1. Reanalyses – WDAC5

2. Overview Atmospheric Reanalyses
MERRA-2
JRA
ERA5
NCEP
Task Team for Intercomparison of Reanalyses (TIRA)

3. MERRA-2 Motivation and Objectives
Produce an ongoing, intermediate reanalysis for the satellite era using a recent version of GEOS-5 to
(1) address known limitations of MERRA (c. 2008), and
(2) provide a stepping stone to a future coupled Earth system reanalysis.
Specifics:
Incorporate modern satellite observation types not available to MERRA
Reduce spurious trends and jumps related to changes in the observing system
Reduce biases and imbalances in the water and energy cycles
Test coupling GOES-5 meteorology with other Earth system components

4. The MERRA-2 data assimilation system
GEOS AGCM/GSI 3D-Var
0.5° x 0.625° x 72 hybrid-eta levels to 0.01 hPa
MERRA MERRA-2 Evolution
Updates to the AGCM and GSI
AGCM
Cubed-sphere dynamics
Updated physics: limited deep convection, re-evap of rain, snow sublimation
Improved glacier model and cryosphere albedos
GSI
Modern observations: GPSRO, NOAA-19, MetOp-A/B, S-NPP, SEVIRI, Aura OMI and MLS, capable for JPSS, MetOp-C
Updated moisture control variable and background errors
Bias correction for aircraft temperature observations
TC Relocation
Aerosol assimilation with radiative coupling to AGCM (direct effects)
Constraints on dry mass and globally integrated water
Corrected precipitation for land surface forcing and aerosol deposition
Climate validation and other docs at:

5. GMAO Reanalysis Plans
Future directions are consistent with the Integrated Earth System Analysis framework (IESA), which calls for coupled reanalyses including atmosphere, aerosols, land, ocean, and trace gas constituents. A prospective target is for a quarter-degree coupled reanalysis using 4D EnsVar atmospheric data assimilation.
Ongoing development would include the following:
Investigation of assimilating observed ice thickness (e.g., freeboard observed from CryoSat 2).
Two-moment cloud physics for improved representation of mixed phase clouds
GRACE land moisture assimilation
Aerosol-on-snow deposition (impact on surface albedo).
Improvement of snow hydrology over ice sheets

6. JRA-55 family
To aid further study of the representation of low-frequency variability and trends in JRA-55, different types of product have been produced with the common base NWP system.
JRA-55 (JMA)
Full observing system reanalysis
Available from JMA ( DIAS, NCAR, NASA/ESGF
S. Kobayashi et al. (2015), Harada et al. (2016)
JRA-55C (MRI/JMA)
Using conventional observations only
Available from DIAS, NCAR
C. Kobayashi et al. (2014)
JRA-55AMIP (MRI/JMA)
AMIP type run
RMS errors of 2-day forecasts of geopotential height (gpm) at 500hPa averaged over the northern hemisphere
Adapted and updated from C. Kobayashi (2014)

7. The next Japanese reanalysis
JRA-3Q (pronounced as “Thank you!” in Japanese)
Japanese Reanalysis for Three Quarters of a Century
Provisional specifications
Higher resolution: TL319L60 -> TL479L100
40 km in horizontal, 100 layers up to 0.01 hPa in vertical
Extending the reanalysis period back in time
Atmospheric reanalysis from 1948 (planned) to present
New boundary conditions and forcing fields
COBE-SST2 (from the beginning to 1981)
MGDSST (satellite-based SST from 1982 onward)
New observations
Observations newly rescued and digitised by ERA-CLIM et al.
Improved satellite observations through reprocessing
JMA’s own tropical cyclone bogus

8. Data assimilation system
JRA-55
JRA-3Q
Base system
JMA’s operational system as of December 2009
JMA’s operational system as of 2018 (planned)
Horizontal resolution
TL319 (~55 km)
TL479 (~40 km)
Vertical levels
60 levels up to 0.1 hPa
100 levels up to 0.01 hPa
Analysis scheme
4D-Var
(with the T106 inner resolution)
(with the TL319 inner resolution)
Radiative transfer model for satellite radiances
RTTOV-9.3
RTTOV-10.2 (used in the current system)
Improved accuracy
Inclusion of the Zeeman effect
Inclusion of GHGs variations
GNSS-RO
Refractivity (up to 30 km)
Bending angle (up to 60 km)

9. Production schedule (as of January 2016)
FY2015
FY2016
FY2017
FY2018
FY2019
FY2020
FY2021
FY2022
FY2023
Acquisition of observational data
Preparation for JRA-3Q production
AMIP
~10-year period
AMIP
JRA-3Q production
Production of TC bogus
( )
Development of TL479 system
JRA-3Q production ( )
Preliminary experiments
Construction of
exp system
JRA-3Q production ( )
Preparation
for connection
JRA-3Q near-real-time production
Product release
Prerelease
Full release
JRA-55 near-real-time production
Parallel production

10. ERA5: the ERA-Interim replacement
What is new in ERA5?
ERA-Interim
ERA5
Period
present
Start of production
August 2006
Jan 2016, complete record end 2017
Assimilation system
2006 technology
Current state of the art
Model input
(radiation and surface)
As in operations,
(inconsistent sea surface temperature)
Appropriate for climate, e.g.,
evolution greenhouse gases, volcanic eruptions, sea surface temperature and sea ice
Spatial resolution
79 km globally
60 levels to 10 Pa
32 km globally
137 levels to 1 Pa
Uncertainty estimate
Based on a 10-member ensemble at 64 km
Output frequency
6-hourly Analysis fields
Hourly (three-hourly for the ensemble),
Extended list of parameters
~ 5,000 Tera Byte
Extra Observations
Mostly ERA-40, GTS
Various reprocessed CDRs, latest instruments
Variational Bias corrections
Satellite radiances
Also ozone, aircraft, surface pressure
ERA5: the ERA-Interim replacement

11. Hourly reanalysis fields
Copernicus Climate Change Service
The ERA5 archive in the make
Hourly reanalysis fields
ERA5 2-metre temperature compared to independent data
Observation feedback archive

12. ESRL Single-Resolution EnKF Only System
Conventional obs only in this system
Possible R1 replacement
Used for forecast IC for next cycle
Generating new ensemble perturbations given the latest set of observations and a first-guess ensemble
member 1
forecast
Uses background error covariances computed from the ensemble
member 1
analysis
EnKF ensemble perturbations
are "re-centered" around the high-res analysis
EnKF
member update
T254L64
member 2
forecast
member 2
analysis
member 3
analysis
member 3
forecast
1) Can you describe what happens in the box entitled "EnKF member update?" The EnKF is run at the late (GDAS) data cutoff, and updates every ensemble member using the EnKF algorithm, using background error covariances computed from the ensemble (no static component).   The GSI cannot be used to update the ensemble, since it only provides a single deterministic analysis (analagous to the mean of the EnKF ensemble). However, since the high-res hybrid analysis then replaces the EnKF ensemble mean analysis, effectively only the ensemble perturbations are cycled using the EnKF update.  You can think of the this update as doing the same thing as the current operational ETR scheme - it's generating new ensemble perturbations given the latest set of observations and a first-guess ensemble.  The ETR does this without using any observational information, just be rescaling and rotating the first-guess perturbations. 2) Exactly what is meant by "re-centering the ensemble around the analysis?" The high-resolution analysis computed by the GSI, using the first-guess ensemble to estimate background error covariances, replaces the ensemble EnKF analysis computed by the EnKF update step.  In other words, the EnKF ensemble perturbations are "re-centered" around the high-res analysis. 3) How are members 1-n of the analysis used (arrows going off page)? They are run forward by the GFS to form the first-guess ensemble for the next analysis time. 4) is it correct that the deterministic forecast is run off the analysis (bottom row)? Yes.
Previous Cycle
Current Update Cycle
Faster than a similar hybrid since no waiting on one branch or the other
Thanks to Jeff Whitaker

13. Global mean temperature diff
CFSR-ERAI
CFSR sat bias correction problems
EnKF-ERAI
Warm bias in low model top
EnKF-CFSR

15. Task Team for Intercomparison of ReAnalysis
White Paper developed with input from
Michael Bosilovich, NASA GSFC GMAO
Jean-Noël Thépaut, ECMWF
Kazutoshi Onogi, JMA
Arun Kumar, NOAA NCEP
Dick Dee, ECMWF
Otis Brown, NCSU and NOAA

16. TIRA Objectives Charged to develop an intercomparison Project
To foster understanding and estimation of uncertainties in reanalysis data by intercomparison and other means
To communicate new developments and best practices among the reanalyses producing centers
To enhance the understanding of data and assimilation issues and their impact on uncertainties, leading to improved reanalyses for climate assessment
To communicate the strengths and weaknesses of reanalyses, their fitness for purpose, and best practices in the use of reanalysis datasets by the scientific community

17. Tangible Activities
Communication (support and continuation of reanalysis.org)
Reanalysis Intercomparison Project
Start with a template based on IPCC
CREATE-IP provides technical support, CORE-CLIMAX provides methods
Other WCRP projects coud provide support
TIRA guides and provides structure
Joint experimentation/work
Consensus inventory of input observations, updated as new versions become available (incl SSTs, and other disciplinary data for eventual integration)
Add standard output (facilitate inclusion in CREATE-IP)
Contribute to deep understanding with center support

18. Potential Standing Members
Atmospheric Reanalysis Developing Centers
ECMWF, JMA, CMA, NCEP and GMAO
WCRP Panels
GEWEX, CLIVAR (likely GSOP), CLiC and SPARC (SRIP)
Disciplinary – Land/Ocean/Cryo reanalyses
Others?
Regional reanalysis projects
Reanalysis user communities? Wind Power

19. End of MERRA Surprise MERRA last data day 29 FEB 2016
Older CRTM not capable of using latest satellite observations, down to one polar orbiter in each of AM/PM.
Updating would have caused discontinuity, and cost significant manpower
Supercomputer was retired, would need manpower to port to new system
Posted on MERRA WWW, announced at meetings and users mailing list, still, some users just finding out when no new data is showing up at the data center
Renewable energy operational assessment of potential wind turbine locations