Broadband PE Research Articles

Seismic and hydroacoustic analysis of four large underwater explosions off the east coast of Florida

During the summer of 2016, the US Navy detonated four large underwater explosions off the east coast of Florida as part of shock wave testing for the Littoral Combat Ships USS Jackson and USS Milwaukee. The water depth at the explosion site is 900 m. Videos of the events provide further insight into the sources. Originally listed as magnitude 3.7 earthquakes in USGS bulletins, the source types were changed to “Experimental Explosion” as the Navy began releasing information to local news stations. These explosions were recorded on US seismic stations to distances of ~2000 km and on the hydroacoustic array at Ascension Island at a distance of ~8100 km. Seismic signals are complex but cepstral analysis reveals a clear bubble pulse period of 1.15 sec, corresponding to source depths of 50 meters. At Ascension the signals arrive at an RMS level of 136 dB at an SNR of 30 dB. We modeled the propagation loss from the source to Ascension using broadband PE and find the propagation loss to be 126 dB. Backing out this loss, we estimate source levels for the two explosions at 262 dB corresponding to explosion yields of 11,000 lbs.

Broadband impulse response modeling using the single frequency parabolic equation solution

The parabolic equation has been considered the workhorse of low to mid frequency propagation modeling in range dependent environments due to its accurate handling of both refraction and diffraction. The frequency domain solution approach of the PE has severely limited its use in broadband pulse synthesis die to the computational cost of computing each Fourier component. In this paper a hybrid approach that blends the PE forward model for field computation and an analytic model for travel time computation is presented. The PE is computed at the center frequency in a range dependent environment. The beam arrival angle for a virtual line array at the receiver depth (as well as sea surface and sea floor) is stored at each range. The average group slowness as a function of phase speed at the source (wave number invariant) is computed using analytic acoustics methods (Snell's law). The impulse response at each range is computed by incoherently adding the pier of each beam at the arrival time of the associated phase speed energy (the range averaged group slowness). This approach reproduces the behavior of the pulse spread, without the cost of the broadband PE.

Passive striation based geo‐acoustic inversion using ships of opportunity.

The coherent interaction of acoustic multipaths leads to an interference pattern in range. As range changes, the locations of the peaks (and troughs) of this interference pattern shift (as the group velocities of the modes change) in a sometimes predictable fashion. The intensity pattern of a moving source, plotted as a function of range and frequency, is commonly referred to as striation patterns. These patterns are commonly observed in the passing of a loud surface vessel in the spectrogram of a single hydrophone. In this paper, a geoacoustic method is presented based upon the quantitative measure of striation slope (wave guide invariant parameter) and the striation spacing in frequency (reciprocal time spread). Both of these parameters are generally sensitive to the geo‐acoustic parameters of the sediment. In 2003 this technique was applied to a passing surface ship [K. Heaney, IEEE JOE 29(1), 43–50 (2003)] and to a range‐dependent benchmark case [K. Heaney, IEEE JOE 29(1), 88–99 (2003)]. The approach in these papers compared the measured acoustic observables with those from a table predicted by the forward computation using the broadband PE. Current implementation of the algorithm uses a Gaussian beam code to efficiently compute the acoustic observables.

Observations of micromultipath formation at very long ranges in shallow water.

Long- and very-long-range shallow-water transmission is observed for low-frequency m-sequences transmissions with center frequencies of 50, 100, and 250 Hz. The source is suspended from a ship that is holding position at each of three ranges: 10, 20, and 80 km. The receiver is a bottomed horizontal array. The array is used to resolve out-of-plane arrival angles to a small fraction of a degree by observing phase rolls of the spatial coherence function of the various arriving wavefronts. At the 10 and 20 km ranges, distinct pulse arrivals are observed in one-to-one correspondence to normal mode arrivals predicted with PE propagation models. But at the 80 km range, a large number of distinct arrivals appear with intensities or arrival times inconsistent with deterministic normal mode or ray arrivals. One possibility is that these are micromultipaths from out-of-plane paths after reflections from facets or sloping bottom contours. In fact, each of the arrivals has a unique horizontal arrival angle, and the steeper the out-of-plane angle, the later the arrival—an observation that is verified with broadband PE propagation model computations for both random faceted and gently sloping bottoms.

Internal wave inversions for one year of Pioneer–Hawaii ATOC transmissions

The ATOC experiment transmitted 75-Hz signals from Pioneer Seamount off the coast of California to a vertical line array (VLA) off Hawaii. The combination of VLA and long-range propagation provides the opportunity to perform mode-based internal wave tomography. Each arrival is mode-filtered and the statistics of the mode arrivals are tabulated. These individual mode arrival time (centroid) and spread are then compared with calculated mode statistics from broadband PE modeling of the propagation path. In this manner a statistically robust inversion for the internal wave strength is performed. The weakness of the inversion is the sensitivity to the bottom scattering near the source. Results for one year of data will be presented for two different bottom types. The long-scale time series may indicate possible sources of internal waves in the ocean. Small time scale variations may be attributable to internal tides. a)The ATOC Group: A. B. Baggeroer, D. Menememlis, and C. Wunsch (MIT); T. Birdsall, K. Metzger (Univ. of Michigan); C. Clark (Cornell Univ.); J. Colosi (WHOI); B. D. Corneulle, M. Dzieciuch, W. Munk, P. F. Worcester (SIO); D. Costa (Univ. of California, Santa Cruz, CA); B. D. Dushaw, B. M. Howe, J. A. Mercer, and R. C. Spindel (APL-U Wash.); A. M. G. Forbes (CSIRO, Hobart).

Modeling of the out-of-plane pulse propagation in the SWARM95 experiment

The acoustical data collected in the shallow water acoustics in a random medium (SWARM95) experiment on the continental shelf off New Jersey in 1995 have shown that the temporal variations of the time series of the airgun signals (center frequency 75 Hz and bandwidth 50 Hz) are both range and azimuthal dependent. A full-wave broadband PE model is used to simulate such shallow water pulse propagation at different horizontal bearings. The time-dependent shallow water internal wave fluctuations constructed with a broad spectrum of excitation and the solitonlike thermocline displacements based on the in situ observations are included in the numerical simulation as well as the bathmetry. The vertical wavefronts and the 2-h time histories of the repeated transmissions from airgun positions above and below the thermocline are calculated and displayed. The simulated data and the experimental data of the receiver arrays at different locations with an approximately 10-km cross-range distance are compared. The results suggest that the azimuthal dependence of the received signals correlate to both the internal wave fields and the bathmetric variations.

Analysis of internal wave and bottom effects on the acoustic field generated by a towed source during the SWARM95 experiment

Thirty-second duration 250- to 375-Hz LFM pulses were transmitted from a towed acoustic transducer to study the effects of internal waves and the sea bottom on an acoustic signal propagating along a 30-km transmission path. The data were taken on the New Jersey continental shelf as a part of the multi-institutional SWARM95 experiment. Environmental data were collected by using CTDs, thermistor strings, a high-frequency acoustic flow visualization system, and a chirp sonar system. Geoacoustic properties of the bottom were inverted from the chirp sonar data with a 1.5-m vertical resolution and 40-m penetration. A deterministic sound speed field in the water column was constructed by imposing the high-frequency visualization measurements on both CTD and thermistor string data. Environmental data indicated that two distinct geological provinces and two major internal wave packets were crossed during the 30-km towed transmission. Using a full-wave broadband PE model, vertical structures of sound field were calculated for comparison to data obtained by a 32-element vertical line receiver array. Preliminary model/data comparisons indicate the dominance of water column effects over the bottom effects. [Work supported by ONR.]

Localization of sources of acoustic transients using a broadband matched-field processing technique

By matching, or correlating, a complex-demodulated (envelope) time series measured at a single hydrophone receiver to predicted envelope response functions obtained by back propagation using a full-physics broadband PE model, it is possible to remove multipath distortion effects and to determine the location of the source of an acoustic transient with good resolution and high processing gain. This ‘‘semi-coherent’’ broadband matched-field processing (BMFP) technique is demonstrated in several deep water and shallow water situations. [Work supported by ONR and ARPA.]

Internal wave effects on long-range ocean acoustic tomography

Full-wave numerical simulations of internal wave scattering of low-frequency sound in the deep ocean are performed with an efficient broadband PE model based on a new parabolic approximation that is second-order accurate, and an internal wave model based on a new representation of the GM spectrum that allows efficient generation of internal wave fields that evolve in geophysical time. Results are displayed as plots of acoustic travel time versus geotime for various source–receiver separations (up to 8 Mm), and various center frequencies and bandwidths. Quantitative examination (coherent and incoherent geotime averages) yields information about the temporal coherence and stability of each multipath. It is found that the later near-axial arrivals are unstable and unresolvable, and therefore not useful for tomography, at ranges and frequencies that are consistent with the predictions of Dozier and Tappert [ 533–547 (1978)] who used a coupled mode model. [Work supported by ONR and ARPA.]

Travel time effects of mesoscale structure on rays and waves at global ranges

A range‐dependent ray trace model and a broadband PE model are used to model sound propagation at fixed bearing through a field of mesoscale baroclinic modes in order to study the effects of mesoscale structure on travel time at long ranges. The ray model exhibits chaos at ranges beyond a few Mm as manifested by an exponentially increasing number of eigenrays and triplications, especially in the late‐time near‐axial arrivals. In addition, the ray model predicts a mesoscale travel time bias, in the direction of later time, of the last axial arrival amounting to about 100–200 ms/Mm. At center frequency 75 Hz and bandwidth 50 Hz, the full‐wave PE model qualitatively confirms the ray trace predictions of chaos, and shows that the later near‐axial arrivals are smeared out into a continuum of unresolvable multipaths characterized by saturated (Gaussian) statistics at ranges of a few Mm, and also quantitatively confirms the mesoscale bias of 100–200 ms/Mm of the last axial arrival. The steeper, early raylike arrivals are relatively stable in the presence of mesoscale structure and may be useful for long‐range tomography. [Work supported by ONR.]

Internal wave effects on shallow water acoustic tomography

Full‐wave numerical simulations of internal wave scattering of sound in shallow water are performed with an efficient broadband PE model that also includes scattering by bathymetric variations. Bathymetric scattering by itself gives an exponentially large number of unresolvable multipaths as predicted by ray chaos theory. When internal waves that evolve dynamically in geophysical time are included, the unresolvable multipaths are found to be temporally unstable with coherence times less than a few minutes at ranges greater than a few tens of water depths in shallow water, except possibly for rare paths that are shielded from internal waves such as within strong surface ducts. Since traditional travel‐time ocean acoustic tomography requires many stable, resolvable and identifiable paths, conditions that are not satisfied in shallow water at longer ranges, some other inversion method will have to be developed for remote sensing of shallow water environments. [Work supported by ONR.]

The Kaneohe acoustic thermometry experiment and waves

A new parabolic approximation is derived and implemented that efficiently gives travel time predictions that are accurate to within a few milliseconds at a range of 1000 km and center frequency 75 Hz. Sensitivity of the full-wave travel time predictions to oceanographic and bathymetric inputs along a 3700-km track is studied by comparisons to acoustic data between a source at Kaneohe having center frequency 133 Hz and a bottom mounted receiver near the northern California coast. Path stability for a time interval of several years is also examined with the new broadband PE model, and full-wave generalizations of eigenrays, called eigentubes, are computed. [Work supported by SERDP, ARPA, and ONR.]

Modeling issues for predicting sound propagation at moderate frequencies in bottom-limited sound channels

Consider sound propagation in the moderate frequency range (depth to wavelength ratio, H/λ, between 20 and 100) in shallow water environments with bottom roughness. Measurements in the Straits of Florida have shown remarkable phase stability together with rapid (minutes-to-hours) unstable amplitude fluctuations that nevertheless yield stable envelopes when averaged over many hours. What physical effects should be contained in an acoustic model in order to predict the observed behavior? It is suggested that a full-wave model that is capable of multiple rough bottom forward scattering and includes temporally slowly varying range-dependent sound-speed profiles will yield observed results, when used together with coherent signal processing techniques. A broadband PE model will illustrate the assertion, at least qualitatively. Absolute level prediction is quite another matter that requires more extensive environmental inputs.