Thursday, August 20, 2015

How is eelgrass mapped?

  • First we look at aerial photos (Google Maps/Earth, Bing, those flown for the state).
  • In some places, we run acoustics to look at deeper waters or cover time periods that we don't have aerial photos for. 
  • Then we get in the water with cameras to confirm (groundtruth) that we're identifying eelgrass and not algae.
  • These steps are illustrated below.

Is the dark shading in the aerial photo eelgrass in the center of Cohasset Harbor?

Step 1: Go out in a boat with sidescan and underwater camera.

The water is shallow where eelgrass grows and instruments are fairly sensitive to waves.  So use a small boat or jetski on a calm day.  Here we're using a Humminbird 698SI (200 kHz) sidescan. The transducer is mounted on a pole attached to the boat with a trolling motor bracket.

This is the Humminbird 698SI sidescan unit. The  photo to the right shows the whole unit as it comes out of the box.  The next photo down shows the transducer on the pole and the processing (and GPS/nav) unit is the smaller computer screen mounted to the table in front of the boat driver.

This is the GoPro Hero2 camera on a line with a shackle for weight.  We set it up to automatically snap images every 2 seconds.  Before each station, we write the station number on a white board, take an image of the board, then toss the camera overboard and lower to seafloor then bring it up slowly. We record the lat/lon of the station on paper and using the waypoint feature on the Humminbird sidescan unit.  This works to about 20 feet.  Then more weight is needed.  Deeper than about 30 feet might need light too.

Step 2: Drive survey lines with the sidescan and drop camera overboard at stations throughout the survey area.  We can cover 150-200 acres in 4 hours.

The survey lines are in yellow.  We space the lines approximately 200 feet apart to achieve about 150% coverage of the sidescan.  We do that by eye using the chart on the Humminbird sidescan unit.  The drop camera stations are the white dots.  These were distributed on the fly in the field to achieve 6-10 stations per planned survey line.  In this survey we added survey lines to the northern section late in the day, on the fly.  We didn’t bring the deeper drop camera, so that area wasn’t included in the photo groundtruthing.

 Step 3: Process sidescan data in SonarTRX or CARIS HIPS-SIPS software, create mosaic.

Here is the mosaic, I don’t know why there are two blank sections –one in the middle toward the top and one in the northwest corner.  The data is there and when I zoom in and out it blinks on and off.
Step 4: Examine groundtruth photos and compare to sidescan and aerial photos.  Identify patterns in aerial photos and sidescan that are consistent with eelgrass (or other habitats like gravel).

The image below has the grass stations coded in green, sand in yellow, and sand-gravel in orange. All data files are available for download as kml/kmz files for opening in Google Earth.
Sidescan sonar png files
Lat/lon of photo stations with notes
Photos geotagged to the correct location (several per station)

Thursday, March 5, 2015


This has been the iciest winter we've seen on the South Coast of Massachusetts in a long time. I haven't been able to quantify this yet -- a NYTimes article said the Coast Guard severe ice bulletins were issued this year "for the first time in more than a decade."  The eastern shore of Buzzards Bay has been completely iced in.  This image was captured on February 25 by the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Aqua satellite (thank you to Greg Bettencourt, MADMF). We're wondering how this will impact eelgrass beds.

Quite a few folks have been taking videos.  Here are some links:
Brian Switzer, Woods Hole
My drone shots of NB Harbor
Peter Perriera drone shots of NB Harbor
Tom Dunlop, Vineyard Gazette story about Chappaquiddick Ferry battling ice
Tyler Fields photographs of Sea Ice (Buzzards Bay ice just south of Mass Maritime/Wareham)
The Buzzards Bay Coalition has some photos on its Facebook page

Monday, November 3, 2014

Camera research

Doing some more camera research.  I tend to do this every year or so and find new things each time.  Right now I'm thinking about how to add a video system to the jetski to groundtruth eelgrass. Good battery life, waterproof, and very compact are all important features.  High def isn't critical.  Light isn't that critical either since we'll be in less than 30 feet of water primarily.  I want to be able to record the images.  Grabbing stills and recording video preferred.

Getting your GoPro to work in odd situations:

How to extend the GoPro wireless underwater (very similar solution as the cam-do website, just homemade):

Very small all-in-one solutions: (search for underwater video fishing for the smaller all-in-ones)
Aqua Vu Micros
MarCum Pan Cam and VS485C
Docooler  Fish Finder (can't record)
Lucky Fish Finder (can't record)
Aquabotix AquaLens and AquaLens Pro

Comparison table:
Cost Camera res Screen size Power Cable length Recording
Aqua Vu Micro Plus  $   500 color 1/4 CMOS 3.5 battery 50' 8 gb SD card
Aqua Vu Micro 5  $   600 color 1/4 CMOS 5 battery 100' 8gb internal
Pyle PFSHCMR1  $   200 color 1/4 CMOS, 30MP still 3.5 battery 15m microsd
MarCum VS485C  $   400 color 1/3 CMOS 7 battery 50' ?
MarCum PanCam  $   350 color 1/3 CMOS smartphone battery 50' smartphone
Aquabotix Aqua Lens  $1,300 color 3.5 battery 25' optional
Vexilar FP100 FishPhone  $   270 color CMOS, stills smartphone battery 50' smartphone

Slightly larger all-in-one solutions:
Aqua Vu 360, 760c, 760cz, and 715c
DB Power (this appears to be a generic version; other vendors were selling this same system)
Snake Mate
Gnom Drop Camera
JD (needs screen and recording device)

Wednesday, October 1, 2014

Lobster on camera

This lobster was captured on camera in 2011 on Jeffrey's Ledge.

Monday, August 18, 2014

Humminbird 698SI survey

Steve and I mounted the Humminbird on a jetski and used it last Friday to check out an eelgrass meadow in Clark's Cove, New Bedford.  It worked well.

Here's the jetski with the Humminbird.  The transducer is on a trolling motor mount so we could easily move it up and down to try and avoid noise.  We could also pop it out of the water for high speed travel.  The set up was fine, though noise is an issue.  With the transducer about 2-3 feet down in the water, the signals were acceptable.  We collected with a 120 foot range and used an enhancement setting of 9.

The processing workflow was straightforward.  I first checked the lines in HumViewer, a free viewer that plays back the sonar data.  This helped me identify which files I wanted to fully process.  Since we were just doing a test run, we didn't keep careful track of when we pressed record!  It also turned out to be very useful because the viewer shows the sounder data as well, and our target habitat, eelgrass, is clearly visible in the sounder data.  I discovered that when we were directly over the eelgrass, it was actually difficult to see in the sidescan image, but it was very easy to see in the sounder data.  The edge of the eelgrass meadow was clear.  The images below are the unprocessed data.  Depth and range are in meters.

The next step was to slant range and beam angle correct the data.  These processing steps remove the water column and make corrections so the sonar image is georeferenced and gain corrected across the range.  I used SonarTRX and it was easy, quick, and resulted in images we could interpret.  I've used Caris HIPS/SIPS in the past, and the batch processing capabilities will probably send us back to Caris, especially for larger surveys.  But the simple and intuitive workflow and easy export to Google Earth (and ArcGIS) made me an instant fan of SonarTRX.

You can see that in some lines it was pretty hard to see the vegetation nearer the sonar -- the best images are at about 1/2 range, or 60 feet.  Even in the 2nd image it might be interpreted as having two beds.  However, as I mentioned earlier, the sounder sonar (downward looking dual beam of 200/83 kHz) clearly showed that there was vegetation under the jetski.

I think the next step is to tweak some of the image settings to try and draw out the vegetation signal more clearly in the sidescan images.

First fusion table: survey map

Here's my first fusion table and Google Map.  If you click on the dot on the map, you'll see information about our latest survey.

Monday, August 11, 2014

Bathymetry and side imaging on a kayak

This webpage has a thorough description of a setup of a Garmin transducer on a kayak.  It looks like the project rapidly advanced to a Humminbird side imaging sonar.  (Thanks to Mike Sacarny, MIT for pointing out the sonar projects on this site; I previously used the same website for information about remote controlled helicopters for mapping.  Just scroll down to the list of projects at the bottom of the home page.)

I liked the idea of using a portable car emergency power pack.  As we learn about operating on smaller vessels, we'll have to move away from the generator.  Also, the generator is heavy and stinky, so for shorter surveys, a battery alternative is ideal.

What mapping software was used for the bathymetry?  I assume something like ArcGIS.  What about the side imaging?  Was it SonarTRX or a higher-end software like Caris or Hypack or was it mosaicing of screen grabs?  Can something like Hugin (which Mr. Illsley provides links to elsewhere in his website) be used to stitch photos together, and then something like ArcGIS or MapKnitter or QGIS be used for the orthorectification?