yet another WRF12km for the Plains

[Janek Zimmer]

Hi everybody,

I hope my first posting in this forum is not misplaced - I didn't come across a subforum especially designated to numerical models, so that's why I'll put it in here.

I'm running a WRF configuration for the Plains region during the storm season. You can view the maps for the 06UTC run (only once a day) here:

WRF12km Plains:
http://www.karstenhaustein.com/Janek/maps/plains-WRF/index.php

It is mainly meant to provide just another ensemble member in the model chain. In my opinion, it is worth looking at different configurations (there are various other WRF suppliers out there), especially regarding the question "initiation or not?".

I am aware that it will not be able to compete with the excellent 4km-versions run by NCEP (see my WRF_hires maps at http://www.superzelle.de/maps.htm , under "Weltwetter", sorry, German only). But occasionally it might help to pinpoint the target area if you are still undecided after having seen all other models ;-)

Some numbers about it:
- 06UTC run, driven by GFS06UTC
- 12km grid cell size
- 0-48h forecasts
- update time: about 12:50UTC (07:50CDT)

Be sure to check out the 1h-loops at the bottom of the table. The cloud forecast seemed to work out quite well during the last days.
You can also find forecast soundings for numerous stations/cities in the Plains.
A complete variable list is linked at the bottom.

As a "quick+dirty verification", I also update the comparison between modeled and observed 24h-precipitation, as soon as the stageIV analysis comes in (~10UTC of following day). The link can be found at the very bottom as well. Looking at the two active days so far, it performed not too bad on April 22nd, but had its problems with the storm tracks on 23rd (see the rainfall ripples in the model over LA and AR - induced by the cumulus scheme, really bad example...).
Another problem is grid-scale convection in situations of strong forcing. The model develops MCS-like structures in those cases, do not verify in all cases. At least it's a good indicator for forcing...

I'm currently checking if the Grell-Devenyi Cumulus scheme performs better on average, so it is possible that there will be a configuration change within the next days. From the last days it seems that both schemes have their advantages, with the Kain-Fritsch being more aggressive regarding convective initiation.

So, that's it for now. If you have questions or suggestions, let me know. Moreover, I archive all the maps, if you need any...

Cheers,
Janek

 

[beaudodson]

Thanks! I will have to check this out for some of the upcoming events.

 

[Jason Persoff]

Well done website. Clean and quick with low bandwidth. Looking forward to putting this through its paces during my trip next month !

Would suggest you get rid of the "Copyright" feature on the bottom left of the images, or display it dynamically with the correct year. Currently all images display as Copyright 2008. This could confuse people thinking that it's a map from 2008 rather than 00z models. Sure, the timestamp in the left upper portion of the image should have most people tuned in to the right time and date, but I caught myself doublechecking .

 

[Greg Blumberg]

This is very nice! I will be using this in the near future!

 

[Janek Zimmer]

The WRF forecast has been reactivated today for this season, until the beginning of July (or so). The 48h forecast can be found here:
Plains-WRF_06Z

The daily update time will be ~9:00-9:20CDT (~14:00-14:20UTC).

I updated some of the plots, esp. the high clouds should be better now.

Please check out the composite plots as well:
TStorm Composite, example +18h:

Cloud Composite:

There are quite a few more plots to be found below the table (loops: the blue boxes), besides 24h-accumulations and forecast soundings.

The RR24h-verification (link at the bottom) for 2011 will be updated every 1 or 2 days then.

Greetz,
Janek

 

[Jeff Duda]

Nice setup, Janek. I'm curious what other model configurations you're using. I'm guessing you're using the Ferrier microphysics. Is this domain nested or is it a single domain? What PBL scheme are you using? I would suggest going with the KF CPS - it's arguably the best convective scheme for a mesoscale model.

I also have a question about the cloud plots you made in Grads. Did you use alpha shading on the colors to make some of them transparent? Are you using OpenGrads for that instead of the Linux package?

 

[Janek Zimmer]

Nice setup, Janek. I'm curious what other model configurations you're using. I'm guessing you're using the Ferrier microphysics. Is this domain nested or is it a single domain? What PBL scheme are you using? I would suggest going with the KF CPS - it's arguably the best convective scheme for a mesoscale model.

I also have a question about the cloud plots you made in Grads. Did you use alpha shading on the colors to make some of them transparent? Are you using OpenGrads for that instead of the Linux package?

Hello Jeff,
I'm using the Lin mp since it combines low computational cost with an acceptable precip bias (the WSM's are too wet in winter here). PBL is YSU, Cumulus scheme is indeed KF (Grell produces slightly better objective scores on average, but I like the KF pattern more, esp. with convective initiation). It's a 2way-nested setup to reduce boundary effects.
I'm using Opengrads because of higher speed and the enabled UDF's. The cloud composite is merged from the 3 cloud level plots (non-transparent png files) via a well-known linux command-line tool ;-)

Cheers into the Plains (where I can't be before next season...),
Janek