Hi All,

I'm new here and currently testing out the Reef Master software. I have a heavy background in Surveying, GIS, and Photogrammetry with lots of experience using different types of sensors. I build custom unmanned aircraft for mapping above ground and am starting to build unmanned "boats" for mapping underwater.

We are using many open-source components that are very inexpensive, compared to other similar hardware/software. We use autopilots that have 6-axis gyros/accels with digital compass, GPS, laser range finders, etc. We also integrate Raspberry Pi's to log data and then merge the data together after it's collected. The GPS we utilize is also very inexpensive for its cm accuracy, which could help immensely with mapping accuracy.

Enough background on me. I wanted to talk with others about some of this new technology and how to implement it with sonar technology for mapping. I have a lot of interest in sidescan sonar and especially with the SL3, which is using PDBS (I believe). Those two technologies combined with the above sensors I mentioned could create a very good bathymetric package at a very affordable price.

Sorry if I left anything out.
Let me know your thoughts and any experience.

Best,
Steve

Welcome Steve

You are right to post here, there are some very good technical people checking in here. I am working on shallow water mapping here in Northland New Zealand, and have a colleague who is very involved in drone work.

Steve could you elaborate a bit more on your thoughts with working with the sl3 data and format.
With my work I need to crack a reliable and accurate method to map algal forests and the degraded condition of urchin grazed zones. A complication for me is that some of the shorelines I am interested in are quite steep, thus the need to look to 3D approaches. Also I need to be able to 'sense' intermediate conditions as well.

Steve I can't contribute to the build developmental issues you are working on, but there are people here who can. What I can do that might be useful is the application of these things in the field and I am happy to test things and share data. I work independently so have no problems with collaborations.

I have a new Lowrance gen3 going in my boat at the moment, but I am a bit in the dark of what can be done with this data stream.

Hi Vince,

How shallow of water are you wanting to map? Also, what do you think is the approximate Secchi depth of the water?

There are currently 3 types (that I have found) of "pro-sumer" 3D bathymetric hardware. The Lowrance SS3D, Garmin Panoptix, and the Humminbird 360. If it was possible to extract the 3D data from any of these systems, for use in 3rd-party software, we could perform some very accurate shallow water mapping. (This is all speaking from my some-what limited sonar sensor knowledge). As for the SL3 format, in particular, I believe it contains the angles where each sonar beam pings back to the transducer and could be used to create a 3Dmodel.

Another method, that we are exploring, is the use of SfM Photogrammetry algorithms to extract 3D data in the overlap of sidescan "images". We need 80-90% overlap sidescan images of high-resolution to experiment with.

Best,
Steve

Hi Steve

Thanks for response. To answer your question. I mostly work on the northeast coast of the North Island of New Zealand. Our region is sub-tropical and for much of the year is a lee shore, however it also has storms that arrive from NE E and SE and these can be major with significant wave energy. Visibility varies a lot but is normally 6m or better. We can get 10-15 viz days quite commonly in some parts of the year. Offshore viz is better yet and beautiful Pacific blue water with tuna and marlin can be as close as a few klms offshore during much of the summer. Is there some good recreational fishing in New Zealand the short answer is YES. While I am a life long fisher, these days I mainly focus on photography and marine conservation work.

My area of interest at the moment is the shallow kelp forest areas typically from the intertidal down to 25 m deep although most of what is important to me is in the band between about 2m and 15m depth. As mentioned earlier these shallow reefs are very diverse in exposure, complexity of the terrain and slope.

I was particularly interested in your comment about the Sl3 format.

"As for the SL3 format, in particular, I believe it contains the angles where each sonar beam pings back to the transducer and could be used to create a 3Dmodel."

When I first started testing my Lowrance Gen 3 it dawned on me that there must be a data stream in there that was somehow portraying the triangle position between center (below boat) distance to left right and angle to transducer. Obviously this gives you depth of any point out in the side scan image. Yet it seems all our bathymetry models are being built on just the downscan cone data below the boat. I don't get it. If what you are saying is correct than surely there is valuable and useable 3D data there. A related question is why can't this triangle of data be derived from any side scan data stream. I guess the simple answer is that the transducer is not programed to receive or process the beam angle of returns from out on the side away from the center line. I have never quite understood this because how can it calculate a distance from the center if it is not working this out from the angle of the return from a given point. How would shadow heights be rendered in an image without this data?

Just at the moment I am planning to change sounders to the Humminbird gen2 mega imaging beast that has recently been released. I think the increase in frequency and transducer output will make a difference to to the quality and resolution of side scan imagery. I am most optimistic about solving the interpretation challenge of various states of algal forest health by making the best SS mozaic possible via Hum mega imaging sounder and Reef Master and then with this mozaic image using some sort of image analysis tool to classify the differences I am after. I am trialing some object orientated image analysis software solutions at present. This is a working idea at this stage there are clearly many ways to attack this particular challenge.

Given my non existent research budget I was planning on selling the Lowrance when moving to the new Humminbird but it worries me a bit to walk away from the Sl3 format data collection option as the logic is to me that this should be a useful 3D mapping tool. Good if others here with the background knowledge could comment on this. Ultimately I will need a good 3D map of the vegetation mapping I am attempting.

vincekerr wrote:
Typically, across track distances are measured on sidescan with the assumption that the surface is completely level. Shadows are not measured so much as they represent a gap in time on the sounding return because the water body isn't level or consistent. A shadow can be a blind spot from the sonar because either something is in the way or something absorbed more of the sound than the neighboring areas.

With depth interpolation, you can likely get more accurate placement but the assumption of a level surface is reasonably accurate (as the hypotenuse increases, the distance equation typically becomes dominated by the across track distance).

Hi Vince,

Very cool place where you live!! I live in the Midwest U.S.; small lakes, rivers, and streams :)

A typical sidescan sonar only has the distance to each ping and not the angle. There are some sidescan units which have multiple sonar elements so they can calculate the angle as well I'm guessing that the Lowrance StructureScan is something of this nature. Although, Navico might only share the distances in the .SL3 file and not the angles in the file format when they unlock it the ReefMaster.

There are three consumer grade sonars I'm aware of which can produce 3D data, the Humminbird 360, Lowrance SS3D, and the Garmin Panoptix. Although, I'm not sure if there is a way to extract the 3D data from any of those products due to the proprietary nature of their business.

We have extensively looked into the sidescan and units by Lowrance, Humminbird, and Garmin as well. We are in the process of testing a few different ways to create an affordable 3D underwater survey using the Humminbird Helix9 Mega. We chose that unit because, as you said, it has the highest resolution.

One way is we are looking into the possibility of utilizing photogrammetry techniques to create 3D data from the sidescan horizontal "images".

Another possibility we are looking into is what we'll call "Sidescan Sonar Triangulation". In theory, we could use simple triangulation to get the 3D data using a SS sonar from two different positions. If we take the precise position of a GPS synced with the transducer from two parallel lines, we can simply intersect the two circles (which are the straight line distances from the transducer) and omit the intersection above the water. This would give us pretty accurate 3D points underwater. All of the SS sonar units we have looked into have built-in GPS. We can easily sync the GPS times between the built-in and a higher accuracy GPS (Emlid Reach). Or, if you want to do real-time kinematic, you could input the NMEA stream from the higher accuracy GPS. It may be more accurate to get a heading from two Reach units as well? That will require some testing to know for sure.

We are in the process of procuring all of the equipment and hope to start some tests this spring.


Best,
Steve

Thanks Greg and Steve
Well that clears up a few basics, but I can see there is a lot more to come from this discussion.

Steve, Matt here at Reef Master will be very interested in exploring what he can extract from Sl3 data, so a great idea for you to compare notes with him as you go.

I took a look at developments with 'Emlid Reach', and their accurate GPS gear. What can I say those guys are on fire, and I can see the logic of where you are going with SS triangulation. Kind of exciting really if we can get really good resolution SS to unleash this technology in accurate 3D.

I think for me I will plunder on with the Hum mega imaging approach and build from there, this means some data starts to come in end of March/ April as the machines are not released in NZ until mid-March. Will be a lot easier to continue this discussion with real examples in hand. As I start to do some field work Steve, I can look at doing some 'parallel SS runs' that could fit your 3D processing protocol.

Hi Vince,

Yes, I have been chatting back and forth with Matt over email quite a bit lately.
I'm trying not to bother him too much so he can work on RM 2.0 :)

I know he is very interested to work on this once 2.0 is released.

My thoughts are this will create some decent 3D bathymetric data (not quite the resolution of SS "images"), but then you could always "drape" the SS image over the more accurate 3D bathymetric data and have a better idea of how things look. Although, this is all in theory. We'll see how it turns out once we start collecting some good data and process it. By the way, if you are collecting parallel SS test data, try and get a good ~80% overlap in the SS. Obviously, we'll need accurate transducer location and heading as well if we are going to try SS Triangulation.

Best,
Steve