Hydroacoustics Technical Working Group: Task 2.3 Fully operationalize auto-adapting ADCPs Interim Operating Procedures for SonTek RiverSurveyor M9/S5 P Campbell, E. Jamieson, F Rainville, A Du Cap, 2014 1
Revision History Revision Date Source Description/Rational for Change D1.1 November 2012 Initial draft for review by NetOps, HATWG and OMC-H 1 March 2013 QMS formatting added. Approved by OMC-H 2 January 2014 Revision to exclude RiverRay with Honeywell Compass Require minimum of 3 SMB Tests for MB Correction with SMBA Revision to clarify best practice for the use of M9 for MBT and GPS reference 2
Table of Contents Acknowledgements... 3 Introduction... 4 Interim Procedure: Use Bottom Tracking as Reference... 4 Probing Transect... 5 Composite Tracks (RiverSurveyor Live)... 5 Stationary Moving-Bed Analysis (SMBA)... 7 Compass Calibration... 8 New Firmware and Software Upgrades... 9 Summary and Conclusions... 9 References... 9 Appendix A: Best Practice for Calibrating SonTek RiverSurveyor M9/S5 Compass... 10 List of Figures Figure 1: Flowchart for assessing and mitigating the effect of moving bed when using the M9/S5 ADCPs. Note that grey items indicate methods not yet approved. *SMBA and LC methods are now integrated within the recent versions of RiverSurveyor Live (v.2.50 and later).... 6 Figure 2: Illustration of non-uniform distribution of moving-bed velocity across a transect, which indicates the benefit of using more than one stationary moving-bed test. (Adapted from a presentation by D Mueller).... 8 List of Tables Table 1: Summary of most recent firmware and software versions for RiverSurveyor Live.... 9 Acknowledgements Thanks to D Mueller, M Rehmel and N Stasulis of the USGS for providing guidance and best practice recommendations for minimizing compass-related errors. 3
Introduction The primary objective of these interim procedures is to reduce the probability of erroneous results resulting from compass errors while using SonTek YSI s RiverSurveyor M9 and S5. Compass performance and compass calibration problems have been identified in both the M9 and S5 during testing. These deficiencies are well described in the USGS memo Best Practice for Calibrating SonTek RiverSurveyor M9/S5 Compass (D. Mueller August, 2012) attached as Appendix A in this document. Given the unpredictable nature of these compass-related problems, Water Survey s Hydroacoustic Technical Working Group (HATWG) recommends the following interim solution to allow measurements to be taken with the M9/S5. Preferential order of referencing boat and water velocities are, 1) bottom tracking unless there is a measured moving bed; if there is moving bed, 2) use GPS. If there are suspected problems with either the GPS or compass, 3) use SMBA with at least 3 sampling points. Even if no moving bed is detected, it is best practice to operate with a GPS as a backup anyway. This lets you monitor GPS and compass performance at these sites so you will aware of potential problems when moving bed is encountered. Compass calibration procedures that are unique to the M9 and S5 are also described below, and also in more detail in the best practice document in Appendix A. Interim Procedure: Use Bottom Tracking as Reference The effect of compass errors on ADCP measurements will depend on the boat-velocity reference. When an external boat-velocity reference is used (such as a GPS) the effect of the compass is substantial. However, when bottom tracking is used, a compass error will cause a rotational error in the measured water velocity, but the magnitude of the velocity is unaffected. Therefore, the compass has no effect on measured discharge using bottom tracking as the boat-velocity reference. When moving bed conditions are present, typical procedure is to perform a loop test and apply a loop correction to account for moving-bed bias. However, the loop method is very sensitive to compass errors. Therefore, it is recommended to use GPS as a reference and monitor closely for GPS and compass problems both during and immediately after the measurements. Should there be any suspected problems with either GPS or compass data quality, the stationary moving bed method is an accepted method to correct for moving bed effects that does not depend on the ADCP compass. Some indicators of poor GPS reception may include multiple satellite changes, HDOP greater than 2 or an elevation change greater than 3.5m among other indicators. Compass problems may be indicated by directional bias (check for differences in flow direction in reciprocal transects), unexpected patterns in water vectors in the stick shiptrack view, noticeable differences between shiptracks referenced to 4
bottom track and GPS. For further details see pages 135,136 Measuring Discharge with Acoustic Doppler Current Profilers from a Moving Boat WSC (qsop-na028-01-2013, WSC library) The flowchart below (Figure 1) indicates the general procedure to follow for an ADCP discharge measurement for either the M9 or S5, for both non-moving-bed and moving-bed conditions. It should be noted that the grey boxes in Figure 1 are methods not yet approved for these instruments. In most cases, additional testing and comparison data sets for these methods are required for validation and approval for use. These non-approved methods are included in Figure 1 for information and to indicate that testing and approval of these methods are planned for the near future. Regional office are encouraged to submit comparison test sets to resolve remaining issues and reduce the operational restrictions on the M9/S5. Contact Paul Campbell or Elizabeth Jamieson for further details on comparison measurement conditions and deployment methods. Probing Transect A probing transect is recommended before performing the moving bed test (but after the compass calibration) in order to check that conditions are suitable for a valid discharge measurement. A probing transect involves simply traversing the measurement cross-section while the ADCP is pinging and viewing the data for obvious problems. Examples of what to look for during the probing transect include: non-uniform cross-sectional shape and velocity distribution, low backscatter (important for RiverRay), and lost bottom track (should not lose more than ~ 5% of bottom track data). Composite Tracks (RiverSurveyor Live) For the RiverSurveyor M9/S5, the default set-up for track reference is for Composite Tracks to be enabled. It is recommended that during the pre-measurement set-up, under the Smart Page set-up in RiverSurveyor Live, Composite Tracks be disabled. When disabled, the track is calculated only using the selected reference (i.e. bottom track) and not a combination of the best possible references. This eliminates the problem of when GPS is inadvertently used as a track reference when bottom track is suspect. 5
Figure 1: Flowchart for assessing and mitigating the effect of moving bed when using the M9/S5 ADCPs. Note that grey items indicate methods not yet approved. *SMBA and LC methods are integrated within the recent versions of RiverSurveyor Live. 6
Stationary Moving-Bed Analysis (SMBA) Stationary Moving-Bed Analysis (SMBA) computes the dot product of the bottom track vector onto the unit water vector and determines the moving bed velocity in the upstream direction relative to the water velocity vector. This program is applicable to all stationary moving bed tests but is required to accurately assess a moving bed test when there is no compass (for example a StreamPro) or when the compass is thought to be unreliable. SMBA was originally developed to correct for biased measurements made with a StreamPro, where the StreamPro was swimming back and forth at the end of the tether. SMBA provides a correction algorithm using one or more stationary moving bed tests. The corrected discharge is computed by distributing the moving-bed velocity correction based on the near-bed velocity in each ensemble of the transects comprising a measurement. Water velocities near the bank are assumed to be sufficiently low as to not cause a moving bed and therefore, no correction is applied to the left and right edge measurements. Additional information on SMBA may be found in the USGS s Techniques and Methods 3-A22 (Mueller and Wagner, 2009) and the USGS Office of Surface Water website, http://hydroacoustics.usgs.gov/movingboat/smba1.shtml While it is possible to use the SMBA method with a minimum of one stationary moving-bed test, new procedures require that a minimum of 3 stationary moving bed tests be used when a moving bed correction is required using the SMBA method. This is to account for non-uniform bed load transport and moving bed velocity that could exists across a cross-section. Figure 2 illustrates an example of how the moving-bed velocity may be poorly represented by a single stationary measurement. In this example, the largest moving-bed velocity is concentrated in the center of the channel. If only one stationary measurement, taken in the center of the channel (test #1), is used, it is possible that the corrected discharge will be overestimated, since in this case, the measured moving-bed velocity is greater than the mean moving-bed velocity for the cross-section. In this example, a better representation of the moving-bed bias would be captured using 3 stationary measurements, distributed across the cross-section. 7
Figure 2: Illustration of non-uniform distribution of moving-bed velocity across a transect, which indicates the benefit of using more than one stationary moving-bed test. (Adapted from a presentation by D Mueller). Compass Calibration It is recommended that the internal compass be calibrated prior to all ADCP measurements. A valid compass calibration is compulsory when GPS is used as the boat-velocity reference, or when the loop test and loop correction method are applied. Although these interim procedures are intended to reduce reliance on an accurate compass, consistent use of best practices for compass calibration will allow easier quality assessment of the measurement and provide a best possible dataset for assessing improvements in compass hardware and instrument firmware over time. Each ADCP requires a slightly different procedure for calibrating the compass. The points below summarize some of these important steps for the M9/S5. Smoothly pitch and roll ADCP +/- 10-15 degrees (ADCP should not be pitched and rolled outside expected range for measurement) Collect data for minimum of 1 min and max 2 min. Note that only the last 2 min of data are used. Complete at least one and preferably two 360-degree rotations the M-score is a poor predictor of successful compass calibration For complete instructions see "Best Practice for Calibrating SonTek RiverSurveyor M9/S5 Compass" by D. Mueller (USGS) dated August 30, 2012 (Appendix A) 8
New Firmware and Software Upgrades In order to best comply with these new recommendations and compass calibration procedures and ensure reliable ADCP operation for the M9 and S5, we recommend that the following (as of January 2014) software and firmware versions be used: Table 1: Summary of most recent firmware and software versions for RiverSurveyor Live. Instrument Firmware Version Software Version SonTek RiverSurveyor M9/S5 v.3.00 (August 2012) RiverSurveyor Live v.3.50 (July 2012) Summary and Conclusions These interim procedures provide guidance and best practice for the use of the new auto-adaptive ADCP, SonTek s RiverSurveyor M9/S5 for measuring discharge. The recommendations contained in this document should be followed until the compass and compass calibration issues are resolved and/or when sufficient testing and validation of these instruments is completed using comparison measurements for under-represented conditions such as, moving-bed conditions and with GPS as the boat-velocity reference. References Environment Canada, 2013, Measuring discharge with acoustic Doppler current profilers from a moving boat, version adapted for Water Survey of Canada, 62 p http://hydroacoustics.usgs.gov/movingboat/smba1.shtml (Information on SMBA and download for Mueller et al., 2007). 9
Appendix A: Best Practice for Calibrating SonTek RiverSurveyor M9/S5 Compass David S. Mueller, OSW August 30, 2012 The calibration procedures for the compass used in SonTek RiverSurveyor M9/S5 ADCPs are evolving as we learn more about the compass and its calibration characteristics and SonTek improves the user interface and feedback for the calibration. Some of what OSW recommended as the best practice in the past has changed, and this document represents the best practice we know as of August 30, 2012 and supersedes previously presented and taught procedures. Background The USGS has been testing the RiverSurveyor M9/S5 since 2008. During this time it has been observed that the compass calibration procedures frequently did not result in a well calibrated and accurate compass. SonTek replaced the compass in early versions of the RiverSurveyor M9/S5 with a new compass they developed in-house. Although this resulted in some improvements, inaccurate compass calibrations which resulted in inconsistent discharges measured using GGA or VTG as the navigation reference and loop tests that did not close accurately were still common. During the 2011 and 2012 there has been a concerted effort by both the USGS, OSW and SonTek to document and identify the calibration issues and develop a solution that will result in consistent accurate compass calibrations. Although a final solution has not been achieved, several findings have resulted in a modification of the best procedures for calibrating the compass. Important Characteristics The following are important characteristics of the SonTek compass and its calibration algorithms: The compass appears to be more sensitive to magnetic interference than compasses used in older ADCPs. The M-score used to determine whether a calibration was successful is highly sensitive to pitch and roll and has been shown during controlled tests to be an overall poor predictor of the accuracy of the calibration. The calibration algorithm only uses 2 minutes of data. Calibrations that collected more than 2 minutes of data only used the last 2 minutes of data, as the memory uses a first in, first out approach. Very short calibrations with high pitch and roll resulted in a low M-score and indicating that the compass was well calibrated, even though it was not. Concentrating the calibration in the range of pitch and roll that the ADCP will be used in results in a better calibration than pitching and rolling the instrument through large angles, as previously recommended. Errors in heading increase dramatically if the ADCP is pitched and rolled outside the range in which it was calibrated. Motion during calibrations needs to be smooth, replicating the natural accelerations typical of the deployment platform (tethered or manned boats). 10
Currently Recommended Procedures There is currently no method available for evaluating the accuracy of a compass calibration for the RiverSurveyor. The best that can be done is to follow good calibration procedures and then carefully observe the collected data for potential compass errors. Given the current situation, use of the stationary moving-bed test will be consistently more accurate than the loop test and is recommended as the preferred moving-bed test for RiverSurveyors until the compass calibration issues can be completely resolved. The following are the procedures that should be used for calibration and use of the compass in a RiverSurveyor: 1. Calibrate the compass in as magnetically clean environment as is possible for the site and deployment platform you are using. Avoid calibrating on or near a bridge or near steel posts or guard rails. On a manned boat move batteries, ferrous objects (tools), and other sources of potential magnetic interference as far from the ADCP as is practical. For tethered boat deployments, mount the PCM box with the connectors towards the ADCP and the batteries away from the ADCP. 2. Collect data for a minimum of 1-minute and a maximum of 2 minutes. RiverSurveyor Live 3.5 will not permit calibrations less than 1 minute nor greater than 2 minutes. Previous versions of the software although allowing calibrations greater than 2 minutes only used the last 2 minutes of calibration data. 3. Complete at least one and preferably two 360-degree rotations of the instrument. These rotations should be smooth and at a uniform speed to allow the compass to obtain a uniform distribution of calibration data. RiverSurveyor Live 3.5 provides graphical aids that should help the user ensure a uniform distribution of data. 4. While rotating the instrument, smoothly pitch and roll the instrument and deployment platform about +/- 10-15 degrees. The goal is to achieve a uniform distribution of the pitch and roll angles for the complete rotation. The range of motion should represent the maximum limits of pitch and roll that the instrument will experience during data collection. On a manned boat, this may be difficult. It is not recommended that the instrument be pitched and rolled independently of the boat, such as by swinging the mount or removing the mount and instrument from the boat. Although pitching and rolling the instrument and/or mount was previously recommended, this can lead to an incorrect calibration. If the boat is the source of magnetic interference, the compass is being artificially moved relative to the source of interference and this is not representative of the magnetic field the instrument will experience during data collection. 5. Once the calibration is complete, any data collected that depends on the compass should be carefully monitored for errors in compass heading. These errors often appear as a difference in the ship track direction based on orientation of the instrument. 6. If the discharge or velocity data being collected will require use of GGA or VTG as the reference, the accuracy of the compass is critical. In this situation, if the compass appears to be inaccurate on one or more headings, try recalibrating the compass or collecting the data such that the heading of the compass changes are minimized. 11