Also shown in the mbnavedit is a "Use Solution" button. Clicking this button tells mbnavedit to save the Dead Reck-oning solution when exiting such that it may be used as the primary navigation source by mbprocess.
4.8. Flag Erroneous Bathymetry Data
Now that we have done some preliminary work regarding SSP's, physical constants for the sonar system, and navig-ation corrections, we can begin automated and interactive editing of the bathymetry data itself. Editing of bathy-metry in MB-System™ consists simply of "flagging" data points that are erroneous such that they can be ignored. In this way, tools that will later grid the data set, can interpolate from the surrounding data without the interference of poor data points.
It may seem odd that MB-System™ does not allow the user to manually manipulate data points. One cannot look at a piece of sonar data with $mbs; and create one's own sea floor profile through an otherwise noisy data set, as can be done with some sonar data editing systems. Rather, one can only chose existing points to ignore, and allow standard interpolation algorithms to fill in the blanks. The distinction is a subtle, but important difference in philosophy re-garding sonar processing on the part of the authors of $mbs;.
Sonar editing, whether automated or manual, creates an ancillary data file, typically with an ".esf" suffix. This bin-ary file contains the list of the data points flagged in the editing process. The depths associated with these data points are given negative values (or zero) during the final processing such that GMT will disregard and interpolate around them.
$mbs; editing processes also check to see if an associated parameter file exits (with the ".par" suffix). If not, one is created. In either case, the parameter file is made to specify the file name of the edits ancillary file, so that it may be applied by mbprocess.
First let us consider automated flagging of Bathymetry.
4.8.1. Automated Flagging of Bathymetry
Lucky for us, many errors in multibeam sonar systems are predictable, such that we can use an automated tool to flag them as incorrect. Toward that end, MB-System™ provides the process mbclean.
mbclean is an extremely useful tool to automate a portion of the bathymetry editing process. Often mbclean can be used in automated scripting to modify sonar data in near real-time such that reasonable quality data can be produced right from the get-go. However, even when processing data from hull mounted, modern sonars, no automated al-gorithms can approach the discretionary ability of the human eye. Interactive editing with mbedit is recommended to augment the automated editing for publication quality data sets.
mbclean provides seven ways to automatically flag bad beams in the bathymetry data. In the order they are applied, they are:
1. Flag specified number of outer beams. (-X option)
2. Flag soundings outside specified acceptable depth range. (-B option)
3. Flag soundings outside acceptable depth range using fractions of local median depth. (-G option) 4. Flag soundings outside acceptable depth range using deviation from local median depth. (-A option) 5. Flag soundings associated with excessive slopes (-C option or default)
6. Zap "rails". (-Q option)
7. Flag all soundings in pings with too few good soundings. (-U option)
The first flags a specified number of outer most beams from each side of the sonar. These are commonly poor beams, as SNR is low and sound speed profile errors are magnified.
The second flags sounding outside a specified depth range. This can be helpful when, for example, the sea floor is relatively flat and the range of depths is well known.
Next mbclean optionally flags beams with depths that are outside bounds generated by some specified fraction of the local median depth. This is helpful, when the sea floor is not particularly flat, and depth range is not well known.
The local median depth is generated as the median of the current beam and those immediately before and after it.
Then mbclean flags beams that are outside some exact specified range of the local median depth. This option is usu-ally preferable to the previous one when the total depth is quite deep, resulting in larger than desirable bounds using the previous method.
The next method employed by mbclean is flagging based on excessive slopes. This is actually the most common way to identify faulty bathymetry. Four optional methods of flagging beams based on slope are provided. The first flags the beam that is furthest from the local median depth. The second flags both beams associated with the excess-ive slope. The third method zeroes the beam furthest from the local median depth rather than flagging it. The final method zeroes both beams associated with the excessive slope.
If specified, mbclean will next try to identify groups of along track beams that have abnormally short across track distances and shallow depths. These kinds of errors cause ridges or "rails" along the track of the ship.
Finally, mbclean will toss out entire pings whose quality is so poor that flagging by other methods has resulted in fewer than a specified number of good beams.
4.8.1.1. An MBclean Example
We can apply mbclean to a portion of our Loi'hi data set and see how it works. mbedit is such a nice tool for
visual-izing (and editing as we will see) sonar data, we can use it in this example to see the effects of mbclean. However, the details regarding mbedit are saved for the next section, so for now just bare with me and look at the pictures.
Here is a screen-shot of mbedit with the first data file from the SumnerLoihi data loaded. All flagging of data has been removed the scaling has been adjusted and the data has been scrolled such that erroneous data is easy to make out.