary key
mtype varchar(20) magnitude type
description varchar(50) description

29
Remarks
• magnitude_type lists the IASPEI and reported magnitude types.

readingmag table
Column Type Description Comment
rdmagid integer reading magnitude id primary key
hypid integer hypocentre id not null
magid integer magnitude id
repid integer data report id
rdid integer reading id not null
sta varchar(6) station code
mtypeid integer magnitude type id not null
magnitude numeric(4,2) magnitude value
magdef integer defining magnitude [0|1]
author varchar(16) author

Remarks
• readingmag stores the magnitudes calculated for a reading from individual amplitude
reports.
• the magdef field indicates whether the reading magnitude contributed to the station
magnitude. It is set to 1 for the median magnitude, otherwise to 0.
• readingmag is populated even if there is an insufficient number of stations to calculate a
network magnitude. In that case magid is null.

ampmag table
Column Type Description Comment
ampmagid integer amplitude id primary key
ampid integer amplitude id not null
hypid integer hypocentre id not null
repid integer data report id
rdid integer reading id
mtypeid integer magnitude type id not null
magnitude numeric(4,2) magnitude value
ampdef integer defining amplitude [0|1]
author varchar(16) author

Remarks
• ampmag stores the magnitudes calculated from individual amplitude reports.
• the ampdef field indicates whether the amplitude contributed to the reading magnitude. It is
set to 1 for the magnitude with max(A/T), otherwise to 0.
• ampmag is populated even if there is an insufficient number of stations to calculate a
network magnitude. In that case magid is null.

30
ms_zh table
Column Type Description Comment
mszhid integer mszh id primary key
repid integer data report id
rdid integer reading id not null
msz numeric(4,2) magnitude MS on vertical component
mszdef integer defining MS(z) [0|1]
msh numeric(4,2) magnitude MS on horizontal component
mshdef integer defining MS(h) [0|1]
hypid integer hypocentre id not null
author varchar(16) author

Remarks
• ms_zh stores the MS values calculated on horizontal and vertical components.
• the mszdef and mshdef fields indicate whether the component magnitude reading magnitude
contributed to the reading MS magnitude.
• ms_zh is populated even if there is an insufficient number of stations to calculate a network
magnitude. In that case magid is null.

report table
Column Type Description Comment
collector varchar(8) ISC personnel who parsed the report
dirname varchar(40) location of data report file
filename varchar(20) data report filename
first_hyp timestamp time of first hypocentre in report
first_phase timestamp time of first phase arrival in report
hyp_count integer # of hypocentres in report
hyp_max_lat numeric(8,5) extremes of hypocentres in report
hyp_max_lon numeric(8,5) extremes of hypocentres in report
hyp_min_lat numeric(8,5) extremes of hypocentres in report
hyp_min_lon numeric(8,5) extremes of hypocentres in report
last_hyp timestamp time of last hypocentre in report
last_phase timestamp time of last phase in report
lddate timestamp load date
moddate timestamp modification date
phase_time_precision integer precision code [0|1|2|3|null]
product varchar(20) data report type (bulletin, catalog, &c)
repid integer data report id primary key
replaceable varchar(1) not used
reporter varchar(8) agency
reporter_id varchar(24) additional notes on reporter
status integer status of data
sta_count integer # of stations in report
sta_max_lat numeric(8,5) extremes of station locations in report
sta_max_lon numeric(8,5) extremes of station locations in report
sta_min_lat numeric(8,5) extremes of station locations in report
sta_min_lon numeric(8,5) extremes of station locations in report
issue_date timestamp time data were parsed
parser varchar(30) program used to parse data to DB

Remarks
• repid called reporter in the other tables

31
site table
Column Type Description Comment
author varchar(16) author
close_date timestamp station close date
country varchar(40)
depth numeric(5,1)
digital varchar(1) digital station [0/1]
elev numeric(8,1) elevation above seal level (m)
lat numeric(8,5) latitude (decimal degrees)
lddate timestamp load date
lon numeric(8,5) longitude (decimal degrees)
moddate timestamp modification date
net varchar(6) network
open_date timestamp station open date
prinet varchar(6) primary network code
prista varchar(6) primary station code to deal with alternate codes
region varchar(80) station specific region info
remid integer remark id
reporter integer data report id
sta varchar(6) station code
staname varchar(80) station name
statype varchar(8) not used
zone varchar(30) station specific zone info
siteid integer station id primary key

32
 Appendix B: Data files

B1. General configuration parameters

The general iscloc configuration parameters are read from $QETC/iscloc/config.txt file. The format
is expected as name = value pairs. The description of the configuration parameters is given in
Table 4.

B2. Phase and velocity model specific configuration parameters

These iscloc configuration parameters are read from $QETC/iscloc/<ttime_table>_model.txt file.
Currently ttime_table = ak135 or jb is supported. The description of the configuration parameters is
given in Table 5.

B3. ak135 ellipticity correction coefficients

The ak135 ellipticity correction parameters (Kennett and Gudmundsson, 1996) are specified in the
$QETC/ak135/ELCOR.dat file. For each phase a block of tau coefficients (Dziewonski and Gilbert,
1976) is given in a format:

phase name, number of distance samples, min dist, max dist

distance (deg)
tau0 (at 0, 100, 200, 300, 500, 700 km)
tau1 (at 0, 100, 200, 300, 500, 700 km)
tau2 (at 0, 100, 200, 300, 500, 700 km)

B4. ak135 travel-time tables

The ak135 travel-time tables (Kennett et al., 1995) are listed in the
$QETC/ak135/ak135.<phase>.tab files. Note that because the Mac OS X file system (HFS+) is
case-insensitive, that is, it cannot distinguish between ak135.pP.tab and ak135.PP.tab, the depth
phases are prefixed with ‘little’ so that ak135.pP.tab becomes ak135.littlepP.tab. The travel-time
tables follow the format:

number of distance and depth samples

delta samples (max 25 in a line)
depth samples (one line)
TT table (rows - delta, columns - depth)
dtdd table (rows - delta, columns - depth)
dtdh table (rows - delta, columns - depth)
bounce point distance table (rows - delta, columns - depth) if depth phase

B5. JB travel-time tables

The jb travel-time tables (Jeffreys and Bullen 1940) are listed in the $QETC/jb/jb.<phase>.tab
files. The travel-time tables follow the same format and naming conventions as the ak135 travel-
time table files.

33
B6. Flinn-Engdahl regionalisation
The Flinn-Engdahl regionalization scheme, 1995 version (Young et al., 1996) is given in the
$QETC/Flinn_Engdahl/FE.dat file. The file follows the format:

For each latitude (from 90N to 90S) a set of longitude ranges is given (first part of the file).
The second part of the file lists the geographic region numbers for each latitude within the
corresponding longitude ranges.

B7. Default depth grid

The 0.5º x 0.5º default depth grid (Bondár and Storchak, 2011) is given in the
$QETC/Flinn_Engdahl/default_depth0.5.grid file.

Columns:
lat, lon: center of the grid cell
depth: median depth in the cell
min: minimum depth in the cell
25Q: 25th percentile depth in the cell
75Q: 75th percentile depth in the cell
max: maximum depth in the cell
N: number of observations in the cell
range: quartile range (75Q - 25Q)
Rows are ordered by descending latitude and increasing longitude.

B8. Default depths for Flinn-Engdahl regions

The default depths for each Flinn-Engdahl region (Bolton et al., 2006) are given in the
$QETC/Flinn_Engdahl/grn_default_depth.ak135.dat and the
$QETC/Flinn_Engdahl/grn_default_depth.jb.dat files.

Columns:
grn: Flinn-Engdahl geographic region number
depth: depth for GRN

B9. ETOPO topography/bathymetry

The ETOPO1 bedrock topography (Amante and Eakins, 2009) resampled to 5’ x 5’ resolution is
given in the $QETC/topo/etopo5_bed_g.i2.bin binary file.

grdfilter -I5m etopo1_bed.grd -Fg15 -D4 -Getopo5_bed.grd

Gridline node registration used
x_min: -180 x_max: 180 x_inc: 0.0833333 nx: 4321
y_min: -90 y_max: 90 y_inc: 0.0833333 ny: 2161
z_min: -10515.5 z_max: 6917.75 name: m
scale_factor: 1 add_offset: 0
The ETOPO parameters are specified in the config.txt (Table 4) file.
Format:
etoponlon * etoponlat 2-byte integers.

34
B10. Generic variogram model
The generic P variogram model (Bondár and McLaughlin, 2009) is given in the
$QETC/variogram/variogram.model file. The file follows the format:

number of samples
sill
max station separation in km
station separation (km), variance (s^2)

B11. Magnitude attenuation correction tables

The Gutenberg-Richter (Gutenberg and Richter, 1956), the Veith-Clawson (Veith and Clawson,
1972) and the Murphy-Barker (Murphy and Barker, 2003) magnitude attenuation tables are given
in the $QETC/magnitude/GRmbQ.dat, the $QETC/magnitude/VCmbQ.dat and the
$QETC/magnitude/MBmbQ.dat files. The files follow the format:

number of distance samples

delta samples
number of depth samples
depth samples
number of distance samples, number of depth samples
magnitude attenuation table Q(d, h)

B12. Station lists

The comma-separated list of registered stations is given in the $QETC/Stations/isc_stalist file. This
file is regularly updated at the ISC website (www.isc.ac.uk).

Columns:
sta, altsta, lat, lon, elevation
where sta and alsta are the station code and the alternative station code; lat and lon are the station
coordinates in decimal degrees; elevation is the station eleavtion above sea level in meters.

The station list that complies with the new standards of the International Registry of Seismographic
Stations is given in the $QETC/Stations/IR2_stalist file.

Columns:
fdsn.network.sta.location:latitude longitude elevation

35