To gain the benefit of the many types of meteorological observations for weather forecasts, the data are analyzed (or assimilated) to provide initial conditions for numerical weather forecast models. The resulting analyses are the merger of model and many types of observed data. Weather forecasts greatly benefitted from the assimilation of satellite remotely sensed observations. After years of these analyses had been produced, researchers hoped they would yield insight on many aspects of the Earth’s climate. For example, a consistent picture of the global circulation was produced (Trenberth and Olson, 1988), though not without uncertainties. Changes and improvements to the operational analyses and forecast models introduced spurious climate shifts in the time series of such analyses, so that the resulting interannual variability was not well represented. Bengtsson and Shukla (1988) proposed a reanalysis, or retrospective-analysis, of the observations, using a fixed analysis/forecast system to provide more consistent time series of the analyzed data products. Since then, retrospective-analyses have been produced at NASA (Schubert et al. 1993), the National Centers for Environmental Prediction (NCEP, Kalnay et al. 1996; Kanamitsu et al. 2002), the European Centre for Medium-range Weather Forecasts (ECMWF, Uppala et al. 2005) and the Japanese Meteorological Agency (JMA, Onogi et al, 2007). These have all been used extensively in climate and weather research.
There are advantages and disadvantages to reanalyses for climate study. The main advantages are that global Earth observations (disparate distributions in space and time) are assimilated leading to uniformly gridded and globally available data. Reanalyses also combine many different types of observations into a single analysis. The model diagnostics include data that would not otherwise be observed, providing insight into the Earth system. On the other hand, the influence of the imperfect global models affects the resulting reanalyses, any improvements in modeling and data quality control all lead to differences in the climate produced by the aforementioned reanalyses. Hence, as models, data assimilation, observational quality control and computing power improves, so shall the climate information in reanalyses (Bengtsson et al. 2007).
The Global Modeling and Assimilation Office (GMAO) at NASA’s Goddard Space Flight Center is producing a satellite era reanalysis called the Modern Era Retrospective-analysis for Research and Applications (MERRA). The objectives are to support NASA’s climate strategies by placing current research satellites in a climate context and improving the representation of the water cycle in reanalyses. MERRA will be run for the modern satellite data era (1979-present). The data assimilation system that will be used to produce MERRA is the Goddard Earth Observing System version 5 (GEOS-5, Rienecker et al. 2008). The process of data assimilation in GEOS-5 is summarized in Figure 1. Utilizing NASA GSFC’s supercomputing systems, MERRA will be run on a 2/3å¡ longitude by 1/2å¡ latitude native grid (540×361 global gridpoints), with observational analyses every 6 hours. Figure 2 provides an overview of the satellite radiances that are assimilated in MERRA. In addition, several remotely sensed retrieved data sets are assimilated (e.g. SSMI surface winds and cloud track wind). Of course, conventional observations from radiosonde, dropsonde, aircraft and surface pressure are also routinely assimilated. Figure 3 provides examples of the observing system at various points over recent decades.
The MERRA output data will include 3 dimensional state fields for every 6 hourly analysis cycle on 42 pressure levels (or 72 terrain following model coordinate levels) from the surface through the stratosphere. Several data products are specifically designed to support chemistry and stratosphere transport modeling. The 2 dimensional surface and atmospheric diagnostics (numbering 259) are being stored on the native grid at 1 hourly intervals (Figure 4 shows an example of 1 hourly global precipitation). These include radiation and vertical integrals of the atmosphere for water and energy budget studies and also surface diagnostics where the diurnal cycle is important. The one hourly surface and near surface data product will also facilitate research on the integrated analysis of Earth system observations in the land, ocean and crysosphere.
Current Status and Future Research
In November 2007, MERRA underwent a final validation review, and was found to perform at a level above or near existing reanalyses, and the science codes were frozen. For example, monthly mean precipitation is very well represented by the system compared to existing reanalyses (Bosilovich et al. 2008). In early 2008, the system engineering codes that run the reanalysis were finalized and data production began in May 2008. To best optimize production, data are being run in three streams (bottom of Figure 2). Each stream is producing data at a rate of 10 days per computing day. It is anticipated that the 1979-present period will be completed in late 2009.
MERRA data volumes are quite large: the full data sets to be made available to the community will exceed 150 Tb. Accessibility of the data is a major concern. The full data set will be online in networked drives (not tape archive). Quick look data previewers and subsetting routines are being developed. Data will be made available during the production period.
This formulation of the system is expected to be a contribution to climate and weather research, but some caveats remain. While a reanalysis uses a fixed modeling system that produces the analysis, the input observing system is still quite variable at climate time scales. For example, sensitivity experiments have shown that we should expect a 10% increase in tropical precipitation in MERRA solely from SSMI becoming available (Bosilovich et al. 2008; WCRP Conf). The changing observing system currently presents an enormous challenge to global trend study. Likewise, the background model is still imperfect and in the absence of observations, the model influence is prevalent. These are research questions that the reanalysis developers face in the next generations of reanalyses. In addition, the analysis of Earth system observations is becoming more integrated, including land, ocean and crysophere models and observations. Higher spatial resolutions will allow better use of the observations, and improved representation of weather features. Many challenges remain in order to produce reanalyses that best utilize the global Earth observations and can accurately represent the changing climate.
GEOS-5 and MERRA are supported by the NASA Modeling, Analysis and Prediction (MAP) program (http://map.nasa.gov).
Bengtsson, L. and J. Shukla, 1988: Integration of space and in situ observations to study global climate change, Bull. Amer. Meteor. Soc., 69, 1130- 1143.
Bengtsson L, and coauthors, 2007: The Need for a Dynamical Climate Reanalysis. Bull. Amer. Meteor. Soc. 88, 495ÛÒ501.
Bosilovich, M. G., J. Chen, F. R. Robertson and R. F. Adler, 2008: Evaluation of precipitation in reanalyses. J. Appl. Meteorol and Climatol., 47, 2279ÛÒ2299.
Bosilovich, M. G. and co-authors, 2008: Evaluation of Variations in the late-80s Observing System and the Impacts on the GEOS-5 Data Assimilation System. 3rd WCRP International Conference on Reanalyses, January 2008, Tokyo Japan. (http://wcrp.ipsl.jussieu.fr/Workshops/Reanalysis2008/abstract.html)
Kalnay, E. and co-authors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437-431.
Kanamitsu, M., W. Ebisuzaki, J. Woollen, S-K Yang, J.J. Hnilo, M. Fiorino, and G. L. Potter, 2002: NCEP-DOE AMIP-II Reanalysis (R-2). Bull. Amer. Met. Soc., 83, 1631-1643.
Onogi, K. and co-authors, 2007: The JRA-25 Reanalysis. J. Met. Soc. Japan, 85, 369-432.
Rienecker, M.M.. and co-authors, 2008: The GEOS-5 Data Assimilation System ÛÒ Documentation of Versions 5.0.1 and 5.1.0. NASA GSFC Technical Report Series on Global Modeling and Data Assimilation, NASA/TM-2007-104606, Vol. 27., pp. 92.
Schubert S. D., R. B. Rood, and J. Pfaendtner, 1993: An assimilated dataset for Earth-science applications, Bull. Amer. Meteor. Soc., 74, 2331-2342.
Trenberth, K. E., and J. G. Olson, 1988: An evaluation and intercomparison of global analyses from NMC and ECMWF. Bull. Amer. Meteor. Soc., 69, 1047-1057.
Uppala, S. M., and co-authors, 2005: The ERA-40 re-analysis, QJRMS, 131 Part B, 2961-3012.
MERRA Information: http://gmao.gsfc.nasa.gov/merra/
MERRA Status and Discussion: http://merra-reanalysis.blogspot.com/
MERRA Data will be provided by the Goddard Eaarth Science Data Information and Services Center (GES-DISC): http://disc.sci.gsfc.nasa.gov/MDISC/
ERA-40 Reanalysis data provided by the European Centre for Medium-Range Weather Forecasts, from http://www.ecmwf.int/
JRA-25 Reanalysis data are provided by the Japan Meteorological Agency (JMA) from http://www.jreap.org/