Gregory L. Duckworth

Crustal structure measurements near FRAM II in the pole abyssal plain

Abstract An extensive refraction profiling program was carried out during the FRAM II experiment (March–May, 1980) in the eastern Arctic Ocean. Two structural areas were covered: north of the ice camp (86°N, 24°W) into the basin of the Pole Abyssal Plain and south onto the flanks of the Morris Jesup Rise. Digital multichannel data on an 800 by 800 m, 24 channel hydrophone array and a single 2-component ocean bottom seismometer (OBS) were recorded for offsets from 2.5 to 100 km. Arrival times, amplitudes and phase velocities of the seismic signals recieved on the hydrophone array were determined using high resolution array processing. From these measurements and the OBS data, preliminary velocity structural models of the crust have been derived. For the purposes of this paper, 2 refraction lines have been analyzed, a 40 km line on a flat region of the Pole Abyssal Plain and an 86 km line on a slightly dipping region taken as the drifting ice camp shoaled on the Morris Jesup Rise. These preliminary analyses yield a sedimentary layer with a gradually increasing velocity 1.5–2 km thick. This cover overlays a crust with a thin layer 2 ( ) and yields a total ocean bottom to mantle thickness of 4–7 km.

Inversion of refraction data from the Fram and Nansen Basins of the Arctic Ocean

Abstract As a part of the Fram 2 experiment (March–May 1980) six refraction lines were shot in the Pole Abyssal Plain or Fram Basin of the eastern Arctic Ocean. During the Fram 4 experiment (March–May 1982), four refraction lines were also shot in the Nansen basin just northeast of the Yermak Plateau. The Fram 2 lines at 86°N, 24°W cover both the abyssal plain region and the northern flank of the Morris Jessup Rise near magnetic anomalies 22 and 23. The Fram 4 lines near 83°N, 15°–20°E are on slightly younger crust at anomaly 12. In this paper we introduce new methods for processing and inversion of Arctic refraction data and contrast the results from these tectonically similar regions on opposite sides of the spreading center at the Arctic Mid-Ocean (Nansen) Ridge. In both experiments the data were recorded digitally on a two-dimensional array of 24 hydrophones located 93 m below the floe camps. These multi-channel data are processed by high-resolution array processing techniques to yield velocity spectra over the short 1 km apertures of the arrays. These velocity spectra are transformed to tau-slowness and offset-slowness parameterizations and are inverted by 1. (1) velocity-depth migration 2. (2) tau-sum 3. (3) extremal inversion techniques to obtain crustal velocity models down to the Moho. Without the ability of the array processing to discern late breaking primaries, multiples, and shear arrivals, the sparsely shot lines with 5–15 offsets over the 30–100 km line lengths would not be invertible by these methods. These results demonstrate the effectiveness of arrays in determining crustal structure under the constraints of Arctic ice pack environment.

Acoustic array sensor tracking system

Multichannel acoustic and geophysical experiments have been conducted recently in the Marginal Ice Zone (MIZ) of the eastern Arctic Fram Straits region using a random array of hydrophones deployed from ice floes with radio linked telemetry across apertures of up to 6 km. Accurate relative positions of hydrophone sensors in the array are required for subsequent data processing. A system of hardware and software has been developed to compute and record relative sensor positions and make plots of the sensor field in real time. Off-line processing uses sound speed profile corrections and model based filtering for greatest accuracy. The system consists of 5 reference sensors in the array field, each co-located with a hydrophone, emitting 5 millisecond tone bursts at a unique frequency approximately every fifteen seconds. Signals are telemetered via a radio link to a receiving station where hardware detects pings and computes time delays between all sensors. These delays are acquired every thirty seconds, to compute an approximate set of array element locations and recorded for later processing. Relative location accuracies of 1 m rms can be obtained at ranges up to 5 km.

Minimum cross entropy seismic diffraction tomography

One difficulty in applying seismic tomography is that seismic sources and detectors cannot be deployed so that the imaging area is probed from all directions. This paper develops two methods to alleviate this difficulty. These methods combine the minimum cross entropy estimation algorithm with two tomographic image reconstruction algorithms: the direct transform reconstruction algorithm and the backpropagation reconstruction algorithm. These two methods were tested by numerical and scaled model ultrasonic laboratory experiments simulating seismic cross‐hole tomography. It was found that (1) the horizontal resolution, which is inherently poor for the cross‐hole geometry, can be improved, and (2) artifacts caused by the sparse sampling and noise can be reduced. Both methods work better when the targets are isolated impulses surrounded by a uniform background medium.

Seismic Exploration in the Arctic Ocean

The Arctic Ocean is one of the least known regions of the world’s oceans. Major questions still remain concerning its origins and the processes the led to its evolution; the principal reason is the paucity of seismic data available. While some of this can be attributed to the harshness of the environment on men and machines, it is the ice cover that is the principal limiting factor. It is not stable or contiguous enough to support the use of land techniques, yet only the sturdiest ice breakers can maneuver within it, thereby preventing the use of routine marine procedures. As a result, seismic exploration in the Arctic has been a peculiar mix of land and marine techniques.

Velocity analysis of multireceiver full waveform acoustic logging data in open and cased holes

Average semblance and maximum-likelihood spectral analyses are applied to synthetic and field full waveform acoustic logging data to determine formation velocities. Of particular interest is the ability of these methods to resolve the P and shear/pseudoRayleigh arrivals in data from poorly-bonded cased boreholes. In synthetic open-hole data the velocity analyses yield results within 4% of the true velocities. Results from synthetic well-bonded cased hole data are generally as good as those from the open hole data. However, if the formation P-wave velocity is within roughly 10% of the plate velocity of the steel pipe (about 5.3-5.5 km/s), then there may be a resonance effect that appears to slow down the P wave slightly (on the order of 6%). For cased-hole models with no steel/cement bonding (the free-pipe situation), the measured P-wave velocities are typically 6 to 8% less than the actual formation velocities. If the formation S-wave velocity is greater than about 2.5 km/s, the S-wave velocity estimate may also be 6 to 8% low. Furthermore, increasing the thickness of either the cement layer or the fluid layer between the pipe and the cement further decreases the formation velocity estimates. Also, if the P-wave velocity is within roughly 15% of the velocity of the steel arrival, the P wave may not be resolved by the semblance method unless the data is first low-pass filtered. Initial tests show that this filtering process may adversely affect the final P-wave velocity estimate, but the details of this type of approach have not been studied. The P wave is resolved. by spectral analysis of the original, nnfiltered data. For cased-hole models with no cement/formation bonding (the unbonded-casing 366 Block et al. situation), formation S-wave velocities are estimated to within 3% relative error, and the formation P-wave velocity is estimated to within 2% error in a slow formation. However, for P-wave velocities between 3.4 km/s and 5.94 km/a, the P wave cannot be resolved by spectral analysis, and it is resolved by the semblance method only in the model with the low velocity (3.4 km/s).

Estimation of Ice Surface Scattering and Acoustic Attenuation in Arctic Sediments from Long-Range Propagation Data

A single explosive shot at a range of 341.3 km in the Pole Abyssal Plain of the Arctic Ocean is used to assess the components of propagation loss for this region. The acoustic energy propagated between a satellite ice camp and the Fram II ice station in a water column of nearly uniform depth. Much of the observed energy interacted with the upper 200 meters of sediments along a path which was nearly parallel to the Arctic Mid-Ocean Ridge. In addition, the upward refracting sound channel of the Arctic Ocean also caused the observed energy to interact extensively with the ice canopy, which was contiguous over the entire path. The deterministic lateral homogeneity of the bathymetry and sediments, and the statistical lateral homogeneity of the ice canopy in this region allow us to attempt to separate the effects of geometrical spreading, ice surface scattering, and effective sediment compressional wave attenuation. The primary data for this work are observations of the signal from a 25.8 kg explosive charge received on a 24 channel two-dimensional hydrophone array with a 1 km aperture. The data are inadequate to resolve attenuation at depth, but provide an estimate for frequency independent Q of 200 to 300 for the upper 200 meters of the sediments and a dependency of surface scattering on frequency over the 5−50 Hz band. It is shown that partially coherent summation of the surface multipaths is required to predict the observations.

Adaptive Array Processing for Acoustic Imaging

The need for high resolution acoustic image formation under the constraints imposed by small (≈ 100 wavelengths) apertures, long wavelength radiation, and sparsely sampled discrete apertures is encountered in many applications. Most techniques currently in use require large arrays for resolution and uniform sampling of the array aperture for sidelobe control to yield adequate performance when imaging specular reflectors. This paper examines the applicability of the data-adaptive array processing technique known as the Maximum Likelihood Method to image formation in the undersea environment, where acoustics and acoustic imaging techniques have long been of interest as a result of their utility in probing where electromagnetic radiation will not penetrate. The approach taken is to suppress the usual deterministic outlook wherein the propagation phenomenon is “undone”, and to look at the problem in a statistical sense. The result is an imaging technique that is essentially spatial and temporal spectral density estimation for a space/time random process which is sampled at a small number of discrete spatial locations.

The Relative Amplitudes of Primary and Multiple Signals Refracted in the Ocean Crust

In deep ocean seismic refraction experiments the energy in the primary signal is often less than that in the multiple arriving at approximately twice the travel time. This has perplexed interpreters since the multiple has a greater path length and interacted with the seafloor once more, so its amplitude should presumably be less. High resolution slowness-travel time spectral analysis of an expanding spread data set suggests an explanation in terms of the differences in the sound speed gradient in the oceanic crust. The gradient in the upper part of the crust is typically much greater than that in the deeper crust; therefore, energy is focused much more and has a decreased geometrical offset. The multiple then should have a lower phase speed and a higher amplitude when it is refracting in the higher gradient part of the upper crust. Synthetic modeling using the WKBJ seismogram method and the slowness-travel time spectra of an ESP data set support this hypothesis.

Tracking a human walker with a fiber optic distributed acoustic sensor

Traditional ground sensors can be used as security systems to detect third party intrusion, but are costly and cumbersome to scale up to cover a large area. Fiber optic distributed acoustic sensing (DAS) offers a solution that economically and instantaneously monitors up to 40 linear km of a border or perimeter with a single system. The output of the OptaSense third generation fiber interrogation unit is spatially dense, coherent, and exhibits characteristics expected of traditional strain sensors. In this paper, DAS data are shown to indicate a walker’s directionality and support beamforming algorithms for localization.