In practice, and in the example that follows, many of the lines of acquired data may be used for two types of calibration. However, the best resolution for the Pitch correction is obtained in the deepest water, whereas the best resolution for the Roll and Yaw calibration will be obtained when the greatest swath width is covered.
There is a "Patch Window" selection, so it is possible to process different parts of the same line for different purposes. For example the "flat" part of two lines could be used for the Roll calibration whereas any steps, ridges or other features can be used for the Pitch calibration.
Follow the BathyPro Guide and create a project that contains all the lines you are going to use in the Patch Test, process them in the normal way, be sure to apply the offset data .GEO file (with zeros in the Roll, Pitch and Latency) and use the highest resolution (in Map and Projection Settings) that gives a continuous seabed. Apply any needed filters (max depth min depth etc). Be careful that you do not use the "outside angle" filter too aggressively, if the lines do not overlap the Patch test will fail initially with a message:
"Image Difference not constructed"
Any and all filtering set in the Bathymetry Correction Processing dialog will be applied during the construction of the surfaces used in the Patch test.
It is very important for the accuracy of the result that each step is taken in the correct order as shown below, as the accuracy of each step depends on the previous one.
Here is the BathyPro Main Window at the end of the initial processing of the lines, click Project Save As and save a project file before proceeding .
And this is the DTM of all lines as seen in TritonMap:
Back in BathyPro click on Patch Test: Roll, Pitch, Yaw, Latency...
Select Roll and click Run, a dialog opens - select the first of the Roll lines:
Select the file and click Open, a second dialog opens - select the second Roll line:
Click Open, BathyPro opens the Patch Test main window and creates images of the two lines:
Each line is displayed in the two image areas, both DTMs are present in both windows, the name of the DTM "on top" is shown at the top left side of each window..
Below each window are the Max and Min depths of the surface and a color key.
Display Style Overlap: Is the default - shows one line "over" the other
Display Style Difference: Shows a color coded numerical map of the difference values between each layer (Black = Zero)
Offset Value: Manually enter an offset.
Apply: Apply the entered value and recomputed the DTMs using that value.
Pan: Left click and drag to move the image.
Zoom: Zoom in + and out -
Reset: Re-center and "un-zoom"
Patch Area: Click and drag to define the area used during Automatic Processing.
Window: Click to have the Automatic test use the selected Area.
The normal procedure is to use the Automatic Calibration as described later, however a description of the manual method will help to better understand the automatic process.
In the upper image click the Difference button:
The upper image now represents the difference between the two layers, the scale shows a minimum difference of zero and a maximum of 1.3m. Because there is an angular offset due to a misalignment between the sonar transducer and the motion sensor the two surfaces only intersect at nadir. So the nadir part of the new image is black. (Recall that the two lines used for the Roll test are acquired in opposite directions.)
Enter a value of 1 deg in the Offset box and hit Apply
The image has not changed very much but the difference is now 0 > 3.05m - clearly we went the "wrong way", try a value of -1deg
Now we have a much darker image (values are closer to zero) however the maximum difference is still around 1.3m, try a value of -2 degs:
The image shows greater overall differences and the maximum difference is now 1.85m, so we know that the offset we are looking for is between 0 and -2 and probably about -1 deg.
For various reasons (tide, positioning errors, variations in sound velocity) the difference between the two DTMs will never be zero (black), the best we can hope for is an adjustment that gives the smallest differences, and this is what the Automatic Calibration tries to acheive.
Roll - is the most critical value in the patch test routine as an error in roll will result in an error in sounding depths. However the Roll offset is also the easiest to detect and correct, the Patch Test routine is very sensitive to roll and should provide unambiguous results as long the data are good.
Use the Zoom, Pan and Patch tools to Zoom in and select a flat area of the seabed, then click the Window button to have the Automatic Calibration use only the selected Patch area. Be careful to select an area that exists in both DTMs Click RUN
The automatic routine uses the default settings initially (From -3.00 to 3.00 in steps of 0.5 degrees), computes the difference at each increment of 0.5deg, and plots a graph of the differences. As you can see the minimum is a little greater than -1 degree.
We now have a better aiming point, set the the Range from -2 to 0 and the step to 0.05 (0.05 is the smallest available)
This seems like a good result, to check you can enter -0.8 in the offset value and hit apply:
When completed enter the computed value in the Offset to Save dialog and click Save and Exit;
Latency - almost all Multibeam sonar installations will incorporate GPS time synchronization and, as such, no latency is expected in the GPS position. However, it is necessary to complete one or two latency tests to prove that the latency, for all practical purposes, is zero.
Select the Latency radio button and hit RUN, select the names of the two files that you want to use for the latency calibration, the Patch test window will appear. Notice that the Roll offset from the previous test is now being used under Current Offset Values. We are expecting a very small or zero latency, set the range to 100mS and the increment to 5mS (5mS is the smallest value), use the Pan, Zoom, Select, and Window buttons to select a steep, well defined feature, and hit the RUN button again:
This is a typical result for a Latency test with data using GPS precise timing, 50mS represents around 0.1m displacement at normal survey speeds and is well within the expect accuracy of most GPS systems. It is debatable whether in fact this value should be used at all, it might be better to accept zero.
Negative Latency - because we are measuring a positional difference to determine the latency it is possible that the latency could be negative - this seems impossible but it could occur if the navigation is not repeatable, the value should still be small (<100mS). The BathyPro patch test does not handle negative latency at this time.
For the purposes of demonstration we will enter 50mS into the Offset to Save dialog: Hit Save and Exit to proceed to the Pitch test:
Pitch - A pitch error will result predominantly in along-track position error, which increases rapidly with water depth.
As for the Latency test we need to select a sharp feature from two lines run in opposite directions, check that all previous calibrations are being used (-0.8 for the Roll and 50mS for the latency)
Notice that the minimum is not well defined, as mentioned above better results for pitch calibration are obtained in deeper water, Save and Exit to continue to the Yaw test:
Yaw - the yaw correction is the most difficult to get consistent results because we need good data at the maximum horizontal distance from the sonar head to get good resolution. But as we move further away the data become more noisy.
Check that the results from the two previous tests are displayed correctly (Roll -0.8, Latency 50mS, Pitch 1.80).
Be careful when selecting the patch area, in this case only the very edges of the two swaths overlap, you need to make sure that you select a sharp feature within the overlap area:
This is the last test, enter the Offset to Save and hit Save and Exit.
BathyPro automatically updates the .GEO file with the values from the Patch test. Go back to BathyPro and hit Settings > Sensor Geometry Settings
The Vessel Geometry dialog should show the values obtained from the Patch Test:
.