POSMV Quality

Comparison between a POSMV and gyro provides information about both systems. If one is known to be good (operating within normal parameters) then problems with the other become clear.

In the interests of understanding the behaviour of one particular POSMV, a collection of figures was generated for several ships with POSMVs. POSMV and gyro data from each ship for an arbitrary recent cruise was binned in 1-hour segments, and the percentage of heading values with GAMS active was calculated for each hour. Similarly, using a threshold just above each instrument's "best" accuracy, the fraction of points with good accuracy was calculated. From these crude statistics we get an idea of the fraction of solutions with GAMS (mean percent GAMS over the cruise) and the fraction of time (fraction of hour-long chunks) that have some kind of accuracy failure.

Summary:

--------------------------------------------------------------------------

ship           accuracy            overall      percent          comment
               cutoff              percent      not perfect
               (from data)          GMS         accuracy
-----          ------------        ---------    ---------    ------------
Kilo Moana(1)   .013                >95             10          very good
Kilo Moana(2)   .013                >99              1          excellent
Knorr           .018                 75             17           good
Ka`imimoana     .015                 60             15           good
Thompson        .055                 80              1          very good
Hi`ialakai(1)   .018                 50             50            bad
Hi`ialakai(2)   .018                 50             35            OK

--------------------------------------------------------------------------

Links to figures, and comments grouped by quality

Excellent

• Kilo Moana(2)

  • after POSMV replacement (amazing!)

Very Good

• Kilo Moana(1)

  • before POSMV replacement (when it worked, it worked pretty well)

  • many GAMS solutions

• Thompson

  • short baseline for antennsa makes for worse stated accuracy?

  • heading accuracy cannot be used as a quality criterion on this ship

  • deviations in heading are successfully avoided by requiring GAMS

Good

• Knorr

  • seems OK; under 20% are degraded in accuracy

  • missing 30% of GAMs solutions

• Ka`imimoana

  • seems OK; under 20% are degraded in accuracy

  • missing 30-40% of GAMs solutions

Not-so-good??

• Hi`ialakai(2)

  • after PCS replacement

  • see below: 35% are not full accuracy

  • missing 30-40% of GAMs solutions

Outliers

• Hi`ialakai(1)

  • before PCS replacement

  • many holes where GAMS was not used (failed accuracy cutoff)

  • evidently rebooting frequently

  • maybe 50% of GAMS were missing

The Hi`ialakai POSMV is used for bathymetric mapping and for ADCP ocean velocities. The POSMV PCS was replaced around Feb 25, and during the week before, data were gathered with the PCS from the Ahi. Indications of POSMV illness prior to mid-Feb included variable POSMV-gyro offset, lack of GAMS solution data, and frequent degradations in stated heading accuracy. Several figures and associated comments follow:

POSMV-gyro difference (after QC)

This figure shows the variability of the POSMV compared to the gyro during the first 3 months of 2006. Before decimal day 50, the POSMV reported GAMS solutions and accuracies better than 0.018, yet still showed large and variable differences from the gyro. During the use of the Ahi PCS, the POSMV-gyro (dday 51-56), difference was centered around 0deg. Very near the time when the new PCS was on line the POSMV-gyro difference jumed to a consistent 2deg. This lasted until until dday 81.3 (2006/03/23 07:12:00), when it drifted back to be closer to the gyro over the course of 10 minutes (details below). The POSMV seems to have gone from "very unhealthy" (often no GAMS for long periods, varying offsets from gyro, and frequent loss of accuracy for short periods) with its old PCS, to perhaps "moderately unhealthy" with the new PCS. Compared to Knorr and Ka`imimoana, it looks slightly worse (lower GAMS percent, more imperfect hours). However, with the new PCS, there is still an unexplained jump in the offset between gyro and POSMV.

shift in POSMV-gyro

POSMV headings were offset roughly +2.5deg prior to decimal day 81.3. POSMV quality deteriorated (GAMS dropped out and accuracy failed the .018 cutoff) starting at 81.3 (2006/03/23 07:12:00 UTC) but after quality was regained, the offset had changed to nearer zero.

zoom in same revion

The Hi`ialakai POSMV shouldn't have drifted like this (from a 2.5deg gyro offset to nearly zero offset). Ship's personnel insist that only one gyro was on line at the time. Acoustic Doppler Current Profiler data also indicate a 2 degree offset in calibration relative to the POSMV at about this time. This drift, combined with the relatively larger number of GAMS and accuracy failures suggests this instrument requires more attention before it should be considered operational.

This link contains a zipped copy of this evaluation.

Jules Hummon Thu Apr 13, 2006