3.1 Surveying technique and fie ld parameters
'l
A 15 cm Sodera watergun was the source for the multichannel survey and a 93.75 m long, 16 channel hydrophone array was used as detector. The watergun was run from a 2000 PSI compressor and fired every other second. At firing the compressed air expels the water from the cham ber o f the watergun which creates a compressional wave radiating from the source. These waves are reflected from the subsurface impedance boundaries and are detected by the hydrophones.
The source and the hydrophone array were towed behind the ship, as shown in Figure 4. As the source and the hydrophone cable were towed behind the laboratory boat the tug boat was far away even from the small offset traces. This together with the quiet weather conditions resulted in improved signal-to-noise ratio compared to marine surveys.
The speed of the ship relative to the ground was maintained between 3 -4 km/h.
Using two seconds firing rate resulted approximately 2 m shotpoint distance. Positioning has been carried out using a differential GPS (DGPS) system. Position readings were made every 10 second and interpolated in-between. Estimated accuracy of the positioning is 1 to 2 m.
Recording has been done with an OYO DAS-1 system on DAT tapes in SEG-D format. Sampling frequency was set to 4 kHz, recording length to 500 ms. High frequency content of the source resulted 1 m resolution in the upper part of the sections (first 100 ms)
and 2 -3 m resolution in the deeper part of the sections (down to 500 ms). This makes sure that a fault displacement of 2 -3 m can be observed with confidence on the sections.
Table 1 shows the most important parameters o f acquisition.
Table 1: Field param eters f o r the m ultichannel survey
Source type Sodera 15 w atergun run at 2000 PSI A verage source depth 0.5 m
Firing rate 2 s
Sensors 16 channel hydrophone-cable, 5 hydrophones serially connected for each channel Group sensitivity 4 m V /m Bar
Special codes were written for geometry installation. The following processing steps were applied to the data.
1. D ata input, sorting and editing:
Field data w as loaded from m agnetic tape.
2. G eom etry installation:
Spherical spreading correction was applied to the data.
5. Frequency filtering:
T im e variant zero phase Butterw orth bandpass filter was applied to rem ove high frequency noise and to suppress low frequency coherent noise present in the field data. Zero phase Butterw orth filter was applied with 80 Hz low cut frequency and 20 dB/o slope to suppress low frequency noise, w hile for suppressing the high frequency noise tim e varying filter w as applied with two tim e gates. A t shallow depths (betw een 0 and 150 ms) zero phase Butterworth high cut filter was applied with 700 Hz high cut frequency and 24 dB/o slope, while in the low er part o f the section (between 200 and 500 ms) 500 Hz high cut frequency and 24 dB/o slope was chosen after filter tests. Betw een the two tim e gates linear com bination o f the tw o filters w as applied. Noticeable absorption o f the high frequency content o f the seism ic signal in the deeper part o f the section made it w orthwhile to apply the tim e variant high cut filter.
128
6. Predictive deconvolution:
M inim um phase predictive deconvolution was applied to suppress the w aterbottom m ultiples in the section. Sim ilar to the bandpass filtering, deconvolution w as applied in a tim e varying m anner. T he tw o tim e gates were the same as the ones selected for bandpass filtering and deconvolution param eters w ere chosen after param eter testing.
In the upper tim e gate 9 ms operator length and 2 ms prediction distance was applied, while in the low er gate 20 ms operator length and 3 ms prediction distance have been chosen. Prew hitening was 0.1% in both tim e gates.
RM S velocities w ere derived from the C D P gathers using constant velocity stack and sem blance analysis.
Velocity analysis w as carried out at least every 500 m and w here abrupt change in the velocity field could be expected m ore densely spaced velocity analysis was carried out.
9. NM O application:
Norm al m oveout correction was applied using the estim ated RMS velocities w ith 30% stretch m ute allowed.
10. Trace M uting
Top muting o f the N M O corrected gathers was perform ed in channel dom ain. Reflections from the w aterbottom could be observed only on the first tw o traces, on all the other traces the first reflections w ere masked by the direct arrival. This direct arrival w as progressively muted out from the traces. However, on the first two traces the substrata can be exam ined for the w aterbottom down to m ore than 500 m depth
11. Predictive deconvolution
M inim um phase predictive deconvolution was applied in tw o steps to rem ove peg-leg m ultiples in the record. The rem oval w as focused on the low er part o f the section, where peg-leg m ultiples w ere stronger. The upper part o f the traces w as virtually left untouched by this deconvolution due to the applied param eters. The same tim e gates w ere used as m entioned before. In the first step w aterbottom peg-legs w ere attacked (relatively small prediction distance) with 100 ms operator length and 7 ms prediction distance, w hile in the second step longer prediction distance w as applied to suppress intrabed peg-legs. In this step 150 ms operator length and 25 ms prediction distance w as used. Prew hitening was 0.1% in both cases.
12. CD P stack:
A fter testing several stacking algorithm s (mean, diversity power, diversity am plitude, m edian, alpha trim) diversity pow er stack was applied using 150 ms long diversity scalar operator.
13. T race mixing:
15. Hand statics:
A static shift o f - 2 4 ms (negative sign means upshift) w as applied to every trace to com pensate for the mechanical delay o f the watergun. The delay was estim ated from the recorded source signature. A fter the static shift applied the w ater depth estim ated from the single channel m easurem ents w as in good agreem ent w ith the w ater depth observed on the multichannel lines.
16. M igration:
Steep dip explicit finite difference tim e m igration w as perform ed using spatially varying interval velocities derived from the RM S velocities.
17. Presentation o f the seism ic sections:
C D P values plotted on the top o f the sections are increasing upstream , from South to North. CD P interval is 5 m. C D P values o f different sections are not independent, the gap betw een two sections can be calculated from the gap in C D P numbers.
Figures 5 -9 show the m igrated and interpreted tim e sections o f seism ic profiles Danube-202, D anube- 203, D anube-205, D anube-207 and Danube-208. The vertical scale o f the sections corresponds to that o f the standard onland sections, and the horizontal scale is 1:20 000.