Table of Contents
Now that we have seen how one can quickly survey a data set, both graphically and statistically, we now turn to the task of post processing sonar data with MB-System™.
MB-System™ supports many types of sonar, each of which as its own data file format. The details of processing data from each are slightly different, however, in general, the core set of steps is the same. We have attempted to capture this core set of steps in the ensuing chapter.
That said, we should quickly note that the devil is in the details. Sonar data formats, vary between sonar models - even from the same vendor. Occasionally the same model sonar differs from one ship installation to the next. In as much as possible, MB-System™ rides atop an input/output layer, MBIO, which handles these differences in the form of format identifiers as we have described. However it is a constant struggle to cater to the evolving needs of the scientific and sonar development communities, and in some instances special processing techniques are required. We have added an appendix entitled "Shipboard Multibeam Sonar Installations" in an attempt to document the varied idiosyncrasies of particular sonar systems we know about and, when appropriate, provide hints regarding how to handle their data.
The processing strategy is to first survey the data and organize it in some meaningful way. Then, when they have not been provided for us by the survey ship, we calculate physical constants such as roll and pitch bias of the sonar system from sections of our own carefully taken data. Next we conduct automated and manual editing of the data to remove unwanted artifacts and erroneous segments. We edit the navigation data to ensure continuity and "reasonableness". We identify high quality sound speed profiles for regions in the data set. For sonars that produce sidescan data, we generate amplitude and grazing angle tables. Finally we apply all our physical constants, bathymetry and navigation editing changes, sound speed profiles and grazing angle tables to the data to create the final, processed, data set.
Rather that creating intermediate incremental files with each processing step, changes are tabulated in a processing parameter file. These changes are applied in the last step to create the final post processed data file. The next section is describes the parameter file in detail.
Interactive tools provided by MB-System™ utilize the editing already recorded in the parameter file to display the current state of the edited data. In this way, while the final processed file is not produced until the final step, one always works with the most recent changes applied. Because the changes are not irrevocably applied to the data files, one can conduct these steps in any order or return to any intermediate step and make changes.
A summary of the processing steps might look like the following:
Determine the make and model of the sonar that recorded the data as well as the format of the data. The combination of these two will map to an MB-System™ format type, explicit delineation of which is required in many of the subsequent steps.
Survey the data to get an idea of its coverage and quality. Organize the data files into directories and create data lists to aid in processing. Also make a qualitative assessment of the quality of the navigation data.
Determine the sonar system characteristics, including the roll and pitch bias. Typically these values are measured periodically and kept on record by the ship's science officer, however it is never a bad idea to verify their values through independent measurement. It is a common misconception that these values do not change over the life of a ship or sonar.
Identify bathymetry data that is erroneous. This is a several step process using both automated and interactive techniques to identify processing errors in the data and data plagued by excessive noise or other interference.
SSP information is assumed by the sonar when calculating bathymetry from its measurements. The SSP used by the sonar may come from historical data of the survey area, direct measurements via a CTD or an XBT, or may be pulled out of thin air by the science party or ship's crew. The next step is to identify the correct SSP 's for the data set. Rarely is a single SSP sufficient for a survey. The SSP should be reevaluated on periodic intervals, with changes in bodies of water, and when the quality of the calculated data indicates a change in the SSP .
Smooth incongruities from the navigation data.
For sidescan data, determine amplitude and grazing angle functions.
Recalculate bathymetry and sidescan data from the system constants, SSP(s), smoothed navigation, amplitude and grazing angle functions and flagged data errors to produce a final processed data file.
Remember that each of these steps are autonomous in that one may go back and recalculate, for example, the system's roll bias, creating a revised parameter file which may be then used to reprocess the original data without recalculating any of the other steps.
The following sections illustrate each of the steps above, providing examples and results along the way. While most of the examples utilize a single data set for continuity, we have included other data sets in the sound speed profile and roll bias discussions. We hope no loss of clarity results.
From Section 1.5, you will recall that these examples will primarily utilize datasets mapping the volcanic seamount Lo'ihi off the southern coast of the Big Island of Hawaii.
Now on to the processing!