Large-scale atmospheric moisture transport as evaluated in the NCEP/NCAR and the NASA/DAO reanalyses

Abstract

Large-scale aspects of the atmospheric moisture transport and the overall moisture budget are studied using data from the National Centers for Environmental Prediction (NCEP) reanalysis. Our objective is to critically evaluate the usefulness of the reanalysis products for studies of the global hydrologic cycle. The study period is from January 1985 to December 1993. Monthly mean water vapor transport, evaporation, and precipitation are compared to the NASA Data Assimilation Office (DAO) reanalysis for roughly the same period and with satellite estimates and station observations. Comparisons of the moisture flux fields form the NCEP and the DAO reanalyses show general agreement in most aspects, but there are regional differences. Discrepancies in tropical moisture transport are largely due to uncertainties in the divergent winds. The DAO reanalysis shows a weaker Hadley circulation and weaker cross-equatorial flow, particularly during the Northern Hemisphere winter. Global patterns of evaporation from the two reanalyses are similar, but the NCEP values are higher over the oceans and lower over the landmasses. In the eastern Pacific, the DAO has less total precipitable water and less rainfall. While the large-scale features of precipitation from the reanalyses agree with each other and are within the envelope of the satellite rainfall estimates, regional differences are large. Both analyses show questionable features in the moisture flux divergence fields over North and South America that are to a large extent terrain related. Interannual variability related to the 1987-1989 ENSO cycle is well captured by both reanalyses. On intraseasonal timescales, the NCEP reanalysis has difficulty capturing the precipitation signal associated with the 30-60 day oscillation, but the moisture flux divergence from both reanalyses produces a more reasonable signal. An examination of the overall moisture budget for rectangular regions over North and South America in both reanalyses reveals large differences in the moisture flux divergence. Both reanalyses overestimate rainfall in the southeastern United States. The largest uncertainties during the spring and summer months are directly related to differences in the topographically bound low-level jets.