Seismic Unix Processing

Seismic Unix is good for modifying header values in a seismic section. The SuHeaderModify page describes how to do this.

Claritas Processing

The SEGY files created by dobs2segy2000 are not the standard IBM SEGY format. It is the Vista format (little endian (intel/amd), IEEE (ascii) encoded header, and (short) integer formatted data). Claritas can read Vista SEGY natively which eliminates the time consuming NSCU conversion process.

To read Vista SEGY in Claritas, setup READSEGY in Claritas (xsje) to read INT_FORMAT as IEEE, and FLT_FORMAT as INTEGER_2. Claritas automatically handles the endian-ness.

You can also write out to a Vista format SEGY (WRITESEGY) with the same parameters as above. These files can be read perfectly by SeisWide.

IBM format SEGY files are the default format for the WRITESEGY modules in Claritas, and therefore are simple to read and write.

The easiest way to read in a SEGY file is to use the SEGY_ANALYSER. launch this program and read in a segy file. Adjust parameters such as the endianness, int/float data format, ascii/ebcdic format to match your data. A new .job file can be launched from the segy_analyser with all of the appropriate values filled in.

SEMBSMOOTH: this is a noise reduction (coherency) filter that can be used to remove incoherent noise from OBS (or other) data. It is easy to use and works well. Choose stacked mode, NTRPANEL 10, DX 13, PMAX 0.3 to start with, and modify these values until a good result is obtained. It appears that it is best to use this filter before other processes, such as an FK filter. Any minor artifacts from an FK filter will be magnified by SEMBSMOOTH.

Another coherency filter worth looking at is FXDECON. It is much more subtle than SEMBSMOOTH. However the FXDECON process can be repeated to give a stronger noise reduction (eg DO/FXDECON/ENDDO). Stronger FX filtering is achieved with shorter filter lengths. It is important to insure that only the last trace in every ensemble has LASTTR set to 1, and all other traces set to 0. If LASTTR is set incorrectly FXDECON will not work.

MISSING: this process will add any missing traces to the SEGY data. follow this up with SETHEADER (reel=recordnum) to set the trace number equal to the shot number. This is handy for programs (such as The Kingdom Suite) where it is convenient to have the trace number equal to the shot number and monotonically incrementing.

ClaritasTraceInterpolation is a page that describes how to fill in missing or dead traces with interpolated traces that do a good job of filling in the gaps. This is specific to Claritas.

DEBIAS: probably a good idea for all OBS SEGY, no need to modify the default settings.

DECONVOLUTION: There are several deconvolution modules in Claritas. Deconvw seems to work well on OBS data, but needs an input file (.dwf). The parameters for the gap and operator need to be carefully set in this file for best results. Here is an example parameter file »upload/deconv_param.dwf.

The gap should be set as the wavelength of one desired wavelet. The operator should be the length of the reverberations which you want to reduce (in other words, the length of the input wavelet minus the gap). These values can be approximated by using the 'auto-correlation functions' command in Xview (under zoom). If the deconvolution works well, the autocorrelation functions should show little or no ringing of the deconvolved data.

Migration: Finite difference migration is fairly easy to do in Claritas. This can be useful for reflection lines that are coincident with wide angle seismic data. This migration simply requires a velocity model containing interval velocities and trace numbers (or shot points), and some parameters to be filled out. If a velocity model has been created in Rayinvr, see ClaritasDepthConv for information on how to convert the velocity model to a Claritas style velocity model.

Filtering a specific part or "window" of a seismic section can be done in Claritas, and is discussed here: ClaritasFilteringWindow.

Claritas seems to do NMO correction much differently that SeisWide. Claritas's NMO seems suited more towards pre-stack multi-channel processing. However, TRFUDGE can be used to calculate the time shift required for this simple correction. This shift is then applied with STATIC.

Step 1: Use trfudge to calculate the required time shift for each trace, and store it in an unused header field (in this case SUBWEATH_VEL). I've made a small Fortran file to do this, so simply point TRFUDGE to use this file »upload/nmo_statics.f. Examine the header values to ensure they are correct. Also ensure that you use SETHEADER to set COORD_SCALE=1 if your offset values are in terms of metres.

Step 2: Use STATIC to apply this time shift to each trace, using the quick method. This is the last option in the STATIC window, and is not difficult. Then continue with other processing modules (eg: writesegy, etc.).