Technical Note 4: Errors in older Kongsberg seabed imagery datagram

What are the changes?

Older versions of Kongsberg .all files have two distinct errors in the Seabed Imagery datagram fields in that some of the Kongsberg real-time TVG parameters are stored with incorrect scaling.

There are two distinct problems that were discovered by the manufacturer at different times, this technical note addresses both of these.

Newer versions of the datagram are correct in .all files now. The Kongsberg CPU software version can be used to identify which models of multibeam sonar are affected, this is recorded in the .all Installation Parameter datagram. Software versions prior to the dates in the table below are affected by these bugs.

1) Incorrectly scaled BS_n and BS_o (backscatter at nadir and backscatter at oblique angle, respectively):
EM710: PSV = 2.1.8, July 6, 2010
EM302: PSV = 1.4.8, Nov. 10, 2009
EM122: PSV = 1.1.5, Nov. 10, 2009

2) Incorrectly scaled R_n (range to normal incidence):
EM710: PSV = 2.0.6, Oct. 16, 2008

Version 7.3.6 and 7.4.0 were insensitive to these problems because the FMGT correction that removed the Kongsberg real-time TVG was incorrectly implemented (it was essentially still the original Geocoder code, which was found to be incorrect in the last few months). We have improved the correction of the Kongsberg TVG to be in accordance with Kongsberg documentation as of 7.4.1, however, this now introduces biases with the older data files that can result in ARA inconsistencies. Mosaic results appear to be affected less since the AVG removes the type of biases that are introduced by both of these problems.

Who does this impact?

Users with data collected with these sensors prior to the appropriate Kongsberg software update will have inconsistent processing results when using 7.4.1a as compared to previous versions. The mosaic results should be relatively close to each other since these variables only affect the response near nadir and will not affect the angular sector that is used to normalize the data in the AVG process. The ARA results will be inconsistent to a larger degree since the shape of the angular response curve used by the ARA process will be incorrect unless the BS_n and BS_o parameters are correctly scaled prior to removal from the signal levels as reported in the raw data file.

Previous ARA results from 7.4.1 (and earlier versions) with older files would have been incorrect for EM710 due to large effect of the incorrectly scaled R_n value affecting the accuracy of the Lambertian correction. ARA results for EM710 prior to PSV 2.0.6 should be recomputed. Previous results should be disregarded.

Users will notice a backscatter level shift in their mosaics between 7.4.1 and 7.4.1a. The R_n parameter that was incorrectly scaled was used to remove the real-time Lambertian correction that is applied by Kongsberg. Since this parameter was incorrectly recorded in the raw data file, the removal of the correction is incorrect, this would result in a dB shift that varies across the swath. This would have been been removed by the AVG but the data would have suffered from a bulk dB shift. Mosaics from 7.4.1 and 7.4.1a will not agree for EM710 systems running with PSV versions earlier than 2.0.6.