Dynamic Range Issues with the Knudsen 320B/P
(8 bit logging limitations)

April 2000. © OMG/UNB

 John E. Hughes Clarke
Ocean Mapping Group, Dept. Geodesy and Geomatics Engineering
University of New Brunswick, CANADA
506-453-4568 (ph.), 506-453-4943 (fax.), jhc@omg.unb.ca

  1. Water Column Structure at 200 kHz.
  2. Subbottom Structure at 28 kHz.
  3. Sidescan imagery at 200 kHz.
The following three examples show the amount of information that is stored within the 8bit data range preserved. In each case the top image shows a linear representation of that data, whereas the bottom image shows a logarithmic representation of the same data.  The linear representation is that which one would view as played to a greyscale recorder, or equivalently viewed through the current Knudsen display software (contrast enhancement is essentially a linear function and does not add to the detail).

Note that, although the data is better viewed in the logarithmic mode, the retained information content is not actually increased (only its visibility). Increasing the information content would only be possible if the original 16? bit data were logarithmically compressed into the 8bit range (or the data retained at the original higher dynamic range).

Water Column Structure at 200 kHz

The example used here shows the abrupt nature and detailed structure of the halocline developed in Kennebecasis Bay. The upper layer is hot fresh water from the St. John River. The lower layer is a plug of cold salt water from the Bay of Fundy, trapped upstream of the Reversing Falls.  The base of the layer occurs at about 10m depth.  The oblique zigzag lines seen in the lower image are just interference with another (unspecified) source. Both images are derived using a 200kHz source with the gain/power level set to total saturation on the seafloor echo.
 

default linear representation of water column echo structure
(click on image to view full resolution)
modified logarithimic representation of water column echo structure
(click on image to view full resolution)


Subbottom Structure at 28 kHz

The second example used here shows the subbottom structure developed in the approaches to the Royal Kennebecasis Yacht Club.  Little is actually known about the details of the subbottom in the area. What one can see is a partially stratified conformal drape over an abruptly faulted and folded layered strata. The contact betwen the two layer is an erosional unconformity.  We are looking at between 10-15m subbottom (depending on the sound speed in the sediment). Both images are derived using a 28kHz source with normal gain/power level settings (I'd like to tell you what they were but I currently can't read those fields in the data telegram!).
 
 
 

default linear representation of 28 kHz subbottom structure
(click on image to view full resolution)
modified logarithmic representation of 28 kHz subbottom structure
(click on image to view full resolution)

For the case of the default prime role of Knudsen echosounders (robust first-arrival bottom detection), there is no need to go to this effort. But because these systems provide a digital file with full sonar parameters retained (hopefully), it represents a powerful scientific tool.

Sidescan imagery at 200 kHz

And here is a sample of the 200kHz sidescan data that you provided showing the advantage for logarithmic (or other equivalent) packing, in bringing out the textural information:
 
 
default linear  representation 
of 200 kHz sidescan image
(click on image to view full resolution)
modified logarithmic representation
of 200 kHz sidescan image
(click on image to view full resolution)

In this case, first arrival detection (for fish altitude) is of secondary importance. It is the spatial variations in the backscatter strength, after the first arrival which contain the useful information. More particularily, many of the apparent cast shadows in the left image turn out, not to be shadows.

April 2000, J.E. Hughes Clarke