Comparison and Satellite Assessment of NASA/DAO and NCEP–NCAR Reanalyses over Tropical Ocean: Atmospheric Hydrology and Radiation

Abstract

This study compares the atmospheric reanalyses that have been produced independently at the Data Assimilation Office (DAO) of Goddard Laboratory for Atmospheres and at the National Centers for Environmental Prediction (NCEP). These reanalyses were produced by using a frozen state-of-the-art version of the global data assimilation system developed at these two centers. For the period 1987?88 and for the tropical oceanic regions of 30°S?30°N, surface and atmospheric fields related to atmospheric hydrology and radiation are compared and assessed, wherever possible, with satellite data. Some common biases as well as discrepancies between the two independent reassimilation products are highlighted. Considering both annual averages and interannual variability (1987?88), discrepancies between DAO and NCEP reanalysis in water vapor, precipitation, and clear-sky longwave radiation at the top of the atmosphere are generally smaller than discrepancies that exist between corresponding satellite estimates. Among common biases identified in the reanalyses, the authors note an underestimation of the total precipitable water and an overestimation of the shortwave cloud radiative forcing in warm convective regions. Both lead to an underestimation of the surface radiation budget. The authors also note an overestimaton of the clear-sky outgoing longwave radiation in most tropical ocean regions, as well as an overestimation of the longwave radiative cooling at the ocean surface. Surface latent and sensible heat fluxes differ by about 20 and 3 W m?2, respectively, in the two reanalyses. Differences in the surface radiation budget are larger than the uncertainties of satellite-based estimates. Biases in the surface radiation fluxes derived from the reanalyses are primarily due to incorrect shortwave cloud radiative forcing and, to a lesser degree, due to a deficit in the total precipitable water and a cold bias at lower-tropospheric temperatures. This study suggests that individual features and biases of each set of reanalyses should be carefully studied, especially when using analyzed surface fluxes to force other physical or geophysical models such as ocean circulation models. Over large regions of the tropical oceans, DAO and NCEP reanalyses produce surface net heat fluxes that can differ by up to 50 W m?2 in the average and by a factor of 2 when considering interannual anomalies. This may lead to vastly different thermal forcings for driving ocean circulations.