Error analysis on normal incidence reflectivity measurement and geoacoustic inversion of ocean surficial sediment properties
Introduction
Normal-incident acoustic reflections from the ocean seabed are often applied in remote acoustic classification and quantitative analysis of ocean sediments (Hamilton et al., 1999; Meredith et al., 2015; Rodríguez-Pérez et al., 2014; Tang and Hefner, 2012). By incorporating with geoacoustic models, the normal-incident acoustic reflectivity estimates can be further used to predict sediment properties. Such measurements are usually performed with commercial acoustic systems such as chirp sonars and sub-bottom profilers. Numerous studies have involved estimating sediment properties using acoustic reflection data from the seabed (LeBlanc et al., 1992; Schock, 2004a, Schock, 2004b; Chiu et al., 2015), and the results demonstrate that the accuracy of acoustic reflection data measurement and of acoustic reflectivity estimation is essential in determining seabed sediment properties (Schock 2004a; Chiu et al., 2015).
Two main sources of uncertainties and errors exist in acoustic reflectivity estimates, namely reflection on rough surfaces and erroneous assumptions in the estimation. Firstly, the random reflection effects induced by the seafloor roughness and bottom curvature can affect the normal incidence reflection measurement, which can be reduced by averaging the reflected values over several pings (Chotiros et al., 2002). However, for significant bedform curvature, the normal-incident energy could be scattered or focused, respectively reducing or enhancing the reflection strength and consequently result in erroneous estimates of sediment parameters (Chiu and Chang, 2014). These uncertainties and errors arise from the characteristics of reflection on a rough surface and could be reduced by signal processing techniques.
Apart from the uncertainties caused by the seafloor roughness and bottom curvature, the uncertainties and errors in acoustic reflectivity estimates resulting from erroneous assumptions in the estimation approaches are non-negligible. Two approaches have been used to calculate the reflection coefficients of the seabed using acoustic reflection data; direct and indirect. The direct approach estimates the reflection coefficients using the reflected strength of the seabed reflections and the source level, while the transfer function of whole chirp sonar system is known. For the direct approach, the normal incidence return data is collected and used to calculate the reflection coefficient estimates of surficial sediments, which are defined as the difference (in decibels (dB)) between the intensity of the source function and the collected reflected wave, given the compensation of the two-way spherical spreading loss, with the known transfer function of the whole system.
However, because the precise transfer function of a commercialized sonar system is generally unknown, an alternative approach was suggested by Schock (2004b). The indirect approach estimates the reflection coefficients by using the seabed reflections and seabed–surface–seabed reflections while the information of the chirp system is not available. When the indirect approach is used to estimate the seabed reflection coefficient, the reflection coefficient of the air–water interface is often assumed to be −1. However, Landrø et al. (2013) used an approximation model to calculate the reflection coefficient of a rough sea surface; the results show that the reflection coefficient decreased with the increasing sea-surface wave height and acoustic frequency. Similar theoretical and modeling results were also presented by Mousavi et al. (2012) and Asgedom et al. (2017).
Whichever approach is adopted, several questions and uncertainties remain regarding the accuracy of the reflection coefficient estimation. First, is it appropriate to compensate for the propagation loss in the reflection coefficient estimate using spherical spreading? The uncertainties and errors that might appear in the spreading loss compensation in both approaches should be better understood and quantified. Second, for the indirect approach, is it appropriate to always assume that the reflection coefficient of the sea surface is −1? That is, the uncertainties and errors in the resulting reflection coefficient estimates should be analyzed and quantified.
In this study, the uncertainty and error that arise during the estimating process of normal incidence reflection coefficient will be explored in depth, and the error propagation effects on geoacoustic inversion will be analyzed. The remainder of the paper is organized as follows. Section 2 introduces the normal incidence experiments and describes the seabed sediment model and inversion technique applied in this research. Section 3 presents the data analysis results and discussions, including the propagation loss and the reflectivity measurements for both the seabed and sea surface. Also, a simulation case and the measured data are used to demonstrate and discuss the error propagation in geoacoustic inversion induced by erroneous assumptions. A summary of the results of this study and their implications are presented in Section 4.
Section snippets
2015 normal incidence experiment
In 2015, the Taiwanese R/V Ocean Researcher 3 (OR3) conducted a normal incidence experiment in a region of the southwestern seas off Taiwan that had a smooth bottom. The average water depth in the region was 220 m. Fig. 1 shows the bathymetry data of the survey regions with survey transects. In this experiment, the source (top unit) and the receiving system (lower unit) were deliberately separated in the water column, as shown in Fig. 2, in an arrangement that differed from typical chirp sonar
Results and discussions
The echograms, shown in Fig. 6(a), obtained by the multi-core platform during its ascent show multiple echoes. Moreover, the pulse-compressed signals lined up with the direct path signal; thus, during the multi-core platform ascent from the seabed, the arriving time interval between and increased. In this section, we analyze the echoes to study the characteristics and uncertainties of propagation loss and sea-surface reflections. The estimation accuracy of the seabed reflection coefficient
Conclusion
This work involved a normal incidence survey in the southwestern sea off Taiwan in 2015 using a multi-core platform with an acoustic recording system capable of photographically surveying the seafloor in near scope and acquiring sound data. Photographs, acoustic reflection measurements, and surficial sediment samples were simultaneously acquired at the same site to explore uncertainties in seabed reflection coefficient estimates. The accuracy of acoustic reflection coefficients and estimates at
Acknowledgement
This research and the experiments were supported by the Ministry of Science and Technology of Taiwan (project number: MOST 106-2221-E-110 -038-MY3 and MOST 108-2218-E-110-004). Special thanks are offered to Dr. Ying-Tsong Lin, Prof. Chi-Fang Chen, and all of the crew members of the research vessel R/V OR3.
References (25)
- et al.
Comparison of RoxAnn and QTC-View acoustic bottom classification system performance for the Cairns area, Great Barrier Reef, Australia
Continent. Shelf Res.
(1999) - et al.
Variability in normal-incidence acoustic response in shallow-water marine sediments
Continent. Shelf Res.
(2015) - et al.
A pulse-length correction to improve energy-based seabed classification in coastal areas
Continent. Shelf Res.
(2014) - et al.
Rough sea surface reflection coefficient estimation and its implication on hydrophone-only pre-stack deghosting
79th EAGE Conference and Exhibition
(2017) Theory of propagation of elastic waves in a fluid-saturated porous solid. II. Higher frequency range
J. Acoust. Soc. Am.
(1956)Mechanics of deformation and acoustic propagation in porous media
J. Appl. Phys.
(1962)- et al.
Normal incidence measurement in a subaqueous sand dune field in the South China Sea
J. Acoust. Soc. Am.
(2014) - et al.
Estimating geoacoustic properties of surficial sediments in the North Mien-Hua Canyon region with a chirp sonar profiler
IEEE J. Ocean. Eng.
(2015) Biot model of sound propagation in water-saturated sand
J. Acoust. Soc. Am.
(1995)An inversion for Biot parameters in water-saturated sand
J. Acoust. Soc. Am.
(2002)