Time-mean precipitation and vertical motion patterns over the United States

Abstract

A post-initialized set of global atmospheric gridded data, generated at NMC, is used to calculate monthly mean vertical motion at 500 mb (ɷ for all months in 1984–1986. Various methods are used to calculate ɷ such as those based on the continuity (the best and easiest) and vorticity equations. The ɷ fields are correlated with colocated monthly precipitation (P) collected in raingauges and averaged over Climate Divisions in the United States. The correlation between ɷ and P for the United States as a whole is negative (to be hoped), the coefficient ranging from about –0.6 in winter to small negative values in summer. Statistical significance is reached in 10 out of 12 months. The vorticity equation is used to decompose ɷ into contributions from mean flow- and transient eddy vorticity fluxes, respectively. Using mean flow fluxes alone leads to highly overestimated ɷ (compared to continuity), but the spatial pattern over the United States is essentially correct. The transient eddy fluxes are smaller by a factor of two and nearly opposite (correlation ͌– 0.8) to the mean flow fluxes, and seem to “force” vertical motions that are roughly out of phase with the precipitation field (correlation ͌ 0.0). Using mean flow fluxes only and accepting geostrophic constraints, the results deteriorate, although P and ɷ remain negatively correlated at modest levels for all months. The above results are true for 3-year monthly means (“climatology”) as well as for departures from climatology (anomalies). The impact of the initialization on ɷ is small, and the impact on the P, ɷ correlation almost negligible. Statistics for the 3-year period, and some examples in map-form are presented. Furthermore the geographical distribution of the P, ɷ correlation is discussed using data for 1981–86.