EMC's Model Evaluation Group

 The Model Evaluation Group (MEG) community is established to interact with the user community on issues related to the forecast systems of NCEP's Environmental Modeling Center (EMC).    It is used to discuss overall model performance and provide feedback on operational and parallel versions of the models which comprise the production suite.

Forums

angle-left Back

Precipitation forecasts in GFSX

GW
Glenn White, modified 1 Year ago.

Precipitation forecasts in GFSX

Youngling Posts: 3 Join Date: 1/12/15 Recent Posts
Email fromAndy E:
 
I have been watching the new GFS on the WR side by side web page -- initially the two models looked pretty much the same -- but now that the synoptic regime has become more active during the last week -- there are some notable differences - especially during the last 3-5 days
 
As a reminder
  • I like to click on the Dprog/dt button -- upper left corner
  • Then step through the animation by hand  --
 
Note:  as always -- please double check the times -- the data acquisition scripts to pull data from NCEP are not infallible
 
Question for NCEP:  This is subjective feedback -- but some of the forecast solution bouncing around and differences between the two models is pretty significant.  
  • Please look at the surface and precip forecast fields, the new GFS model  precipitation amounts can vary a lot run to run and be much different at times from the current GFS.   The new GFS solutions  seems to bounce around more than the op GFS version.   I believe more than you can write off as natural model variability.  
Some of the WR STID folks have noticed similar differences this week.   
 
I am suspicious that the  data assimilation first guess fields are  more different (run to run) between the two models than we expected. 
 
 NCEP -- you have some better diagnostic tools...what is your interpretation?   Is there an issue with moisture data assimilation?
 
Hopefully this gets a conversation started 
GW
Glenn White, modified 1 Year ago.

RE: Precipitation forecasts in GFSX

Youngling Posts: 3 Join Date: 1/12/15 Recent Posts

Equitable threat scoeres for the GFSX seem higher than the operational, but so is the standard deviation of 120 hr forecasts of precip every 6 hours.  Th latter is consistent with less continuity.

EMC will try to look at this.

 

Glenn White

 

GW
Glenn White, modified 1 Year ago.

RE: Precipitation forecasts in GFSX

Youngling Posts: 3 Join Date: 1/12/15 Recent Posts
Based on 13 months of retrospective so far, for thresholds of 0.2 to 15 mm and forecast lengths of 0-24 to 84-108 hrs, seventy of 120 categories  of precipitation forecasts show statistically significant improvement in the GFSX compared t the GFS.

Navigation Menu

What's New

Intro to MEG Evaluation of FV3GFS Retrospective Cases

Hi all, For those who may have missed yesterday's MEG meeting, we discussed an overview of our plans for evaluating the FV3GFS retrospective cases. The covered information included a...

RE: Better handling of high-level Hurricane moisture outflow?

Hi Tim, Thanks for posting this case!  It's good to see that the FV3 performance was better overall here, and given it was consistently drier than the GFS for several runs, it...

Better handling of high-level Hurricane moisture outflow?

Here is something completely different.  For Fire Weather in the Intermountain West, it is often very subtle model differences that are VERY important.    When Fabio was decaying off the...

fabioclouds.gif

Image

FabioVerifa.png

Image

FabioVerif.png

Image

Fabio10a.png

Image

Fabio09a.png

Image

Fabio08a.png

Image

Fabio07a.png

Image

Fabio06a.png

Image

Fabio05a.png

Image

Fabio04a.png

Image

Fabio03a.png

Image

Fabio02a.png

Image

Fabio11a.png

Image

Fabio01a.png

Image

RE: More evidence of FV3 being a bit wetter...

I believe I spoke too soon , and too inconclusively, in stating that this Southeast TX event was a clear indication that the FV3 was wetter in its lead-up forecasts than the GFS. ...

8plusinches.PNG

Image

GEFS_QPFplume_00Z_Jul2_2018.png

Image