Abstract:
Deterministic source signature deconvolution is applied to the
processing of marine wide angle and vertical profiler data with air-gun
sources. Optimum results are obtained with a source signature
measured by stacking the signal reflected from a relatively homogeneous
abyssal plain sedimentary environment. This eliminates the need for
the unstable inverse source-receiver ghost filter. Improved resolution
of reflection event timing allows the computation of more reliable
interval velocities by the T² - X² method, provided the layer thickness
limitation of the method is not exceeded. Accurate timing of
primary reflection events in the deconvolved vertical profiler
data permits computation of frequency dependent attenuation by
univariate least-squares regression in the Fourier transform domain.
The technique successfully extracts input amplitude attenuation
functions from model reflection coefficient sequences with additive
random noise. This success is attributed to the stability of
singular value analysis in solving the least-squares regression model.
Statistical tests on the solution vectors for model and field data
give criteria for evaluating their reliability. The model data studies
suggest that multiple and primary events not included in the
regression may be considered part of the noise term without seriously
affecting the accuracy of the computed spectral ratios.
The method is tested on field data from the following sedimentary
environments off the coast of Oregon and northern California: a
continental shelf basin, an abyssal plain environment, the base of
the continental slope and two locations on the Astoria sea fan, one
near the Cascadia sea channel and one north of DSDP site 174. Velocity
versus depth and frequency dependent spectral ratio plots are determined
for each environment. The computed surface layer interval velocity
of 1.77 km/sec over a thickness of 455 m for the station north of
DSDP site 174 is in good agreement with the average material type
found in the drill core (sandy-silt with greater than 60% sand).
Maximum attenuation coefficients are estimated from the spectral
ratios for the upper sediment intervals of the study areas using
typical acoustic impedance values of surface sediment types determined
from nearby piston cores. Some maximum attenuation coefficients
are too high suggesting the possibility of a stratigraphic component.
The maximum attenuation in the upper interval for SB 46 over the Tufts
abyssal plain where fine-grained material (silts and clays) is
expected is 0.025 dB/m at 127 Hz compared with 0.004 dB/m at 80 Hz
for the upper interval of the turbidite environment north of DSDP
site 174.
