Py-ART ideas from the 2015 AMS radar conference

[jjhelmus]

This is a meta-issue for keeping track of ideas that were discussed with Py-ART users and supporters at the 2015 AMS radar conference. Expect this top level comment to be edited to reflect changes as development progressed. Some of these issue already have their own issue in the tracker, I'll add links to these when possible.

Please provide any additions to this list. Feel free to share opinions as to the priority of any of these ideas.

Idea for Py-ART from the conference broken down by area

pyart.io

• UF writing (UF File Writing #186, completed in ENH: write UF files #388)
• Reading NEXRAD Level II message 1 format (Read NEXRAD Level 2 files without Message 31 #352, complete in ENH: Support for reading message 1 NEXRAD Level 2 files #455)
• Reading single field/sweep from NEXRAD Level II (complete in ENH: Support for reading message 1 NEXRAD Level 2 files #455)
• Interpolating NEXRAD data to a uniform gate spacing when not-constant (NEXRAD files with non-uniform gate locations #152, Method for storing a volume in which the range gate differ between rays. #252, complete in ENH: Support for reading message 1 NEXRAD Level 2 files #455)
• Full support for reading ODIM_H5 files (ODIM_H5 reader #131)
• Writing ODIM_H5 files.

pyart.correct

• Correct unfolded velocities in the region based dealiasing algorithm using a sonde profile or previously unfolded field.
• Updated phase processing code using self consistency or other additional constraints (Need to enhance LP code with a constraint that limits KDP when Z is low #125)

pyart.graph

• Default colormaps based on field name (completed in ENH: Default field specific colormaps and luminosity when plotting #464).
• Default luminosity limits based on field name (completed in ENH: Default field specific colormaps and luminosity when plotting #464).
• Categorical colorbars. (See discussion below and ENH: Add norm parameter to plotting methods #468)
• Expose calculation of inputs to pcolormesh in the public API. (See discussion below and MAINT: Refactor of Display classes #463)

pyart.grid

• Three dimensional linear interpolation based gridding as used by SPRINT.
• Nearest neighbor gridding (New potential mapping routine #182)
• Distance weighted gridding where the weighting distance in x, y and z can be specified independently
• Support for creating non-uniform grids (Support for visualizing and creating non-uniform Grids #350)

pyart.retrieve

• QPE equaltion, R(A), R(KDP), etc.
• VAD, VVT and DVAD retrievals. Some of these are available in SingleDop

Other

• Quasi vertical profiles, QVP.
• Object for iteration over radar objects by specifying a list of files (useful for QVP).
• Cross section tool for making building RHI like scans from PPI data (completed in ENH: Add pyart.util.cross_section_ppi function #380)
• Radar composibility (add, remove, insert sweeps or rays) (related join_radar function is broken #127 and Object oriented, hierarchical Radar, Sweep, and Ray objects #242)
• Create conda packages for other radar tools (RSL, wradlib, RadX, PyTDA, SingleDop, ARTView, etc). (Tracking progress in Conda packages for various open source radar packages #505)
• GridFilter class.
• GateFilter method to include/exclude based upon height of gate.

Documenation

• Add links to short course materials. (complete DOC: Updates to the documentation and README #391)

[jjhelmus]

Should mention that this list is in no way a indication that all of these ideas will be implemented. Rather that they are being considered.

[gamaanderson]

I can't find this anymore, but in some issue it was discussed about Grids with non-uniform gates. You may want to add this as a topic there, I think it is worth considering.

[scollis]

Do you mean radars or grid objects?

On 10/1/15 12:38 PM, Anderson wrote:

I can't find this anymore, but in some issue it was discussed about
Grids with non-uniform gates. You may want to add this as a topic
there, I think it is worth considering.

—
Reply to this email directly or view it on GitHub
#390 (comment).

[gamaanderson]

I mean grid objects, i.e. non uniform "x_disp", "y_disp" and "z_disp".

Radar objects being based in CfRadial should already accept non-uniform range and time axes; there may be some function that presuppose uniformity in Radar, but I would classify that (if not documented) as a Bug.

[scollis]

Yes. This is work that WILL be done. Either part of Py-ART or something
compatible with Py-ART. I am about to start a new project that will work
with grid data in a variety of coordinate systems and attempt to make
them inter-compatible. eg take a lat lon grid in sigma (terrain
following) coordinates and easily project to x/y in height
etc...

On 10/1/15 12:56 PM, Anderson wrote:

I mean grid objects, i.e. non uniform "x_disp", "y_disp" and "z_disp".

Radar objects being based in CfRadial should already accept
non-uniform range and time axes; there may be some function that
presuppose uniformity in Radar, but I would classify that (if not
documented) as a Bug.

—
Reply to this email directly or view it on GitHub
#390 (comment).

[gamaanderson]

Starting a new progress is also good, them we can make a fresh start and extend py-ART later. Of course you are following issues #285 and #294, there are important considerations there.

[jjhelmus]

Issue #350 covers non-uniform grids. I'll add a link in the top comment.

[jjhelmus]

Closing a few NEXRAD related ideas with the merging of PR #455

[jjhelmus]

Expose calculation of inputs to pcolormesh in the public API.

Once PR #463 is merged this will be possible using the get_field and get_gate_x_y_z methods of the Radar class. For example to plot a PPI:

import matplotlib.pyplot as plt
import pyart

radar = pyart.io.read('XSW110520105408.RAW7HHF')
data = radar.get_field(0, 'reflectivity')
x, y, z = radar.get_gate_x_y_z(0, edges=True)
plt.pcolormesh(x, y, data, vmin=-32, vmax=64.)
plt.show()

You still need to choose the correct pair from x, y, or z, but I think this should be sufficient.

[jjhelmus]

With PR #468 merged it is possible to create categorical or non-linear colorbars by specifying the norm keyword. For example:

import matplotlib as mpl
import matplotlib.pyplot as plt
import pyart

radar = pyart.io.read('KLOT20130418_043539_V06.gz')

bounds = [-7.5, 12.25, 22.5, 35.5, 77.5]
cmap = mpl.cm.get_cmap('pyart_NWSRef')
norm = mpl.colors.BoundaryNorm(bounds, cmap.N)
display = pyart.graph.RadarDisplay(radar)
display.plot('reflectivity', norm=norm, cmap=cmap)
plt.show()

[nguy]

This is great! I just implemented something similar in AWOT using a 'discrete_cmap_levels' keyword.

[jjhelmus]

Updated list to indicate that field specific colormaps and luminosity limits were added by PR #464