Abstract
The moisture budget of the central United States during May is examined using multiyear (1985–89) assimilated datasets recently produced by NASA/DAO and NCEP/NCAR. Intercomparisons and comparisons with station observations are used to evaluate the limitations of the assimilated products for studies of the atmospheric component of the U.S. hydrologic cycle. Attempts are made to reconcile differences in terms of disparities in the analysis systems.
Both reanalyses overestimate daily mean precipitation rates by a factor of almost 2 over the southeastern United States. This is associated with much larger than observed afternoon convective rain and a substantial overestimate of the number of days with precipitation. Both products capture the transition to the much drier conditions over the western United States, though the NCEP/NCAR product extends moderate rain rates too far to the northwest. Over the Great Plains, the reanalyses capture observed synoptic-scale precipitation events quite well, but the variability of the daily mean precipitation is underestimated; this is particularly true for the NASA/DAO analysis, which has difficulty capturing the extreme rain rates. The NCEP/NCAR product shows generally higher correlation's with the observed precipitation, though the fluctuations in the two assimilation products are more similar to each other than they are to the observations.
The moisture transport in the reanalyses compares favorably to gridded rawinsonde data though there are some significant regional differences particularly along the Gulf Coast. Examination of the overall moisture budget for the central United States shows that the observations act as a significant local source of moisture, reflecting model bias in the first-guess fields. In both products the analysis increments act to remove water over much of the northern and western part of the country, apparently counteracting excessive evaporation in those regions, especially in the NASA/DAO. Perhaps most disturbing are the substantial differences between the two reanalyses in the moisture divergence fields since these are the most strongly constrained by the observations.
Both reanalyses capture the basic temporal and structural characteristics of the Great Plains low-level jet (LLJ) documented in previous observational studies. Composites of the nocturnal fluxes of moisture during LLJ events reveal a horizontally confined region of strong southerly transport to the east of the Rocky Mountains that is sandwiched between well-defined synoptic-scale cyclonic (anticyclonic) circulations to the northwest (southeast). Low-level inflow from the Gulf of Mexico increases by more than 50% over nocturnal mean values in both reanalyses, though the excess inflow is more than 30% stronger in the NCEP/NCAR reanalysis. While both analyses underestimate the nocturnal maximum in precipitation over the Great Plains, the pattern of precipitation anomalies associated with LLJ events compares favorably to observations.
