Data Processing
ODP logging
contractor: Delaware Geological Survey
Hole: Oceanview
Leg: 174AX
Location: Sea Isle City, Dennis Township (New Jersey)
Elevation: 2.87 m
Latitude: 39¡ 10' 43.826" N
Longitude: 74¡ 43' 31.643" W
Logging date: October, 1999
Total penetration: 480.18 m
Total core recovered: 387.53 m (80.7 %)
Logging
Runs
Logging string 1: Gamma Ray
Logging string 2: Gamma Ray-Resistivity-SP-Temperature
Logging string 3: Full Wave Sonic Tool
Logging string 4: Caliper
Logging string 5: Gamma Ray
On October 8 the first logging string (Gamma Ray) was run through
the rods from land's surface to 342.4 m with sampling at 3.05 cm. Log quality
was excellent as judged by the match between gamma-ray log kicks and lithologic
changes. With the rods pulled, the second logging string was run from land' s
surface to 342.3 m. Tools on the sonde included natural gamma ray, 16 and 64-in.
normal resistivity, fluid resistivity, lateral resistivity, spontaneous
potential, single point resistance, temperature, and delta temperature. A third
logging string consisting of a full wave sonic tool (run to 342.3 m) was run
next, followed by a fourth logging string (3-arm caliper), also run to 342.3 m.
Logging operations were concluded on October 28, with a gamma ray log recorded
through the rods.
Processing
Depth shift: No depth match between runs has been performed.
Gamma-ray processing. The Natural Gamma Ray logs recorded uphole on October
28 were smoothed in IgorTM using a 21-pt binomial function. Binomial smoothing convolves
the data with a (discrete) triangular function. The values of this function are
normalized coefficients derived from Pascal's triangle at a level equal to the
Smoothing parameter. The
coefficient is that of a binomial (a + b)N , where N is the smoothing parameter, normalized
to preserve the signal amplitude. At the given sampling rate (1/10 ft) and with
N = 21, the width of the filter function is 2 ft. The algorithm used is derived from Marchand and Marmet
(1983).
In frequency domain binomial
smoothing is a Gaussian filter that removes higher frequency signals that are
random (i.e., 'noise'). This is
illustrated (Figure below) by the frequency response of the binomial smoothing
algorithm applied to the Ocean View logs.
Frequency cuts are not sharp and the high frequency signal is reduced in
amplitude but not completely removed.
The benefit of this filter is that it is very smooth (i.e., there is no
frequency "ringing").
Such are filters superior to the moving average smoothing because the
latter are 'boxcars' in the frequency domain; such filters do not smooth in the
frequency domain and thus simply reduce amplitude without removing high
frequency noise. Binomial
smoothing is similar to Gaussian convolution filtering used by Miller et al.
(1987) to smooth and present Cenozoic stable isotopic record. Although more sophisticated means exist
to remove noise from data sets (i.e., those that maintain the full amplitude
and gain that are lost in binomial smoothing), the filtering that we used is
simple, statistically valid, and visually pleasing.
P. Marchand and L. Marmet, (1983). Binomial smoothing filter: A way to avoid some pitfalls of
least-squares polynomial smoothing.
Review of Scientific Instruments, 54(8) 1034-1042
For further information about the logs, please contact:
Kenneth G. Miller
Department of Geological Sciences, Wright Labs
Rutgers University
Piscataway, NJ 08854-8066
Phone: (732) 445 3622
Fax: (732) 445 3374
Peter P. McLaughlin, Jr.
Delaware Geological Survey
University of Delaware
DGS Building
Newark, DE 19716-7501
Phone: (302) 831-8263
Fax: (302) 831-3579