Clutter filtering

[Sylvain Watelet]

Dear wradlib users,

Looking at this function https://docs.wradlib.org/en/stable/generated/wradlib.clutter.classify_echo_fuzzy.html , I noticed that the trapezoidal membership function for the decision variable rhohv (rho2) is very strict in the sense that it very easily regards not so low values of rhohv as clutter, although such values could simply correspond to the bright band. See here https://github.com/wradlib/wradlib/blob/main/wradlib/clutter.py , line 409: "rho2": [-9999, -9999, 0.95, 0.98].

Looking at Vulpiani et al. (2012) on which the function classify_echo_fuzzy is based, there is no sign of rhohv as a decision variable (only its texture). So, what is the scientific basis for the above implementation?

Thank you in advance,

Cheers,

Sylvain

Sylvain WATELET +32 (0)2 890 98 80 Koninklijk Meteorologisch Instituut Institut Royal Météorologique www.meteo.be / Weather app / Social Media

[Kai Muehlbauer]

Dear Sylvain,

thanks for bringing this to attention.

I'd need to hand that question over to Maik Heistermann. He
added/changed that "rho2" back in the days:

https://github.com/wradlib/wradlib/commit/163c3e272b6f5484e640673a1c00072b680445ae

It looks like incorporating RHOHV directly is an enhancement to the
Vulpiani algorithm. I'd guess it should discard high RHOHV values from
clutter, but I agree that the current default seems far to strict.

If you are not happy with the current setup, you might just override
that trapezoid functions with your own (via "trpz" kwarg) or set the
weight to zero (via "weights" kwarg).

Do you think an update of the docstring would help users?

Best,
Kai

Am 11.05.22 um 10:22 schrieb Sylvain Watelet:

> Dear wradlib users,
>
> Looking at this function
> https://docs.wradlib.org/en/stable/generated/wradlib.clutter.classify_echo_fuzzy.html
> <https://docs.wradlib.org/en/stable/generated/wradlib.clutter.classify_echo_fuzzy.html>
> , I noticed that the trapezoidal membership function for the decision
> variable rhohv (rho2) is very strict in the sense that it very easily
> regards not so low values of rhohv as clutter, although such values
> could simply correspond to the bright band. See here
> https://github.com/wradlib/wradlib/blob/main/wradlib/clutter.py
> <https://github.com/wradlib/wradlib/blob/main/wradlib/clutter.py> , line
> 409: "rho2": [-9999, -9999, 0.95, 0.98].
>
> Looking at Vulpiani et al. (2012) on which the function
> classify_echo_fuzzy is based, there is no sign of rhohv as a decision
> variable (only its texture). So, what is the scientific basis for the
> above implementation?
>
> Thank you in advance,
> Cheers,
> Sylvain
>
> Sylvain WATELET
> +32 (0)2 890 98 80

> /Koninklijk Meteorologisch Instituut
> Institut Royal Météorologique/
> www.meteo.be <https://www.meteo.be> / Weather app
> <https://www.meteo.be/en/info/faq/products-services/the-rmi-weather-app>
> / Social Media
> <https://www.meteo.be/en/info/faq/products-services/rmi-social-media>
>
>
> --
> You received this message because you are subscribed to the Google
> Groups "wradlib-users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to wradlib-user...@googlegroups.com
> <mailto:wradlib-user...@googlegroups.com>.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/wradlib-users/376538498.1616371.1652257332575.JavaMail.zimbra%40meteo.be
> <https://groups.google.com/d/msgid/wradlib-users/376538498.1616371.1652257332575.JavaMail.zimbra%40meteo.be?utm_medium=email&utm_source=footer>.

--
Kai Muehlbauer
Institute of Geosciences, Meteorology Section, University of Bonn
Auf dem Huegel 20 | +49 228 739083
D-53121 Bonn | kai.mue...@uni-bonn.de

[Sylvain Watelet]

Hi Kai,

Thank you for the quick answer! I think the documentation should at least include the default values for the weights and for the trapezoids. A nice addition would be to cite the source for each default parameter. This is just my opinion of course.

Thanks for reminding me the possibility of using the trpz argument, I think I will use that.

Cheers,

Sylvain

---

De: "Kai Muehlbauer" <kai.mue...@uni-bonn.de>
À: "wradlib-users" <wradli...@googlegroups.com>
Envoyé: Mercredi 11 Mai 2022 10:48:41
Objet: Re: Clutter filtering

To unsubscribe from this group and stop receiving emails from it, send an email to wradlib-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/wradlib-users/933da9ad-1210-eb44-e3ec-bd317d7b5954%40uni-bonn.de.

[Kai Muehlbauer]

Hi Sylvain,

you're welcome.

Would you be able to change the docstring via a PullRequest? That would
be very much appreciated.

Best,
Kai

Am 11.05.22 um 11:18 schrieb Sylvain Watelet:

> Hi Kai,
>
> Thank you for the quick answer! I think the documentation should at
> least include the default values for the weights and for the trapezoids.
> A nice addition would be to cite the source for each default parameter.
> This is just my opinion of course.
>
> Thanks for reminding me the possibility of using the trpz argument, I
> think I will use that.
>
> Cheers,
> Sylvain
>

> ------------------------------------------------------------------------
> *De: *"Kai Muehlbauer" <kai.mue...@uni-bonn.de>
> *À: *"wradlib-users" <wradli...@googlegroups.com>
> *Envoyé: *Mercredi 11 Mai 2022 10:48:41
> *Objet: *Re: Clutter filtering

[Sylvain Watelet]

Hi Kai,

Ok, I will do that for the default values of the weights and the trapezoids, then Maik could add something on the sources if he wants to.

Cheers,

Sylvain

---

De: "Kai Muehlbauer" <kai.mue...@uni-bonn.de>
À: "wradlib-users" <wradli...@googlegroups.com>

Envoyé: Mercredi 11 Mai 2022 11:21:43
Objet: Re: Clutter filtering

To unsubscribe from this group and stop receiving emails from it, send an email to wradlib-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/wradlib-users/86b7371d-c2a3-8616-c4e5-5a230d6e5c8d%40uni-bonn.de.