Wireline Sonic Data Processing

ODP logging contractor: LDEO-BRG

Hole: 878A

Leg: 144

Location: MIT Guyot (tropical NW Pacific)

Latitude: 27° 19.143' N

Longitude: 151° 43.028' E

Logging date: July, 1992

Bottom felt: 1337.6 mbrf

Total penetration: 910 mbsf

Total core recovered: 265.9 m (25.5 %)

TOOL USED: SDT (Digital Sonic tool, also known as Array Sonic)

Recording mode: depth-derived, borehole-compensated mode

Remarks about the recording. Like in all of the past legs, the sonic waveforms are not on depth with the other logs: in Hole 878A they show an offset of about 38 m downward. This offset is very clearly visible at the base of the bottom hole assembly in the main section. In the repeat section, the data was recorded open hole and therefore, without the reference of the base of the bottom hole assembly, it is difficult to confirm the amount of offset; the waveforms should be processed and the output compared to the other logs.

Acoustic data were recorded in LIS format. Each of the four waveforms consists of 500 samples and is recorded at a sampling rate of 0.1524 m. The original data is first loaded on a Sun system using GeoFrame software. The waveform data files are then converted into ASCII and finally binary format.

Each row of the binary file is composed of the entire waveform set recorded at each depth, preceded by the depth. In the general case of 4 waveforms with 500 samples per waveform, this corresponds to 1 + 4x500 = 2001 columns. In this hole, the specifications of the files are:

Number of columns: 2001

Number of rows: 4958 (main)

Number of rows: 455 (repeat)

All values are stored as ' IEEE floating point numbers' (= 4 bytes).

Any numerical software or programing language (matlab, python,...) can import the files for further analysis of the waveforms.

The following files were converted:

SDT from DIT/SDT/HLDT/CNTG/NGT (main, bottom hole assembly at ~ 1352 mbrf)

878A-m.bin: 1335.33-2090.78 mbrf

SDT from DIT/SDT/HLDT/NGT (repeat, recorded open hole)

878A-r.bin: 1536.19-1605.38 mbrf

The sonic waveform files are not depth shifted to a reference run or to the seafloor. For depth shift to the sea floor, please refer to the DEPTH SHIFT section in the standard log documentation file. Also, please refer to the "Remarks about the recording" note above.

NOTE: For users interested in converting the data to a format more suitable for their own purpose, a simple routine to read the binary files would include a couple of basic steps (here in old fashioned fortran 77, but would be similar in matlab or other languages):

The first step is to extract the files dimensions and specification from the header, which is the first record in each file:

  open (1, file = *.bin,access = 'direct', recl = 50) <-- NB:50 is enough to real all fields

  read (1, rec = 1)nz, ns, nrec, ntool, mode, dz, scale, dt

  close (1)

The various fields in the header are:
      - number of depths
      - number of samples per waveform and per receiver
      - number of receivers
      - tool number (0 = DSI; 1 = SonicVISION; 2 = SonicScope; 3 = Sonic Scanner; 4 = XBAT; 5 = MCS; 6 = SDT; 7 = LSS; 8 = SST; 9 = BHC; 10 = QL40; 11 = 2PSA)
      - mode (1 = Lower Dipole, 2 = Upper Dipole, 3 = Stoneley, 4 = Monopole)
      - vertical sampling interval *
      - scaling factor for depth (1.0 = meters; 0.3048 = feet) *
      - waveform sampling rate in microseconds *

All those values are stored as 4 bytes integers, except for the ones marked by an asterisk, stored as 4 bytes IEEE floating point numbers.

Then, if the number of depths, samples per waveform/receiver, and receivers are nz, ns, and nrec, respectively, a command to open the file would be:

  open (1, file = *.bin, access = 'direct', recl = 4*(1 + nrec*ns))

Finally, a generic loop to read the data and store them in an array of dimension nrec × ns × nz would be:

  do k = 1, nz

    read (1, rec = 1+k) depth(k), ((data(i,j,k), j = 1,ns), i = 1,nrec)

  enddo

For further information about the logs please contact:

Cristina Broglia
Phone: 845-365-8343
Fax: 845-365-3182
E-mail: Cristina Broglia