A reanalysis of
Introduction The goal of SODA is to reconstruct the historical physical (and eventually biogeochemical) history of the
ocean. As its name implies, the Simple Ocean Data Assimilation ocean/sea ice
reanalysis (SODA) uses a simple architecture based on community standard
codes with resolution chosen to match available data and the scales of motion that are resolvable. Agreement with direct
measurements (to within observational error estimates) as well as unbiased statistics are expected.
While SODA remains a university-based research project, we want to be helpful. Our goal is to provide a reliable source of seasonal climate time-scale ocean reanalysis
to complement the atmospheric reanalyses available elsewhere (NOAA/EMC, NASA/GMAO, and ECMWF, for example).
SODA Version 3 represents a major change from the previous SODA2. It now uses GFDL MOM5/SIS numerics at
finer 1/4°x1/4°x50lev (28km at the Equator down to <10km at polar latitudes)
resolution, similar to the ocean component of the
GFDL CM2.5 coupled climate model, and includes an active sea ice component.
The Optimal Interpolation filter has also been augmented (relative to previous releases) with bias correction to reduce bias in estimates of long term trends of variables such as heat content.
News 8 October, 2016
forced by MERRA2 and spanning the 36-year period 1980-2015,
is now available from this website along with a parallel simulation
SODA3.3.0. We have
also made available the 5-day and monthly fields remapped onto a uniform 1/2°x1/2°
Mercator horizontal grid similar to SODA2, but with expanded vertical resolution along with variables on the native
1/4°x1/4°x50levx5dy tripolar grid and variables remapped onto an isopycnal grid.
A manuscript describing SODA3.3.1 will be put on this website in the next few weeks to provide users with a more detailed system description,
some error analysis, and some simple statistics.
News 10 December, 2016 Production of soda3.4.1, based on ERA-Int, is underway
and should be available through this website in January, 2017. soda3.4.1 encorporates some improvments as well such as the Bamber (2012) Greenland discharge estimates
and improvements of the error covariances in the Arctic. For soda3.3.1 we are adding some ancillary variables such as monthly dynamic height, corrected monthly sea level,
and heat content.
News 1 April, 2017 We have interrupted production of the soda3.X.1 reanalyses to evaluate
impact of choice of bulk parameterizations for heat, mass, and momentum. The GFDL CM2.5 coupled model, from which SODA3 derives, uses the Large and Yeager (2004) parameterizations.
To address errors we were seeing in the reanalyses we have implemented the
COARE 4 parameterizations of Fairall et al. (2003). We find improvements (for example, stronger stresses under the storm tracks)
however, we have been frustrated by our inability to match the stresses actually provided by
ECMWF (e.g. ERA-I). After some struggles Senya has found that matching those stress at the 0.05N/m2 level requires using the same bulk formulas as included in the ECMWF AGCM.
In the end we have decided to specify daily stress directly instead of deriving it from U(z=10m) by bulk formula in the flux coupler software. In contrazst, we use COARE4
to compute turbulent flux terms. The advantage is that we are using what the atmospheric reanalysis products provide. The disadvantage is that we ignore feedback associated with
movement of the ocean surface layer and changes in SST, and also the impact of the use of different bulk parameterizations (forced by variables specified at different vertical levels)
by different atmospheric reanalysis centers.
We have also decided to focus on production of reanalyses with surface flux bias correction (soda3.X.2). The first three of these will be: soda3.3.2, soda3.4.2, and soda3.7.2.
Once those are released we will gradually fill in the other four from the list below. The only reanalyses we will release without bias-correcting the forcing are: soda3.3.1,
soda3.6.1, and 3.9.1. We will release the latter two because NOAA CORE2 and WHOI OAflux have already undergone bias-correction.
Coming soon We expect to complete an ensemble of SODA3 reanalyses (1980-2015) with the following:
The spread in these provides information regarding the uncertainty due to errors in surface forcing. Version 1 (e.g. soda3.7.1) uses forcing as provided by the reanalysis centers.
Version 2 (e.g. soda3.7.2) includes bias-correction of the surface forcing fields based on consistency with the ocean observing systems. A series of SODA3 system upgrades and analysis studies will
keep us busy for much of 2017. We will add the analysis of year 2016 for all of these by July 2017. By late-2017 we expect to return to the problem of
creating a reanalysis with quantified error estimates for the first half of the 20th century.
Further forward: SODA4 will encorporate quasi-isopycnal dynamics,
a new data assimilation scheme, enhanced resolution, and ocean biogeochemistry. The next major revision after that (provisionally SODA5) will be, or be part of, a fully
coupled atmos/ocean/ice reanalysis.
For additional information send us email:
Gena Chepurin and
Jim Carton. If you don't get a response we've gotten distracted. Please keep trying!
Join the SODA3 email list
SODA relies on extensive collaborations.
In addition to the National Science Foundation Physical Oceanography Program we owe debts to: NOAA/GFDL, NOAA/NCEP, NOAA/NESDIS (especially
the Laboratory for Satellite Altimetry and NCEI), NASA/GMAO, and the NASA MAP and Physical Oceanography programs. Many individuals have
contributed to SODA including: Tim Boyer, Gil Compo,
Dick Dee, Eric Hackert, Sirpa Hakkinen, Sasha Ignatov, Eugenia Kalnay, Syd Levitus, Matt Maltrud, Julie McClean, Laury Miller,
Steve Penny, R. Raghunath, James Reagan, Tony Santorelli, Mike Steele, and most notably Ben Giese, Xianhe Cao, and Hank Seidel.
To report problems with this site please contact
SODA website admin.