Data assimilation research
Data assimilation is the process of merging data from observations with computer models. It can transform diverse and incomplete observations to gridded estimates that can easily be used and interpreted. The assimilation process also produces quantitative information on model error, forecast skill, and observational errors, all of which allows us to improve models.
Data assimilation is providing rapid advances in geophysical studies. The Data Assimilation Research Section (DAReS) of IMAGe performs fundamental research on ensemble data assimilation methodologies for application across a wide range of geophysical problems. DAReS develops and maintains the Data Assimilation Research Testbed (DART), a software facility for doing ensemble data assimilation. DAReS also provides support to a growing community of NCAR, university, and government laboratory partners who are interested in applying ensemble data assimilation methods.
Doing data assimilation with DART/CAM
has led to significant improvements in CAM. The analyzed 266 hPa meridional
velocity at 12 GMT on 13 September 2007 is shown in the top panel. Comparing
to the lower panel shows that the longitudinal grid scale noise along 67N
and 87N was eliminated by introducing a revised polar filter in CAM.
Latitudinal noise along 220E is due to a finite volume dynamical core
numerical issue that is being studied by CAM developers. Both noise problems
are present in climate runs of CAM, but were only identified through the use
of DART assimilations.
DAReS supports three of NCAR's strategic priorities: "Developing community models," "Developing and providing advanced services and tools," and "Enhancing science education." The DART user community includes members from many NCAR divisions, more than a dozen universities, and many government labs. NCAR projects supported by DART during 2008 included:
- Year-long reanalyses with several versions of CGD's CAM climate model identified and led to the correction of errors in the polar filter.
- The impact of COSMIC GPS radio occultation measurements was evaluated for large scales in CAM and for hurricane predictions in MMM's WRF model.
- Researchers in ACD used DART with the CAM/CHEM model to provide real-time analyses of weather and carbon monoxide for the ARCTAS field campaign.
- Researchers in CGD are using DART/CAM analyses and forecasts to explore mechanisms for the rapid loss of Arctic sea ice observed in 2007.
- Researchers in MMM have developed DART-based forecast sensitivity tools and used them to address questions about the impact of observations on forecasts of hurricane position and strength.
A growing number of university groups are using both DART/WRF and DART/CAM for research and instruction. For instance, a joint DAReS/University of Wisconsin project is investigating the impact of advanced hyperspectral infrared retrievals on hurricane prediction. Research partners at Cal Tech have completed implementing a version of DART/WRF configured for prediction on Mars. At the University of Colorado, DART has been used to explore assimilating velocity observations of the flow exiting a laboratory slit jet.
DAReS has supported the incorporation of two new large geophysical models this year: a version of the MIT ocean GCM in partnership with Scripps Institution of Oceanography and the Navy's COAMPS® model in partnership with NRL Monterey. DART/WRF is also being evaluated for typhoon predictions in non-operational tests by the Central Weather Bureau in Taiwan. DAReS continues to incorporate feedback from all partners to develop more powerful and generic assimilation tools.
Fundamental ensemble data assimilation research has focused on dealing with non-Gaussian distributions of observational error. An algorithm developed last year has been further refined, used to produce year-long assimilations in CAM, and is being tested for assimilation of radar reflectivity in WRF. Research on designing filters that can tolerate both non-gaussianity and non-linearity will be the key focus during the next year. Such tools would improve analysis and prediction using radar and remote sensing observations in hurricanes and severe convection.
Data assimilation research in IMAGe is supported by NSF Core funding and NASA Grant NNX08A23G.