NAME
orbassoc - spatial grid search based real time associator/locator
SYNOPSIS
orbassoc [-start {pktid|time|OLDEST}]
[-select expr] [-auth auth] [-dbforeign dbname]
[-target_orb2dbt torb2dbt] [-delay delay_seconds]
[-pf pfname] [{-v|-vv|-vvv}] [-leavetmps]
orbin orbout ttgridf1 [ttgridf2 [...]]
DESCRIPTION
orbassoc is an ORB client program that reads candidate
arrival detections, produced by
orbdetect(1), and attempts to find associated event locations by searching
over one or more spatial grids for a candidate hypocenter
that produces theoretical time moveout to each station that most
closely matches the observations. Both P and S moveout velocities
can be used. If a suitable match is found to the detections, then
the associated arrivals, including time and phase information, are written
to an output ORB as
arrival,
assoc,
event and
origin
database rows that are encapsulated into special parameter file
packets destined for further processing by
orb2dbt(1).
Candidate detections are first assimilated into an internal pick list,
time sorted, and then candidate pick lists are sent to the grid processing
based upon a set of parameters that control the time window of the candidate
pick list and how often the list is processed. The grid processing of the
internal detection pick list can be time delayed to account for
latent data channels. There is a primitive
builtin capability to initiate pick list processing based upon detections
over a number of stations within a specified time window, similar to the network
trigger algorithm used by the old deprecated orbtrigger program.
A feature of orbassoc is that you can specify multiple search grids.
You can also specify a teleseismic global search grid. All search grids
must be pre-computed with ttgrid(1). By using multiple grids and teleseismic
grids, it is usually easy to automatically discriminate between local and
teleseismic events from local/regional network data. One can also tailor
the grids and node densities to match the prevailing seismicity patterns
in the region. By using pre-computed grids, it will be possible to do
associations/locations using 3-dimensional velocity models and complex
event location and phase dependent station corrections.
The processing from the detection-based internal pick list is done
recursively so that multiple closely spaced, or even overlapping,
events can be extracted from the picks. Each time an event is found
its associated picks plus all other picks that fall close to the predicted
arrival times are removed from the list and the remaining picks are reprocessed.
This is done until no more events can be extracted.
OPTIONS
-
-start {pktid|pkttime|OLDEST|NEWEST}
Where to start in the ORB that contains the detection and parameter object packets.
This argument is optional.
If this option is not specified, then orbassoc starts with the next
new packet on the orb.
-
-select expr
An ORB select expression that is applied to all ORB reads.
The default is /db/detection, the
packets that are produced by orbdetect(1).
-
-auth author
A name to be filled into the database auth fields.
If this argument is not specified, then the default is "orbassoc".
Note that this will override the auth parameters in the orbassoc
parameter file.
-
-dbforeign dbname
The name of a database that contains the foreign keys mapping table, snetsta,
for mapping CSS sta codes into SEED net_sta codes. This is only used when
using the station_weights parameter (see below) for specifying individual
station weights and maximum association residuals.
-
-target_orb2dbt torb2dbt
A target name for the output parameter file packet that is targeted
for orb2dbt(1). The default is "orb2dbt".
-
-delay delay
This is a time delay applied to the processing of the internal pick
list, assimilated from orbdetect(1) detection row packets,
that can be used to account for time delayed data channels. This
can be specified in seconds or in the forms mm:ss.s or
hh:mm:ss.s.
-
-pf pfname
Name of program parameter file.
The actual parameter file name is pfname.pf. If this argument
is not specified, then the default pfname is "orbassoc".
-
-v|-vv|-vvv
Increasingly verbose log output.
-
-leavetmps
If this flag is specified then all of the temporary databases created for running the
relocation codes will not be deleted. This is used for debugging purposes and would
normally never be specified.
-
orbin
Name of input ORB. This argument is required.
-
orbout
Name of output ORB. This argument is required.
-
ttgridf1 ttgridf2 ...
Names of travel time grid files, as produced by ttgrid(1).
At least one file name must be specified.
PARAMETER FILE
An example of the parameter file is:
schema css3.0 # database schema for temp databases - not used in dbgrassoc
process_time_window 500.0 # Main detection processing time window
process_ncycle 20 # how often to do detection processing, in detections
process_tcycle 0.0 # how often to do detection processing, in delta time
process_timeout 60.0 # timeout for processing detections
trigger_number_stations 5 # number of stations for trigger initiation
trigger_time_window 20.0 # time window in seconds for trigger initiation
use_associated_stations_for_best_grid no # Should the number of associated stations
# be used for determining the best grid?
detection_reject_expressions &Tbl{ # an optional list of expressions applied to
# incoming detection rows for rejecting particular
# rows based on any fields in the detection row
snr < 3.0
sta == "ABCD"
sta == "EFGH" && chan == "BHE"
}
grid_params &Arr{
local &Arr{
nsta_thresh &Tbl{ # Minimum allowable number of stations
# expressed as a function of maximum
# source-receiver distance
3.0 4
5.0 5
180.0 6
}
nxd 11 # Number of east-west grid nodes for depth scans
nyd 11 # Number of north-south grid nodes for depth scans
cluster_twin 1.5 # Clustering time window
try_S no # yes = Try observations as both P and S
# no = Observations are P only
associate_S yes # yes = Try to associate observations as both P and S
reprocess_S yes # yes = Reprocess when new S-associations found
drop_if_on_edge yes # Drop if solution is on the edge of the grid
auth oal # Set auth field in origin table
algorithm orbassoc_l # Override algorithm field in origin table
P_deltim 0.1 # Default deltim value for arrival table rows for P arrivals
S_deltim 0.2 # Default deltim value for arrival table rows for S arrivals
phase_sifter l # Iphase value for phase sifting, match to iphase in dbdetect.pf
P_channel_sifter ..Z|..Z_00 # Only do P associations with channels that
# match this expression
S_channel_sifter ..[NE] # Only do S associations with channels that
# match this expression
P_det_tmin 10.0 # This is a time window in seconds for culling closely
# spaced detections before initial P association processing
priority 5 # Grid priority - higher value selects this grid over lower priority grids
use_dwt yes # yes = Use source receiver distance weighting factor (or no)
# this one is for P associations
dwt_tbl &Tbl{ # distance weights can be input using a table or with
# the dwt_dist_near, ... parameters
0.0 2.0
5.0 1.0
10.0 0.0
}
dwt_dist_near 7.0
dwt_wt_near 1.0
dwt_dist_far 10.0
dwt_wt_far 0.1
use_dwts yes # yes = Use source receiver distance weighting factor (or no)
# this one is for S associations
dwts_dist_near 5.0
dwts_wt_near 1.0
dwts_dist_far 5.0
dwts_wt_far 0.0
closest_stations 20 # Use only the 20 closest stations to a particular source
# node in the search for defining phases
sta_defining_pxmin 0.0 # Use only stations that are to the east of the
# west edge of the search grid for defining phases
sta_defining_pxmax 100.0# Use only stations that are to the west of the
# east edge of the search grid for defining phases
sta_defining_pymin 0.0 # Use only stations that are to the north of the
# south edge of the search grid for defining phases
sta_defining_pymax 100.0# Use only stations that are to the south of the
# north edge of the search grid for defining phases
nondefining_association_P_maxresid 2.0 # maximum residual for
# non-defining P arrival associations
nondefining_association_S_maxresid 5.0 # maximum residual for
# non-defining S arrival associations
relocate rundbgenloc # Run this relocation script to refine the final
# solution
relocate_params &Arr{ # Parameters passed to relocation program
depth_floor 24.0 # This parameter is a depth floor.
}
dont_apply_weights_for_relocation yes # If yes, then do not adjust deltim according to weights
use_only_relocation yes # If relocation converges, just output the relocation
drop_original_if_relocation_fails yes # If relocation does not converge, dont output anything
drop_relocation_depth 20.0 # If relocation depth is greater than this, then dont output relocation
}
regional &Arr{
nsta_thresh 6 # Minimum allowable number of stations
nxd 11 # Number of east-west grid nodes for depth scans
nyd 11 # Number of north-south grid nodes for depth scans
cluster_twin 1.5 # Clustering time window
try_S no # yes = Try observations as both P and S
# no = Observations are P only
# phase_sifter l # Iphase value for phase sifting, match to iphase in dbdetect.pf
auth oar # Set auth field in origin table
# algorithm orbassoc # Override algorithm field in origin table
priority 2 # Grid priority - higher value selects this grid over lower priority grids
use_dwt no # yes = Use source receiver distance weighting factor (or no)
dwt_dist_near 2.0
dwt_wt_near 1.0
dwt_dist_far 6.0
dwt_wt_far 0.1
nondefining_association_P_maxresid 2.0 # maximum residual for
# non-defining P arrival associations
station_weights &Tbl{ # individual station weights and maximum association residuals
# sta_expr p_weight s_weight max_p_resid max_s_resid
PFO 1.0 0.0 5.0 0.0
AZ_.* 0.0
}
}
tele &Arr{
nsta_thresh 8 # Minimum allowable number of stations
cluster_twin 1.5 # Clustering time window
try_S no # yes = Try observations as both P and S
# no = Observations are P only
# phase_sifter t # Iphase value for phase sifting, match to iphase in dbdetect.pf
# sta_weight_radius 10.0 # Weighting value in km for mixed station spacing situations
auth oat
priority 1
number_threads 8 # Number of simultaneous threads to use for this grid
}
}
The processing of the detection-based internal pick list is controlled
by the following parameters.
-
schema
orbassoc makes a number of temporary databases which get encapsulated into ORB parameter file packets.
Programs downstream, such as orb2dbt, need to know the schema used for these encapsulated
databases. This parameter defines the temporary database schema. It should be either css3.0, or css3.1
or some other schema that was derived from these two.
-
process_time_window
This is the size of the time window in seconds that will be used to assimilate
the internal candidate pick list, assimilated from the orbdetect(1) detection packets,
for grid processing. This time window should correspond to
the total phase moveout time difference between the closest grid source node-station
distance and furthest grid source node-station distance for all grids. The larger this value,
the more picks will be included in the internal pick list and the more CPU time
and RAM memory it will take to process the picks through all of the grid searches.
This number should be set no larger than several minutes for most local/regional networks
but can be set as high as 1212, the IASPEI91 P-wave travel time for 180 degrees distance,
for processing global distributions of stations.
-
process_ncycle
This specifies how often the running internal pick list is processed through the grid
searches, as a number of picks. The grid processing will take place every process_ncycle
new picks that are added to the internal pick list. If this is set to a low number, then
the internal pick list will be grid processed often at the expense of computer resources
and the generation of many origin rows for each event. If this is set to a high number,
then the internal pick list will be grid processed infrequently resulting in less intermediate
origin rows per event but a higher time latency in determining an event. Note that the application
of the process_time_window to the pick list takes place AFTER the grid processing so that
a large value of process_ncycle will not result in gaps between successive processing cycles.
-
process_tcycle
This specifies how often the running internal pick list is processed through the grid
searches, as a time increment in seconds. The grid processing will take place whenever
the latest new detection time is greater than process_tcycle seconds plus the last
latest time when the pick list was last processed. If process_tcycle is set to 0.0, then this
method of triggering the pick list processing is disabled. This type of pick list triggering
is most useful when running the database version, dbgrassoc(1), since the detections are
always time sorted before processing. In the real-time world, detections may arrive out of
time order and using this method for triggering the pick list processing may produce unexpected
results.
-
process_timeout
This is a wall clock timeout value in seconds that is used to force grid processing of the internal
pick list as long as any new picks have been added to the list since the last grid processing
cycle. If this is set to 0, then the timeout is disabled.
-
trigger_number_stations
-
trigger_time_window
This is the trigger_number_stations number of stations with detection pick times within trigger_time_window
seconds that will initiate immediate pick list processing. This provides a simplified method for rapid
event processing similar to the old and deprecated orbtrigger program. If these parameters are not specified, then
the internal pick list processing is performed strictly according to the process_... parameters
given above.
-
use_associated_stations_for_best_grid
If this is not specified, or set to no, then the number of defining stations is used to
determine which of the grids is considered to produce the best solution. If this is set to yes,
then the number of defining plus non-defining but associated stations is used instead. In both cases
the tie breaker, in cases where two or more grids produce the same numbers of defining or associated stations,
is the grid with the lowest standard deviation of residuals.
-
detection_reject_expressions
This is an optional table with a list of regular Datascope expressions that
are each applied to every detection row as it is read from the ORB. If any of the expressions
match the new detection row, it is thrown away and not used for further processing. This provides
a means to reject detections with low snr or detections from known noise stations and/or channels.
The syntax for these expressions is the same as would be used in the subset expression applied to
a detection table in dbsubset(1).
The travel time grids are read from
the file(s), as specified by the ttgridf1 ... command line
arguments. The grids must be pre-computed with the program ttgrid(1). The
parameters for each grid are specified in the pf array grid_params.
Within the grid_params array there must be pf array entries for each
of the grids used in the association. The individual
grid_params array entries must have key names that are the
unique grid names, as assigned when running ttgrid(1), and values
that represent the association parameters for that grid. The parameters within each
of the grid_params entries are as follows.
-
nsta_thresh
If this is an integer number, then a solution for a particular grid is declared if the number of
stations corresponding to the associated arrivals is greater than
or equal to nsta_thresh. If this is a table, then it must consist of entries which are
distance-number of stations pairs. The distance refers to the maximum source-receiver distance
in degrees for defining associations. A solution for a particular grid is declared by
comparing the number of stations with the numbers of stations in the table. If the actual maximum
source-receiver distance is greater than the distance in the table, then the solution is not allowed.
This can be used to define low numbers-of-stations solutions only for situations where the stations
are relatively close to the event. In the table form, both the distance and the number of stations
must monotonically increase. Any solution with number of stations less than the first entry
in the table is dropped. Any solution with number of stations that is between two of the
table entries will use the distance value from the previous (lower) entry value.
-
nxd, nyd
For regular 3-dimensional
grids, the parameters nxd and nyd specify the number of east-west
and north-south horizontal grid nodes that are searched during the depth
scan. For regular rectilinear edp grids with multiple depths, orbassoc first searches the full
horizontal plane for the first depth (usually depth=0) and finds a best
solution. It then scans over depth, but instead of searching the full
horizontal plane for each depth, it instead searches only over nxd by
nyd horizontal grid nodes centered at the best solution for the last
depth scanned. When a best depth solution is found for all depths in this manner,
the entire horizontal plane is searched one last time at the best depth.
-
cluster_twin
A clustering time window in seconds that
is used to determine whether an observed arrival associates with a
hypothetical event location. All of the candidate observed arrivals are
shifted back to an equivalent origin time for each grid node using the
predicted travel times. All arrivals that cluster within cluster_twin
seconds are considered to be associated with the corresponding event
location in the grid and the corresponding phases that were used to compute
the travel times. 2 x cluster_twin is also used to associate both P and S
phases as non-defining associations to an event. Note that the specification
of this time window is critical for proper grid processing. This window must
be large enough to account for both uncertainties in the picks, as well as
velocity model errors and the effects of finite source grid spacing. In
particular, the cluster_twin should be set no lower that about 10 seconds
for the fixed universal teleseismic grid that can be produced by ttgrid(1).
On the other side, making this time window too large will result in spurious
associations.
-
try_S
If try_S is set to "yes" and the S-arrival times have been computed for
that grid, then both P and S travel times are tried in the original grid searches
for each observed arrival.
-
associate_S
If associate_S is set to "yes" and the S-arrival times have been computed for
that grid, S travel times are used only for non-defining S-arrival associations
with the picks.
-
reprocess_S
If reprocess_S is set to "yes" and non-defining S-arrival associations were
found, then all of the associated picks are reprocessed through the grid
search with the try_S parameter temporarily set to "yes" to try to relocate
the event using both P and S arrivals. Note that this parameter should be set to "yes" ONLY if
the original try_S parameter is set to "no" and the associate_S parameter
is set to "yes".
-
phase_sifter
The incoming picks from the pick list can be pre-sifted with the phase_sifter
parameter which should be a regular UNIX expression that is matched against
each detection phase code in the pick list (this code is set with the iphase
parameter in the orbdetect parameter file, see orbdetect(1)). Only picks that
pass this match are used for a particular grid association. If this is not specified,
then all picks are used.
-
P_channel_sifter
-
S_channel_sifter
Like the phase_sifter, these are UNIX regular expressions that are matched
against the pick channel codes for P and S associations. This can be used, for instance, to restrict
the P associations to use only vertical component picks and the S associations
to only use horizontal component picks.
-
P_det_tmin
This is a time window in seconds that can be specified (a value of 0 disables this option)
to cull out picks from the pick list, on a per channel basis, that follow closely in time,
before the grid search processing. The idea behind this parameter is to eliminate the multiple
picks that can occur after the true initial P-arrival due to multiple detection filter bands and/or
closely spaced secondary arrivals that follow after the initial P-arrival. These secondary picks
can often result in spurious associations, especially in cases with few associating stations.
This only effects the P associations.
-
auth
This can be used to set the auth field in the origin database rows.
If this specification is omitted, then the -auth command line argument
is used in its place.
-
algorithm
This can be used to set the algorithm field in the origin database rows.
If this is not specified, then the default algorithm field value in the
new origin rows is set to the grid name with the F value performance factor appended.
-
dtype
This can be used to set the dtype field in the origin database rows.
If this is not specified, then the
default dtype field value in the new origin rows is set to either r, when
the grid contains a single depth, or f, when the grid contains more than one depth.
-
P_deltim
This is a default deltim value in seconds to be used in arrival table rows
for the time uncertainty value for P arrivals. Note that this default deltim value is
divided by the total weighting factor for each arrival so that the weighting can be applied
by any relocation code.
-
S_deltim
This is a default deltim value in seconds to be used in arrival table rows
for the time uncertainty value for S arrivals. Note that this default deltim value is
divided by the total weighting factor for each arrival so that the weighting can be applied
by any relocation code.
-
drop_if_on_edge
If the solution for a particular
grid is on the map-view edge of the grid and the drop_if_on_edge
parameter is set to "yes", then the solution for that grid is
disallowed.
-
priority
This is a numerical grid priority that is used to disallow certain grid solutions
from larger and coarser grids that fall within the boundaries of smaller and finer
grids. This can be used to mitigate the problem of solutions for large events from teleseismic
grids that may have large numbers of associated stations that fall within a local
grid that ends up with a solution with a smaller number of associated stations due to the smaller
number of stations that were used to compute the local grid. Whenever a particular grid
passes its other solution rejection tests, its location is compared against the grid
boundaries of all of the other grids. If the priority of one of these other grids that
contain the solution is higher than the priority of the grid being tested, then the
grid solution with the lower priority is rejected. In general, the teleseismic grids should
have the lowest priority numbers and the local grids should have the highest. Non-overlapping
grids of similar grid density can have the same priority number.
-
sta_weight_radius
This is a radius in degrees that is used to compute a station density weighting
factor used to weight picks from each station. When a pick from a station is processed
for a particular grid, the distance to the other station locations for all of the other
picks in the list are computed and a count of unique stations is made for distances
that are less than sta_weight_radius. The inverse of the number of stations within sta_weight_radius
is used to weight that pick in computing a weighted performance function for that grid.
This sort of weighting should mitigate the problem of extreme heterogeneous spatial
station density that can occur when array or small aperture local network data is
processed together with much larger aperture network data, like data from global network stations.
Without this type of weighting a group of closely clustered stations can dominate the
grid search solution. If this is set to 0 or not specified, then no station density
weighting will be done.
-
use_dwt, dwt_dist_near, dwt_dist_far, dwt_wt_near, dwt_wt_far
If use_dwt is set to "yes", then a source-receiver distance weighting factor
will be used to weight picks for P associations only. Note that distance weighting also depends on
having distances computed in the ttgrid files. Distances are always computed
in the 1D grid files but will only be computed for the 3D grid files if
requested in the ttgrid parameter file. The distance weighting factor for each pick is
a function that is equal to dwt_wt_near, for source-receiver distances that are less
than or equal to dwt_dist_near, or equal
to dwt_wt_far, for source-receiver distances that are greater
than or equal to dwt_dist_far, or linearly interpolated for distances between
dwt_dist_near and dwt_dist_far. All distances are in degrees. Distance weighting generally
should be used primarily to increase the computational speed by specifying zero weight at the dwt_dist_far
distance. When weights for picks are zero, they are completely removed from the grid search for
defining associations. However, picks with zero weights are used in the final search for non-defining associations,
so that they can be properly removed for succeeding recursion processing.
-
use_dwts, dwts_dist_near, dwts_dist_far, dwts_wt_near, dwts_wt_far
These are the same distance weighting parameters, except for S associations. Usually, S detections are only
reliable fairly close to the source, so these parameters can be used to prevent spurious defining S associations
for stations that are distance from the source (by specifying zero weight). As with the P distance weighting, zero
S distance weighting only affects the defining S associations and not the non-defining S associations.
-
dwt_tbl, dwts_tbl
Distance-weight functions can alternatively be input via two tables; dwt_tbl for P associations
and dwts_tbl for S associations. If either of these tables is defined, then they take precedence
over the dwt_dist_near, ... parameters. The tables consist of distance-weight entries and the entries need
to be specified in order of increasing distance.
-
closest_stations
This is another way to make processing more efficient. It automatically allocates zero weights to stations beyond
a certain distance from each source grid node. In this case the distance is adjusted automatically to include
the closest_stations closest stations to each source node.
-
sta_defining_xpmin
-
sta_defining_xpmax
-
sta_defining_ypmin
-
sta_defining_ypmax
This is another way of restricting picks for defining associations to come only
from stations that fall within a bounding box relative to the search grid
boundaries. The four values are all percentages relative to the search grid
width and height respectively and have their origins at the southwest corner of the
search grid. A bounding box that was exactly coincident with the search grid boundaries would
be sta_defining_xpmin = 0.0,
sta_defining_xpmax = 100.0,
sta_defining_ypmin = 0.0,
sta_defining_ypmax = 100.0.
-
nondefining_association_P_maxresid
-
nondefining_association_S_maxresid
These define time tolerances in seconds for determining if non-defining picks should be associated with a
particular grid solution, for both P and S associations. After a particular grid solution has been found,
each pick in the internal pick list is compared
against the predicted P and S arrival times at every station, regardless of the various sifter parameters and whether or not
the pick was assigned a zero weighting. Picks that fall within the tolerances are all marked as "used" and are
not processed further in subsequent recursion processing. The pick at a particular station-channel with the lowest
residual is also output as a non-defining but associated arrival with the proper phase.
-
station_weights
This defines a table of individual stations with corresponding P and S weights and maximum P and S residuals
for non-defining associated arrivals. This provides a mechanism for assigning weights to stations on a per-station
basis. Each line in the table consists of at least two white-space separated fields. The first field is sta_expr,
an expression string that is matched against each of the pick list station codes. All stations that match the expression
will use the related p_weight, etc. Also, association maximum residuals can be specified in the optional max_p_resid
and max_s_resid fields. If the sta_expr field contains a '_' character, then it is assumed that the
SEED net_sta codes should be compared against sta_expr instead of the CSS sta. When matching
the sta_expr expressions with SEED net_sta codes, the -dbforeign command line argument
must be specified, in order to do the CSS to SEED name mappings. Each station is compared against each of the
lines in the station_weights table. The first line that matches is the one that is used.
-
no_output
If this is set to yes, then a grid solution results in no actual output. This is used to define an exclusion grid
for the purpose of finding poorly determined locations in regions where a particular network is incapable of producing
well determined locations, but not actually outputting the poorly determined locations. Without exclusion grids to cover
regions where a particular network performs poorly, the locations in these regions would tend to alias incorrectly into
more local grids where the network locations would normally be good.
-
relocate
If this parameter is specified, it should be the name of an executable event relocation script or program. Once
a final grid solution has been found and all of its non-defining associations have been determined, orbassoc
will execute this program/script, if it is specified. orbassoc first sets up a temporary database and populates
it with the tables that the relocation program/script will need. orbassoc then executes the program/script with a predefined
set of command line arguments. The program/script itself must then compute a refined event location, using the initial event
location provided by orbassoc's grid solution and all of the arrivals and associations from that solution.
The program/script can optionally compute error statistics and other event-related parameters and write these out to
new rows and tables in the temporary database along with the refined event solution. When the
relocation script/program is finished, orbassoc then reads back the refined solution and other parameters and
encapsulates these into event ORB objects for output, just like the original grid-based event solution.
This provides a simple approach for refining the locations of events off of the grid nodes and it also provides
a method for automatically determining error statistics. By using this, the search grid can be specified as coarsely
as possible, thus increasing overall efficiency, without unduly affecting the accuracy of the event locations.
A description of how the relocation script/program needs to be written can be found in rundbgenloc(1),
a relocation script in perl that runs Gary Pavlis' "generic" location code, dbgenloc(1). Note that normally
the temporary databases are deleted. The temporary databases are retained when the -leavetmps command argument is specified.
-
relocate_params
This is an optional associative array with other parameters that are passed unmodified
to the event relocation script or program.
-
dont_apply_weights_for_relocation
Normally the weights used in the grid search are used to adjust the arrival deltim attributes (arrival
time uncertainty) to reflect the weighting so that the relocation is weighted in a similar fashion. If this
parameter is set to yes, then the P and S arrival deltim values are all set the same according to
the P_deltim and S_deltim parameters.
-
use_only_relocation
This is a boolean that flags whether or not to output the original grid-based solution when the relocate
option is being used. If this is yes, then only the relocated solution is output as long as it
converges and produces a solution. If this is no, then both solutions are always output.
-
drop_original_if_relocation_fails
Normally the original grid search hypocenter is output if the relocation does not converge, regardless of how the use_only_relocation parameter is set. If
this parameter is set to yes, then the original grid search hypocenter is not output if the relocation does not converge.
-
drop_relocation_depth
This is a depth value in km that is used to screen out relocated hypocenters. If the relocation depth is greater than this value, then the relocated hypocenter is not output.
-
gridfilename
This is the name of a file that will be created and filled with the association grid performance function values as bugrid objects (see bugrid(3)). This allows for display of the association grid performance function for debugging and parameter tuning. Note that only
regular 3-D grid meshes can be currently captured in this way. If this is not sp[ecified, then no grids are saved.
-
number_threads
This is the number of simultaneous threads to use in processing this grid. If the host
computer has multiple processors, then the grid processing can be split up among the processors
and run in parallel to reduce the time it takes for each processing cycle. If this parameter is not
specified for a particular grid, then it defaults to 1. Note that each grid can use different
numbers of threads for the processing.
EXAMPLE
orbassoc -v -select /db/detection $ORB $ORB ttgrid-local_regional ttgrid-tele
DIAGNOSTICS
The output from orbassoc shows the grids used, test results, and
the possible solutions. In detail:
The first set of detections did not generate an
event association.
For the second set of detections, two grids passed
all tests and had a valid solution for the second
set of detections. The local_no_CA grid solution was
the best solution and was not screened out due to a
priority setting.
SEE ALSO
AUTHOR
Danny Harvey
Boulder Real Time Technologies, Inc.