Correct the noise in radial velocity field

[Vanessa Ferreira]

Dear All,

I am plotting radial velocity from a .UF file, but there is a lot of noise in the velocity field.

And I would like to know if there is some way to correct or smooth this noise.

I attached a .png file with the velocity field. The .UF file is here https://drive.google.com/file/d/0B__zqcn0rbv4VFFPWFR0bUpWUk0/view?usp=sharing

And this is the code:

import numpy as np
import matplotlib.pyplot as plt
import pyart

filename = 'STI-250--2015-10-06--22-50-23.mvol.uf'
radar = pyart.io.read(filename)
radar.latitude['data'] = np.array([-29.22499999])
radar.longitude['data'] = np.array([-54.929999993333])

display = pyart.graph.RadarMapDisplay(radar)
display.plot_ppi_map('velocity', 0, vmin=-30, vmax=30,
                     lon_lines=np.arange(-60, -50, 2),
                     lat_lines=np.arange(-31, -26.5, 2), resolution='h',
                     cmap=plt.get_cmap('seismic'))

display.plot_point(radar.longitude['data'], radar.latitude['data'])

plt.savefig('velocity.png',)


Thank you very much!
Vanessa Ferreira

[Scott Collis]

Hey Vanessa,

Is this radar using Dual PRT?

Also does it look so bad at higher elevations?

Take a look at this notebook for how you can smooth a field. 

https://github.com/openradar/AMS-Short-Course-on-Open-Source-Radar-Software/blob/master/5_Simple_proccessing_and_adding_a_field.ipynb

Dr Scott Collis
Purveyor of lofty observations (Atmospheric Scientist) 
Environmental Science Division
Argonne National Laboratory
Mb: +1 630 235 8025
Of: +1 630 252 0550
Become a Py-ART user today! http://arm-doe.github.io/pyart/

--
You received this message because you are subscribed to the Google Groups "Py-ART Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pyart-users...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
<velocity.png>

[Vanessa Ferreira]

Hi Scott,

Thank you for your prompt response.

"Is this radar using Dual PRT?"

No, this radar isn't using Dual PRT.

"Also does it look so bad at higher elevations?"

Yes, the velocity field look so bad at higher elevations too. Also, I have other .UF files from this radar for other events and they are also very noise.

I'll take a look at this link and see if I can smooth the velocity field.

Thank you.
Vanessa Ferreira

[Vanessa Ferreira]

Hi All,

I corrected the radial velocity field by using the dealiasing method pyart.correct.dealias_region_based.
The pyart.correct.dealias_region_based has produced nice results. But there is a problem: in some cases the signal of some regions of the radial velocity field
 changes from negative to positive (or vice versa) from one image to another (e.g, the SW region in the figure that I have attached) and then, in the very next image, it changes back to the correct signal.

An example of this error is attached here: http://syno-meso.com/images/juntas.png

Why does this happens? And is there some way of correcting this?

Thank you!

Vanessa Ferreira

Em quinta-feira, 12 de maio de 2016 21:31:58 UTC-3, Vanessa Ferreira escreveu:

[Jonathan Helmus]

Vanessa,

    I can't seem to view your linked image but from the description I believe I understand the issue.  The cause of the issue is that the region based dealiasing algorithm produced smooth velocities fields which may be globally offset from the true velocities by integer multiples of the Nyquist interval. 

    The reason this has to do with how the algorithm operates.  The algorithm find regions of similar velocities and then unfolds these regions against each other to minimize the velocity differences between neighboring regions.  No region is marked as "good" so if the unfolding begins at a region which was folded this first pass can produce a solution in which all gates in the sweep are over or under-folded.  The algorithm tries to correct for this by "centering" the sweep, which unfolds the entire sweep so that the average number of folds applied to each gate is zero.  This correction can be disabled by setting the "centered" parameter to False but this is not recommended.

At no point are the velocities evaluated for reason-ability or compared to a reference wind profile/pattern.  Therefore, if the average wind field in the sweep is above the Nyquist limit it is possible for the "centered" velocities to be offset from the true velocities by an multiple of the Nyquist interval. 

The root of the problem is that the algorithm needs some method to identify at least one region which is not folded.  This can then be used to anchor the solution. Unfortunately. I have not been able to determine how to do this without an external source of wind information.  It is on the Py-ART roadmap to allow users to provide such information as either a a previously unfolded volume which is known to be "good" or sounding data.

Cheers,

    - Jonathan Helmus

--