Page 1 of 2 1 2 >
Topic Options
#1462591 - 26/04/2018 14:18 FV3-GFS
ashestoashes Offline
Weather Freak

Registered: 22/10/2017
Posts: 629
Loc: Voyager Point (South West Sydn...
Hi all,
I noticed on tropical tidbits that they had the new experimental FV3-GFS model available although for only the US. I was wondering when it will become operational on our side of the planet, also how it will change the game in respect to weather modelling. Because from some research I saw that it can create resolutions of as high as 2km, but it counters it by dropping the resolution in other areas.

Top
#1463480 - 09/05/2018 21:59 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
A2A, I don't think Fv3-GFS is operational anywhere yet, but happy to be corrected.
Some experimental outputs are available for our region but only viewable at global scale.
http://data1.gfdl.noaa.gov/fvGFS/fvGFS_p...ow_slp_wind_6hr
I occasionally look at this for the total water path output.
Double check what the options are when using this site.

Top
#1465391 - 11/06/2018 06:55 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
I found this website where it is possible to view FV3-GFS experimental forecasts for Australia.
https://lab.weathermodels.com/models/gfs/fv3gfs.php

Top
#1465392 - 11/06/2018 08:11 Re: FV3-GFS [Re: ashestoashes]
retired weather man Offline
Meteorological Motor Mouth

Registered: 01/07/2007
Posts: 5016
Loc: Wynnum
Thanks Flowin.
_________________________
Wyn Nth 2019-Jan11.4(160),Feb47.0(146),Mar220.8(130)YTD279.2(436)

Top
#1465398 - 11/06/2018 09:10 Re: FV3-GFS [Re: ashestoashes]
ashestoashes Offline
Weather Freak

Registered: 22/10/2017
Posts: 629
Loc: Voyager Point (South West Sydn...
Thanks mate.

Top
#1465400 - 11/06/2018 10:16 Re: FV3-GFS [Re: ashestoashes]
ozone doug Offline
Weatherzone Addict

Registered: 06/11/2006
Posts: 1908
Loc: Roma SW QLD Eye to the West...
That,s great Flowin. I have book marked it.
_________________________
Cheers Doug. 491 Doug/ uhf ch40 When severe weather
BOM Stormspotter G0388 Roma S W Queensland Formerly Redcliffe.

Top
#1469661 - 28/08/2018 18:56 Re: FV3-GFS [Re: ashestoashes]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
I only just stumbled across this thread so thought I might throw my two cents worth in.

The FV3 (or more broadly, the name of the dynamical core of a bunch of models, one of which is due to replace the existing version of GFS later next year) is still technically in experimental stage so it's not operational yet.

It's been providing data for the whole globe for awhile now but there's only a few websites I know of that offer forecasts that cover our area and even then, it's usually restricted to a limited range of forecast products.

I've recently written some scripts to display realtime FV3 data for our region - see example below. One of its advantages over GFS is that it greatly reduces the often unrealistically intense rainfall bullseyes in convective setups that's plagued GFS for years.
Also, objective skill score stats for FV3 during its testing shows that it's performed significantly better than the current GFS (on average) in most parameters and regions including hurricane tracks and intensities although still not as good as EC and UK. Of course there'll always be some setups when another less skillful model or models will beat it.



Top
#1469758 - 30/08/2018 20:15 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
Thanks Ken,
That is a useful comparison. Your scripts to process the FV3 model outputs are good and would be helpful as while the FV3 is in the experimental / testing phase it is hard to get visualisation of it at meaningful scale for Australia.
I have been watching the precipitable water outputs of FV3 for a while and find the smaller resolution of FV3 more helpful to understand the likely moisture feed source and direction for some weather systems. I have an interest in this for atmospheric rivers which I have discussed in another thread.

Top
#1469761 - 30/08/2018 21:05 Re: FV3-GFS [Re: ashestoashes]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD
You'd have to hope FV3-GFS does better than GFS in the tropics too because I've lost count how many typhoons it's bombed to ridiculous pressures this year (I'm talking as low as 900 hpa for some) while no other model suggests such a thing. It has been extremely bad in that respect and what's so embarrassing is all of those supposed upgrades to the model haven't seemed to have helped in that area at all.

Top
#1471909 - 03/10/2018 07:27 Re: FV3-GFS [Re: ashestoashes]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD
Found this on Twitter, it shows the general timeline for the upgrade from GEFS (GFS Ensemble) to FV3GEFS through next year and into 2020:



All I can say is it's about bloody time.

Top
#1471910 - 03/10/2018 07:29 Re: FV3-GFS [Re: ashestoashes]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
You can also find the latest updates to that implementation timeline on the official website for the FV3 here: http://www.emc.ncep.noaa.gov/users/Alicia.Bentley/fv3gfs/

Top
#1471915 - 03/10/2018 08:12 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
Thanks for that link Ken. Some good information there.
I noted this presentation: http://www.emc.ncep.noaa.gov/users/Alicia.Bentley/fv3gfs/updates/MEG_9-27-18_FV3GFS_EVAL.pptx
Seems to suggest that FV3GFS may better than current GFS in relation to not overcooking cyclone intensity.
_________________________
Models are for estimating and gauges are for knowledge.

Top
#1474546 - 23/10/2018 20:51 Re: FV3-GFS [Re: Flowin]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
Originally Posted By: Flowin
I found this website where it is possible to view FV3-GFS experimental forecasts for Australia.
https://lab.weathermodels.com/models/gfs/fv3gfs.php

I discovered today that the link I posted earlier to FV3GFS is now not easily accessible.
If anybody knows where FV3GFS forecasts for Australia are available lets us know

Top
#1474547 - 23/10/2018 20:57 Re: FV3-GFS [Re: ashestoashes]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
That reminds me, Iíll post a skill score graph of FV3 vs other models tomorrow.
I was meaning to post it a few weeks ago but never got around to it.

Top
#1474568 - 24/10/2018 11:01 Re: FV3-GFS [Re: ashestoashes]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
The 1st image below is the skill scorecard for the current version of FV3 compared to the current GFS using the commonly used anomaly correlation technique. I probably wouldn't read too much into the MSLP comparison because MSLP scores can be quite erratic and often isn't the best representation of the overall skill of a model.

The 2nd image below is a skill score graph comparing FV3 to some other global models - it shows the die-off in skill of the models as a function of forecast lead time out to 10 days for runs averaged across the last 30 days (up to the 22nd Oct) for 850hpa winds in the southern hemisphere extratropics covering the latitudes between 20 and 80 degrees South. Note that the rainfall skill of overseas models isn't usually verified here in Australia until they've become fully operational for a sufficient period of time.

The higher the skill values (using anomaly correlation), the better.

Red = EC, orange = UK, purple = FV3, black = current GFS, blue = NAVGEM, JMA = grey, green = CMC, turquoise = CFSR.

The relative skill rankings of these models are generally similar across most regions, time of year, and weather parameters.

As you can see, there's been a big improvement in accuracy of the current version of FV3 over the current GFS. Even though it still isn't up there with the likes of EC, it looks promising so far.
I've been subjectively comparing FV3 forecasts with those from the current GFS for rainfall distribution and amounts in SE QLD/NE NSW for a little while now and from what I've noticed so far, the better skill of FV3 has been quite noticeable.
The fact that the FV3 almost eliminates unrealistically intense CAPE and rainfall bullseyes that have plagued the current version of GFS is also a bonus.

The FV3 doesn't seem to be as good for TC track forecasts outside of the Atlantic basin but that's just my subjective impression from some recent typhoons and eastern Pacific hurricanes so it could be wrong. As for inside the Atlantic basin, here's some skill graphs (red = FV3, green = GFS): http://www.emc.ncep.noaa.gov/gmb/emc.glopara/vsdb/prfv3rt1/www/track/track.html

But as always, there'll always be some individual cases when the FV3 will flop and other model/s will beat it... just like any model or forecast in general.
A person from the FV3 modelling team told me a few weeks ago that while they'll now try and keep the coding of the FV3 frozen until it becomes operational, they can't guarantee it until the code is officially frozen. So I hope they don't mess around with it too much (don't ruin a good thing) before it becomes operational unless they can make another substantial improvement in accuracy.








Edited by Ken Kato (24/10/2018 11:08)
Edit Reason: added stuff

Top
#1474570 - 24/10/2018 11:33 Re: FV3-GFS [Re: ashestoashes]
ashestoashes Offline
Weather Freak

Registered: 22/10/2017
Posts: 629
Loc: Voyager Point (South West Sydn...
Ken on TC tracks FV3-GFS has the ability to alter the resolution over a region , with it being a US model it's very likely to focus on the Atlantic Basin. But come our TC season it's is very likely there will be improved resolution across the Indian Ocean and South Western Pacific, time will tell though.

Top
#1474572 - 24/10/2018 11:48 Re: FV3-GFS [Re: ashestoashes]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
ashestoashes: The subjective impression of non-Atlantic systems I was referring to was actually based on forecasts from the global FV3 that essentially had the same resolution for the western and eastern Pacific basins as the Atlantic basin (not a higher resolution nested version dedicated for the Atlantic basin).

Top
#1474587 - 24/10/2018 19:46 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
Ken thanks for your input, much appreciated.
While forecasting is a skilled art, verifying is equally challenging.
I look forward to seeing FV3 becoming operational as I would like to see what it generates at a regional scale and whether it can get the location of rain more accurate (spatial displacement error reduced)

Top
#1475122 - 02/11/2018 14:25 Re: FV3-GFS [Re: ashestoashes]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD
So, does anyone have a public link to FV3 yet since Weathermodels decided to paywall it?

Top
#1475143 - 02/11/2018 17:13 Re: FV3-GFS [Re: ashestoashes]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD
Well, I found one...no precip and only 00z and 12z runs though: http://www.pivotalweather.com/model.php?m=gfs_fv3&p=sfcwind_mslp&rh=2018110200&fh=0&r=au&dpdt=

Top
#1481973 - 21/12/2018 11:36 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
I thought one of the changes with FV3 was that the unrealistically deep minimum lows forecast by GFS would be improved. Some have commented that FV3 may have gone too far to the opposite extreme and not allowing enough formation of deep lows. We saw with TC Owen that FV3 deviated significantly from GFS.

Well now look at what FV3 has cooked up in the 20 Dec 12Z run at about 300 hours onwards, a deep low NE of New Caledonia, when the 12Z run and forecast hours from GFS are much more moderated.
http://www.pivotalweather.com/model.php?m=gfs_fv3&p=sfcwind_mslp&rh=2018122012&fh=324&r=au&dpdt=

Before anybody says it, I do acknowledge that anything beyond about forecast 96-120 hours can often be quite speculative.

Top
#1490486 - 15/02/2019 16:07 Re: FV3-GFS [Re: ashestoashes]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD

Top
#1490488 - 15/02/2019 16:20 Re: FV3-GFS [Re: Mega]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
That's about the only advantage of our region not regularly getting large amounts of snow laugh

If we did, it'd be a problem but from the many comparisons I've made between FV3 and current GFS for rainfall amounts and placement in our region, I've found the FV3 has been far superior to GFS especially in terms of not putting in the unrealistically intense precip bullseyes all over the place.
But that's for short range rainfall forecasts (24hrs or less) so I'm not sure if the same level of better skill applies to longer range forecasts for our region.

One thing I will say though is that I still have reservations about how well the FV3 does with tropical systems in the Australian region and their associated heavy rainfall. I'd prefer to make more comparisons before I can draw any conclusions from that aspect of it.

The FV3 was originally due to become fully operational on the 29th Jan but the government shutdown interfered with that (at least all parties were given ample warning of that though).

Top
#1490490 - 15/02/2019 16:32 Re: FV3-GFS [Re: Ken Kato]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
Originally Posted By: Ken Kato

One thing I will say though is that I still have reservations about how well the FV3 does with tropical systems in the Australian region and their associated heavy rainfall. I'd prefer to make more comparisons before I can draw any conclusions from that aspect of it.

Agree. I have been looking at it for a while but not done any objective fair comparisons. But what I have seen does not inspire confidence. frown
_________________________
Models are for estimating and gauges are for knowledge.

Top
#1490497 - 15/02/2019 17:00 Re: FV3-GFS [Re: Flowin]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD
Originally Posted By: Flowin
Originally Posted By: Ken Kato

One thing I will say though is that I still have reservations about how well the FV3 does with tropical systems in the Australian region and their associated heavy rainfall. I'd prefer to make more comparisons before I can draw any conclusions from that aspect of it.

Agree. I have been looking at it for a while but not done any objective fair comparisons. But what I have seen does not inspire confidence. frown


It's been worse than GFS, hands-down.

And this is just my opinion, but GFS splattering convective rainfall all over the place is a pretty small price to pay if in the end the model is still overall more accurate for our part of the world. Some of the FV3 solutions I've seen over the past few months have been mind-boggling to say the least and it makes you wonder how it's so close to release.

Top
#1490499 - 15/02/2019 17:11 Re: FV3-GFS [Re: Mega]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969
Originally Posted By: Mega
And this is just my opinion, but GFS splattering convective rainfall all over the place is a pretty small price to pay if in the end the model is still far more accurate for our part of the world.

Dunno which model you were referring to by "the model" but personally, I'd much rather prefer the FV3 for Australia than GFS (but mainly for precip for short range forecasts) since rainfall amounts and placement are one of the vital bread and butter variables we need to routinely forecast here compared to high snowfall amounts. The extreme rainfall bullseyes that current GFS puts down are a pain and they also have flow-on effects such as increasing the model-averaged rainfall amounts in products like WATL rainfall and OCF.

Top
#1490504 - 15/02/2019 17:27 Re: FV3-GFS [Re: Ken Kato]
Mega Offline
Meteorological Motor Mouth

Registered: 03/02/2003
Posts: 7783
Loc: Maryborough, Wide Bay, QLD
Originally Posted By: Ken Kato
Originally Posted By: Mega
And this is just my opinion, but GFS splattering convective rainfall all over the place is a pretty small price to pay if in the end the model is still far more accurate for our part of the world.

Dunno which model you were referring to by "the model" but personally, I'd much rather prefer the FV3 for Australia than GFS (for things like precip for short range forecasts) since rainfall amounts and placement are one of the vital bread and butter variables we need to routinely forecast here compared to high snowfall amounts. The extreme rainfall bullseyes that current GFS puts down are a pain and they also have flow-on effects such as increasing the model-averaged rainfall amounts in products like WATL rainfall and OCF.


That is a good point and I can't argue with it at all. It will at least (hopefully) help clean up those silly bulls-eyes during big storm events. However, I was under the impression that FV3 did this by reducing the convective feedback for large storm clusters like tropical lows etc. From what I've seen it hasn't done that at all, and theoretically speaking, you'd think that issue would carry through to our part of the world in the sub-tropics during big storm events too.

The other thing is that I guess I'm just not convinced FV3's issues lie only on tropical activity through the Australian region. While it might end up being in the okay short-term, I've seen nothing yet to convince me it's any better in the mid to long-range unfortunately. This winter and '19 spring/summer will be a good test for it so hopefully it proves me wrong.

Top
#1490507 - 15/02/2019 17:35 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
For TC and TS weather events, I prefer HWRF but it only goes out to 120 hours. The range of detailed outputs is also much more useful like the Wind swath and rain swath.

Ken do you know if they have ever run HWRF longer than 120 hours, or plans to extend its lead time ?

Top
#1490526 - 15/02/2019 18:16 Re: FV3-GFS [Re: ashestoashes]
Ken Kato Offline
Meteorological Motor Mouth

Registered: 05/03/2012
Posts: 5969






Flowin: Note sure if they've ever run it out past 126hrs and I also don't know of any immediate plans to extend it beyond 126hrs either.

Mega: Its skill in the longer range depends on what parameter you're looking at and how you measure it really. For example, the above graphs are for anomaly correlation (one of the most common and widely used forms of statistical analysis of model skill) for MSLP in the southern hemisphere (1st graph) and the global tropics (2nd graph). They show the natural die-off in skill with forecast lead time. The higher the better using that particular skill measuring technique.
I don't believe anyone's done any objective skill analyses specifically for the Australian region yet though, let alone any QPF skill scores for our country. Once it becomes fully operational for awhile, it might be done.

Top
#1490781 - 18/02/2019 07:40 Re: FV3-GFS [Re: ashestoashes]
Flowin Offline
Weather Freak

Registered: 15/10/2017
Posts: 890
Loc: Pinjarra Hills, Qld
I noticed this morning that the FV3 model is now available on Meteologix. Good to have some access to it on a page where the map area can be changed.

https://meteologix.com/au/model-charts/gfs/1602-e-268-s/sea-level-pressure/20190217-2100z.html

Top
Page 1 of 2 1 2 >


Who's Online
16 registered (Rainbow Spirit, Snapper22lb, cold@28, mithrandir, Weary, bundybear, Dr Philosophy, wet sky, Farra, kizz, GringosRain, Birdie, Weathergrrl, 3 invisible), 104 Guests and 1 Spider online.
Key: Admin, Global Mod, Mod
Today's Birthdays
Dale Small, Michael Bath, Ronfishes
Forum Stats
29890 Members
32 Forums
24117 Topics
1521815 Posts

Max Online: 2985 @ 26/01/2019 12:05
Satellite Image