New datasets from the National Center for Atmospheric Research's Scientific Computing Division (NCAR/SCD) will make it easier for oceanographers, atmospheric researchers, and climate modelers to use ERA-40 reanalysis data from the European Centre for Medium-Range Weather Forecasts.

The ERA-40 reanalysis, a comprehensive set of global analyses describing the state of the atmosphere, land, and ocean from September 1957 to August 2002, is one of the largest and most complete collections of conventional and satellite observations ever assembled. SCD's new datasets make selected segments of this collection available for the first time on a standard transformation grid, used by default in one of the world's leading general-circulation climate models, the Community Climate System Model (CCSM).

CCSM is a powerful high-performance computing application designed to model Earth's past and future climate. It was developed by NCAR in collaboration with a number of universities and government institutions, with funding from the National Science Foundation, the Department of Energy, the National Oceanic and Atmospheric Administration, and the National Aeronautics and Space Administration. NCAR is releasing CCSM model results and the underlying computer codes to atmospheric researchers and other users worldwide.

HPC resources

David Stepaniak, a software engineer in SCD's Data Support Section, developed the new datasets by running code 24 hours per day for several months on “tempest,” a computing system at NCAR especially configured for data analysis. Tempest is a 128-processor SGI Origin 3800 with 64 gigabytes of distributed shared memory and 2.8 terabytes of disk space. The software Stepaniak wrote performed computations on data from the 35-terabyte ERA-40 archive on NCAR's Mass Storage System.

“This is a project too large to do on smaller systems,” says Steven Worley, manager of the Data Support Section. “It wasn't something that could have been done by small research teams that need these data.”

The ERA-40 reanalysis

The ERA-40 reanalysis project applies a modern data assimilation technique to past surface, radiosonde, and satellite observational data. The data assimilation technique works in conjunction with a general circulation model to produce an output representing many variables (such as temperature, relative humidity, vorticity, and ozone mass mixing ratio) on many levels of the atmosphere, every six hours, over the 45-year period. The result is an environmental record that can be used to study climate trends, historic storms, and weather patterns over time.

Significantly, the reanalysis also eliminates inconsistencies found in operational model output caused by changes in the models over the years, creating a more accurate simulation by using data that have been improved or were not originally available when the operational models were run.

The high-resolution (T159) ERA-40 outputs are relatively complex to use because they are in spectral coefficient or reduced Gaussian grid form. Stepaniak's software ingests these complex outputs and creates lower-resolution datasets, all on a regular Gaussian grid (T85). These datasets are easier to use and match the output resolution from the CCSM. His software takes advantage of ECMWF's Ensemble Model Output Statistics (EMOS) math library and NCAR's Spherepack math library.

Available to the community

DS124.1, the first of 12 datasets Stepaniak has completed, consists of data on 23 pressure levels at six-hour intervals. This dataset, which also includes horizontal wind components, is now available for non-commercial research in the U.S. and by member institutions of the University Corporation for Atmospheric Research, a consortium of more than 100 universities and affiliates that manages NCAR.

Joey Comeaux, also a software engineer in the Data Support Section, is working to make the next 11 datasets available online by mid-April. A total of 15 datasets will be ready by late spring. Information and access instructions are on the web at http://dss.ucar.edu/datasets/ds124.1

Stepaniak predicts there will be widespread interest in the new SCD datasets at universities, research institutions, and government laboratories. “The ERA-40 collection is really a critical element in climate studies,” he says. “To have that now on a standard grid goes a long way to helping the climatological community, not to mention the modeling community.”

Stepaniak expects the datasets to be utilized by Lawrence Livermore's Program for Climate Model Diagnosis and Intercomparison (PCMDI). “PCMDI takes the output from models around the country and compares them for accuracy and internal consistency,” he says. “To do so, they prefer to have models on a standard grid. So when the ERA-40 becomes available at T85, it will facilitate intercomparisons of model output. CCSM, in fact, is one of the main models in their diagnostic set.”

Supporting the geosciences

SCD's Data Support Section provides historical scientific data for meteorological and oceanographic research and is the focal point for distribution of the ERA-40 reanalysis data throughout the U.S. “We're continually extending and enhancing our data archives,” says Worley. “Creating T85 ERA-40 datasets is an example of the work we like to do to support science throughout the community.”

Data provision and management is part of the larger mission of SCD, whose goal is to advance research in the geosciences by providing and advancing high-performance computing technologies.

SCD is a division of the Computational and Information Systems Laboratory at NCAR, which is operated by the University Corporation for Atmospheric Research and sponsored by the National Science Foundation.

For more information

To learn more about the new T85 ERA-40 datasets from SCD, contact Dave Stepaniak ([email protected]).

Additional information is available on the web:

Dataset 124.1: http://dss.ucar.edu/datasets/ds124.1
SCD Data Support Section: http://dss.ucar.edu
Community Climate System Model: http://www.ccsm.ucar.edu
National Center for Atmospheric Research: http://www.ucar.edu