Jump to content

  • Member Statistics

    17,584
    Total Members
    7,904
    Most Online
    LopezElliana
    Newest Member
    LopezElliana
    Joined

Major upgrade coming to the Canadian Global Model

Ellinwood

Recommended Posts

Ellinwood

Ellinwood

Posted
Posted

http://dd.weatheroffice.ec.gc.ca/doc/genots/2013/02/07/NOCN03_CWAO_071810___00179

 

MAJOR UPGRADE TO THE GLOBAL DETERMINISTIC PREDICTION SYSTEM GDPS-
VERSION 3.0.0) AT THE CANADIAN METEOROLOGICAL CENTRE
ON WEDNESDAY FEBRUARY 13, 2013, STARTING WITH THE 1200 UTC RUN, THE
CANADIAN METEOROLOGICAL CENTRE (CMC) OF THE METEOROLOGICAL SERVICE
OF CANADA (MSC) WILL IMPLEMENT VERSION 3.0.0 OF ITS GLOBAL
DETERMINISTIC PREDICTION SYSTEM (GDPS), HEREAFTER REFERRED TO AS
GDPS-3.0.0. THE MAIN CHANGES OF THIS UPDATE ARE:

- CHANGES TO THE 4D-VAR DATA ASSIMILATION SYSTEM: - THE USE OF
ADDITIONAL REMOTE SENSING DATA: - AN INCREASE IN HORIZONTAL
RESOLUTION FROM 33 TO 25 KM: - IMPORTANT CHANGES IN THE GEM MODEL
PHYSICS: - AND 1200 UTC RUNS NOW DONE TO 240 HOURS AS FOR 0000 UTC
RUNS.
OBJECTIVE SCORES DONE IN DEVELOPMENT AND PARALLEL RUN PHASES SHOW
IMPROVEMENTS IN THE FORECASTS WITH MOST METRICS THROUGHOUT MOST OF
THE ATMOSPHERE, IN PARTICULAR OVER NORTH AMERICA IN WINTER. THESE
IMPROVEMENTS ARE OF AN ORDER OF MAGNITUDE USUALLY SEEN ONLY ONCE IN
A DECADE. A SUBJECTIVE EVALUATION BY OPERATIONAL METEOROLOGISTS AT
CMC CONFIRMED THOSE IMPROVEMENTS. OTHER FORECAST SYSTEMS WHICH
DEPEND ON GDPS OUTPUT ALSO BENEFIT FROM THE GDPS-3.0.0. CHANGES
WERE MADE TO THE REGIONAL DETERMINISTIC PREDICTION SYSTEM (RDPS) TO
HARMONIZE IT WITH THE NEW GDPS, AND THE RDPS FORECASTS ARE ALSO
IMPROVED AS A RESULT, SO ITS VERSION NUMBER IS ALSO INCREASED TO
3.1.0, AND IS HEREAFTER REFERRED TO AS RDPS-3.1.0.

Link to comment
Share on other sites
dtk

dtk

Posted
Posted

It's not like we don't WANT to... just need the resources (or at least this is from what I gather from the situation).

 

Actually, we a working toward a (hybrid) 4D-Ensemble-Var system (not exactly the same as 4DVAR, since it uses a 4D ensemble in stead of linearized model and its adjoint in the minimization process).  Lack of resources (computing AND manpower) is a huge issue (I think there are other threads on this and I have made various comments/replies along the way).

 

The 4D-ensemble-var type development is something also being embraced/pursued by the Canadians and UKMet Office.

Link to comment
Share on other sites
Mallow

Mallow

Posted
Posted

At least from the tropical/impact weather point of view, it's hard to argue the new radiation scheme introduced to the GFS in 2010 was a microscopic tweak

 

Haha, well sure. But increasing the resolution significantly, completely overhauling the data assimilation scheme, increasing the amount of satellite data ingested, and altering the physics parameterizations all in one update... comparatively speaking, that makes a simple update to a new radiation scheme look microscopic. :P

Link to comment
Share on other sites
dtk

dtk

Posted
Posted

Haha, well sure. But increasing the resolution significantly, completely overhauling the data assimilation scheme, increasing the amount of satellite data ingested, and altering the physics parameterizations all in one update... comparatively speaking, that makes a simple update to a new radiation scheme look microscopic. :P

 

They aren't completely overhauling their DA scheme (they've had 4DVAR for a while now).  This implementation is similar in scope to the one we did in 2010 (the one Adam is referring to).  That particular change was not just a "simple update to a new radiation scheme", but instead was a resolution increase AND physics overhaul ("all in one update")

http://www.nws.noaa.gov/os/notification/tin10-15aab_gfs.txt

 

 RADIATION AND CLOUD OVERLAP  GRAVITY WAVE DRAG HURRICANE RELOCATION NEW PLANETARY BOUNDARY LAYER SCHEME NEW MASS FLUX SHALLOW CONVECTION UPDATED DEEP CONVECTION SCHEME POSITIVE DEFINITE TRACER TRANSPORT SCHEME

These types of changes generally occur in coordination with computer upgrade cycles, since they require better (or more) hardware.  The Canadians got a big computing upgrade in the last year, and this fits right in with that.

 

The GFS/GDAS also had major upgrade last year to the DA scheme with the hybrid ensemble-var implementation:

http://www.nws.noaa.gov/os/notification/tin12-22gfs_hybridaab.htm

 

Our new operational supercomputer goes live this summer, and accordingly, we are planning a significant change to the GFS (including a new dynamical core [semi-Lagrangian], resolution increase, many physics changes, etc.).

Link to comment
Share on other sites
Mallow

Mallow

Posted
Posted

They aren't completely overhauling their DA scheme (they've had 4DVAR for a while now).  This implementation is similar in scope to the one we did in 2010 (the one Adam is referring to).  That particular change was not just a "simple update to a new radiation scheme", but instead was a resolution increase AND physics overhaul ("all in one update")

http://www.nws.noaa.gov/os/notification/tin10-15aab_gfs.txt

 

 RADIATION AND CLOUD OVERLAP  GRAVITY WAVE DRAG HURRICANE RELOCATION NEW PLANETARY BOUNDARY LAYER SCHEME NEW MASS FLUX SHALLOW CONVECTION UPDATED DEEP CONVECTION SCHEME POSITIVE DEFINITE TRACER TRANSPORT SCHEME

These types of changes generally occur in coordination with computer upgrade cycles, since they require better (or more) hardware.  The Canadians got a big computing upgrade in the last year, and this fits right in with that.

 

The GFS/GDAS also had major upgrade last year to the DA scheme with the hybrid ensemble-var implementation:

http://www.nws.noaa.gov/os/notification/tin12-22gfs_hybridaab.htm

 

Our new operational supercomputer goes live this summer, and accordingly, we are planning a significant change to the GFS (including a new dynamical core [semi-Lagrangian], resolution increase, many physics changes, etc.).

 

I stand corrected. ^_^

Link to comment
Share on other sites
Jim Marusak

Jim Marusak

Posted
Posted

talking about computing "horsepower" of the NWS compared to the CMC, Japan, the ECMWF, etc; in the US we're at a big power dis-advantage when it comes to weather computing.

 

according to top500.org, as of last october, here's how it ranked...

 

NWS - main model computer: #369

NWS - research side : #138

CMC - #90

Korean Weather Agency: #55

ECMWF: #35

 

as for who was #1, it was the Department of Energy at Lawrence Livermore

 

 

I have to admit, what if the NWS has the same mainframe horsepower as say the ECMWF? or even Canada? what would they be able to do model-wise with that? because right now it's like we're going with a big block V-8 against V-10 intercooled twin-turbos and W-16 intercooled quad turbos, if we were using racing equivalents between the met agencies.

Link to comment
Share on other sites
ohleary

ohleary

Posted
Posted

talking about computing "horsepower" of the NWS compared to the CMC, Japan, the ECMWF, etc; in the US we're at a big power dis-advantage when it comes to weather computing.

 

according to top500.org, as of last october, here's how it ranked...

 

NWS - main model computer: #369

NWS - research side : #138

CMC - #90

Korean Weather Agency: #55

ECMWF: #35

 

as for who was #1, it was the Department of Energy at Lawrence Livermore

 

 

I have to admit, what if the NWS has the same mainframe horsepower as say the ECMWF? or even Canada? what would they be able to do model-wise with that? because right now it's like we're going with a big block V-8 against V-10 intercooled twin-turbos and W-16 intercooled quad turbos, if we were using racing equivalents between the met agencies.

 

And we run about 50 models instead of a handful.

Link to comment
Share on other sites
brettjrob

brettjrob

Posted
Posted

Wouldn't it be better to run a handful that were excellent, instead of the 50 that many of are terrible.

 

It's not quite that simple. Many of them serve disparate purposes, such as the RAP and HRW for higher-resolution short-term forecasts. Then you have ensembles in the GEFS and SREF which provide a lot of value to forecasters, even if they aren't as popular with hobbyists/weenies due to their interpretation being more cumbersome and time-consuming.

 

The efficiency of running the global model out to 384 hours four times a day might be up for debate, but even if it were slashed to twice a day, I doubt the resources freed up would be enough for a revolutionary increase in the resolution and computational cost of the GFS.

Link to comment
Share on other sites
donsutherland1

donsutherland1

Posted
Posted

talking about computing "horsepower" of the NWS compared to the CMC, Japan, the ECMWF, etc; in the US we're at a big power dis-advantage when it comes to weather computing.

 

according to top500.org, as of last october, here's how it ranked...

 

NWS - main model computer: #369

NWS - research side : #138

CMC - #90

Korean Weather Agency: #55

ECMWF: #35

 

as for who was #1, it was the Department of Energy at Lawrence Livermore

 

 

I have to admit, what if the NWS has the same mainframe horsepower as say the ECMWF? or even Canada? what would they be able to do model-wise with that? because right now it's like we're going with a big block V-8 against V-10 intercooled twin-turbos and W-16 intercooled quad turbos, if we were using racing equivalents between the met agencies.

 

IMO, NOAA needs a strong director who will advocate effectively for the NWS. Looking at the list, the DOE has the #1, #2, #4, #18, #19, #22, and #47 supercomputers, not to mention additional supercomputers in the top 100. If public safety, aviation safety, and disaster preparedness are high priorities for the NOAA's director, there is no reason that the NWS should not have at least a top 100 supercomputer, even one comparable to the ECMWF's. Other federal institutions have such computers. IMO, given its mission and the need for complex computational workloads, NWS should, as well.

Link to comment
Share on other sites
weatherwiz

weatherwiz

Posted
Posted

Where can we find the new 12z data out to 240 hours?  I don't see on the met office site I have bookmarked.

 

I'll second this question.  Now that it seems like this model could be even more valuable to look at, what's the best link to view this?  I see they have a link to it on EWAL but just wondering if there are other places?  

Link to comment
Share on other sites
Chicago WX

Chicago WX

Posted
Posted

Where can we find the new 12z data out to 240 hours?  I don't see on the met office site I have bookmarked.

 

 

I'll second this question.  Now that it seems like this model could be even more valuable to look at, what's the best link to view this?  I see they have a link to it on EWAL but just wondering if there are other places?  

 

http://meteocentre.com/models/models.php?mod=gemglb&map=na&run=12〈=en'>http://meteocentre.com/models/models.php?mod=gemglb&map=na&run=12〈=en

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Go to topic listing

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...