Session I- Part B Image Interpretation

The approach we take to mapping habitat involves

manual digitization and classification of features based
on visual interpretation of sonar imagery.
Photographic interpretation of imagery as a basis for
map development is a long-standing, tried-and-true
approach in the field of cartography.

A very relevant question, couched in terms of reducing

potential subjectivity in the manual process, is whether
classified habitat maps can be generated in an
automated fashion? Automated, computer-based
approaches to segmentation and classification of side
scan sonar imagery are currently in various phases of
development and evaluation. Few demonstrations of
such approaches can be found in the literature, and
most are limited to benthic marine settings with open,
flat topography and reduced substrate complexity.
User input is often required for computer “training” on
the front end, and editing and correcting errors in draft
maps generated from automated routines is typically
needed on the back end. One could argue these
inputs are user specific and potentially subjective as
well. Automated, computer-based approaches are not
widely available, and require additional image
processing software packages and specific expertise.

Indeed, one of the hurdles for development of reliable,

automated approaches to mapping with side scan
sonar imagery is the inherent complexity of side scan
data and sonar image products. Making sense of this
complexity is the foremost topic of this chapter. 1
Visual interpretation Creating a Habitat Map
The process of creating habitat feature layers by visual
interpretation of sonar imagery is much like tracing a
scene from a photograph. High quality imagery, and Truism #1- Image (Data) Quality and
the ability to critically examine, identify, and
differentiate patterns (i.e., sonar signatures) common Interpretation are the foundation of low-cost,
to the surveyed system are essential inputs to this
process. Sonar interpretation and map making skills
sonar habitat mapping
can be improved through training and experience, yet
also draw upon a set of human aptitudes that includes
keen observation, powers of discrimination, attention
to detail, and consistency. These aptitudes serve both
art and science!

The ability to accurately interpret sonar images is of

such great importance that we devote the remainder
of this session to the topic.

2
Large underwater features Interpreting Sonar Imagery
Let’s begin our discussion of image interpretation with
this raw sonar image captured shortly after passing
beneath a bridge spanning the lower Flint River in
Albany, Georgia. Across the top of all HSI screen
snapshots is a display of the range setting. A setting
of 150 feet per side was used to create this image- this
represents the distance from the centerline (i.e., the
boat path) and the edge of the image. As we see
here, the river bank was much closer than 150 feet on
the right hand side of the image. The well defined
150 feet
dark margin along this edge represents the river bank.
The rather large, blocky shapes in the middle of the
image are submerged, concrete bridge abutments.
These structures reflect the sonar signal, casting sonar Broad St.
shadows behind them. We have circled a few of the
very large boulders that are resting on the riverbed in
Bridge
this reach. These boulders somewhat resemble cotton abutments
balls and also cast shadows- an indication that these
objects are protruding up into the water column. A log
can be seen resting next to the boulders in the middle
of the image. If you look closely, you can find what
appears to be part of a log sticking out from the edge Image
of the upper right side bridge abutments.
Centerline
The information panel along the left of the image can
Large (boat path)
be manipulated to display a variety of information rocks
available at the exact time of capture, like GPS
coordinates (not shown here). The depth, 10.8 ft, is
the depth at the point of capture, which is the position
located at the top center of the image (behind the blue
boat icon). 3
Same bridge abutments... Different Imaging Conditions
An interesting exercise in comparison can be made
using the adjacent image. This image comes from the
same reach of the Flint River approximately 1 year
later- it has been rectified or transformed, unlike the
previous raw image. Some rip rap was added around
the bridge abutments on the west (right) side of the
river, and around one abutment on the east side that
did not exist when the earlier image was captured.
The bridge abutments are more defined, and the
shadows are well defined and narrower. The reason
for this difference is that the river stage was higher
when this image was made, and the water was
completely covering the foundational elements. As we
will see on the next page, the bridge uprights are
narrower than their concrete bases. Another reason
why this image is sharper and more well defined is
that a front-mounted transducer was used to create
this image rather than a rear-mounted transducer. The
importance of this deployment will be discussed
shortly.

A second truism of sonar imaging is that no two sonar

images of the same area look exactly alike, even if
captured on the same day just minutes apart. It is
impossible to replicate the sonar imaging conditions
experienced during the creation of an image.

Truism #2- No two sonar images

of same area look exactly alike 4
Low water conditions The Bridge Abutments
Here is a digital photo of the west side bridge
abutments of the bridge over the Flint River taken
during low water conditions. The difference between Sonar Shadows
abutment base and uprights is plain to see, as is the
representative signature of these structures in the
sonar image.

5
Texture difference Rip Rap
This photo was taken while looking at rip rap (i.e.,
boulders) added to the bridge abutment area. The
sonar signature of this material is clearly different from
the surrounding riverbed.

6
The water column Interpreting the Water
One of the unusual features of a sonar image that
bothers a lot of people is the centrally located, dark
area that represents the water column. The width of
Column
the water column is a direct representation of the
depth of water beneath the transducer. A wide water
column represents deep water, and a narrow column Dead Zone
represents shallow water. Water column Water column
Many mistakenly assume that this dark area represents
missing data. The truth of the matter is that there is
very little missing data within this region of the sonar Deep
image. If missing data exists, it will occur in a very
narrow band, here identified as the “dead zone”.

If we are not actually missing data due to the water

column, how then should a sonar image be properly
interpreted?

Aren’t we
missing
data
beneath
the boat!?
Shallow
7
The water column Objects directly beneath
To properly interpret sonar imagery with the water
column displayed we must imagine that both sides of
the sonar image actually join together right down the
transducer
middle of the image, as if the water column does not -Appear as mirror images on either side
exist. Visual proof of this concept is occasionally
obtained when the boat happens to pass directly over
an object, or set of objects, like these boulders. The
boulders appear as mirror objects on either side of the
image. Imagine mentally removing the water column
and stitching the two halves of the image together
down the center- the modified image would have a
series of three or so boulders that were located directly
beneath the boat during the survey.

8
Mirror objects Objects directly beneath
Here is another example of an object that was directly
beneath the transducer during the survey. In this case
a log, oriented parallel to the river channel, appears on
transducer
both sides of the water column in the center of the -Appear as mirror images on either side
image. We are not, however, looking a two separate
logs, but rather one log that was perfectly split down (the log in this example)
the middle during the survey (this is quite a rare
occurrence!). If we imagine removing the water
column, and stitching the two halves of the image
together, we are left with one log directly beneath the
boat.

A skeptic may think this phenomenon is limited to

shallow waters- that objects located beneath the boat
in deeper water would not appear as mirror objects as
seen here.

Depth
~4-5 feet

9
Objects in deep water Objects directly beneath
Here is visual proof that objects located directly
beneath the transducer, even in deeper water, will
appear as mirror objects on either side of the image.
transducer
-Appear as mirror images on either side
So, if the water column does not represent missing
data, what effect does the inclusion of the water another log, this time in deeper water
column have on the representation of objects and
features in this image?

Depth
~14-15 feet

10
Image compression How is the image affected?
The water column occupies some of the space
available for image creation, and in doing so leads to
some compression of objects and features appearing
Slant Range Distortion
in the near-field (near water column) portion of the
image (red boxed area). Compression of near-field Near-field Compression
features increases as the width (depth) of the water
column increases relative to a fixed range setting; the
compression effect dissipates with increasing distance
from the water column. Likewise, the positional error
of features attributable to compression increases with
increasing depth (i.e., more water column showing).

Due to compression, objects or features in the near-

field portion of the image are smaller and closer to
each other than in reality, and these features are not
truly in their proper spatial location, as pointed out
during our previous discussion of mirrored objects.
The image distortion created by near-field compression
is also called slant range distortion. The processing
required to remove the water column from the image
and undo the compression is called slant range
correction.

When interpreting images that include the water

column you (the reader) must mentally perform slant
range correction by imagining the removal of the water
column and the slight adjustment of size and position
of features appearing in the near-field portion of the
image. The process of digitizing features when the
water column is showing will again be addressed in the
chapter on habitat mapping. 11
Water column present Slant range correction now
For several years since the release of the
Humminbird® SI series it was not possible to perform
available!!
slant range correction on sonar screen snapshots. All
screen snapshots, by default, displayed the water
column. A recent firmware update (Jan 2011) has
added a feature called “Contour Mode”, an option Boat track with
under the SI Enhance tab of the control head settings water column
menu. Contour mode enables the sonar operator to
choose between screen imagery that either includes displayed
the water column, or performs on-the-fly slant range
correction to remove the water column from the image
display. We call this an “on-the-fly” process because
the control head is incorporating slant range correction
into the internal processing that occurs in real-time
during field scanning.

In other, high-end sonar systems, slant range

correction is applied after the raw sonar data has been
recorded, during the data processing phase. This
approach to slant-range correction is often partially
automated, with user input required to edit and correct
output wherever necessary (i.e., where the computer
fails to properly identify the true bottom).

On the right are two rectified sonar image layers

obtained using parallel transects to cover the entire
channel across a river. The imagery obtained during
this survey included the water column in the display.

12
Water column removed Slant range corrected imagery
When slant range correction is applied via the contour
mode setting, the near-field portion of the image is
decompressed, bringing both halves together along the
Boat track with water column removed
survey path (centerline). When performing optimally,
slant range correction produces imagery that
seamlessly covers the survey swath. In the pair of
slant range corrected mosaics to the right the boat
path is barely perceptible as a faint line down the
middle of each layer.

The ability to remove the water column on-the-fly is

pretty slick, and water column haters will rejoice at this
development. Let’s briefly discuss, however, some of
the costs and benefits of enabling this feature.

13
Pros and cons Slant range corrected imagery
Not unlike several other options and settings that can be
manipulated during data capture, the choice to remove
the water column by slant range correction has its pros
and cons. One of the notable benefits of removing the
water column is the improvement in spatial positioning
and dimensionality of features located in the near-field
portion of the image. In the slant range corrected image
to the right we find a well-defined outcrop of hard
bottom substrate (perhaps clay) that crosses the
Center line of
centerline in this image. Slant range correction has survey path,
cleanly brought the two halves of the image together,
making it easier to digitize the apparent boundaries of this i.e. directly
substrate patch. Note that the water depth at the point of
image capture was nearly 19 feet. At this depth, and with under the
a range setting of 110 feet, a total of 34% of the upper
portion of this image would have been occupied by water
transducer
column if slant range correction had not been applied.
This amount of water column would have compressed the
near-field portion of the image and made the work of
accurately digitizing the boundaries of this patch a bit
more difficult.

By removing the water column, however, we have lost a

very useful, and easily referenced source of information
on depth, and depth changes, as we undertake the
process of sonar image interpretation. It is true that we
can reference depth data from other sources (e.g.,
trackpoint data- to be described later), yet the water
column provides a continuous record of this information
displayed front and center in the image. Changes in
substrate composition often accompany changes in depth, 14
making this information quite useful during mapping.
Image artifacts Strange distortion forms
Unfortunately, the use of on-the-fly slant range
correction can lead to some very unusual image
artifacts. In this example a strange, saucer-shaped disk
has appeared in the middle of the image. These shapes
sometimes appear when imaging undulating bedforms,
such as ripple and dune sequences on sand bed rivers,
although the bottom in this image appears relatively flat.
It is not practical to attempt removal of these artifacts
from raw imagery.

Possible causes?

15
Image distortion Strange distortion forms
One of the features associated with a lot of image
distortion when applying on-the-fly slant range
correction is large woody debris. The distorted tree Deep, outside bend of large Coastal
shapes and shadows in the image on the right are better
suited to a Tim Burton movie than a sonar habitat map. Plain river with large submerged trees

16
Image distortions Strange distortion forms
The distortion present in this image is downright
horrible. If you had to spend more than a few minutes
trying to map habitat from imagery like this you might Deep, outside bend of large Coastal
end up puking on your shoes!
Plain river with large submerged trees
What is going on here, and what might we learn from
these examples regarding the judicious use of slant
range correction with the Humminbird system?

17
Making sense of distortions What is going on here!?
In a simplified channel setting (i.e., flat, open bed), the
sonar signal first contacts the bottom directly below the
boat (solid black arrow). As a result, the first signal In a simplified setting, first contact
returns to the transducer are also coming from points
directly below the boat. This is not the case if you have occurs directly below boat.
large woody debris suspended above the bottom, near
the boat path. As illustrated on the right, first sonar
returns are instead coming from the suspended, lateral Here, first sonar returns are instead from
branches of the sunken timber (dashed black arrow),
rather than from a point directly below the boat.
suspended lateral objects, causing distortions
during slant range processing.
When slant range correction is being applied, the
computer assumes the first returns to the transducer are
coming from features that should be spatially relocated
to a position directly below the boat. Thus, the pixels
representing the returns off the branches of the
submerged tree on the right are repositioned directly
below the boat, and the pixels representing the open,
sandy area below the boat are repositioned somewhere
out in space to the right or left of the boat path. The
result is an image with varying degrees of distortion; an
image that does not make sense visually or spatially.
(By the way, these issues also plague automated
processing routines used by high-end sonar processing
software). By preserving the water column in sonar
imagery, this source of distortion is eliminated.

When planning a survey it is advisable to perform some

field tests of the Contour mode setting to determine
whether slant range correction is suitable to your survey
situation. In most cases, this author (Adam) prefers not 18
to use this feature.
Classic substrates Rocky shoals and sand bar
Now that we’ve covered some of the bases on water
column and slant range correction, let’s look at some of
the typical substrates we’ve encountered in surveys of Deep, outside bend of large Coastal
streams of the Southeast Coastal Plain. The image from
the right was captured in the lower Flint River. This river Plain river with large submerged trees
is characterized by extensive rocky shoals (primarily
cobble to boulder sized material), sand flats, and
reaches of flat, limestone bedrock exposures. On the
right, we can see that the survey boat approached a
rocky shoal, and charted a course over the shoal. The
transducer came close, but did not strike, a few of the
large, shallow boulders present in the shoal. As the boat
approached this shoal, the shallow water and rock pile
blocked and reflected the signal back to the boat,
casting sonar shadows. These shadowed areas
Boulder
represent missing data that can be quantified during
mapping. Note the difference between the large, coarse Cobble
material predominant along the left side of the image,
and the finer (yet still textured) rocky material on the
right hand side of the image. This finer textured
material is cobble-sized rock (according to the modified
Wentworth particle size scheme). In the lower left hand
corner of the image appears a smooth sand bar. The
boundary between the sand and rocky shoal is strikingly
obvious.

19
Limestone bedrock Limestone rock formations revealed
Here’s a mosaic of several raw sonar images that reveals
an extensive outcropping of hard limestone rock. In
several places like this on the Flint River, large pinnacles
of limestone emerge from the bottom. During this An
survey I charted a course directly over one of these
pinnacles, though I never knew it existed lurking below
extensive,
in murky water. With some familiarity and experience submerged
with sonar signatures from substrates such as limestone,
it becomes possible to discriminate this type of material limestone
from other rock types.
outcrop
An interesting side note regarding deep holes containing
massive limestone blocks- This reach was known to
consistently produce large flathead catfish during annual
surveys. Once sonar survey work began in this river, we
quickly associated deep holes that contained limestone
boulder structure with abundant, large flathead catfish.

20
Sand formations Sand Dunes
In rivers sandy substrate is often sculpted into beautiful
dune and ripple patterns. Like winds that carry sand
across the desert, currents carry sand downstream. This
process creates characteristic bedforms that reveal the
Sand dunes along the
nature of the substrate. bottom of the Flint River

21
Sand formations Sand ripples
In the adjacent image the sandy bedform might best be
described as rippled. Note, however, that within the
series of ripples on the left side of the image are 2 very
distinct leading edges of what might be called sand
Sand ripples along the
waves. Hydrogeomophological processes and bottom of the Flint River
mechanisms are responsible for sand dune, ripple, and
wave formation. Particle size, stream velocity, and shear
stress at the sediment surface are among variables
involved. For purposes of habitat mapping, these
bedform features are not only common, but also
extremely valuable for discrimination of sand in lotic
systems.

22
Large Woody Debris Large Woody Debris
Whether you call it large woody debris, large woody
material, coarse woody debris, or something
else…submerged wood can be imaged and quantified Large woody debris along the outside,
using side scan sonar. In sand bed, Coastal Plain rivers,
wood is found in predictable locations. The adjacent right bend of the river
image comes from the Altamaha River in Southeast
Georgia. Here, submerged wood has accumulated along
the right bank, which happens to be the outer bend, and
erosional side of the channel. To the left we have
smooth, shallow sand. You might imagine it possible,
however painstaking, to attempt to count the individual
number of pieces of wood in this image.

23
Large Woody Debris Wood Accumulations
In some systems, the idea of counting pieces of LWD or
putting points on the map for each piece may not be
feasible due to the large quantity and dense An accumulation of snags along
accumulations of wood as seen here along another outer
bend of the Altamaha River. Perhaps a more suitable outside bend of river
alternative when mapping wood in this case would be to
draw a polygon around the accumulation to derive areal
estimates of woody cover.

If we were interested in identifying suitable mussel

sampling locations within this reach, we might avoid
sending divers down into this snag fest.

24
In search of deadheads Deadhead Logs
Virgin, pre-cut submerged timbers can be found in most
navigable Coastal Plain river systems. These logs have
rested on the river bottom for 100 years or more in most Several deadhead logs nested at base
cases. Deadhead logs are sometimes easily
distinguished from other pieces of LWD by their pole- of limestone wall
straight, cylindrical forms. Here we see a few potential
deadhead logs positioned at the base of a deep, oddly
structured bank of the Flint River. These logs are hiding
along the base of a towering limestone rock wall. In
many places along the Flint River these rock walls are
visible above the water’s surface during low water
conditions. Diving down along these limestone walls is
like exploring the dark side of the moon. Now that’s
truly a gig for a Georgia deadheader!

*Note- removing deadhead logs from state navigable

waters in Georgia is unlawful.

25
In search of deadheads Large (Deadhead) Log
Sometimes it’s hard to miss a deadhead log when it is
perched along the bank of a drought-stricken creek. A solo deadhead log along the bank of
Ichawaynochaway Creek

26
In search of deadheads Large (Deadhead) Log
Here’s a close-up photo of this same log, with intern
Josh Hubbell posing to provide reference on the massive
size of this log. The canoe is 16 feet long.

27
In search of deadheads Same log in sonar image
Here we show a raw sonar image of the creek where the
deadhead log in the previous slide was resting. Of
course we created this image when stream flow was
much higher and the log was completely submersed. -Double image is an artifact
Interestingly, there appear to be two identical log-like
objects adjacent to one another. This double image is
actually an artifact. In other words, there is only one
log present in this location, and the mirrored object does
not exist. This example was specifically chosen to
demonstrate that artifacts can and do occur in sonar
images, just as they do in digital photographs. This type
of artifact is sometimes associated with logs, and
although the cause is uncertain, we suspect it involves a
deflection of some of the sonar signal off of the log and
reflection from the water surface above.

28
Caches of deadhead logs Log Caches
Deadhead logs do not always appear as isolated objects.
In these examples we find caches (piles) of sunken logs
resting on creek bottoms. Caches are common in areas
formerly used for launching and landing log rafts.

29
The remains of a raft Log Raft on Flint River
During a reconnaissance survey for deadhead logs in the
Flint River we encountered the curious feature circled on
the right. Although water was too deep and swift at the
time to confirm its identity, we returned later that year
to have a look.

30
The remains of a raft Log Raft on Flint River
What we found during this groundtruthing expedition
was a regularly arranged group of logs now exposed
along the right bank of the river. Rather than remain
preserved underwater, these logs were in various states
of decay due to repeated exposure and drying during
low flow periods. We suspect this log pile may be the
remains of a large log raft that never found its final
destination.

31
Features in context Interpreting Features
In the following series of slides we will work on
interpreting complex features in context. During our
early work with sonar mapping we seized the
opportunity to visit local creeks during periods of
extreme drought and obtain photos, like the one shown
here, of study areas. The time spent examining these
areas during low, clear water, and the opportunity to
study the relationship between field photographs and
sonar images of the same areas proved invaluable for
honing our skills of interpretation. Let’s spend some
time doing the same for a few of these images.

In the scene to the right our intern Josh is standing atop

a large boulder in the middle of the stream channel,
diligently studying the area. To his right an old cypress
tree snag stands rooted in the channel. In front of Josh
we see another large boulder. Almost touching this
boulder is a deadhead log that is oriented parallel to the
channel. The topside of this log is just above the water
surface. Let’s see if we can pick out each of these
objects in the corresponding sonar image.

3232
Features in this scene Interpreting Features
Let’s point out the features discussed in the previous
slide. The two boulders are located on the left side of
2 boulders log
the image, just left of the path the boat took during the
survey. The rock Josh was standing on was tipped up,
and here we see the sonar shadow being cast by this
object. The second boulder in front of Josh is identified
here as the roundish object located near the edge of the
water column. Note that this sonar image boulder
appears to be smaller than the one Josh was standing
on, yet they appeared to be about the same size in the
digital photograph. The reduced size of this second
boulder that is close to the center of the image is a good
example of the effect of object size compression in this
region of the image. The boulder artificially appears
smaller than it is in reality.

The cypress tree stump cannot be seen, except for the Cypress
vertical leading edge that reflected signal back to the
transducer. Instead, we clearly see the tapered sonar snag
shadow that was cast by the buttress of this tree. The
shadow extends all the way to the bank indicating this
casting
object indeed protruded all the way through the water tapered
column. The sunken deadhead log is quite difficult to
identify in this image, but if we look closely behind the sonar
second boulder we find an object that represents this
log. The log was almost directly underneath the boat
shadow
during the survey, as we can almost see part of the
object mirrored on the right side of the image.

33
Features in this scene Boulder field with 3 logs
Here’s another photograph of a drought-stricken creek in
South Georgia. This shoal contains many large
boulders, in addition to three noticeable deadhead logs
exposed above the water surface. Let’s have a look at
the sonar image captured for this area during higher
water.

34
Features in this scene Boulder field with 3 logs
The field of boulders is quite evident extending across
the lower half of this sonar image. Many of these large
rocks are casting long shadows, especially because the
water was fairly shallow over the shoal during the sonar
survey. The three deadhead logs have been identified
by red boxes in this image. The survey boat passed
over the log in the middle of the image; the log was
oriented at an angle to the boat path and as a result we
see portions of the log on either side of the image.

35
Features in this scene Bridge pylons and Concrete ramp
In this scene, our trusty survey vessel sits next to a
concrete boat ramp that extends underwater into the
creek. Just upstream is a bridge span with submerged
narrow abutments. Let’s see if we can identify these
features in the corresponding sonar imagery.

36
Features in this scene
In the upper half of the mosaic we find four submerged
bridge pylons. We can only visualize the edge of these
structures. Each has a narrow sonar shadow behind
that extends all the way to the bank; a tell-tale
indication that each pylon fully protrudes through the
water column.

In the lower left hand corner of the mosaic we find the

submerged concrete boat ramp. In our live program we
are able to toggle the yellow lines defining this feature
on and off to help illustrate the difference in overall
texture and tone of this smooth, hardened area relative
to the creek bottom substrate above and below the
ramp. These differences are subtle, but with experience
these they become more pronounced and recognizable.

37
Features in this scene Mid-channel Boulder Shoal
In this scene, intern Wes Tracy and I visited an exposed
boulder-strewn shoal along Ichawaynochaway Creek for
some groundtruth work. The aspect of this photograph
reveals how steep the banks of this entrenched creek
were in some places. Would you believe we successfully
navigated down the center of this creek during a survey?
Not at this flow, of course! The yellow line provides an
indication of the downstream path taken by our vessel
during the sonar survey.

Boat path
in sonar image

38
Features in this scene Mid-channel Boulder Shoal
Here we show a portion of this shoal as it was imaged
during the survey. Note the large boulders that were
located along the upstream, leading edge of the shoal.
The red, hand-drawn line illustrates the apparent
boundaries of this shoal in the sonar image. A yellow X
has been placed in the location of the large boulder I
believe Wes was standing on in the previous slide.

39
A complex scene Pilings, Cypress, and Humps
Here is a portion of lower Spring Creek in Southwest
Georgia. This area was long ago used as a landing for
deadhead logs. A mill and lumber company were
located on the adjacent property. Along the far side of
the creek we find a series of upright wooden pilings.
Note the gap between two adjacent pilings identified
with a red line. In the middle of the channel are two
small cypress trees identified with red dots. In front of
the marked cypress tree closest to the camera is a log
oriented perpendicular to the channel (red x). Along
the right side of the image (left bank) are two exposed
bars or humps. Although it is difficult to tell from this
photograph, these bars are actually outcrops of solid
limestone bedrock. The surrounding substrate was mud
and silt.

40
Features in this scene
Here we display this reach of the creek in raw sonar
image mosaic form. On the left is the raw mosaic, and
on the right we have added reference markers for the
features identified in the previous slide. Notice how
each upright piling casts a long narrow shadow that
reaches the right bank. The gap between the two
pilings is clearly evident. It is somewhat difficult to
identify the two upright cypress trees because each was
rooted on a raised portion of the bed. The approximate
positions of these trees have been identified with red
circles. And lastly, the two limestone outcrops have
been identified with the shaded polygons on the right
hand mosaic. The difference in texture between these
outcrops and the surrounding muddy substrate are quite
subtle. Interpreting these substrate distinctions is
challenging in this scene.

41
A popular question Can fish be seen?
The question of whether fish can be imaged by side
scan sonar is relevant and intriguing. Until now we have
focused on inert physical features and objects, yet all of
the streams we have scanned have resident fish Gulf sturgeon: a large, reflective target
populations. What does it take to image a fish?

Several important factors having to do with operation of

the sonar equipment, such as the range setting used,
will likely play a role in imaging fish. At higher range
settings, smaller objects (fish) are less likely to appear
as distinguishable targets in an image. As far as
subjects go, fish that are larger and more reflective
(hard-bodied) like the Gulf sturgeon to the right,
probably stand a better chance of being imaged by side
scan sonar. Fish with soft bodies, like catfish, may
absorb too much of the sonar signal to be detected,
although much work remains to explore differences
among fishes.

At least one study demonstrating the use of the

Humminbird system for detecting manatees (see
reference below) appears in the primary literature, and
others are underway examining the effectiveness of
detecting and counting sturgeon with side scan sonar.

Gonzalez-Socoloske, D., L. D. Olivera-Gomez, and R. E.

Ford. 2009. Detection of free-ranging West Indian
manatees Trichechus manatus using side-scan sonar.
Endangered Species Research, 8: 249-257.
42
Sturgeon in resting area Gulf Sturgeon
Here is a single image captured in a known resting area
for Gulf sturgeon that typically holds dozens if not
hundreds of sturgeon. Many of the suspended targets
seen here are likely sturgeon. Below we zoom in on the
target just right of center to examine its shadow profile.
Note the sloping forehead, heterocercal tail, pectoral and
pelvic fins. Interestingly, it is the shadow of sturgeon
rather than the target itself that often appears so well
defined in the sonar image.

43
What’s lurking below Alligator
Here is a sonar image captured from a reservoir cove
that harbored a small alligator. This gator was at the
water surface until I approached it with the sonar boat.
The animal sank down to the bottom and rested there
as I passed overhead. It even appears from the
shadow that his head was turned up toward the
surface, perhaps wondering where I was going.
Alligators, like sturgeon, are large, reflective targets.

44
Fish on the move Groups of fish
One of the realities of imaging fish is that they can
move. If targets are in motion as the sonar beam
passes over them, their representative sonar returns
can be distorted, either stretched or shrunken or
perhaps not visible at all. In the image on the right I
believe we see a school of fairly large fish off to one
side of the boat (left). Although the targets are close
to the bottom, we can see that the shadows are
slightly offset from the object, a clear indication that
these objects are not logs. To me, these targets
appear stretched, or longer than normal. I suspect this
effect occurred because the school of fish were moving
in the same direction as the boat, however not moving
as fast as the survey vessel. Perhaps these fish were
common or grass carp. Their identity remains
unknown given the water was very muddy and I was
unprepared to do any fishing.

45
More fish schools Groups of fish
This is an interesting image that shows another school
of fish, located both left and right of the boat. Again,
the offset shadows indicate these are suspended
objects. The left side school appears to have more
fish, and the top end members appear to be on the
move as the boat passed over them. The odd circular
pattern in this image is something like the sonar
equivalent of a crop circle. On this day, several
tournament level bass anglers had their shiny boats
out on the water for spring test-runs and were doing
donuts, leaving nice circular prop scars in the muddy
flat.

46
A cratered lakebed Fish Beds
Side scan sonar may not reveal everything that exists
underwater, but there seems to be a endless number
of potential features and applications that could be Applications-
explored with this technology. The adjacent image
provides a good example of a potential application for identifying
validation and development in lentic settings. The
crater-like depressions on the bottom of this reservoir
location, size, and
cove are centrarchid spawning beds. Not only do we timing of
see the beds in this example, but some have male fish
guardians (identified with red circles)! spawning activity
How might this information be used? We once began
a pilot study to assess our ability to use side scan
sonar to detect and quantify fish beds and monitor
trends in the production of fish beds over time in
several reservoir coves using time lapse sonar surveys.
In a time lapse approach repeat sonar surveys of the
same transects are conducted to examine changes in a
parameter of interest. This work showed great
promise but was never completed. So many potential
applications remain for development; we hope to
encourage our readers to join the effort!

47
A cratered lakebed Colonial bed formations
Here are a few other images showing bed aggregations
along reservoir shorelines. Unlike the scattered beds in
the previous slide these are very tightly spaced,
suggesting colonial spawning aggregations of bream.

29 m2

76 m2

25 m2
48
What about plants? Submerged Aquatic Vegetation
In every workshop we’ve ever hosted, someone has
asked about aquatic vegetation. Does sonar reveal
submerged aquatic vegetation and what applications
Plant growth through water column to surface
exist for the study of SAV with side scan sonar? Our
experience with SAV is limited, having worked primarily
will reflect (block) sound
in streams where vegetation does not exist. However,
when taking sonar into reservoirs and lakes we have
captured images like this one. Here, we navigated our
sonar boat over a vast bed of hydrilla, a troublesome
invasive plant that forms thick colonies in shallow
waters. The stems of this plant can grow to reach the
water surface, and in doing so these plants block and
reflect much of the sonar signal as shown here. This
hydrilla bed appears to have a sonar signature unlike
any of the “substrates” previously examined in this
chapter. If other types of SAV also provide unique or
distinguishable sonar signatures in imagery, then the
idea of mapping and monitoring SAV with side scan
sonar has great promise, and should be investigated.

49
What about plants? Submerged Aquatic Vegetation
Clearly, one issue with imaging hydrilla or other surface
level plants is that signal blocking can occur. Note
that as the boat passed beyond the hydrilla bed in the
Plant growth through water column to surface
adjacent image, the signal was no longer blocked and
the open lake bottom was visible. We can easily
will reflect (block) sound
delineate the boundary between hydrilla and the open
lake bed in this image. Another interesting point about
this transition is the change in depth (look at the width
of the water column) at the point of transition between
hydrilla and open lake bottom. The edge of this
hydrilla bed appears to be tracking the bathymetric
contour of the lake, where growth is limited beyond a
certain depth threshold by light availability.

50
Seagrass signatures Seagrass Beds
To provide a few other examples of submerged aquatic
plants we did some pilot sonar survey work on St.
Andrews Bay (Panama City, FL) to take a look at
seagrass signatures. Although this bay is often crystal
clear, the water this year was very tea stained from the
heavy volume of summer rain, and visually locating
seagrass beds was not possible. On the right is an
image that shows a clear boundary between a
seagrass bed known to exist in the survey area and the
sand/mud bottom.

51
Seagrass prop scarring Seagrass Scarring
Unfortunately, this seagrass bed was located near a
shallow, high traffic area of the bay. Note the
transition from deeper, sand/mud bottom to the
shallower flat inhabited by seagrasses. When the tide
is low at this location, boaters apparently plow right
through the seagrass bed as evidenced by the many
crossing prop scars left behind.

This concludes Session I-Part B on sonar image

interpretation. A groundtruthed image library can be
an invaluable training and reference tool for a sonar
mapping workgroup. We encourage you to consider
developing a library specific to aquatic systems of your
region.

52