What does the number calculated after classifying the surface object mean?

[pjw0727]

Dear everyone

hello. I am a researcher who is using PYART to make good use of radar data processing.

Recently, I am coding the data processing for classifying the surface object using the "pyart.retrieve.hydroclass_semisupervised" module.

However, the calculated HCI data only shows numbers ranging from 0 to 9.
I've tried a lot to find the meaning of that number, but I can't find it, so I'm writing this.

Can you guess what each number means by the class name of the spheroid?
I want to know which numbers are strong precipitation, snow, ice crystals, etc.

Then we look forward to your valuable comments.
thank you

[sherma...@gmail.com]

Hello!

So the method is based on https://amt.copernicus.org/articles/9/4425/2016/amt-9-4425-2016.pdf

For the corresponding values, the code pulls from a mass centers table found here:

https://github.com/ARM-DOE/pyart/blob/2a79d6b6ebfda36bbeb1528135c3dc248144c945/pyart/retrieve/echo_class.py#L364

rain (RN) rimed ice particles (RP) vertically aligned ice (VI) wet snow (WS) ice hail and high density graupel (IH/HDG) metaling hail (MH) crystals (CR) aggregates (AG) light rain (LR)

Hope that helps,

Zach S.

[pjw0727]

In the reference paper, the PID [rain (RN) rimed ice particles (RP) vertically aligned ice (VI) wet snow (WS) ice hail and high density graupel (IH/HDG) metaling hail (MH) crystals (CR) aggregates (AG) light rain (LR) ] was confirmed.
However, the PID specified in the _mass_centers_table() function in the actual code is [DS, CR, LR, GR, RN, VI, WS, MH, IH/HDG].

Here I am curious about two things.
First, I want to know the full name of the PID.
And there is only WetSnow in the reference paper, but _mass_centers_table() seems to have WS(WetSnow) and DS(DrySnow). How could you apply the centroids value to that lookup table with a different PID than the paper? Is there a function in pyart that can change the value of that table constant?

Please clear my doubts. thank you

2022년 7월 15일 금요일 오전 12시 53분 45초 UTC+9에 sherma...@gmail.com님이 작성:

[Maxwell Grover]

Hello,

We encourage you to please open a discussion on our discussions page, specifically the Q and A section (https://github.com/ARM-DOE/pyart/discussions/categories/q-a) since we are phasing out this email list.

I attached two images, the first being Figure 6 from the referenced paper, where they define the centroid locations, which are used for the clustering, with their associated classes. On the right is a plot generated from the mass values in Py-ART, for a C-band radar.

You'll notice the two classifications you brought up before, Dry snow (DS) and GR (graupel) do not have a direct 1 to 1 correspondence to the paper. Based on these visualizations, it appears that dry snow corresponds to the aggregate class, and graupel corresponds to the rimed particle class. We will need to clarify here with the original developers of this implementation in Py-ART (the Meteoswiss folks), but this seems like the most likely case based on this exploration.

You can input your own centroid mass centers and weights using the hydroclass semisupervised function, with the default values coming from pyart.retrieve.echo_class._mass_centers_table(). That default table has values for X-Band and C-Band radars, with the difference explained within the paper (specifically the appendix). I encourage you to dig into the paper a bit, and try to reproduce similar figures based on the values so you can decide what is best for you, and if our implementation in Py-ART suits your use-case.

Thanks,

Max

--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/pyart-users/f61b5059-a79a-40fa-8055-4c3f8c1d705an%40googlegroups.com.