11-Year solar cycle in the stratosphere extracted by the empirical mode decomposition method K.T. Coughlin * , K.K. Tung University of Washington, Box 352420, Seattle, WA 98195, USA Received 19 October 2002; received in revised form 26 February 2003; accepted 26 February 2003 Abstract We apply a novel method to extract the solar cycle signal from stratospheric data. An alternative to traditional analysis is a nonlinear empirical mode decomposition (EMD) method. This method is adaptive and therefore highly efficient at identifying embedded structures, even those with small amplitudes. Using this analysis, the geopotential height in the Northern Hemisphere can be completely decomposed into five non-stationary temporal modes including an annual cycle, a QBO signal, an ENSO-like mode, a solar cycle signal and a trend. High correlations with the sunspot cycle unambiguously establish that the fourth mode is an 11-year solar cycle signal. Ó 2004 COSPAR. Published by Elsevier Ltd. All rights reserved. Keywords: Solar cycles; Stratosphere; Empirical mode decomposition method 1. Introduction Although there have been many reports of the 11- year solar cycle in atmospheric data, there is consider- able debate on the spatial and temporal extent in which the atmosphere is influenced and on the validity of the statistical significance of these claims. A major problem is the shortness of the data record, which prevents a straightforward extraction of an 11-year signal in the energy spectrum of dynamical variables. In chemical records, like those for ozone in the stratosphere, the solar cycle signal is much clearer (Hood et al., 1993; Haigh, 1994; McCormack and Hood, 1996; Hood, 1997). Ozone heating involves wavelengths shorter than 200 nm. Irradiance at these frequencies changes signifi- cantly between solar minima and solar maxima (WMO, 1987). Changes of UV irradiance thus represents a po- tentially important influence on the upper stratospheric circulation. In tropospheric and lower stratospheric dynamical variables, however, the Fourier mode with an 11-year period has very little power. This small amplitude solar cycle signal has never- theless been found in the quiescent regions of the data set. For the extratropical middle atmosphere, this means during summers (away from the time periods of most variability), over the mid-latitude Pacific (away from the spatial areas of the most variability), and during the westerly phases of the QBO (away from the periods of the most dynamical disturbance). Originally, Labitzke (1987) discovered an association between the 30-mb winter mean polar temperature and the Sunspot numbers during the westerly phase of the equatorial QBO, with the phase defined by the equatorial 50 mb zonal wind. During the more disturbed easterly phase of the QBO though, she found that the correlations are inexplicably negative. Many other studies have shown similar results (Labitzke and van Loon, 1988; van Loon and Labitzke, 1990; Ruzmaikin and Feynman, 2002). During these winter months, the solar cycle signal is weak compared to large atmospheric varia- tions and the signal is therefore more difficult to ex- tract (Labitzke, 1987; Labitzke and van Loon, 1988; Kodera, 1991; Dunkerton and Baldwin, 1992; van Loon and Labitzke, 1990). However, during summer, Labitzke and van Loon (1989, 1990) demonstrate that stratospheric data has large positive correlations with * Corresponding author. Tel.: +1-206-543-0319; fax: +1-206-685- 1485. E-mail address: katie@amath.washington.edu (K.T. Coughlin). 0273-1177/$30 Ó 2004 COSPAR. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.asr.2003.02.045 Advances in Space Research 34 (2004) 323–329 www.elsevier.com/locate/asr