Surveying the Seafloor: Looking Ahead to See Below

Motivation

What is under the water in front of your ship right now? With the use of only charts, radar, and other traditional methods, this can be a difficult question to answer. This is where FarSounder’s forward looking navigation sonar comes in, and through both research and collaboration, FarSounder is able to conduct numerous survey projects in the Northeast.

Introduction

Figure 1 - Argos 1000 attached to the pole mount being lowered into the water

The main purpose of FarSounder’s 3D-FLS is for navigation in real-time, however, with the recent introduction of Local History Mapping™, users can save bathymetry data from areas which they’ve recently visited. Naturally, questions about the accuracy of bathymetry data collected using 3D-FLS arise, and FarSounder’s efforts to benchmark the performance of 3D-FLS against reference data began with a technical paper and presentation at OCEANS2017, where two surveys were compared to reference data in Narragansett Bay with relative errors of less than 5%. FarSounder is continually collecting more data, from customers around the world and through university partnerships.

In August of 2018, FarSounder joined the University of New Hampshire (UNH) on a collaborative boat trip surveying an area near Portsmouth harbor. Along with single beam echosounder and survey grade inertial measurement system on board the RV Gulf Surveyor, a FarSounder-1000 3D-FLS was installed on the vessel’s hydraulic pole mount (Figure 1).

Summary of Results

UNH provided us with the data from the sensors onboard the RV Gulf Surveyor from that day, in addition to data from a survey in the same area, using a standard survey grade Multibeam Echosounder (MBES) recorded just 6 weeks before this trip. With this information we looked at comparing how our sonar performs against their multibeam sonar with regards to bottom bathymetry. The multibeam data was an xyz file CUBE grid with 0.5 meter resolution consisting of over 10 million points!

QGIS is an open source tool used to “create, edit, visualise, analyse and publish geospatial information” . It can handle xyz files with ease and is able to load in our gridded bottom data from a Postgres server. Simply add a base map of Portsmouth harbour under this and now we can start comparing the two data sets.

Figure 2 - Interpolated bathymetric data from: left - MBES in June 2018, right - Argos 1000 3D-FLS in August 2018

Both the multibeam data and the gridded bottom data, generated using the standard Local History Mapping™ algorithm from the FarSounder, are vector layers, but they have different spacing (0.5m and 4m respectively). It is easier to first convert them into discrete raster layers to compare them. To do so you must interpolate values between the points. QGIS provides this functionality, Figure 2 shows the result of interpolating the datasets.

To show the contours of the bottom better, a hillshade was applied and placed underneath the layers shown in Figure 2. Figure 3 shows a comparison flipping back and forth between the two datasets. In looking at both figures, the qualitative similarity of the bathymetry between the two layers is immediately evident.

A more in depth study based on this data is still to come, where quantitative metrics will be considered and the methods of analysis described in detail. However, because the datasets are raster layers, the average absolute percent error between the two layers can be calculated taking the UNH MBES data as a reference, and the resulting value is ~4%. It is exciting that this level of accuracy is obtained in real-time without performing sophisticated correction for orientation, tide height, sound speed profile, or any manual post-processing or data cleaning.

Figure 3 - Surveys performed using each sonar in Portsmouth Harbor. The Argos 1000 is toggled on and off

Conclusion

The main application of FarSounder’s 3D-FLS systems will always be real-time navigation and obstacle avoidance, and of course, one cannot expect to obtain more accurate bathymetric data than with a survey grade MBES. However, there are many situations where it would make sense to use a 3D-FLS instead. For applications in which a navigation sonar is already present for real-time navigation, basic survey capabilities can be added without adding any extra equipment. SonaSoft™, our 3D-FLS processing software, requires no operator input after initial configuration, making it easy to use for navigation in real-time. In addition, our algorithms for both real-time target detection and seafloor mapping, and storing bathymetric data using Local History Mapping™ function without user intervention.

Many applications do not require strict IHO performance standards, and for those applications, especially when a 3D-FLS is already equipped for navigation in real-time, the relative error in the bathymetric data collected using the 3D-FLS is more than sufficient. With this said, the low relative error, without requiring other advanced sensors onboard the vessel, manual post-processing, and/or user intervention, makes the FarSounder system an easy to use alternative to traditional MBES, when accuracy isn’t paramount, and a complement to traditional MBES when a FarSounder system is already present for real-time navigation.

Here at FarSounder, in addition to collecting data, we continue to be hard at work developing new 3D FLS technology in order to better understand how forward looking navigation sonar can continue to be used to supplement traditional down looking survey measurements.