velocities measured at different vertical locations are not over lapping

It's a little difficult to say without knowing quite a bit more about the data quality, number of samples and experiment conditions. We have had several users link profiles like this and achieve very good results. So, I would recommend the following to start:

1. Verify the stationarity of your flow and that you are sampling long enough to develop good estimates of the mean.

2. Perform some basic data quality screening to remove outliers and low correlation values. Outliers could be influencing your statistics and causing some of the differences you've observed.

P.J.

The other things to watch out for are weak spots.  A very reflective bottom (e.g. metal) combined with a small physical cross section (e.g. a flume) can result in persistent echoes that interfere with the entire profiling region.  Try and add some sort of sound absorptive material on the bottom (acryllic or glass).  It could also be worthwhile turning on the "adaptive" algorithm to try and remove weak spots ("Adaptive:Once" should be sufficient).

     Robert

Thanks for replies. We have sampled for 5 minutes duration at each location. Locations 1, 2, 3 and 4 are vertically spaced at 1 cm where 1 is lowest point and 4 is highest point. We have used two sampling frequencies (25 Hz as well as 100 Hz) and we plotted both sets of data independently to check which frequency is good. We have removed spikes from the data using Wahl (2003). Data points with SNR values below 25 and correlation values below 74 are removed.  All the precautions are taken to get good data. I am enclosing the velocity profile plots. I am unable to upload data files because they are above maximum attachment size limit. Please let me know how to get good data and how to make to them to fall on one another.

Thanks

Prashanth

Hi Prashanth,

How much agreement are you expecting in these profiles? The results look pretty good to me with a few exceptions and for not knowing much about the facility and experiment.

How much data are you removing with the screening? Have you looked at an uncertainty analysis to see if the measured differences between profiles are within error bounds?

I'm sorry I don't have an easy answer for you on how to get these to exactly overlap, but I would not necessarily expect this in the first place. There does look to be a problem in the first two bins of any profile, and this is probably related to the calibration and a near field effect of the probe. 

The position 1 profile at 100Hz and position 4 profile at 25 Hz both seem problematic. The very different results seen there suggest the instrument setup probably needs adjustment.

P.J.