03/11/2020 — admin
- R/V Atlantis EM122 multibeam data includes intermittent depth artifacts across the swath that do not appear to be correlated with roll, pitch, sound speed, latency, or other sensor integration errors. These were first brought to the attention of the Multibeam Advisory Committee (MAC) in late 2016, beginning a process of investigation to better define the problem and rule out a variety of potential causes.
- Symptoms of the artifact (commonly described as a ‘wobble’) include intermittent depth biases that increases in amplitude with angle from nadir, alternating in shoal/deep direction with a period of 1-10 minutes. These are distinct from other ‘wobbles’ that are higher in frequency (correlated with roll and pitch periods of a few seconds) that may stem from incorrect/outdated sound speed data. Likewise, these do not follow directly from the increased scatter of soundings observed for FM transmit sectors. The ‘wobble’ under investigation is described briefly in this document and in more detail in other documents provided by the MAC.
- Data from multiple calibrations, transits, surveys, and other troubleshooting steps strongly suggest that the ‘wobble’ is driven primarily by the EM122 yaw-stabilization feature when subject to drifting errors in the heading data feed provided by the existing motion sensor. The installed primary motion sensor provides roll, pitch, and heading from inertial measurements only; this is notable in that all other multibeam mapping systems in the UNOLS fleet employ GNSS-augmented heading data. The importance of GNSS-augmented heading becomes particularly clear over long, straight transits where inertial-only heading data may drift (even as roll and pitch appear reliable). The problem is compounded in deep water because errors of even a small fraction of a degree in any component of the attitude feed can cause significant depth artifacts when propagated out over acoustic ray lengths of many kilometers (i.e., scaling with beam angle from nadir).
- In early 2019, opportunities developed to take several important steps in testing this heading hypothesis:
- A.Technicians installed a secondary GNSS-augmented motion sensor for simultaneous testing against the primary inertial-only system, with attitude time series from both units logging in the EM122 raw data
- B. A new vessel survey was completed in dry dock, including all transducer arrays, GNSS and C-Nav antennas, and both motion sensors to try to eliminate potential sensor integration errors stemming from inaccuracies or misinterpretations of the original vessel survey
- C. A 48-hour window was granted to perform calibrations (‘patch tests’) and collect swath accuracy assessment data for the EM122 using each motion sensor, following established MAC procedures applied throughout the UNOLS fleet
- The limited window available for testing faced complications from:
- A. Ship scheduling (MAC personnel could not be on board due to Atlantis rescheduling and other at-sea commitments, so the MAC provided planning and real-time calibration support from shore; the science crew on board made this possible through constant communication, rapid file sharing, and EM122 updates with calibration results)
- B. Sea state (large swell and associated bubble sweep degraded the calibration and accuracy tests, which were re-planned underway to accommodate swell direction and try to minimize this impact)
- Despite these challenges, the installation, survey, and test data clearly establish:
- A. Sensor offset and integration errors may reasonably be eliminated as primary sources of the artifacts
- B. The inertial-only heading data show significant, time-varying differences from the simultaneous GNSS- augmented (benchmark) heading; these drifting biases present fundamental problems for swath stabilization (TX and RX beam steering based on real-time attitude), leading to depth errors that cannot be corrected in post-processing
- C. Though degraded by elevated sea state throughout testing, swath accuracy generally improved with GNSS-augmented heading data
- The process of clarification and elimination since late 2016 (culminating in AT42-10), and the direct comparison of concurrent inertial-only and GNSS-augmented heading time series in this report, demonstrate the need for GNSS-augmented heading data to address stabilization-related artifacts and improve data quality from the deepwater multibeam mapping system aboard R/V Atlantis.
- As part of the pending mid-life refit period for R/V Atlantis, the MAC recommends installation of a GNSS- augmented position/motion system (e.g., one of a handful of models installed throughout the UNOLS fleet) as the primary position and attitude feed to the EM122
- The EM122 was installed in 2011 and will be approaching the expected hardware service life of ~10-12 years shortly after the mid-life refit; the TX and RX impedance testing performed by Kongsberg on 6 March 2019 should be followed up with routine monitoring to detect early signs of degradation and provide ample time for replacement planning and budgeting