4.2.14. Underway Bias caused by low Percent Good

The standard way we calculate ship speed is to (essentially) use the start and end ping times for an ensemble, and divide by the distance travelled in that time.

nav calculation for ship speed

If whole profiles contain no data (eg. when bubbles knock out a profile but do not bias the rest of the profiles) then this calculation is wrong. Instead we need the ship speed for each ping, and then calculate the average ship speed for the ensemble as only counting those times when velocities were measured.

nav calculation for ship speed

When things work well (frequent pings, high-quality GPS) implementation of this algorithm is not difficult. However it has not been tested sufficiently to roll it out as “the standard” for ship speed calculations. Therefore it is recommended in the final stages of postprocessing, after all the other issues have been dealth with.

Application of this algorithm should be to

  • copy your processing directory, eg:

    cp -a os75nb os75nb.uvship

  • change directories into the new one, and apply the correction:

    quick_adcp.py --steps2rerun navsteps:calib --refuv_source uvship --auto

  • change directories back by one, and compare:

    cd ..
dataviewer.py -c os75nb os75nb.uvship

You should see something like this:

comparison of "nav" and "uvship" for ship speed