Tests of models for high-frequency seafloor backscatter

The interaction of high-frequency sound with the seafloor is inherently a stochastic process. Inversion techniques must, therefore employ good stochastic models for bottom acoustic scattering. An assortment of physical models for bottom backscattering strength is tested by comparison with scattering strength data obtained at 40 kHz at three shallow water sites spanning a range of sediment types from fine silt to coarse sand. These acoustic data are accompanied by sediment physical property data obtained by core sample analysis and in situ probes. In addition, stereo photography was used to measure the power spectrum of bottom relief on centimeter scales. These physical data provided the inputs needed to test the backscatter models, which treat scattering from both the rough sediment-water interface and the sediment volume. For the three sites considered here, the perturbation model for scattering from a slightly rough fluid seafloor performs well. Volume scattering is predicted to be weak except at a site having a layer of methane bubbles     

Expendable penetrometer for seafloor classification

A design is presented for an expendable seafloor penetrometer system for real time classification of marine sediments and estimate for shear strength. An accelerometer in the nose section of an expendable probe senses deceleration after impact with the ocean bottom and provides this data to the underway launch platform via a two-conductor trailing wire umbilical. Signals are amplified, digitized and processed by microcomputer to provide seafloor classification in virtual real time.     

A methodology for acoustic seafloor classification

A seafloor classification methodology, based on a parameterization of the reverberation probability density function in conjunction with neural network classifiers, is evaluated through computer simulations. Different seafloor provides are represented by a number of scatterer distributions exhibiting various degrees of departure from the nominal Poisson distribution. Using a computer simulation program, these distributions were insonified at different spatial scales by varying the transmitted pulse length. The statistical signature obtained consists of reverberation kurtosis estimates as a function of pulse length. Two neural network classifiers are presented with the task of discriminating among the various scatterer distributions based on obtained acoustic signatures. The results indicate that this approach offers considerable promise for practical, realizable solutions to the problem of remote seafloor classification     

Multisource multibeam backscatter data: developing a strategy for the production of benthic habitat maps using semi-automated seafloor classification methods

The establishment of multibeam echosounders (MBES) as a mainstream tool in ocean mapping has facilitated integrative approaches towards nautical charting, benthic habitat mapping, and seafloor geotechnical surveys. The bathymetric and backscatter information generated by MBES enables marine scientists to present highly accurate bathymetric data with a spatial resolution closely matching that of terrestrial mapping, and can generate customized thematic seafloor maps to meet multiple ocean management needs. However, when a variety of MBES systems are used, the creation of objective habitat maps can be hindered by the lack of backscatter calibration, due for example, to system-specific settings, yielding relative rather than absolute values. Here, we describe an approach using object-based image analysis to combine 4 non-overlapping and uncalibrated (backscatter) MBES coverages to form a seamless habitat map on St. Anns Bank (Atlantic Canada), a marine protected area hosting a diversity of benthic habitats. The benthoscape map was produced by analysing each coverage independently with supervised classification (k-nearest neighbor) of image-objects based on a common suite of 7 benthoscapes (determined with 4214 ground-truthing photographs at 61 stations, and characterized with backscatter, bathymetry, and bathymetric position index). Manual re-classification based on uncertainty in membership values to individual classes—especially at the boundaries between coverages—was used to build the final benthoscape map. Given the costs and scarcity of MBES surveys in offshore marine ecosystems—particularly in large ecosystems in need of adequate conservation strategies, such as in Canadian waters—developing approaches to synthesize multiple datasets to meet management needs is warranted.     

Modern swathe sounding and sub-bottom profiling technology for research applications: The Atlas Hydrosweep and Parasound Systems

This chapter describes two separate but complementary research echosounder systems originally developed by Krupp Atlas Elektronik GmbH for the new German oceanographic research vessel Meteor.The Hydrosweep is a Hydrographic wide-swathe sweep survey echosounder for both shallow and deep water applications providing accurate bathymetric surveys and terrain-following navigation capabilities.The Parasound system is a hull-mounted dual channel parametric narrow-beam deep sea survey and sub-bottom profiling echosounder enabling particularly high vertical and horizontal resolution of seabed features.     

An attenuation-based sediment classification technique using Chirp sub-bottom profiler data and laboratory acoustic analysis

A remote sediment classification technique based on attenuation measurements from Chirp sub-bottom profiler data is described. This differs from previously published work in that attenuation measurements are obtained for each stratigraphic unit within a complex, thinly interbedded sedimentary sequence. Compressional wave attenuation measurements are obtained for a wide variety of lithologies, including muds, silts, sands, clayey sands, silty clays and gravel lags, with grain sizes ranging from 8 Phi to -4 Phi. In addition, attenuation measurements from sub-bottom profiler data were calibrated against laboratory acoustic measurements of vibracores and seabed samples from corresponding geographic locations, under simulated in-situ conditions using a Pulse Tube method. We adapt an instantaneous frequency matching method using a causal attenuation filter to model the decay of the Chirp transmitted waveform. From this modelling, a relationship between t* (a causal attenuation operator) and change in instantaneous frequency is established. The Hilbert transform is used to extract instantaneous frequency information from Chirp seismic, which is used to derive attenuation information for selected individual stratigraphic layers imaged by the sub-bottom profiler. This paper draws attention to the limitations in comparing attenuation measurements derived from Chirp sub-bottom profiler data against previously published literature on experimental attenuation measurements, which are limited by the wide variance of these data, and the difficulty in finding a meaningful best fit to these data. This demonstrates the importance of calibrating remote sediment classification observations using complimentary acoustic analysis of seabed samples to generate a site-specific geoacoustic database. A positive correlation between laboratory and sub-bottom profiler attenuation measurements was obtained, with a correlation coefficient of 0.885. Poorly sorted gravels with a mixed lithic and biogenic pebble component are characterised by very high attenuation with values of Q from 4 to 19. These sediments are considerably coarser-grained than those typically described in previously published experimental studies.     

Processing of high-frequency multibeam echo sounder data for seafloor characterization

Processing simultaneous bathymetry and backscatter data, multibeam echosounders (MBESs) show promising abilities for remote seafloor characterization. High-frequency MBESs provide a good horizontal resolution, making it possible to distinguish fine details at the water-seafloor interface. However, in order to accurately measure the seafloor influence on the backscattered energy, the recorded sonar data must first be processed and cleared of various artifacts generated by the sonar system itself. Such a preprocessing correction procedure along with the assessment of its validity limits is presented and applied to a 95-kHz MBES (Simrad EM 1000) data set. Beam pattern effects, uneven array sensitivities, and inaccurate normalization of the ensonified area are removed to make possible further quantitative analysis of the corrected backscatter images. Unlike low-frequency data where the average backscattered energy proves to be the only relevant feature for discriminating the nature of the seafloor, high-frequency MBES backscatter images exhibit visible texture patterns. This additional information involves different statistical distributions of the backscattered amplitudes obtained from various seafloor types. Non-Rayleigh statistics such as K-distributions are shown to fit correctly the skewed distributions of experimental high-frequency data. Apart from the effect of the seafloor micro-roughness, a statistical model makes clear a correlation between the amplitude statistical distributions and the signal incidence angle made available by MBES bathymetric abilities. Moreover, the model enhances the effect of the first derivative of the seafloor backscattering strength upon statistical distributions near the nadir and at high incidence angles. The whole correction and analysis process is finally applied to a Simrad EM 1000 data set.     

Preference of echo features for classification of seafloor sediments using neural networks

Selection of a set of dominant echo features to classify seafloor sediments using a multilayer perceptron neural network is investigated at two acoustic frequencies (33 and 210 kHz). Several sets of inputs with different combinations of two, three, four, five, and six echo features are exploited with three-layer neural networks. The performances of the networks are analyzed to assess the most discriminating set of echo features for classification of seafloor sediments. The results of the overall average performances reveal that backscatter strength and time spread are the two most important echo features at 33 kHz, whereas backscatter strength has higher discriminating characteristics at 210 kHz for seafloor sediment classification. In addition, a set of four echo features consisting of backscatter strength, time-spread, statistical skewness, and Hausdroff dimension gives the highest success at both the acoustic frequencies.     

Recommendations for improved and coherent acquisition and processing of backscatter data from seafloor-mapping sonars

Multibeam echosounders are becoming widespread for the purposes of seafloor bathymetry mapping, but the acquisition and the use of seafloor backscatter measurements, acquired simultaneously with the bathymetric data, are still insufficiently understood, controlled and standardized. This presents an obstacle to well-accepted, standardized analysis and application by end users. The Marine Geological and Biological Habitat Mapping group (Geohab.org) has long recognized the need for better coherence and common agreement on acquisition, processing and interpretation of seafloor backscatter data, and established the Backscatter Working Group (BSWG) in May 2013. This paper presents an overview of this initiative, the mandate, structure and program of the working group, and a synopsis of the BSWG Guidelines and Recommendations to date. The paper includes (1) an overview of the current status in sensors and techniques available in seafloor backscatter data from multibeam sonars; (2) the presentation of the BSWG structure and results; (3) recommendations to operators, end-users, sonar manufacturers, and software developers using sonar backscatter for seafloor-mapping applications, for best practice methods and approaches for data acquisition and processing; and (4) a discussion on the development needs for future systems and data processing. We propose for the first time a nomenclature of backscatter processing levels that affords a means to accurately and efficiently describe the data processing status, and to facilitate comparisons of final products from various origins.     

一种 ReliefF 和随机森林模型组合的多波束 海底底质分类方法

海底底质的快速探测和精细划分对海洋工程建设 、海洋资源开发等具有重要意义。多波束探测是目前声学底质遥测 的有效手段之一, 通常提取多波束反向散射强度图像和地形数据中的多维特征结合分类器进行底质分类。一方面, 若特征空 间维数过高, 分类效率会显著降低; 另一方面, 个别特征容易放大原始数据处理过程中仍存留的异常现象。针对这一问题, 本文提出了一种结合 Re1iefF 算法和随机森林 (Random Forest, RF) 算法的多波束底质分类方法。提取反向散射强度和地形 共 16 维特征, 利用Re1iefF 算法进行特征筛选, 排除低相关性特征, 降低特征空间维数, 结合采样点数据进行模型训练以构 建多波束底质分类模型。试验结合随机森林算法对未经特征筛选 、经主成分分析 (Principa1 Component Ana1ysis, PCA) 特征 优化后的特征进行分类实验作为对比。本文方法 Kappa 系数达到 85%, 分类总精度高于 90%, 精度具有明显优势, 耗时也 比较短。可见, 本文提出的结合 Re1iefF 和随机森林模型的多波束底质分类方法可以在保证分类精度的同时对多维特征进行 优化, 有效地提高了分类效率, 可对海底底质分类研究提供参考。     

Improved detection and mapping of deepwater hydrocarbon seeps: optimizing multibeam echosounder seafloor backscatter acquisition and processing techniques

Marine seep hunting surveys are a current focus of hydrocarbon exploration surveys due to recent advances in offshore geophysical surveying, geochemical sampling, and analytical technologies. Hydrocarbon seeps are ephemeral, small, discrete, and therefore difficult to sample on the deep seafloor. Multibeam echosounders are an efficient seafloor exploration tool to remotely locate and map seep features. Geophysical signatures from hydrocarbon seeps are acoustically-evident in bathymetric, seafloor backscatter, midwater backscatter datasets. Interpretation of these signatures in backscatter datasets is a fundamental component of commercial seep hunting campaigns. Degradation of backscatter datasets resulting from environmental, geometric, and system noise can interfere with the detection and delineation of seeps. We present a relative backscatter intensity normalization method and an oversampling acquisition technique that can improve the geological resolvability of hydrocarbon seeps. We use Green Canyon (GC) Block 600 in the Northern Gulf of Mexico as a seep calibration site for a Kongsberg EM302 30 kHz MBES prior to the start of the Gigante seep hunting program to analyze these techniques. At GC600, we evaluate the results of a backscatter intensity normalization, assess the effectiveness of 2X seafloor coverage in resolving seep-related features in backscatter data, and determine the off-nadir detection limits of bubble plumes using the EM302. Incorporating these techniques into seep hunting surveys can improve the detectability and sampling of seafloor seeps.     

The effect of a layer of varying density on high-frequencyreflection, forward loss, and backscatter [seafloor acoustics]

This paper examines the impact that a thin layer of varying density would have on high-frequency reflection, forward loss, and backscattering of acoustic plane waves from the seafloor. A functional form for density stratification was found by examination of several high-resolution density profiles obtained from X-ray computed tomography (CT) scans of seafloor cores. A solution based on these general profiles was used to estimate the reflection coefficient. The influence of the density profile on reflection loss and backscatter was then calculated using the estimated reflection coefficient. Parameter values used in simulations were obtained from the literature or from the CT scans of cores. It was found that inclusion of a density profile adds a strong frequency dependence to estimates of the reflection coefficient and forward loss. The largest effect on total scattering strength is near normal incidence where returns are dominated by interface scattering. The effect of the density profile on the strength of acoustic returns suggests that care should be taken when using high-frequency systems for measuring sediment properties, especially near normal incidence     

Recent structures in the Alboran Ridge and Yusuf fault zones based on swath bathymetry and sub-bottom profiling: evidence of active tectonics

The seafloor of the Alboran Sea in the western Mediterranean is disrupted by deformations resulting from convergence between the African and Eurasian plates. Based on a compilation of existing and new multibeam bathymetry data and high-resolution seismic profiles, our main objective was to characterize the most recent structures in the central sector, which depicts an abrupt morphology and was chosen to investigate how active tectonic processes are shaping the seafloor. The Alboran Ridge is the most prominent feature in the Alboran Sea (>130 km in length), and a key element in the Gibraltar Arc System. Recent uplift and deformation in this ridge have been caused by sub-vertical, strike-slip and reverse faults with associated folding in the most recent sediments, their trend shifting progressively from SW–NE to WNW–ESE towards the Yusuf Lineament. Present-day transtensive deformation induces faulting and subsidence in the Yusuf pull-apart basin. The Alboran Ridge and Yusuf fault zones are connected, and both constitute a wide zone of deformation reaching tens of kilometres in width and showing a complex geometry, including different active fault segments and in-relay folds. These findings demonstrate that Recent deformation is more heterogeneously distributed than commonly considered. A narrow SSW–NNE zone with folding and reverse faulting cuts across the western end of the Alboran Ridge and concentrates most of the upper crustal seismicity in the region. This zone of deformation defines a seismogenic, left-lateral fault zone connected to the south with the Al Hoceima seismic swarm, and representing a potential seismic hazard. Newly detected buried and active submarine slides along the Alboran Ridge and the Yusuf Lineament are clear signs of submarine slope instability in this seismically active region.     

改进的模糊C均值聚类算法及其在海底热液硫化物组分分析中的应用

在聚类分析中,模糊C均值(FCM)聚类算法有着广泛的应用。在实际应用中,该算法存在着很多缺陷,如最优聚类数目的确定完全依赖于数据的数目,算法易收敛到局部极值点以及收敛速度慢等。本文针对这些缺陷提出了2点改进方法:首先,利用减法聚类确定聚类数目的范围,提出一个新的聚类有效性指标函数,实现最优聚类数目的自适应确定。在此基础上,提出了基于粒子群(PSO)的模糊C均值混合聚类算法,以解决已有原始FCM聚类算法容易陷入局部极小点和收敛速度慢的问题。仿真测试结果表明:改进后的FCM聚类算法能够有效减少迭代次数,并以较快的收敛速度获得更加准确的聚类结果。最后,将改进的FCM聚类算法应用到冲绳海槽热液硫化物矿物组分分析中,准确地反映出了其矿物化学组分中主要金属元素的分布特征及矿石分类状况。     

Analysis of simrad EM12 multibeam bathymetry and acoustic backscatter data for seafloor mapping,exemplified at the Mid-Atlantic Ridge at 45° N

The EM12 multibeam echosounder can record acoustic backscatter information as well as high resolution bathymetry. The dataset presented, from the axis of the Mid-Atlantic Ridge at 45° N, was the first EM12 survey of a mid-ocean ridge. This paper presents methods for utilising the backscatter information. Data processing enables the production of a mosaic of acoustic backscatter, and visualisation techniques are investigated to provide initial qualitative views of the combined backscatter and bathymetry datasets. The co-registration of the backscatter and bathymetry data enables quantitative analysis of their relationships. Various sites of different geological type have been selected and their angular acoustic backscattering relationships estimated, including the effect on backscatter of incidence angle, its regional variability with bottom type and the influence of bottom slope. Incidence angles and bottom type are shown to affect backscatter to a similar degree, while slopes appear to contribute little. The geometry of hull-mounted systems, such as the EM12, is significantly different from that of conventional sidescan sonars, such as GLORIA, and the backscatter images from the two types differ in various respects. Because of the wide variations in incidence angle that are common with hull-mounted systems, and the importance of incidence angle in determining backscatter strength, it is vital to consider the effect of incidence angle during interpretation.     

Evaluation of image-based multibeam sonar backscatter classification for benthic habitat discrimination and mapping at Stanton Banks,UK

In recent years, efforts have increased to develop quantitative, computer-directed methods for segmentation of multibeam (MBES) backscatter data. This study utilises MBES backscatter data acquired at Stanton Banks (UK) and subsequently processed through the QTC-Multiview software environment in a bid to evaluate the program's ability to perform unsupervised classification. Statistical comparison with ground-truth data (grab, stills and video) enabled cross validation of acoustic segmentation and biological assemblages observed at the site. 132 unspecified variables were extracted from user-specified rectangular patches of the backscatter image, reduced to three vectors by PCA, then clustered and classified by the software. Multivariate analyses of ground-truth data were conducted on 75 stills images and 51 grab samples. Video footage coincident with the stills was divided into 30 s segments and coded by dominant substrate and species. Cross tabulation determined the interrelationship between software classifications, multivariate analysis of the biological assemblages and coded video segments. Multiview optimally identified 19 classes using the automated clustering engine. These were revised to 6 habitats a posteriori, using combined analysis of ground-truth data and Multiview data products. These habitats broadly correspond to major physiographic provinces within the region. Multivariate statistical analysis reveals low levels of assemblage similarity (<35%) for samples occurring within Multiview classes, irrespective of the mode of acquisition. Coded video data is more spatially appropriate than the other methods of ground-truthing investigated, although it is less well suited to the extraction of truly quantitative data. Multivariate analysis indicates assemblages within physiographically distinct Multiview classes have a low degree of biological similarity, supporting the notion that abiotic proxies may be contraindicative of benthic assemblage variations. QTC-Multiview performs well as a mechanism for computer-assisted segmentation of MBES backscatter imagery into acoustic provinces; however a degree of caution is required prior to ascribing ecological significance to these classifications.     

Wavelet analysis of bathymetric sidescan sonar data for the classification of seafloor sediments in Hopvågen Bay - Norway

In this paper we examine the use of bathymetric sidescan sonar for automatic classification of seabed sediments. Bathymetric sidescan sonar, here implemented through a small receiver array, retains the advantage of sidescan in speed through illuminating large swaths, but also enables the data gathered to be located spatially. The spatial location allows the image intensity to be corrected for depth and insonification angle, thus improving the use of the sonar for identifying changes in seafloor sediment. In this paper we investigate automatic tools for seabed recognition, using wavelets to analyse the image of Hopvågen Bay in Norway. We use the back-propagation elimination algorithm to determine the most significant wavelet features for discrimination. We show that the features selected present good agreement with the grab sample results in the survey under study and can be used in a classifier to discriminate between different seabed sediments.