Reanalysis data sets have become an important and integral resource for studying the water cycle at regional to global scales. Since the first NCEP/NCAR reanalysis data sets were released in the mid-1990s, their use has been cited over 6,000 times. Reanalysis data sets offer homogeneous, gridded data products that have found wide application as input forcing data for off-line hydroclimate models and regional coupled models, for carrying diagnostic analysis for climate studies and for bias correcting and downscaling coarse-scale climate models. Thus it is critical that CFS-R and 20CR be evaluated both in absolute terms against observational data on specific variables and for consistency with previous reanalysis products. Very recently, the global high resolution Climate Forecast System Reanalysis (CFS-R) has been completed at NCEP/NWS for 1979-2009, as well as the 20th Century Reanalysis (20CR) at ESRL.
It is proposed to evaluate CFS-R and 20CR data products for the terrestrial water cycle over 32 global river basins, and for selected variables globally. The project will focus on three important evaluations: The first will be traditional error and bias analysis using regional and global data sets developed, compiled and analyzed by the PI over the last decade. Variables will include precipitation, temperature, river discharge, changes in terrestrial water storage, surface radiation and heat flux, and atmospheric profiles of temperature and humidity. For all variables, the data sets are already archived with the PI and have been used for water cycle studies. In most cases multiple data sets are available for each variable, permitting uncertainty assessment for CFS-R and 20CR data products. The second tier evaluation focuses on the consistency between CFS-R (or 20CR) and recently completed reanalysis, which include ERA-40, ERA-interim, MERRA and NARR. It is important for the community have an assessment of CFS-R and 20CR data products relative to currently available reanalysis products. The third tier evaluation focuses on higher-level products important to hydroclimate analysis and modeling. Evaluations will include such variables as the trends and variability in CFS-R and 20CR time series, rain-day frequency statistics (critical to hydrologic modeling and applications), reproduction of large-scale climate events (like drought) and the frequency of heavy precipitation, and reproduction of observed teleconnections such as ENSO and western U.S. precipitation.
Carrying out the proposed comprehensive CFS-R and 20CR evaluation within the one year project is only feasible due to our ongoing and previous work in evaluating/validating currently available reanalysis products (e.g. ERA40, ERA-Interim, MERRA, NARR), with in-situ (e.g. gauge or radar precipitation, tower-based ET, station-based radiation), remote sensing retrievals (e.g. MW/IR derived precipitation, remotely sensed ET, GRACE derived terrestrial water storage and AIRS derived atmospheric water vapor and temperature), and land surface modeling (VIC) surface water and energy cycle variables. Our current research on estimating the terrestrial water cycle against multiple observational and modeling products will allow an evaluation of CFS-R and 20CR products in an important, broader context.