Quantifying the variability of the annular modes: reanalysis uncertainty vs. sampling uncertainty

Abstract. The annular modes characterize the dominant variability of the extratropical circulation in each hemisphere, quantifying vacillations in the position of the tropospheric jet streams and the strength of the stratospheric polar vortices. Their representation in all available reanalysis products is assessed. Reanalysis uncertainty associated with limitations in the ability to constrain the circulation with available observations, i.e., the inter-reanalysis spread, is contrasted with sampling uncertainty associated with the finite length of the reanalysis records. It is shown that the annular modes are extremely consistent across all modern reanalyses during the satellite era (ca. 1979 onward). Consequently, uncertainty in annular mode variability, e.g., the coupling between the stratosphere and troposphere and the variation in the amplitude and timescale of jet variations throughout the annual cycle, is dominated by sampling uncertainty. Comparison of reanalyses based on conventional (i.e., nonsatellite) or surface observations alone with those using all available observations indicates that there is limited ability to characterize the Southern Annular Mode (SAM) in the presatellite era. Notably, prior to 1979, surface-input reanalyses better capture the SAM at near-surface levels than full-input reanalyses. For the Northern Annular Mode, however, there is evidence that conventional observations are sufficient, at least from 1958 onward. The addition of 2 additional decades of records substantially reduces sampling uncertainty in several key measures of annular mode variability, demonstrating the value of more historic reanalyses. Implications for the assessment of atmospheric models and the strength of coupling between the surface and upper atmosphere are discussed.