Saturday, November 18, 2017
Published: Jul 2015
Authors
  • Eli Leblanc
    • Eli Leblanc
    • Company: CARIS
  • Burns Foster
    • Burns Foster
    • Company: CARIS

New Algorithm for Multibeam Imagery Processing

Over the past years, multibeam echo sounders are increasingly being used not only to obtain water depth information, but also to record backscatter response. This information recorded by the sonar is very useful in studies on marine geology (Anderson et al., 2008; Harahap et al., 2010), underwater works and military applications, as it is possible to relate the acoustic intensity response with ocean floor properties(Applied Physics Laboratory, 1994; Eleftherakis et al., 2012; Huang et al., 2011).

Introduction

Over the past years, multibeam echo sounders are increasingly being used not only to obtain water depth information, but also to record backscatter response. This information recorded by the sonar is very useful in studies on marine geology (Anderson et al., 2008; Harahap et al., 2010), underwater works and military applications, as it is possible to relate the acoustic intensity response with ocean floor properties(Applied Physics Laboratory, 1994; Eleftherakis et al., 2012; Huang et al., 2011).

Part of the challenge of existing processing algorithms is to diminish or weaken the influence of local bottom slope and near nadir reflection on backscatter strength data. Furthermore, the well-known method to estimate the sonar beam pattern over a patch of flat homogenous and sandy seabed is not fully adequate, as most surveys are conducted on unknown bottom type, and the data includes the angular dependency signature of this sediment. With all of this in mind, a new robust, documented and standards-compliant algorithm for backscatter processing, currently being integrated into CARIS HIPS and SIPSTM, is presented in this paper.

Material and methods

When implementing the algorithm, special care was taken to minimize user input. Most of the online parameters (frequency, along and across-track beamwidths, transmit power, receive gain, pulse duration, nominal spreading and absorption, sound speed) are read directly from the raw sonar files, along with the necessary raw data (beam angles, range and intensity). These parameters are combined with the processed bathymetry. Local temperature and salinity (on a broad scale) are the only parameters that are required from the user.

The new algorithm (Figure 1) is based on the assumption that after rigorous corrections have been applied to the individual raw backscatter intensities, the resulting fluctuations should be due to the sonar spatial angular spectrum (beam pattern) and the seabed geoacoustical properties and their relation with the grazing angle. The beam pattern is characteristic to the sonar on a particular installation so it is appropriate to use an in-situ approach to estimate it (De Moustier and Kraft, 2013). We can assume that the modulation introduced by the different angles of incidence and bottom types is fluctuating around a mean value. When enough data is fed to the beam pattern estimation, these variations should average out and a clear beam pattern should emerge. The resulting beam pattern can then be removed from the corrected intensities. After an angular varying gain (AVG) is applied to remove the angular dependency, or the typical response of the sediment to the angle of incidence, the resulting values are used to create a mosaic that should only represent sediment reflectivity.

Geometric corrections

The geometric corrections are necessary to translate two way travel time, angle and geographic positions in x-y-z coordinates, and to compensate for vessel motion, tide and refraction in order to compute precisely georeferenced soundings. These calculations are done through traditional means using long-standing functionality in CARIS HIPS and SIPS. The final positions will ultimately be used to create the mosaic. The resulting bathymetry will also be needed to compute the surface normal when correcting for the ensonified area.

Radiometric corrections

Radiometric corrections remove intensity fluctuations that are due to the sonar characteristics and processing, as well as the signal degradation and interaction with the environment that occurs between transmission and reception. These processing steps include correcting for the time-varying gain (TVG) applied at acquisition, for the transmission power and reception gains, for the ensonified area, and for the sonar spatial beam pattern and angular dependency.

TVG

A TVG is used to optimize the dynamic range and compensate for transmission loss that increase with range across the swath. The algorithm used to perform this real-time correction is specific to each sonar and normally uses a nominal value for absorption. The TVG applied at acquisition must therefore be computed and removed from the raw measurement and replaced by a transmission loss correction, taking a real local absorption. The local absorption is computed from the Ainslie and McColm model (1998), which requires the user to enter local values for temperature and salinity.

Transmission and reception gains

During acquisition, the transmitter power, and receiver gain typically can be set by the user or automatically adjusted by the sonar acquisition software. In either case, it can vary from ping to ping so it is important to remove them and replace them by a common reference.

Ensonified area

The area ensonified by the beam at the time of bottom detection is the intersection of the combined transmit and receive footprint with the pulse footprint on the bottom. The geometry and size of this intersection will vary with range, angle, transmit and receive beamwidths, pulse length and local slope. The instantaneous ensonified area is computed assuming an elliptical cone for the beam footprint and a spherical shell for the pulse (de Moustier and Alexandrou, 1991) and considering the surface from the neighboring soundings or bathymetric surface. Because the energy that scatters back to the receiver is proportional to the area of the reflector, the intensity corrected for gains is then normalized by the ensonified area.

Normalization

At this point, the intensities corrected for TVG, transmit and receive gains as well as ensonified area for a particular pointing angle would still fluctuate not only with the reflectivity of the sediement, but also because a variety of bottom types will be ensonified at various incident angles. To limit the effect for these variations, the intensity corrected for the ensonified area is normalized by its projection on a plane perpendicular to the beam vector.

To account for small variations in transmit power or receive gains that some sonars apply even for a fixed sonar setting, the total power in the ping is also removed from each normalized intensity.

Beam pattern

The normalized intensities versus steering angle relative to broad side curve is then interpolated at a 0.1 degree interval for each ping and averaged over the whole survey area. The resulting beam pattern is then removed from the intensities corrected for ensonified area.

Despeckle

At this step, the outliers are removed to avoid the angular dependency curve becoming buried in noise. For each sample, the intensity is replaced by the mean value computed from its 8 neighbors from the previous and next pings and beams, if the difference between the current and mean value exceeds a threshold. The mean value is computed in linear units.

Angular variable gain

A simple AVG algorithm is implemented to compensate for the angular dependency. For a current ping, the previous and next number of pings are stacked. Each sample in the stack is indexed according to its signed incidence angle (negative for negative pointing angle) and the mean value is computed for each bin in linear units. The curve is then smoothed using a moving average and is normalized by the mean value in the degree range. Each beam from the current ping is then corrected for the resulting curve.

Results

The new algorithm was tested on the common dataset provided for the 2015 Shallow Survey Conference. The beam pattern was computed over about 500 000 pings on various relief and bottom types.

The results for a portion of the dataset are presented in Figure 2. The raw mosaic is shown in (a) with a zoom on the orange box in (b). We can see (1) a strong discontinuity on the overlap between lines, (2) poor contrast on the features, and (3) a very strong angular dependency curve. While the eye can partly compensate for these artifacts and interpret the features, it would be very hard for an automatic classifier to obtain consistent results.

The fully compensated mosaic is shown in © and (d). We can see that the applied workflow is successful at correcting the artifacts observed on the raw mosaic. The lines blend smoothly together, the features are clearly revealed and the angular dependency is not visible.

Conclusion

A new algorithm for backscatter processing for CARIS HIPS and SIPS was presented in this paper. This method incorporates industry-standard corrections based on tried and tested acoustic principles as well as a new approach to estimate the sonar beam pattern more accurately. The new workflow was tested on real data and the preliminary results are promising, showing good agreement on line overlap, distinct features and few visible artifacts.

Bio

Eli Leblanc joined CARIS as a developer in 2014. She has been involved in underwater acoustics R&D for almost a decade, from marine mammal vocalizations and shipping noise, to port infrastructure inspection, data quality assessment, macroalgae mapping and bottom classification. She holds a Master’s degree in Engineering from the University of Quebec in Rimouski.

Burns Foster is the Product Manager for CARIS HIPS and SIPS. Burns has been with CARIS for 7 years, originally in support and training for processing-related CARIS products. He holds a Bachelor of Science in Engineering degree (Geomatics) from the University of New Brunswick.

References

  1. Anderson, J.T., Holliday, D.V., Kloser, R., Reid, D.G., Simard, Y., 2008. Acoustic seabed classification: current practice and future directions. ICES J. Mar. Sci. 65,
  2. Applied Physics Laboratory, 1994. APL-UW High-Frequency Ocean Environmental Acoustic Models Handbook (Technical report No. APL-UW TR 9407). University of Washington, Seattle, WA.
  3. De Moustier, C., Alexandrou, D., 1991. Angular dependence of  seafloor acoustic backscatter. J. Acoust. Soc. Am. 90,
  4. De Moustier, C., Kraft, B.J., 2013. In situ beam pattern estimation from seafloor acoustic backscatter measured with swath mapping sonars, in: Proceedings of Meetings on Acoustics. Presented at the ICA, Montreal, p. 070013.
  5. Eleftherakis, D., Amiri-Simkooei, A., Snellen, M., Simons, D.G., 2012. Improving riverbed sediment classification using backscatter and depth residual features of multi-beam echo-sounder systems. J. Acoust. Soc. Am. 131,
  6. Harahap, Z.A., Manik, H.M., Pujiyati, S., 2010. Acoustic backscatter quantification of seabed using multibeam echosounder instrument. Presented at the Third International Conference on Mathematics and Natural Sciences, Bandung, Indonesia.
  7. Huang, Z., Nichol, S.L., Siwabessy, J.P.W., Daniell, J., Brooke, B.P., 2011. Predictive modelling of seabed sediment parameters using multibeam acoustic data: a case study on the Carnarvon Shelf, Western Australia. Int. J. Geogr. Inf. Sci. 26,
  8. Stephens, D.D., Diesing, M.M., 2014. A comparison of supervised classification methods for the prediction of substrate type using multibeam acoustic and legacy grain-size data. PLoS One 9.
Back to Special Reports and Whitepapers Library
Marine Technology ENews subscription Marine Technology ENews is the subsea industry's largest circulation and most authoritative ENews Service, delivered to your Email three times per week
Subscribe for MTR E-news
Figure2-Teledyne-Reson.jpg
page 2

Introducing Teledyne Marine Multibeam Echosounder and Sonar Solutions

Teledyne Marine is a leading provider of advanced multibeam echosounder and sonar solutions servicing a variety of markets and applications. By delivering the most advanced acoustic technology solutions Teledyne Marine provides high resolution, accurate, dependable, long lasting solutions to our customers.

Port-of-Rotterdam-image.jpg
page 12

How to produce bathymetric products within one day after surveying.

The Port of Rotterdam is an open deep-water port in theriver estuary Maas. The location of the port in an estuarineenvironment necessitates maintenance dredging due tosiltation ’attacks’ from both tidal current and riverdischarge. Siltation from the sea, coming in during theflood tide and eroded materials from the hinterlandtransported in the river flow results in a total of 6.000.000m3 of maintenance dredging works in the port basinsevery year.

Spillway-fmSky-300dpi.jpg
page 16

Implementation of a Multibeam Echosounder and Terrestrial LiDAR in Support of a Dam Spillway Construction Project

Nestled Northeast of the city of Folsom, California, sits the 11,450-acre Folsom Lake—formed by an existing dam on the North Fork and South Fork of the American River. It is responsible for providing irrigation, drinking water, and electricity to parts of California, as well as flood prevention for the areas situated south of the lake. Despite the low risk of a damaging occurrence, it was decided (under the Reclamation Safety of Dams Act) that preventive measures needed to be taken to protect those living downstream. Construction of a new auxiliary spillway is in progress, which will provide a means for a steady water outflow during an increased water flow event. Carried out in phases, this project requires surveys from land and water to ensure successful completion. The $900-million cooperative project between the U.S. Army Corps of Engineers, U.S. Department of the Interior, and Bureau of Reclamation will assist the Sacramento region to achieve the 200-year level of flood protection.Completion of the auxiliary spillway is scheduled for late 2017.

QPS-Fig-1.jpg
page 22

High Quality Surveys Using hybrid AUVs and state of the art hydrographic data acquisition software

The acquisition of QPS by Saab opened up for the possibility for closer integration between the Saab AUV platforms and the data acquisition software from QPS. This resulted in a joint project in 2014 to demonstrate a high end survey system that was capable of very accurate combined sonar, multibeam bathymetry, video and subbottom profiling. The system is based on Saabs hovering AUV capable of very stable autonomous close up scanning of both structures and bottom features. The QPS data acquisition software ensures that all data is collected together with navigation data in an optimal way for precision and further processing. The result is merged data of high quality that is easily evaluated and exported.

Graph-1--hypack.jpg
page 34

HYPACK Inc.: Experts in Hydrographic Software

HYPACK, Inc. has been a world leader in software development for the hydrographic and dredging industry since 1984. HYPACK® is the most widely used hydrographic software package with over 10,000 users worldwide. It provides everything necessary to fulfill your singlebeam, side scan, and multibeam survey requirements. This is why?

Fig_1_Gavia_OffShore_Surveyor_Two_Batteries.jpg
page 45

LOW-LOGISITICS MARINE SYSTEMS ENABLE NEW APPROACHES TO SURVEY

Unmanned maritime vehicles (UMVs) have been under development for decades. Since the late 1990s, several systems have evolved beyond research labs and have become commercial realities. These tools are now in routine use for scientific, commercial and military applications.

;