Variable cell sizes for current profilers?

Mounting an instrument near the bed looking upward, I would like to have a small cell size near the bed/instrument to measure the bed boundary layer, and still a good velocity precision higher up in the water column, but there the cell size may be larger.

With instruments like the Aquadopp Profiler, the velocity uncertainty (or velocity precision) depends on cell size and average interval (and measurement load), but also on a sufficient signal/noise ratio. The signal returning from the far profile section will be weaker and the signal/noise ratio lower. Therefore, a small cell size may be possible for short profiles, whereas a larger cell size may be recommended for longer profiles depending on the concentration of scattering particles.

One possibility is to record data with a small cell size on a long profile (thank cheap computer memory this is now possible), and then during the post-processing to average in the far profile section several cells into one larger cell. The result is a profile with cells increasing in size further away from the instrument.

Would with such a post-processing the velocity uncertainty of averaged (for example) pairs of cells be similar to velocity uncertainty data recorded initially with the double cell-size?

Would the same be valid for the Aquadopp Profiler HR?

Urs Neumeier

Hi Urs,

Please check this thread: http://www.nortek-as.com/en/knowledge-center/forum/current-profilers-and-current-meters/810127891 for an answer to your question. In short, creating larger cells from smallers ones will result in higher noise.

I looked in details at the thread http://www.nortek-as.com/en/knowledge-center/forum/current-profilers-and-current-meters/586576292 suggested by Peter Rusello.

If I interpret it correctly, especially the final comment by Atle Lohmann, the cell size influences the velocity precision by two ways, by the length of the receive window but also by the length of the transmit pulse (which is adjusted by the instrument for optimal results). I guess each has an influence of proportional to 1/sqrt(cell-size), which combined give an influence proportional to 1/cell-size. Therefore, averaging together several cells during post-processing could compensate the short receive windows, but never the length of the transmit pulse.

From this I deduce that increasing the cell size in the instrument setup by a factor n would improve the velocity precision by a factor 1/n, but averaging n cells in post-processing would improve the velocity precision only by a factor 1/sqrt(n), still an interesting improvement! Is this correct?

If this method is valid, would it help in my special case, where the main reason to increase the cell size is poor signal/noise ratio in far profile section (at 4-6 m with 0.1 m cell size)?

Urs Neumeier

Hi Urs,

I'm always hesitant when dealing with questionable data. While averaging the bins together will reduce noise, it's not necessarily going to rescue measurements with poor initial data quality.

The ideal solution is two instruments optimized for the two different measurements you'd like to make. If that's not an option, and you don't need the spatial resolution further from the boundary, there's no reason not to average bins together assuming the initial data quality is acceptable.

P.J.