Global Parallel Experiment Log

Experiment Name

T254 L64 runup to implementation

Parallel Slot

PRX

Control Slot

GFS

Start date of parallel experiment

2002/09/30/06

End date of parallel experiment

2002/10/31/06

Implementation date if any

2002/10/29/12

Environmental Modeling Center scientists

Mark Iredell, Hua-Lu Pan, Shrinivas Moorthi, Bob Kistler, John Derber, Weiyu Yang, Russ Treadon, Dave Parrish, Peter Caplan, Suru Saha, Glenn White

Abstract (including Motivation, Hypothesis and Method)

This experiment combines features from two recent parallel experiments:

• T254 L64
• Fall 2002

The following comments are drawn from their descriptions.

The model and analysis are run in T254 L64 resolution. The T254 resolution is associated with a 768 x 384 quadratic Gaussian grid used by the model and a 512 x 256 linear Gaussian grid used by the analysis. The horizontal resolution of the quadratic grid is about a 50 km mesh. The 64 sigma levels are still computed so that the sigma thicknesses are a quadratic function of sigma. The bottom sigma layer is 0.99732 or about 20 m above the ground. The top sigma layer is 0.000266 or about 60 km above the ground. The vertical resolution at 250 hPa is about 700 m.

Most of the changes to the operational system are associated with the resolution changes. The motivations for these changes are

• to improve the global data assimilation system
• to improve the short term forecast

OBSERVATIONAL AND ANALYSIS CHANGES

• The domain decomposition and communication pattern in the analysis are better optimized.
• The background error is newly computed using the NMC method.
• AMSU-A channels 12 and 13 from NOAA-15 and NOAA-16 and HIRS from NOAA-16 are used.

FORECAST AND POST PROCESSING CHANGES

• The forecast resolution is T254 L64 to hour 84, T170 L42 to hour 180, and T126 L28 to hour 384.
• The horizontal diffusion is made more general, using 6th order diffusion.
• Several minor bugs are corrected.

The following changes are made to the analysis system after the original T254 parallel experiment:

• update OPTRAN coefficient files for future use of NOAA-17 TOVS data
• reorganize code (much faster, unify IBM and SGI versions)
• use BUFR format SBUV ozone
• update precipitation observation error and quality control
• slow time evolution of angle dependent radiance bias correction term
• assimilate METAR surface pressure observations
• turn off divergence tendency constraint
• use BUFR format TOVS files (necessitates new data selection algorithm)
• update background error statistics

Experiment changes log

Implementation changes list

Product changes list

Background links

Evaluation of parallel results

Daily forecast maps by Pete Caplan

Daily forecast stats by Pete Caplan

Verification against sondes and aircraft timeseries by Suru Saha

Verification against rawinsondes horizontal maps by Suru Saha

Verification against rawinsondes vertical maps by Suru Saha

Conclusion

CAFTI presentation

Technical Procedures Bulletin or TPB

Global Parallel Home Page

Disclaimer