NAME
horvel3d - 3D automatic horizon velocity or gamma (residual
nmo) picking
SYNOPSIS
horvel3d [ -Nntap ] [ -Ootap ] [ -Pstkpik ] [ -Pstkpik ] [
-Pstkpik ] [ -gamma ] [ -vvelpik ] [ -snord ] [ -tthr ] [
-pnext ] [ [ -gigate ] [ -B ] [ -A ] ] [ -x1x1 ] [ -y1y1 ] [
-x2x2 ] [ -y2y2 ] [ -x3x3 ] [ -y3y3 ] [ -x4x4 ] [ -y4y4 ] [
-cldmcldm ] [ -ildmildm ] [ -llog ] [ -vsvmin ] [ -vevmax ]
[ -liminlimin ] [ -limaxlimax ] [ -dimindimin ] [ -dimaxdi-
max ] [ -didmididmi ] [ -lidmilidmi ] [ -didmodidmo ] [
-lidmolidmo ] [ -L ] [ -FT ] [ -ordiord ] [ -iteriter ] [
-itypityp ] [ -SM ] [ -nfnf ] [ -debug ] [ -f2m ] [ -m2f ] [
-xsd ] [ -V ] [ -? ]
DESCRIPTION
horvel3d consumes semblance records from 3D CDP gathers, two
optional pick segments defining a velocity fairway, and hor-
izons picked from a stacked volume and emits a set of
automatically picked velocity or gamma functions in special
TDFN format suitable for reading by vi3d which generates a
velocity tape format volume.
The default pick format is Landmark horizon file format (5
colmuns: line #, trace #, X coordinate, Y corrdinate, time).
If xsd picks are used they must be in extended trace header
pick format with at least DphInd and LinInd toggled (LI and
DI numbering of the cells making up the survey). Currently
these are the only indeces the program uses to locate the
input semblances within the survey. The LinInd and DphInd
trace header entries of the semblance data must be correct
and correspond to the LI and DI numbers of the gathers from
which they were computed. This is the only way a semblance
record can be located on the survey grid.
If picking horizons must be done using xsd it can be done on
either inline or crossline cuts through the stacked volume,
e.g. every 10th inline. A horizon is defined by its pick
color only which currently means a limitation of 12 horizons
(an xsd limit). When picking each LI or DI slice fromt he
volume care should be taken to assign the correct color to
each horizon picked since these will have to correspond to
previously picked slices and will have to conform with sub-
sequently picked slices.
Optionally, the event horizons can be either smoothed using
a median smoother or can be fitted with a surface. It is
also a good idea to smooth the semblance gathers before
input to the picker. This can be done using fkkstrip -pass
-l3 which should produce the required smoothing and still
maintain enough resolution for a precise pick.
The stack picked horizons are used for each semblance spec-
trum to restrict the automatic picker to a single time,
rather than have it wander all over the spectrum. Lateral
(velocity) wandering is further restricted by a fairway
(generated by picking two RMS velocity curves on one of the
semblance spectra or on a vertical stack of all available
spectra. For each time the semblances are stacked vertically
over the time gate to build up signal and the resulting sem-
blance is then searched for maxima. For each horizon the
picker chooses an initial semblance peak and checks to see
if there is a velocity inversion or the peak sits at the
edge of the fairway. If so another semblance peak (along
the same curve but at a higher velocity) is chosen, and so
on up to a specified number of trials, each cycle checking
for velocity inversion. If no legitimate peak (velocity) is
found that horizon is bypassed for that CDP and the user
notified in the printout file. In the case of no horizons
producing a velocity for that location the location is
skipped.
Provision has been made for using semblances computed from
3D gathers that have been generated using a much coarser
survey grid. This is often useful in land data where the
normal bin size results in gathers having low fold and ulti-
mately produces crappy semblances. By building super gathers
much better quality semblances can be used. This coarse grid
then must be related back to the finer bins used in the
stack.
Between control points where velspec has been run the vi3d
will interpolate velocities so it's a good idea in areas of
structure that you run as many velspecs as you can stand.
The output from horvel3d is a TDFN-like format file: there
is a 0TDFN line containing the LI and DI number of the
analysis location; this is followed by regular 1TDFN ...
9TDFN lines containing the time-velocity pairs (a 9TDFN ter-
minates the sequence for that analysis location). vi3d is
used to read this file (with a suitable 1MC3D header line st
the start - see the vi3d documentation for details).
horvel3d gets both its data and its parameters from command
line arguments. The picks are in named files created by xsd
using the extended trace header format.
Command line arguments
-N ntap
Enter the input semblance set name or file immediately
after typing -N unless the input is from a pipe in
which case the -N entry must be omitted. This input
file should include the complete path name if the file
resides in a different directory. Example
-N/b/vsp/dummy tells the program to look for file
'dummy' in directory '/b/vsp'.
-O otap
Enter the output velocity TDFN function data set name
or file immediately after typing -O. The default is
stdout.
-P stkpik
For Landmark horizon files each horizon will have its
own -P[] command line entry, i.e. each horizon must be
in a separate file. For xsd horizon mode enter the
file name containing the horizon picks made on the LI
or DI xsd-plotted slices through the stacked volume.
Be sure to plot the slices using the correct sample
interval. Each horizon will be defined by a pick color
selected from the pick color palette (which you will
have to do for each horizon and for each new slice).
Save the picks using the extended trace header pick
format. Horizons do not have to go all the way across
the section but will be automatically extended to the
survey boundaries at whatever time the last pick was
made.
-gamma
Enter the command line argument '-gamma' to auto-pick
gamma or residual nmo generated semblances, e.g. sem-
blance from rnmospec
-v velpik
Enter the file name containing the fairway picks.
These consist of 2 picks made on a semblance spectrum
(again left to right, top to bottom) which represent
the lowest and highest RMS velocity-time functions
allowed, i.e. the picker will not be allowed to look
beyond these limits at each time horizon. When saving
the xsd picks use the extended trace header format and
toggle DstSgn - the trace header slot in which velspec
writes the velocity. Not used in gamma (residual) mode.
-s nord
For automatic mode enter the order of smoothing of the
semblance curves. Default = 5
-t thr
Enter lower semblance threshold to accept. Semblances
below this value will be set to zero and ignored by the
peak picker. Default: for X-T case = 0.0, for gamma-T
case = 0.5
-l log
For automatic mode enter the sequential cell cumulative
count increment to output the semblance curves between
the fairway for all horizons. Obviously only the con-
trol semblance analysis points are referred to here.
If you have say 4 analysis locations throughout the
survey and you want to generate curves for all of them
then set log to 1. The curves are output to stderr so
the horvel3d command line should have >& at the end of
it and just before the name of the file to store the
curves. The curves may then be plotted using xgraph.
(See example below). Default is no curve output. Warn-
ing: this can generate lots of output in a hurry.
-g igate
For automatic mode enter the number of samples (in
time) in the semblance gate. This is the number of
isotime semblance curves that are stacked together for
each event before the resulting semblance curve is
searched for peaks. Default = 8 samples.
-B Enter the command line argument '-B' to hang the sem-
blance time gate below the horizon time. Default is to
center the time gate on the horizon time.
-A Enter the command line argument '-A' to hang the sem-
blance time gate above the horizon time. Default is to
center the time gate on the horizon time.
-p next
For automatic mode enter the number of semblance peaks
to test at each horizon. If a velocity inversion is
found for a peak then the program searches the semblace
curve for a peak at a higher velocity (but lower sem-
blance value). Default = 3
-vs vmin
Enter optional minimum velocity to peg the zero time of
each trace. This will have no effect on the picking,
only on the output velocity tape. Default is to ignore
this option. Not used in gamma mode.
-ve vmax
Enter optional maximum velocity to peg the end time of
each trace. This will have no effect on the picking,
only on the output velocity tape. Default is to ignore
this option. Not used in gamma mode.
-limin limin
Enter the minimum LI (x-line) number of the stack. The
default is to use whatever is in the semblance line
header which may or may not be correct. It's best to
input this and the other 3 limits below. When using
semblances generated using one set of cell dimensions
(usually very coarse) and horizon stack data using
another set this command line parameter must be set to
the minimum LI number of the actual stack volume.
-limax limax
Enter the maximum LI (x-line) number of the stack. The
default is to use whatever is in the semblance line
header which may or may not be correct. It's best to
input this and the other 2 limits below. When using
semblances generated using one set of cell dimensions
(usually very coarse) and horizon stack data using
another set this command line parameter must be set to
the maximum LI number of the actual stack volume.
-dimin dimin
Enter the minimum DI (n-line) number of the stack. The
default is to use whatever is in the semblance line
header which may or may not be correct. It's best to
input this and the other limit below. When using sem-
blances generated using one set of cell dimensions
(usually very coarse) and horizon stack data using
another set this command line parameter must be set to
the minimum DI number of the actual stack volume.
-dimax dimax
Enter the maximum DI (n-line) number of the stack. The
default is to use whatever is in the semblance line
header which may or may not be correct. It's best to
input this and the other 2 limits below. When using
semblances generated using one set of cell dimensions
(usually very coarse) and horizon stack data using
another set this command line parameter must be set to
the maximum DI number of the actual stack volume.
-lidmi lidmi
Enter cell dimension in the LI direction (x-line) of
the stacked volume from which the horizon picks were
made. Default is to assume the semblances were taken
from gathers with the same cell dimensions as the stack
volume.
-didmi didmi
Enter cell dimension in the DI direction (inline) of
the stacked volume from which the horizon picks were
made. Default is to assume the semblances were taken
from gathers with the same cell dimensions as the stack
volume.
-lidmo lidmo
Enter cell dimension in the LI direction (x-line) of
the semblance data. Default is to assume the semblances
were taken from gathers with the same cell dimensions
as the stack volume.
-didmo didmo
Enter cell dimension in the DI direction (inline) of
the semblance data. Default is to assume the semblances
were taken from gathers with the same cell dimensions
as the stack volume.
-FT Enter the command line argument '-FT' to fit a surface
of specified order to each horizon.
-ord iord
Enter the order of surface to fit. The higher this
value the more detail will be preserved in the fitted
surface. Default is 14 (the maximum order).
-iter iter
Enter the number of iterations used in the computations
to obtain the fit. Default is 5
-ityp ityp
Enter the type of fit: 1 = robust fit (default); 0 =
least squares. The robust fit will be less dominated
by isolated large changes in the input surface than
will least squares.
-SM Enter the command line argument '-FT' to smooth each
horizon using an nf x nf median smoother.
-nf nf
Enter the length of one side of the box used in the
median smoother. Must be at least 3 but beware that run
time goes up as the square of this number.
-L Enter the command line argument '-L' to turn off (for
velocity) / tuen on (for gamma) calculation and removal
of linear trend line through each horizon's semblance
curve. The default mode (not -L) is some times helpful
in cases where the semblance curves are characterized
by strong trends either toward the lower fairway limit
or toward the higher limit to give the peak picker a
better chance of scoring on the first try.
-V Enter the command line argument '-V' to get additional
printout.
-? Enter the command line argument '-?' to get online
help. The program terminates after the help screen is
printed.
BUGS
No automatic picker is foolproof. Failure to provide a
tight fairway or sloppiness in picking horizons on the
stacked section will result in bad velocities even given
sophisticated logic to determine bad initial picks. Also
increasing the igate parameter increases signal to noise but
making it too large runs the risk of associating a velocity
with a strong nearby reflector you didn't want. For fairway
picking it's a good idea to take all the velspec spectra and
vertically stack them ( rstak -nnrec, where nrec is the
total number of spectra in the data set. This shows the
velocity limits overall better than looking at a few indivi-
dual spectra.
SAMPLE COMMAND LINE
1) picking velocities and diagnostics
horvel3d -Nspectra -Phorzn1 -Phorzn2 -vfwaypiks -Ovout -l50
>& curve
This generates a set of velocity functions, vout, and a set
of semblance curves for the horizons of sequential record 50
and store these in the file curve.
xgraph < curve
This plots the curves.
2) picking and velocity tape generation
horvel3d -Nspectra -Pstackpiks -vfwaypiks -Ovout
which generates the velocity functions, vout. The velocity
tape volume is then generated by
vi3d -e2800 -dt4 -Cvi3d.cards
where an end time of 2800ms with a sample interval of 4ms
has been chosen. The file vi3d.cards contains the file out-
put by horvel3d with a 1MC3D card inserted at the top.
3) picking using different cell sizes
horvel3d -Nspectra -Pstackpiks -vfwaypiks -Ovout
-limin1 -limax76 -dimin1 -dimax76
-lidmi1000 -lidmo200 -didmi600 -didmo200
where the stacked volume has LIs and Dis that range from 1 -
76 on a grid with a bin size of 200x200. The semblances on
the other hand were computed on gathers that were on a bin
1000x600.
4) picking residual nmo (gamma)
fkkstrip -Nsembs -l3 -pass | \
horvel3d -gamma -g 4 -Ph1 -Ph2 -Ph3 -Ph4 \
-x1 5600 -y1 0 -x2 5600 -y2 16000 -x3 22400 \
-y3 16000 -x4 22400 -y4 0 -Ovout
where we have smoothed the semblance gathers before input
into the picking code and where we have defined the survey
in terms of the corner XYs. The file vout is then ready for
input to vi3d or slvr (horizon driven velocity volume). The
corections are then made using rnmo.
SEE ALSO
velspec, rnmospec, vi3d, slvr
AUTHOR
Paul Gutowski, EPTG, (422) 3146, zprg03@trc.amoco.com
COPYRIGHT
copyright 2001, Amoco Production Company
All Rights Reserved
an affiliate of BP America Inc.
Man(1) output converted with
man2html