The NCEP/DOE AMIP-II Reanalysis (Reanalysis-2) is based on the widely used
NCEP/NCAR Reanalysis. The goal of
Reanalysis-2 is to improve upon the NCEP/NCAR Reanalysis by fixing
the errors and by updating the parameterizations of the physical
While we think that Reanalysis-2 is better; we know that it
should not be considered the "next generation" of reanalysis.
Improvements such as higher horizontal and vertical resolution,
direct assimilation of radiances, (proper) use of SSM/I data,
assimilation of rainfall data, need to be incorporated.
9/2014: nomad3 and nomad4 were taken off line. Current R2 can be obtained from
Older analyses can be obtained from
ESRL and NCAR.
12/22/2011: bad snow analyses was assimulated for 12/18/2011 and 12/19/2011. In the process of rerunning.
Expect the rerun to be completed early next year.
7/15/2008: A compile problem with R2 source code was discovered.
The April-June 2008 analyses and associated data have been redone.
The SSI code was recompiled April 3, 2008 to increase the size of the
observation arrays. The source code changes were checked by two other people.
However, it is believed that the code was compiled using an old BUFR
library. The fields had the right large scale structure (warm tropics,
cold poles) but had larger than expected deviations from CDAS. The code was
recompiled May 28, 2008 with the new BUFR library. After May 28th,
the deviations from CDAS returned to normal.
Please replace all R2 analyses downloaded from NOMADS before July 15th with
the updated versions. Further details will be posted later.
4/1/2005: Sea-Ice Problem
required a rerun (00Z 8/1/2005 to 12Z 3/28/2005). This same
problem affected the NCEP/NCAR Reanalysis, GDAS and the operational
global forecasts (GFS).
11/16/2004: The following files on nomad3.ncep.noaa.gov were
revised on 11/16/2004:
Please update these files. Was a glich when ownership of the official R2 was moved to
NCO operations. The script that downloaded data to nomad2 stil used my copy of R2.
5/2003: new R-2 data
Now both sigma and sfcanl files are on-line for people who want
to run models (GSF/MRF and RSM).
11/2002: R-2 summary paper published in BAMS.
NCEP-DOE AMIP-II Reanalysis (R-2): M. Kanamitsu, W. Ebisuzaki,
J. Woollen, S-K Yang, J.J. Hnilo, M. Fiorino, and G. L. Potter.
1631-1643, Nov 2002, Bul. of the Atmos. Met. Soc.
10/1/2002: new R-2 data
has 4x daily analyses.
port of the R2 system to our IBM/SP. Redid runs starting
from 1/2001 (missing files with the Cray analyses).
Schedule: expect to run previous month 1-2 weeks after the month
NCEP users: 1979-2000 6-hourly sigma and flux files are on HSM at
/u/wx51we/hsm/r2/sigma (or flux). The 6-hourly first-guess sigma
files are in the process of being downloaded (needed for RR).
Outside users can get the data from NCAR and CDC (Boulder, CO).
1/11/01: A problem was found in the program that created the monthly
means. The mean values are correct but the GRIB header is flawed.
Naturally GrADS and wgrib users are unaffected by the problem. If
they were, the problem would have been fixed during the software
development stage. For those who want the gory details, the
scan mode is wrong for the 2.5x2.5 grids and the longitude
increment is wrong for the Gaussian grids. A program to
fix the headers will be made available.
12/22/00: NCAR (Roy Jenne's group) has R2 analyses.
12/19/00: Computed 30 year (79-98) climatologies for the flx and pgb
monthly means. These climatologies are available for download and
have been integrated into the web plotting system. Now you can plot
R2 anomalies from climatology.
12/1/2000: Analyses from August 1998 to February 1999 have been
redone. By mistake, automatic programs have downloaded the old
March 1999-December 1999 analyses. Please eliminate all copies
of August 1998+ data that were obtained prior to Nov 16, 2000 and
March 1999+ data that were obtain prior to Dec 2, 2000.
10/17/2000: All analyses from August 1998 are being redone.
More problems found with converting the surface fields (snow and sea-ice)
to the R2 assimilation system.
7/27/2000: A computer glich was uncovered in 06Z 19 March 1982,
flx, dg3, znl files. The PRATE was corrupted with a unrealistic
wavey pattern and the problem went away when the model was rerun.
We suspect a hardware glitch. The continuity of the assimilation
was not affected as 3/19/82 is part of the "rerun" assimilation (1/79-9/82).
(A change in the stratiform cloud tuning was done mid-way through the
assimilation, Since the change in the cloud tuning would have only a
small effect on the pressure level analyses, it was
decided to keep the sigma/pgb analyses and just run the forecast
model with the new cloud tuning.)
6/22/2000: All analyses from August 1998 to March 2000 are being redone.
The sea-surface temperature data set was not converted from Celcius
1/1/2000: The analyses for 1996 are being redone. An analysis
of the observation counts found that all the observational data were
not being incorporated into the analyses. Please discard all 1996
analyses obtained prior to 1/7/2000. (Between 1/1/00 and 1/6/2000,
the cron jobs were downloading the old analyses.)
Reanalysis-2 is running in near real time.
Obtaining the Analyses
The recent 4x daily data is available on-line from
The 1979-present data can be obtained from
Version C: The cloud temperatures should have been set to undefined in cloud
free regions. Unfortunately the cloud temperatures were set
to zero. When the monthly means were computed, the zero
were used making the cloud tempertures too low. A similar problem
also affects the cloud top pressure, cloud base pressure,
The only on-line cloud field that is unaffected by this problem is
the cloud fraction (TCDC).
(7/99) The flux files have a problem with two bad entries in the GDS section
of the GRIB message. Most programs do not use these entries but
copygb does. We have a program to rewrite the entries if anyone
wants it. BTW we plan to fix the GDS.
(7/99) The monthly averages for 199205 were affected by a computer
problem. The files have been updated.
The cloud temperature can be calculated from the 4x daily file by
The precipitation adjustments are stored the fields "BGRUN". If
the adjustment adds water, BGRUN is positive.
For more details (primary paper to cite), see,
Kanamitsu, M., W. Ebisuzaki, J Woollen, S.-K. Yang,
J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002:
NCEP-DOE AMIP-II Reanalysis (R-2),
Bull. Amer. Meteor. Soc., 83, 1631-1643.
If you wish mention R2 in the acknowledgements, you
should mention that R2 was produced with the support of the US National Weather Service and of PCMDI (US Dept. of Energy).