Unrealistic velocity corrections

[pandasambit15]

Hi!

I had contacted @vlouf earlier regarding unrealistic velocity corrections performed on a sample tropical cyclone data. (SAMPLES ATTACHED).
I am using data from a C-band radar at a Tropical Station, with obvious velocity aliasing. Following is a code snippet.

`
###select only dprf files###
if mode == "dprf" and num_bins == 1600 and num_elev == 11 and az.size==3960:
radar = convert_to_pyart(sortedfiles[i])
velcor = unravel.unravel_3D_pyart(radar, velname='VEL', dbzname='DBZ', nyquist_velocity=8.0)

`

[vlouf]

The input data looks quite noisy, Unravel might not work very well here. What I don't understand is that there's part without any data south to the center that is absent from the dealiased data. Has the noise been filtered out of the DBZ field?

[pandasambit15]

This is raw data from the lower elevation angles, and the observations are from Tropical Cyclone Ockhi. The blank region you are referring towards the South is the eye of the cyclone, which is void of any scatterers. What do you suggest should be done to smooth out the data?? I can remove the clutter pixels from the velocity field and a median filter/speckle filter to smooth out the data. Will that work? Regards, Sambit

…

On Wed, Aug 5, 2020, 5:17 AM Valentin Louf ***@***.***> wrote: The input data looks quite noisy, Unravel might not work very well here. What I don't understand is that there's part without any data south to the center that is absent from the dealiased data. Has the noise been filtered out of the DBZ field? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#9 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/APYEBN2SQOCW24W6VFZHPQTR7CM25ANCNFSM4PUNJ4JQ> .

[vlouf]

No scatterer in the eye of the cyclone is normal, the problem is why the output (badly) dealiased velocity shows data where there should be none.

Yes, despeckeling the data is a good thing.

[pandasambit15]

The output giving values in the "no-data" region was precisely my concern and the reason why I raised this issue. I will try de-speckling the data and use it again for the algorithm, to see if it improves the results. In the meantime, I wanted to know whether we are supposed to pass only one single sweep (data for one elevation) as input or can we pass the entire PPI volume as input to the UNRAVEL dealiasing algorithms? If possible, Please have a look at my code which I had sent you; just to make sure what I am doing is write. Since, there is no proper documentation or demo notebooks available for its usage, I am not quite sure how to go about it. Thanks and regards, Sambit *Sambit Kumar Panda* *Scientist/Engineer-SC, ASD/AOSG/EPSA,* *SAC(ISRO), Ahmedabad*

…

On Wed, Aug 5, 2020 at 9:44 AM Valentin Louf ***@***.***> wrote: No scatterer in the eye of the cyclone is normal, the problem is why the output (badly) dealiased velocity shows data where there should be none. Yes, despeckeling the data is a good thing. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#9 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/APYEBN5LR4GAOK662HQO5NLR7DMBBANCNFSM4PUNJ4JQ> .

[vlouf]

Here my comments:
#8 (comment)