NOTE: The nws-changes mailing list is no longer active. The list archives are made available for historical reasons.
--==========1644559653=========
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
---------- Forwarded Message ----------
All,
There is a change to the AVN/MRF suite which will be transparent to
all users in the beginning but is the first step of a process NCEP is
taking to simplify the runs of the NCEP global model. What will happen
for the 12Z run on 5 March 2002 is that the AVN runs will be extended to
the same length as the MRF runs (out to 384 hours). While people who
directly download the GRIB files will benefit immediately to this
change, it will eventually affect all people who use the MRF forecasts.
As the AVN forecasts have become more popular, there is more demand
on the length of the forecast. We currently run the AVN to 126 hours at
00Z and 12Z and to 84 hours at 06Z and 18Z. The weather service is going
toward 4-times daily update of weather to seven days both for the NCEP
service centers but also to the WFOs. Given that the MRF runs high
resolution forecasts to seven days and low resolution to 16 days, the
extension of the AVN beyond five days would make the MRF product
duplicating the AVN for the most part (the low resolution part of the
MRF uses much less resource compared to the high resolution part). We
compared the AVN forecasts and the MRF forecasts for one whole year to
determine if replacing the MRF forecasts with the AVN forecasts will
degrade the forecast skill and we found that the skills of the two
forecasts are practically the same. The only difference between the AVN
forecast and the MRF forecast is that we wait 2 hours and 45 minutes for
observations to arrive at NCEP before starting the AVN runs and 6 hours
for the MRF. Since we start both forecasts with the best first guess
from our global data assimilation system, this one-time difference in
the cutoff time is not crucial to the forecast skill (the explanation
would take too much space here). Furthermore, we will be issuing the 06Z
AVN not long after the normal MRF forecasts are issued. The later
forecasts are usually better than the earlier ones. For NCEP, once we
start to issue AVN forecasts to seven days, it makes no sense to keep
the MRF. It also makes little difference if we extend the AVN to 16 days
in low resolution.
Many people would ask if we have any skill credential to issue two
week forecasts every six hours. Our strategy is to make these forecasts
part of our ensemble strategy. Given that week-two forecasts can only be
used as outlooks and the skill of any individual forecast is very low,
we can only hope to use the ensemble strategy to provide information
about the atmosphere regimes. Running all the ensemble at 00z or running
them at 4 different times are just different strategies. Later this year
when we get more computer resource, we plan to run the global ensemble
four times daily with ten members per run (a doubling of our ensemble
members).
As for the MRF, we plan to stop running that in April. We will make
all the files currently made by MRF from the 00Z AVN runs so users will
not be inconvinienced and will have lots of time to change over to the
AVN products.
Finally, there has been a lot of confusions about the names AVN and
MRF. By streamlining the forecasts to one single suite, we hope to
reduce that confusion. We plan to rename our forecast suite from AVN to
the Global Forecast System (GFS) at a later time and we will provide
ample lead time before we make the change. Please feel free to write to
me with any question you may have about this change.
Hua-Lu Pan
---------- End Forwarded Message ----------
--==========1644559653=========
Content-Type: message/rfc822; name="avn/mrf merger"
Received: from loft.ncep.noaa.gov (loft.ncep.noaa.gov [140.90.193.45])
by unidata.ucar.edu (UCAR/Unidata) with ESMTP id g24McOK26173
for <lmiller@xxxxxxxxxxxxxxxx>; Mon, 4 Mar 2002 15:38:24 -0700 (MST)
Keywords: 200203042238.g24McOK26173
Received: from noaa.gov ([140.90.192.169]) by loft.ncep.noaa.gov
(Netscape Messaging Server 4.15) with ESMTP id GSH07W00.MMD;
Mon, 4 Mar 2002 17:38:20 -0500
Sender: wd23ph@xxxxxxxxxxxxxxxx
Organization: DOC/NOAA/NWS/NCEP/EMC
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.0.35 i586)
X-Accept-Language: en, zh-TW
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="==========1644577048=========="
--==========1644577048=========
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
All,
There is a change to the AVN/MRF suite which will be transparent to
all users in the beginning but is the first step of a process NCEP is
taking to simplify the runs of the NCEP global model. What will happen
for the 12Z run on 5 March 2002 is that the AVN runs will be extended to
the same length as the MRF runs (out to 384 hours). While people who
directly download the GRIB files will benefit immediately to this
change, it will eventually affect all people who use the MRF forecasts.
As the AVN forecasts have become more popular, there is more demand
on the length of the forecast. We currently run the AVN to 126 hours at
00Z and 12Z and to 84 hours at 06Z and 18Z. The weather service is going
toward 4-times daily update of weather to seven days both for the NCEP
service centers but also to the WFOs. Given that the MRF runs high
resolution forecasts to seven days and low resolution to 16 days, the
extension of the AVN beyond five days would make the MRF product
duplicating the AVN for the most part (the low resolution part of the
MRF uses much less resource compared to the high resolution part). We
compared the AVN forecasts and the MRF forecasts for one whole year to
determine if replacing the MRF forecasts with the AVN forecasts will
degrade the forecast skill and we found that the skills of the two
forecasts are practically the same. The only difference between the AVN
forecast and the MRF forecast is that we wait 2 hours and 45 minutes for
observations to arrive at NCEP before starting the AVN runs and 6 hours
for the MRF. Since we start both forecasts with the best first guess
from our global data assimilation system, this one-time difference in
the cutoff time is not crucial to the forecast skill (the explanation
would take too much space here). Furthermore, we will be issuing the 06Z
AVN not long after the normal MRF forecasts are issued. The later
forecasts are usually better than the earlier ones. For NCEP, once we
start to issue AVN forecasts to seven days, it makes no sense to keep
the MRF. It also makes little difference if we extend the AVN to 16 days
in low resolution.
Many people would ask if we have any skill credential to issue two
week forecasts every six hours. Our strategy is to make these forecasts
part of our ensemble strategy. Given that week-two forecasts can only be
used as outlooks and the skill of any individual forecast is very low,
we can only hope to use the ensemble strategy to provide information
about the atmosphere regimes. Running all the ensemble at 00z or running
them at 4 different times are just different strategies. Later this year
when we get more computer resource, we plan to run the global ensemble
four times daily with ten members per run (a doubling of our ensemble
members).
As for the MRF, we plan to stop running that in April. We will make
all the files currently made by MRF from the 00Z AVN runs so users will
not be inconvinienced and will have lots of time to change over to the
AVN products.
Finally, there has been a lot of confusions about the names AVN and
MRF. By streamlining the forecasts to one single suite, we hope to
reduce that confusion. We plan to rename our forecast suite from AVN to
the Global Forecast System (GFS) at a later time and we will provide
ample lead time before we make the change. Please feel free to write to
me with any question you may have about this change.
Hua-Lu Pan
--==========1644577048=========
Content-Type: text/x-vcard; charset=us-ascii; name="hualu.pan.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: "Card for Hua-Lu Pan"
Content-Disposition: attachment; filename="hualu.pan.vcf"; size=391
begin:vcard
n:Pan;Hua-Lu
tel;fax:301-763-8545
tel;work:301-763-8000x7234
x-mozilla-html:FALSE
org:EMC/NCEP/NWS/NOAA;Commerce
version:2.1
email;internet:hualu.pan@xxxxxxxx
title:Chief, Global Modeling Branch
adr;quoted-printable:;;Hua-Lu Pan=0D=0AW/NP23, RM 204 WWBG=0D=0ANOAA=0D=0A5200
Auth Road=0D=0A;Camp Spring;MD;20746-4304;USA
x-mozilla-cpt:;19904
fn:Hua-Lu Pan
end:vcard
--==========1644577048==========--
nws-changes archives: