Revelle 50kHz HDSS sonar unit -- ocean velocities ---------------------------------------------------- The 50kHz HDSS instrument consists of two forward facing beams (original design, now dead), and two newer aft-facing beams. The forward-facing beams cannot be repaired; a fully-functional 4-beam 50kHz HDSS instrument would require replacement of the forward-facing failed beams. The 140kHz HDSS instrument has four beams, but beam #3 has failed. (Pinkel's group is optimistic that it can be repaired during the next dry-dock period without replacing the transducer). During a Jan/Feb 2005 cruise on the Revelle ("P16S"), software was adapted to allow single-ping averaging of HDSS data and its subsequent post-processing using the CODAS processing machinery. (CODAS code, installation instructions, and processing documentation are available at currents.soest.hawaii.edu; aforementioned HDSS adaptation will be available shortly). Single-ping HDSS 50kHz and 140kHz data were processed using the same techniques applied to the NB150 single-ping data (collected on a linux laptop, running UHDAS for data acquisition). A comparison between the final HDSS and NB150 ocean velocities was very disappointing, with differences (in a mutually-overlapping reference layer) of over 10cm/s, and strong differences between HDSS ocean velocities on-station and underway. A previous cruise (HOT149, June 2003) was used to explore data from the 50kHz HDSS instrument on a cruise with a shorter duration. In June 2003, the 50kHz HDSS instrument had the two new aft-facing beams, and the two original forward-facing beams still worked (with reduced performance). Processing variations included 4-beam solutions or two-beam solutions, and including a pitch/roll correction (or setting pitch and roll to zero). Comparison between CODAS processing of NB150 5-minute pingdata and the 50kHz HDSS CODAS velocities was also disappointing. Including the pitch/roll correction had a negligible effect on the 4-beam solution, and actually made the 2-beam solution worse. An example of the comparison for the HOT cruise is shown in figure 01_twobeams_hprtss_u.png. The ship speed (green) and the difference between 50kHz HDSS and NB150 eastward ocean velocity from CODAS processing are plotted in the top panel. Ocean velocities were averaged over a 100-250m reference layer. The bottom panel shows the actual reference layer velocities. Although the values agree on occasion, the overall patterns are: (1) on station, HDSS is 10cm/s lower than NB150 (2) HDSS ocean velocities jump by 10cm/s between on-station and underway. The same reference layers were transformed into ship's forward and port coordinates and the HDSS-NB150 differences were plotted as a function of ship speed (figure 02_fpdiff_HOT.png). Observations include: (1) two-beam solutions (left panels) - (bottom) in the port direction, have a speed-dependent bias - (top) in the forward direction, have a speed-dependent bias (the best is actually the solution with zero pitch and roll) - (top) in the forward direction, have a non-zero bias at low speed - bias for forward beams (1,2) is about 0.1m/s - bias for aft beams (3,4) is about -0.1m/s (2) four-beam solutions (right panels) - (top) there is still a small bias at low speeds - (both) pitch/roll correction has little effect The same comparison for the Jan/Feb 2005 cruise shows an even greater on-station bias in the forward direction with the 2-beam 50kHz data; again, including the tss pitch and roll in the calculation degrades the data. The 140kHz data, with its 3-beam solution, still contains a bias at low ship speed. Because the 50kHz instrument currently allows ONLY a 2-beam solution, an effort was made to understand why the errors are so much greater than for the 4-beam solution, and whether any correction is possible. The driving question is "is it worth trying to get ocean velocities from this instrument, given that it currently has only two beams?". The short answer is "The 50kHz HDSS instrument can return shear, but the barotropic component must be provided by the NB150". A four-beam Doppler sonar system takes the four measured velocities along the beams and solves an over-determined system to obtain u,v, and w. The error in the solution can be used as an editing criterion. A three-beam instrument gives an exact solution, without this (useful) editing criterion. A two-beam solution is required for the 50kHz HDSS, since beams #1 and #2 are dead. The two beams form a plane oriented at about 20deg up from the vertical, aft of the ship. Horizontal ocean velocities are obtained by projecting the measured velocities (on this sloping plane) onto the horizontal. The horizontal velocity caused by this projection of heave is fwd_meas = w*cos(angle)/sin(angle_est) where "angle" is the actual angle of the plane formed by the beams, and "angle_est" is the estimate of the angle. For instance, two beams at 45deg to port and stbd of the ship's center line, each pointing 30deg up from the vertical, will form a plane at about 20deg from the vertical. If there were no ocean velocity at all and the ship simply moved up and down with pitch and it's correlated w, "angle" would be 20deg+oitch and "angle_est" would be 20deg or 20deg+pitch. Although the average w might be zero, the nonlinearity of the expression means that an average of the horizontal velocity induced by w is not zero, and can become quite large as w increases. Correcting for this effect requires a high resolution of accurate w at the transducer. An illustration of this is shown in figure 04_heave_fmeas_pitch.png. The top panel shows 200 seconds of on-station 50KHz HDSS measured velocity in the ship's forward direction, pitch, and the vertical velocity from heave (from tss). The second panel shows the same data, with simulated pitch and w fields (as labeled). The third panel shows the forward measured velocity that results from the geometry of the two beams and the simulated w. In the mean, this forward velocity is not zero, and we think some of the biases are coming from this source. The fact that ocean velocities from HOT 4-beam 50kHz data and P16 3-beam 140kHz data have a remaining velocity bias in the forward direction at slow speeds indicates that even effectively addressing heave rectification is not sufficient to calculate good ocean velocities. One more piece of information comes from the shear in the 5-minute averages of 50kHz HDSS ocean velocities. Figure 05_hr50_deepuv.png shows the on-station and underway averages between 520-620m for 2 weeks of data collected during calm sea conditions. Each 5-minute profile had the average 100-250m velocity subtracted. The left panels are the mean velocities; the right panels are the concurrent standard deviations. When the ship was underway, it was heading south: there is a 4cm/s bias in the 520-620m layer compared to the 100-250m layer, which is consistent with a bias towards zero of 1cm/s per 100m depth in forward measured velocity when underway. We have no idea as to the cause of this bias in shear. (In fact the underway shears might be fine, and it may be the on-station shears that are in error). In more detail then, the bottom line is, "The learning curve is high. To obtain ocean velocities, you can combine the shears from the 50KHz HDSS with the barotropic component from the NB150 to create a referenced 50kHz ocean velocity field, but there will still be errors whose size, cause, and solution are not known at this time". ---------------- Jules Hummon March 17, 2005 hummon@hawaii.edu