This paper has some explanation that it is a calculation based on the vertical velocities, but perhaps more searching or contact of authors who discuss how these are actually measured or calculated from velocity directional vectors is probably appropriate.  I have not used the Doppler system, but perhaps that might be used and recorded in some way to give rotational detail or indicators.  From what I have learned but not measured, these circulation patterns are most measurable on the surface and near the bed.  It is an interesting subject, and I suppose you locate the helix rotation by where the velocity vectors at the surface and across the bottom where the direction changes.  In clear wadable waters, I suppose you should be able to see this directional shift with current meters.  I have read and will search for an article of a flume study where secondary currents were measured and this article showed velocity jumps with areas of channel configuration changes such as the transition from bankfull to floodplain or sharp channel edges so I will try to upload that when I find it.  I am not sure if that used pitot tube for point velocity.  This is an interesting subject that if measurement could be accomplished, would help further define the apparent circulation dynamics as it may apply to channel features such as bar development, but also the added complexity as these secondary velocity vector patterns may shift in varying degrees with stage and channel roughness and dimension change.   

http://onlinelibrary.wiley.com/doi/10.1029/2008WR007428/pdf

I was able to find paper by Keane et al, 2009 with flume study on circulation pattern relative to near bank stress, etc.  

Thank you William for your comment

But, there is no any detail for calculation of secondary velocities and even it wasn't mentioned the Rozovskii method. 

Sorry, I didn't see you first comment. Thank you again for the papers. 

Dear Farhad,

These papers could help you.

Rhoads, B.L. y Kenworthy S.T. (1999), On secondary circulation, helical motion and Rozovskii-based analysis of time-averaged two-dimensional velocity fields at confluence, Earth Surface Processes and Landforms, 24, 369-375.

Lane, S. N, Bradbrook, K. F., Richards, K. F., Biron, P. M. y Roy A. G. (2000), Secondary circulation cells in river channel confluence: measurement artefacts or coherent flow structures?, Hydrological Processes, 14, 2047-2071.

Best regards

Ricardo

Thank you Dr. Szupiany for your comment

I read these papers already. There are some different views on calculating the velocity components based on Rozovskii method.

How honored and fortunate are we to get remarks by Dr. Szupiany, lead author of first paper sent!  

I added a USGS paper on the Doppler utility in mapping 3 D flow pattern such as in a GIS format.  As mentioned, I do not have experience in trying to measure or calculate this, or apply this in some way on a project level.  More than ever before, the technology to measure and map, display and document is better than ever.

https://pubs.usgs.gov/fs/2013/3028/pdf/fs2013-3028.pdf

The paper in ESPL by Parsons et al may also help answer your question. The paper  and the VMT matlab code is available from the USGS website. You could look at the code to see how the Rozovoskii circulation is calculated.  A link to the VMT page is: 

hydroacoustics.usgs.gov/movingboat/VMT/VMT.shtml

Thank you Kevin for your comment.

Yes, I read all documents and also the paper you mentioned. I also read VMT source codes on GitHub. But still, I have some problems with decoding.