The Hadley circulation: assessing NCEP/NCAR reanalysis and sparse in-situ estimates

We present a comparison of the zonal mean meridional circulations derived from monthly in situ data (i.e. radiosondes and ship reports) and from the NCEP/NCAR reanalysis product. To facilitate the interpretation of the results, a third estimate of the mean meridional circulation is produced by subsampling the reanalysis at the locations where radiosonde and surface ship data are available for the in situ calculation. This third estimate, known as the subsampled estimate, is compared to the complete reanalysis estimate to assess biases in conventional, in situ estimates of the Hadley circulation associated with the sparseness of the data sources (i.e., radiosonde network). The subsampled estimate is also compared to the in situ estimate to assess the biases introduced into the reanalysis product by the numerical model, initialization process and/or indirect data sources such as satellite retrievals. The comparisons suggest that a number of qualitative differences between the in situ and reanalysis estimates are mainly associated with the sparse sampling and simpli"ed interpolation schemes associated with in situ estimates. These di!erences include: (1) a southern Hadley cell that consistently extends up to 200 hPa in the reanalysis, whereas the bulk of the circulation for the in situ and subsampled estimates tends to be con"ned to the lower half of the troposphere, (2) more wellde"ned and consistent poleward limits of the Hadley cells in the reanalysis compared to the in-situ and D. E. Waliser (| ) ) Zhixiong Shi1 Institute for Terrestrial and Planetary Atmospheres, State University of New York, Stony Brook, NY 11794-5000, USA E-mail: [email protected] Present address: 1SAIC/GSC MODIS Support O$ce, 7501 Forbes Blvd., Suite 101, Greenbelt, MD 20706, USA J. R. Lanzante 'A. H. Oort Geophysical Fluid Dynamics Laboratory, Princeton University, Princeton, NJ 08542, USA subsampled estimates, and (3) considerably less variability in magnitude and latitudinal extent of the Ferrel cells and southern polar cell exhibited in the reanalysis estimate compared to the in situ and subsampled estimates. Quantitative comparison shows that the subsampled estimate, relative to the reanalysis estimate, produces a stronger northern Hadley cell (&20%), a weaker southern Hadley cell (&20}60%), and weaker Ferrel cells in both hemispheres. These di!erences stem from poorly measured oceanic regions which necessitate signi"cant interpolation over broad regions. Moreover, they help to pinpoint speci"c shortcomings in the present and previous in situ estimates of the Hadley circulation. Comparisons between the subsampled and in situ estimates suggest that the subsampled estimate produces a slightly stronger Hadley circulation in both hemispheres, with the relative di!erences in some seasons as large as 20}30%. These di!erences suggest that the mean meridional circulation associated with the NCEP/NCAR reanalysis is more energetic than observations suggest. Examination of ENSO-related changes to the Hadley circulation suggest that the in situ and subsampled estimates signi"cantly overestimate the e!ects of ENSO on the Hadley circulation due to the reliance on sparsely distributed data. While all three estimates capture the large-scale region of lowlevel equatorial convergence near the dateline that occurs during El Nino, the in situ and subsampled estimates fail to e!ectively reproduce the large-scale areas of equatorial mass divergence to the west and east of this convergence area, leading to an overestimate of the e!ects of ENSO on the zonal mean circulation.