AURORA AUSTRALIS ADCP DATA STATUS Mark Rosenberg, Antarctic CRC, November 1999 1. INTRODUCTION A vessel mounted acoustic Doppler current profiler (ADCP) was installed in the hull of the RV Aurora Australis during dry-docking of the ship in mid 1994. The ADCP unit is a high power 150 kHz narrow band ADCP produced by RD Instruments. The four transducer heads are mounted in a concave Janus configuration, with the beams 30 degrees off vertical, and with the transducers aligned at 45o to fore and aft. The transducers are mounted in a seachest ~7 m below the water surface, behind a 81 mm thick low density polyethylene window, with the window flush to the ships hull. The inside of the seachest is lined with acoustic tiles (polyurethane with barytes and air microsphere fillers), and filled with hypersaline water. ADCP data are logged on a Sparc 5 Sun workstation, using software developed by CSIRO Division of Marine Research, Hobart (J. Dunn and H. Beggs). An array of sounders is mounted on the ship for use in hydroacoustic biology surveys (T. Pauly and I. Higginbottom, Australian Antarctic Division). When these sounders are in operation, firing of the ADCP is synchronised with the sounder trigger pulses, to avoid interference between the two systems. When this synchronisation is active, the ADCP ping rate is lowered by ~35%. When the ADCP system bottom tracking is active, the ping rate is decreased by ~50 %. For cruises prior to September 1997, gyrocompass data were used for ships heading in the processing of ADCP data, logged on the workstation through a synchro to digital converter at a one second sampling frequency. Prior to 13/07/1995, GPS positions and velocities were obtained from a Lowrance unit, with positions and velocities received alternately every second. The Lowrance unit was replaced by a Koden unit on 16/07/1995, with GPS positions and velocities both being received every second. An Ashtech 3DF GPS system was installed on the ship in mid 1997, with the data coming online in September 1997, replacing gyrocompass and Koden GPS data in the logging and processing of the ADCP data (note that gyro and Koden data are still logged as a backup to 3DF failure). The 3D GPS data are logged by the workstation every second (every half second for cruise au9701). A Fugro Starfix differential GPS system was installed in mid 1999, with differential corrections applied to the 3DF GPS data. Data processing is discussed in more detail in Dunn (a and b, unpublished reports). Data file formats are detailed in section 4. The ADCP data are summarised in the following tables and figures: Table 1- summary of ADCP logging parameters Table 2 - summary of data status for each cruise Table 3 - summary of data calibration information Figure 1 - cruise tracks Figure 2 - absolute currents at 90 m for on station data only (see section 4) Figure 3 - absolute currents at 90 m for all data In general, on station data are the most reliable. Data collected while the ship is underway should be treated with great caution, as discussed in section 3. 2. CRUISE NOTES In general, ADCP data are bad when the ship is underway breaking ice, due to obstruction from ice passing under the transducers. As extended periods of ice breaking occur on many cruises, large data gaps may result. For each cruise, periods of rapid ship acceleration are used to obtain the two calibration coefficients for the ADCP data: ? (the alignment angle between the ADCP transducers and the gyro or GPS heading) and 1+? (a scaling factor) (Dunn, 1995a and b). ? and 1+? can alternatively be found from bottom tracking data, however as little time was spent on all the cruises in water depths within bottom tracking range, the calibration coefficients obtained from the acceleration method have been applied to the data sets. Average values of ? and 1+? for each cruise are summarised in Table 3. Note that for cruises au9404, au9501, au9604 and au9601, a time series of ? (see section 3.2 and Figure 6) was applied in the data calibration. 3. DATA ERRORS Two serious problems exist for Aurora Australis ADCP data, as follows. Note that both problems are minimised when the ship travels at very low speed, and are negligible when the ship is stationary. 3.1 Vertical current shear For each cruise, uncorrected ADCP data (i.e. ship motion not removed) was divided into different speed classes, according to ship speed during the 20 or 30 min. ensemble average. For each speed class, the speed difference between each vertical bin and bin2 was averaged over the whole cruise. In most cases the results reveal an erroneous vertical current shear, with magnitude up to ~0.15 m/s over the current profile (Figure 4) (in general, speed classes with only a low sample set should be ignored in the figure). The shear most likely results from: (a) sampling through the bubble layer (caused by ship motion) beneath the ships hull, a problem noted by other workers (e.g. New, 1992) - unfortunately on the Aurora Australis it is not possible to extend the ADCP transducers below the bubble layer (as done on other vessels), as the transducers are sealed behind an acoustic window flush with the hull, to maintain the ice breaking integrity of the ship; and (b) water being dragged along by the ship - this effect is increased by the wide and flat-bottomed hull shape of the Aurora Australis. When the ship is underway, a positive vertical current shear (i.e. current increasing with depth) is typically seen over the first 10 to 15 bins (figure 4), most likely due to a combination of both (a) and (b) above. In addition, a negative vertical shear often occurs below bin ~35. The cause of this deeper negative shear is not clear, although it may in part be attributable to the bubble effect. The best data where the shear is very small occurs when the ship is on station i.e. 0-1 m/s ship speed in Figure 4 - clearly the bubble layer is minimised when the ship is stationary or moving very slowly. At ship speeds above 1 m/s, the shear effect is marked. However from Figure 4, as ship speed increases there is no obvious increase in vertical shear. An attempt was made to relate vertical shear magnitude to roll and pitch of the ship. Roll and pitch data were available for most cruises from the standard underway data set logged by the Aurora Australis (Ryan, 1995). For cruises au9404, au9501, au9604 and au9601, roll/pitch data were measured by a Trim two axis electrolytic gravity sensor. For cruises au9706, au9807 and au9901, roll/pitch data were obtained from the Ashtech 3DF GPS array. In all cases, the data were available at 10 sec. intervals. Roll/pitch data were not available for cruise au9701. For each cruise, standard deviation of both roll and pitch was calculated for each 20 or 30 min. averaged ADCP profile. ADCP data were divided into different speed classes according to the uncorrected ADCP speed (i.e. ship motion not removed) at bin 2. For each profile, the speed difference bin2 - bin20 was used to represent the vertical shear magnitude, and for each speed class these shears were fitted to the roll/pitch standard deviations (Figure 5). The linear bestfit and linear correlation coefficient for each speed class are shown in Figure 5. In the lowest speed class (0-1 m/s) there is no significant correlation, an expected result as there is no significant vertical shear for this speed class (Figure 4). At speeds above 1 m/s, there is in most cases a significant correlation, with the bin2 - bin20 difference decreasing (i.e. positive vertical shear increasing) with increasing roll and pitch values. In conclusion, several mechanisms appear to be simultaneously contributing to the erroneous vertical shear in the ADCP measurements, including motion of the vessel and entrainment of water by the moving vessel. It is however difficult to separate these different mechanisms. It seems reasonable to conclude that more bubbles occur below the ADCP transducers with increased rolling and pitching of the ship, resulting in a greater erroneous vertical shear (Figure 5). However the specific mechanism whereby bubbles cause this effect is not known. 3.2 Gyro error The final corrected ADCP data represent absolute water current i.e. ships velocity removed. For cruises where gyro data were used for ship heading (i.e. prior to au9701, Table 2), varying gyro behaviour throughout each cruise, and gyro error, result in a calibration mismatch between the ADCP transducer alignment and the ship heading. The end result is varying components of the ships velocity contaminating the corrected ADCP data i.e. faster ADCP currents when the ship is moving. For these cruises (i.e. au9404, au9501, au9604 and au9601), only the on station data are trustworthy. For cruises au9701 onwards where 3D GPS data were used for ship heading, the error still occurs but is minimised compared to the earlier cruises. An attempt was made to minimise this gyro error by applying a time series of ? in the calibration of cruises au9404, au9501, au9604 and au9601 (Figure 6). Overall, application of ? time series did not significantly improve the data, and ADCP data while the ship is underway remains suspect for these cruises. Note that a constant 1+? value for each cruise was still applied, as time variation of 1+? was only small. 4. DATA FILE TYPES 4.1 ASCII format files The complete ADCP data for each cruise are in the ascii files auxxxx01.cny, where auxxxx is the cruise number (e.g. au9604). The "on station" ADCP data (specifically, the data for which the ship speed is less than or equal to a particular value) are in the ascii files auxxxx_slowyy.cny, where auxxxx is the cruise number and yy is the relevant ship speed (e.g. au9604_slow35.cny = all data for which ship speed is less than or equal to 0.35m/s). * The data are 20 or 30 minute averages; shorter time averages would suffer too much from GPS/gyro errors. * ADCP currents are absolute - i.e. ship's motion has been subtracted out. * All data are GPS corrected; no bottom track data are included. * Files start with a 3 line header. Then comes each tt min. ensemble (where tt=20 or 30 minutes, Table2), as follows: First, a 1 line header, containing date (UTC) (dd-mmm-yyyy) time (UTC) (hh:mm:ss) (Each tt minute averaging period starts from the time indicated. So, e.g., a 30 min. ensemble with time 120000 is the average from 120000 to 123000). % of tt min. average covered by acceptable 3 min. ensembles deepest accepted vertical bin in the profile ship's E/W velocity over the ground over the tt min. (m/s, +ve=current towards the east) ship's N/S velocity over the ground over the tt min. (m/s, +ve=current towards the north) P= GPS position-derived velocity (D=direct GPS vel.; B=bottom track vel.) mean longitude over the tt min. (decimal degrees) mean latitude over the tt min. (decimal degrees) % of interfix period for which there was bottom depth information mean bottom depth over the tt min. 0 0 Next, the data, from the shallowest bin to the deepest bin: for each bin, there's 4 parameters: u = east/west current (m/s, +ve=current towards the east) v = north/south current (m/s, +ve=current towards the north) qc = quality control value - see below ipcok = percentage of the profile period for which there was good data in this bin Note that the data are written left to right across each line, then onto the next line, etc. (so 4 bins on a full line). The quality control value qc is given by qc = %good / (Verr+0.05) where: %good = percent good pings after logging system screening Verr = RMS error velocity (m/s). Possible range of qc is 0-20, with an expected range of 0-10; values of 0-4 indicate very poor data; values above 8 indicate very good data. * The file bindep.dat shows the water depths (in metres) that correspond to the centre of each vertical bin. * The file READMEdopxxxx (where xxxx is the cruise number, e.g. 9404) summarises ADCP data information for each cruise. 4.2 Matlab vectors and matrices The complete ADCP data for each cruise are in the matlab files auxxxxdop.mat, where auxxxx is the cruise number (e.g. au9604). The "on station" ADCP data (specifically, the data for which the ship speed is less than or equal to a particular value) are in the matlab files auxxxxdop_slowyy.cny, where auxxxx is the cruise number, and yy is the relevant ship speed speed (e.g. au9604dop_slow35.mat = all data for which ship speed is less than or equal to 0.35m/s). header info in matlab files For header info, column number in each matlab vector corresponds to tt minute average number (where tt=20 or 30, Table 2). botd = mean bottom depth (m) over the tt minute period cnav = GPS info: don't worry about it cruise = cruise number date = ddmmyy (UTC) ibcover = a bottom track parameter: don't worry about it icover = percentage of tt minute averaging period covered by acceptable 3 minute ensembles lastgd = deepest accepted bin in this profile lat = mean latitude over the tt minute period (decimal degrees) lon = mean longitude over the tt minute period (decimal degrees) nbins = no. of bins logged (=60 in most cases, Table 1) shipspeed = scalar resultant of shipu and shipv shipu = ship's E/W velocity over the ground over tt minute period (m/s, +ve=current towards east) shipv = ship's N/S velocity over the ground over tt minute period (m/s, +ve=current towards north) time = hhmmss, time (UTC) at start of tt minute averaging period adcp data in matlab files For adcp data matrices, row number corresponds to vertical bin number; column number corresponds to tt minute average number. bindep = depth (m) to centre of each vertical bin in the profile (will be the same for all profiles) ipcok = percentage of the profile period for which there was good data in this bin qc = a quality control value for each bin - see above speed = scalar resultant of u and v u = east/west current (m/s, +ve=current towards the east) v = north/south current (m/s, +ve=current towards the north) 5. ADCP SERVICE HISTORY November 1994 ship tied up in Hobart - seachest filled with salt/milliQ water, salinity ~100 June 1996 ship in drydock, Sydney - transducers inspected, seachest refilled with salt/tap water, salinity ~100 June 1998 ship in drydock, Sydney - transducers inspected and found to have tide marks halfway up, implying seachest had only been half filled at last drydock; unit in good condition, added 2 more zinc electrodes and cleaned transducer faces; seachest refilled (and checked) with salt/tap water, salinity ~100 July 1998 cables damaged due to ship fire - cabling replaced ~August January 1999 cables damaged by second ship fire - cabling replaced prior to July 1999 REFERENCES Dunn, J., 1995a. ADCP processing system. CSIRO Division of Oceanography (unpublished report). Dunn, J., 1995b. Processing of ADCP data at CSIRO Marine Laboratories. CSIRO Division of Oceanography (unpublished report). New, A. L., 1992. Factors affecting the quality of shipboard acoustic Doppler current profiler data. Deep-Sea Research, Vol. 39, No. 11/12, pp1985-1996. Ryan, T., 1995. Data Quality Manual for the data logged instrumentation aboard the RSV Aurora Australis.. Australian Antarctic Division, unpublished manuscript, second edition, April 1995. Table 1: ADCP logging parameters. ens=ensemble averaging duration; RLA=reference layer averaging; XROT=transducer alignment angle (milliradians) used for logging (affects value of the alignment angle correction factor a calculated and applied later in the data processing). A pitch/roll correction was applied during logging for cruise au9706 and onwards. cruise -----------profiling ping parameters---------- ---------bottom track ping parameters-------------- ens RLA bins XROT no.of bin pulse delay ping no.of bin pulse ping bins length length interval bins length length interval au9404 50 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 3-6(13/12/94-13/1/95) 785 3-10(13/1/95-21/1/95) 3-13(21/1/95-1/2/95) au9501 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 3-6 785 au9502 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 3-6(15/9/95-14/10/95) 785 8-13(14/10/95-2/11/95) au9604 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-20 785 au9601 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-13 785 au9701 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-13 785 au9702 50 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 3-6 785 au9706 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-20 822 au9807 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-20 822 au9801 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-20 748 au9901 60 8 m 8 m 4 m minimum 128 4 m 32 m same as profiling pings 3 min. 8-20 822 Table 2: ADCP data status for Aurora Australis cruises. av=averaging period for each ensemble in final data; slow=maximum ship speed allowable for on station data file. cruise dates nature data GPS/heading av slow of cruise status source au9404 13/12/94-02/02/95 WOCE sections S4, SR3 processed Lowrance+gyro 20 min. 0.35 m/s au9501 17/07/95-02/09/95 WOCE section SR3, FORMEX processed Lowrance/Koden+gyro 20 min. 0.40 m/s au9502 15/09/95-02/11/95 resupply Macquarie Island, Mawson, Davis not processed Koden+gyr0 - - au9604 19/01/96-31/03/96 MARGINEX processed Koden+gyr0 30 min. 0.35 m/s au9601 22/08/96-22/09/96 WOCE section SR3 processed Koden+gyro 30 min. 0.35 m/s au9701 09/09/97-22/09/97 SAZ processed Ashtech 3DF 20 min. 0.35 m/s au9702 24/09/97-11/11/97 resupply Casey, Mawson, Davis not processed Ashtech 3DF - - au9706 28/02/98-01/04/98 SAZ processed Ashtech 3DF 30 min. 0.45 m/s au9807 03/04/98-22/05/98 Mertz Polynya, Davis, Casey, Macquarie Is. processed Ashtech 3DF 30 min. 0.35 m/s au9801 15/07/98-31/07/98 Macquarie Is., Mertz Polynya not processed Ashtech 3DF - - au9901 13/07/99-07/09/99 Mertz Polynya Experiment processed Ashtech 3DF+differential 30 min. 0.35 m/s Table 3: Calibration information from processing of ADCP data. n=number of calibration sites; XROT as defined in Table 1. cruise XROT ? (1 standard dev.) 1+? (1 standard dev.) n ? used (cruise average) (cruise average) au9404 785 0.750 1.0500 160 time series au9501 785 1.172 1 1.814 1.0653 1 0.028 83 time series au9604 785 1.165 1 1.144 1.0557 1 0.013 70 time series au9601 785 1.144 1 1.173 1.0566 1 0.016 77 time series au9701 785 2.152 1 0.506 1.0554 1 0.019 17 cruise average au9706 822 4.511 1 0.709 1.0578 1 0.017 78 cruise average au9807 822 4.402 1 1.056 1.0521 1 0.015 66 cruise average au9901 822 2.399 1 0.691 1.0778 1 0.022 93 cruise average Figure 1a Figure 1b Figure 2a Figure 2b Figure 2c Figure 2d Figure 2e Figure 2f Figure 2g Figure 2h Figure 3a Figure 3b Figure 3c Figure 3d Figure 3e Figure 3f Figure 3g Figure 3h Figure 4a Figure 4b Figure 4c Figure 4d Figure 4e Figure 4f Figure 4g Figure 4h Figure 5a Figure 5b Figure 5c Figure 5d Figure 5e Figure 5f Figure 5g Figure 5h Figure 5i Figure 5j Figure 5k Figure 5l Figure 5m Figure 5n Figure 6a Figure 6b Figure 6c Figure 6d 15