Authors Accepted Manuscript

Assessment of the relationship between the combined solar Cycle/ENSO forcings and the tropopause temperature Alfred M. Powell, Jianjun Xu PII: DOI: Reference: To appear in: Received date: Revised date: Accepted date: S1364-6826(12)00081-8 doi:10.1016/j.jastp.2012.02.023 ATP3588

www.elsevier.com/locate/jastp

Journal of Atmospheric and Solar-Terrestrial Physics 21 September 2011 15 February 2012 29 February 2012

Cite this article as: Alfred M. Powell and Jianjun Xu, Assessment of the relationship between the combined solar Cycle/ENSO forcings and the tropopause temperature, Journal of Atmospheric and Solar-Terrestrial Physics, doi:10.1016/j.jastp.2012.02.023 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Assessment of the Relationship between the Combined Solar Cycle/ENSO Forcings and the Tropopause Temperature

ALFRED M. POWELL, Jr. NOAA/NESDIS/ Center for Satellite Applications and Research (STAR), Camp Springs, MARYLAND

JIANJUN XU* Environmental Science and Technology Center, College of Science, George Mason University, Fairfax, Virginia

15 February 2011

revised Journal of Atmospheric and Solar-Terrestrial Physics

*Corresponding author contact information: Dr. Jianjun Xu, Environmental Science and Technology Center, College of Science, George Mason University, Fairfax, Virginia, USA. Email: Jianjun.xu@noaa.gov Dr. ALFRED M. POWELL, Jr., NOAA/NESDIS/STAR, 5200 Auth Road, WWB, Camp Springs, MD 20746. Email: Al.Powell@noaa.gov

Abstract
The tropopause region of the atmosphere shows large variability over time and by region. The complex changes near the tropopause are not fully understood, especially in terms of interdecadal and interannual forcings. The purpose of this paper is to investigate forcings in the tropopause region by using microwave sounder observations and comparing the results to previous analyses. On the basis of the satellite retrieved temperatures from the Microwave Sounding Unit (MSU) Channel 3 (CH3) measurements which began in 1981 and continue to the current time, this analysis will assess the solar forcing and the El Nino Southern Oscillation (ENSO) forcing within the tropopause layer (300 - 100 hPa). The temperature variability from the combined downward solar forcing and the upward ENSO forcing have been investigated using wavelet, multiple linear regression and lag correlation analyses. The results show that the temperature variability within the tropopause layer was dominated by 3.5-7 and 14-28 year oscillations. The temperature responses to the two forcings apparently depend on the location, season and time scale of the measurements. The temperature response to solar forcing can be found over the Arctic and Antarctic zones in winter. On the interdecadal time scale, the temperature response to solar forcing was markedly amplified with a lag of 1-2 years or 5-7 years and was out of phase between the Arctic, and all other latitudes. Interestingly, the statistically significant response to solar forcing was only identified over the tropical central and western Pacific in summer. 1

The temperature response to the ENSO forcing is much stronger than the solar forcing based on the magnitude of the regression coefficients. A significant positive response occurs over most of the tropical ocean areas in winter and a negative temperature response is confined to the tropical western Pacific in summer. On the interannual time scale, the temperature response is observed within the tropical areas and reaches a positive maximum 4-5 months later, and can be identified up to 10 months later with statistically significant values. After 10 months, the response is negative.

1. Introduction The tropopause is defined as the boundary between two very different atmospheric layers, namely, the turbulently mixed troposphere and the stably stratified stratosphere. Accurate knowledge of tropopause change is very important to understand the interaction between the stratosphere and troposphere. Currently two datasets including radiosonde and reanalysis data have been used to determine the trends in the tropopause layer in many previous studies. Based on the radiosondes, Seidel et al (2001) found a multidecadal decrease in temperature of 0.5 K per decade in the tropopause layer. Randel et al. (2000) found a downward trend in tropopause temperature of 0.5 K per decade but could not find it in the reanalysis of the National Centers for Environmental Prediction (NCEP) during 19791997. These previous studies lead us to expect that similar trends will be found in the analysis of a new observational dataset. In addition, we will also investigate forcing mechanisms responsible for interannual and interdecadal variability in the tropopause temperature. The satellite retrieved temperatures from the Microwave Sounding Unit (MSU) Channel 3 (Ch3) were first created by the Center for Satellite Applications and Research (STAR) (Zou et al., 2009) from the TIROS-N, NOAA-10, 11, 12, and 14 satellites operated by the National Environmental Satellite, Data, and Information Service (NESDIS). The Ch3 satellite observations provided a useful data set for the study of the tropopause layer (300-100 hPa) temperature (Zou and Wang 2010). The causal mechanism responsible for the interannual and interdecadal temperature variability in the tropopause layer is very complicated. Santer et al. (2003) pointed out that natural variability alone cannot explain the observed decreases in tropopause temperature, and mainly attributed the changes to anthropogenic activity. Gage and Reid 3

(1981) suggested that the tropopause temperature was initially interpreted as being caused by changes in total solar irradiance associated with the solar cycle. Another study (Hatsushika and Yamazaki 2001) indicated the tropopuase temperature is lower over the equatorial eastern Pacific during El Nino compared with La Nina. Although the total solar irradiance variations from the minimum to the maximum of a solar sunspot cycle are only approximately 0.1%, evidence has been presented in the literature that variations in solar radiation can play a significant role in earths climate variability. Two possible mechanisms may explain the response of the magnitude seen in the observations. The first involves a top down response of the earths atmosphere due to the absorption of ultraviolet (UV) radiation by ozone (Haigh 1996, Shindell et al., 1999, Ziemke et al. 2000; Hood, et al. 2010, Powell and Xu, 2011). The variability in the UV and shorter wavelengths is significantly higher (a few percent) than the integrated or net change across the spectrum which is only about 0.1% of the total solar variation. However, if the variability is amplified by constituent absorption, then perhaps a solar signal can be ascertained as a result of both absorption and the resulting dynamic changes. Why is this a possibility? First, Haigh (1996) demonstrated a top-down

mechanism in a modeling study showing a broadening of the Hadley cells in response to enhanced UV that increased as the solar-induced ozone change was included. Second, a bottom up mechanism that can enlarge the response to an initially small solar forcing involves air-sea interaction. In a recent review article, Gray and her collaborators (Gray et al., 2010) pointed out that the solar influence on tropical circulations through direct Total Solar Irradiance (TSI) effects at the surface involves solar absorption over relatively cloud-free subtropical oceans. In addition, this effect increases during solar maximum (Meehl et al, 2008). This mechanism increases evaporation, and the increased 4

moisture converges into the precipitation zones, which then intensifies the climatological precipitation maxima and associated upward vertical motions. This results in stronger trade winds, greater equatorial Pacific ocean upwelling and colder SSTs, which are consistent with stronger Hadley and Walker circulations. This strengthened circulation also enhances the subtropical subsidence, resulting in a positive feedback that reduces clouds and thus further increases solar forcing at the surface. A recent study (Meehl et al, 2009) claimed that these two mechanisms are not mutually exclusive and suggest that both could be acting together to enhance the response. However, the bottom-up coupled air-sea forcing from El Nino/Southern Oscillation (ENSO) events can operate independently from the bottom-up coupled air-sea mechanism associated with the 11 year solar cycle since there are some differences in the climate responses to ENSO and solar (van Loon and Meehl, 2008). For the purpose of understanding the impacts of the combined solar cycle downward and coupled air-sea upward forcings on the temperature variability in the tropopause layer, in this study, we seek answers in two areas: i) what is the interannual and interdecadal temperature variability within the tropopause layer and the lag in the atmospheric response; ii) in addition, we will also address the possible multiple mechanism approach of combining the solar cycle downward and the ENSO upward forcing by using multiple linear regression and lag correlation analysis. Note, this is different from the combined top-down and bottom-up mechanism proposed by Meehl et al. (2009) where these two mechanisms may work together to produce an amplified SST, precipitation and cloud response in the tropical Pacific from a relatively small solar forcing. The data and analysis techniques are described in the following section. Section 3 shows the temperature variability based on the monthly satellite 5

observations. The combined mechanism discussion in terms of the regression analysis is presented in section 4. The final summary and discussion will be given in section 5.

2. Data and Methodology 2.1 Data To understand the temperature variability within the tropopause layer (300-100 hPa) and the mechanisms responsible for the variability, three datasets will be used to assess the impacts. The Microwave Sounding Unit (MSU) Channel 3 (CH3) provides calibrated direct brightness temperature measurements of the tropopause layer and was created by STAR (Zou, et al. 2009). The other two datasets include the solar F10.7-cm radio flux obtained from http://www.ngdc.noaa.gov/stp/SOLAR and ENSO index defined by U.S. Climate Prediction Center (CPC) based on the East Central Tropical Pacific (Nino3.4) SST (5N-5S,170-120W) which are representative of the forcing caused by their respective sources. All datasets spanned the period from 1981 through present.

2.2 Methodology Wavelet Transforms Many previous studies (e.g., Weng and Lau, 1994, Varotsos and

Kirk-Davidoff 2006; Varotsos et al. 2009) have indicated a significant non-stationary feature in the climate time series. In order to effectively discover the non-stationary signal, a powerful tool Wavelet Transforms (WT) is used in this analysis. Unlike the Fourier transform, WT can generalize local base functions (e.g., Morlet wavelets) that can be stretched and translated with a flexible resolution in both time and frequency (or timescale). It describes the instantaneous phase and amplitude of measurements at a specific frequency during its evolution. This feature provides the ability to identify the 6

temporal behavior of a particular component and to ascertain the variations including their association with other components. In the transformed frequency-time space, the relative significance of a component to the total variance is defined by its amplitude. For details of the complex Morlet WT, the readers are referred to Weng and Lau (1994), Lau and Weng (1995), Torrence and Compo (1998), and Weng (2005).

Multiple linear regression analysis For the temperature (T), a multiple linear regression equation can be expressed as follows: T = a0+a1TRD + a2 F10.7 + a3 ENSO + (1) where a0 is the long term mean, TRD is a linear trend and F10.7 represents the solar forcing quantified by the solar 10.7 cm radio flux. ENSO is the El Nio/Southern Oscillation index that indicates the forcing coming from the coupled air-sea system, is a residual error term. The coefficients a0, a1, a2, and a3 are determined by least squares regression. Note that each forcing term index was normalized before the calculations.

3 Temperature variability within the tropopause layer The time series of the global mean temperature retrieved from MSU CH3 measurements (Fig. 1a), which represents the temperature of the tropopause layer from 300 hPa to 100 hPa , displayed strong interannual and interdecadal temperature variations. The warm anomalies appeared in 1983, 1988, 1992, 1998, and 2002-2006, and cool anomalies occurred in 1984-1986, 1989-1991, 1993-1997 and 2000-2001. To identify the multiple time scale variation of the temperature anomalies, the wavelet transform method is used to obtain the possible oscillations and is then used to reconstruct two components using a threshold of 10 years to obtain the interannual (<10 7

years) and interdecadal oscillation (>10 years) characteristics. The frequency-time section (Fig. 1b) and the wavelet frequency spectra (Fig. 1c) show that the temperature anomalies in the tropopause layer have two identifiable oscillations: 3.5-7 years and 14-28 years. The amplitude of the quasi-biennial oscillations (QBO) also shows noticeable interdecadal variations. Note, at present, one of the shortcomings is the lack of a statistical significance test for the non-stationary processes. However, the results from the wavelet analysis of the temperature time series revealed a possible universal interpretation that may explain a portion of the climate variability. For example, the QBO identified in many previous studies (Holton and Tan, 1980, Labitzke et al., 1987; Varotosos, 1989; Baldwin et al., 2001; Powell and Xu, 2010) was strong during the period of 1982-1992, and 1996-1999, but weak at other times. The quasi 22-year oscillation of the temperature has a peak during 1983-1989, and a minima in the 2000s. Solar forcing has a 22-year cycle (Hale Cycle) based on the reversal of the Sun's magnetic field every 11 years, creating a 22-year cycle. This suggests a combined solar and ENSO forcing may be causing these features. A more detailed discussion will be provided in section 4.2.

4. Combining of the solar UV radiation and ENSO events forcing 4.1 Multiple regression analysis a. Solar forcing For the tropopause layer, the multiple linear regression analysis of temperature with the normalized 10.7-cm solar flux in winter (December-February) shows (Fig. 2a) a large positive regression coefficient (warming) in the Arctic zone north of the EuropeanAsian continent with negative values (cooling) in the middle latitudes of the North 8

American continent and a portion of the southern middle-high latitudes. However, few regression coefficients exceeded the statistical significance test at the 95% level except for a narrow region of the Southern Ocean. Note, the response to solar forcing in tropical and subtropical areas is almost zero. In contrast, the regressed temperature anomaly in summer (Figs. 2b) has a positive regression coefficient in the Southern Ocean but did not pass the 95% significance test. The response exceeded the 95% significant test over most of the western tropical Pacific Ocean and southern Indian Ocean, albeit with small regression coefficients. Briefly, the response of temperature to solar forcing within the tropopause layer shows a strong seasonal variation. The responses to the solar forcing are reversed between summer and winter. Unfortunately, the change does not pass the 95% statistical significant test. However, a significant response is observed over the tropical central and western Pacific in summer. This implies it is possible for solar forcing to impact the temperature of the tropopause layer through air-sea interaction over the Pacific sector. This is consistent with the location of the tropospheric temperature anomalies identified by many previous studies (see Meehl et al., 2009; Gray et al., 2010, and references therein).

b. ENSO Forcing Given the solar forcing results, what temperature anomaly impacts could the ENSO forcing be responsible for? To address this question, the ENSO forcing term in equation (1) will be analyzed in this section. The ENSO results for winter are shown in Fig. 2c. A strong response to the ENSO forcing is found in the tropical Pacific and Indian oceans. These areas correspond to the 9

vertical motion branches of Walker cell and Hadley cell. This pattern is quite different from the solar forcing results (Fig. 2a). A pronounced response is observed for the ENSO forcing and is symmetrical about the equator over most of the tropical/subtropical ocean areas with the strongest amplitude appearing over the tropical eastern Pacific. The shaded areas indicate that the temperature response to ENSO forcing exceeds the statistical significance test at the 95% confidence level. It is obvious that the response over both middle-high latitudes cannot pass the statistical significant test although the regression coefficient is not small. For example, the regression coefficient is as large as 0.5 over the Northeast Asian region. Compared to the boreal winter ENSO forcing, the boreal summer temperature response to the ENSO forcing is significantly reduced (Fig. 2d), especially in the tropical areas. The strongest response is shown over the tropical and subtropical southern/northern Pacific, but the area exceeding the 95% significance test is markedly smaller compared to the winter forcing case. In addition, the values of the regression coefficients are generally lower in the summer case even though they pass the 95% significance test. In summary, it can be easily seen that the response of temperature within the tropopause layer due to ENSO forcing is much stronger than that for solar forcing, and the response pattern apparently depends on seasonal variation and regionally localized impacts. This observation is based on the magnitudes of the regression coefficients with the solar and ENSO forcings respectively. The strongest positive response to ENSO forcing occurs over the tropical eastern Pacific in winter. In contrast, the significant negative response to the solar forcing can be identified over the tropical western Pacific in summer. 10

4.2 Lag correlation analysis The wavelet analysis in section 3 clearly shows that the temperature within the tropopause layer has two predominant time scale oscillations: 3.5-7 years corresponding to the ENSO time scale and 14-28 years corresponding to the decadal solar cycles. For the purpose of understanding the impact of the different time scale oscillations on the relationship between temperature variability and the solar cycle/ENSO forcings, we reconstructed two components using a threshold of 10 years to obtain the interannual (<10 years) and interdecadal (>10 years) oscillations based on wavelet spectral analysis. To quantify the similarities and differences among latitudes, the zonal mean temperature in the six latitudinal bands (60-90N, 30-60N, 0-30N, 30S-0, 30-60S and 60-90S) are calculated. The lag correlation coefficient is categorized into three types, e.g. 1) Comprehensive base: MSU CH3 monthly data on the full time scale; 2) Interdecadal base: reconstructed monthly data on the interdecadal time scale; 3) Interannual base: reconstructed monthly data on the interannual time scale. The 0.5 value correlation coefficient is drawn using a dashed line in Fig. 3. The correlation exceeding the threshold value of 0.5 indicates that 25% or more of the variance in the datasets is explained by the correlation, and satisfies the statistical significant test at the 99% confidence level. For solar forcing, the lag correlation between the time series of the solar radio flux and the MSU CH3 monthly temperature data shows (Fig. 3a) that all correlation coefficients are below 0.5. The highest correlation value of ~0.4 can be found in both tropical bands when the solar forcing is 12 months (or one year) earlier than the tropopause temperature variation. This is consistent with previous studies showing that the lagged response of a year or two in tropical Pacific sea surface temperatures (SSTs) is 11

from solar forcing (Meehl and Arblaster, 2009). In general, the cooler equatorial eastern Pacific SSTs at solar maximum tend to be followed a year or two later by warm equatorial eastern Pacific SSTs (White, et al., 1997; Roy and Haigh (2010). In contrast, the lag correlations between solar forcing and the reconstructed interdecadal component of the temperature time series are dramatically amplified (Fig. 3b). The solar forcing has a positive impact on the temperature within the tropopause layer over the Arctic (6090N) in two years, but it becomes a negative forcing impact in the 5-7 year timeframe. The temperature response over the rest of the latitudes for solar forcing shows an effect opposite to the Arctic zone. Except for the northern middle latitude (30-60N), the lag correlation obviously reflects the high impact of solar forcing on the interdecadal variation of the global temperature within the tropopause layer. For the ENSO forcing, the lag correlations on the comprehensive base (Fig. 3c) or the reconstructed interannual base (Fig. 3d) show quite similar features. This similarity implies that temperature variability of the tropopause is dominated by the ENSO forcing on the interannual time scale. Based on the statistical test at the 95% confidence level, the ENSO forcing persists in driving on the tropical tropopause temperature (30S-30N) for up to 10 months. The response of the tropical temperature within the tropopause layer to ENSO forcing reaches a maximum after 4 or 5 months. In addition, the lag correlation analysis shows a relatively weak correlation outside of the tropical areas, which indicates that the tropical areas are the key to establishing the bridge between ENSO forcing and tropopause temperature variation. To summarize, the lag correlation shows that the response of temperature within the tropopause layer to the solar cycle and ENSO forcings depend on the time scale and location. The interannual response mainly driven by ENSO forcing is confined within the 12

tropical areas and reaches a maximum 4-5 months later. The interdecadal response largely driven by solar forcing displayed an out of phase relationship between the Arctic and all other latitudes with a lag of 1-2 years or 5-7 years. This result is supported by previous findings (White and Liu, 2008, Meehl et al., 2009) where the largest warming signal at the surface and throughout the troposphere and stratosphere occurs with a lag of a couple of years after peaks in the 11-year solar cycle.

5. Summary and Discussion 5.1 Summary Using the satellite retrieved temperatures within the tropopause layer (300-100 hPa) from the Microwave Sounding Unit (MSU) Channel 3 (CH3), the temperature variability associated with solar ultraviolet (UV) forcing and ENSO events have been investigated based on wavelet methodology, multiple linear regression analysis and lag correlation analysis. The results can be summarized as follows. 1) The temperature response to solar forcing shows a strong seasonal variation. A negative response can be identified over the Arctic and Antarctic zones in winter, but does not meet the statistical significant test. However, a significant response in summer is observed over the tropical central and western Pacific. 2) The temperature response to ENSO forcing is much stronger than the solar forcing based on correlation values. The significant positive response to ENSO forcing occurs over most of the tropical areas with the strongest value observed over the tropical eastern Pacific in winter. In contrast, the significant negative response to the solar forcing can only be identified over the tropical western Pacific in summer.

13

3) The response of temperature to the solar cycle and ENSO forcing depend on the time scale. The interannual response mainly driven by ENSO forcing is confined within the tropical areas and the amplitude reaches a maximum 4-5 months later. The interdecadal response, largely driven by solar forcing, displayed an out of phase response between the Arctic and all other latitudes with a lag of 1-2 or 5-7 years.

5.2 Discussion The results of this study appear to have provided answers in two areas mentioned in section 1. Although the current analysis with the wavelet transform (WT) lacks a proper statistical significance test for the temperature variability, the results from new temperature time series within the tropopause layer have revealed variations similar to those identified in the previous studies using a different dataset. For example, the 3.5-7 year oscillation corresponds to the strong climate signal from ENSO events and the 1428 year oscillation may be connected with the 22-year Hale Cycle (solar magnetic activity cycle). Although the significance test for the wavelet analysis is undoubtedly important, its value is debated due to the simple fact that there always seems to be patterns (e.g. peaks) in the wavelet scalogram even if the analysed signal is noisy. The WT significance levels first established by Torrence and Compo (1998) were based on a series of Monte Carlo simulations. Although simple and straightforward, their approach failed to disclose many interesting and essential properties that could have been shown through an analytical approach (Ge, 2007). Lau and Weng (1994) pointed out that almost all traditional significance tests are derived from the assumption of identical repeated cycles for non-stationary processes and therefore are inappropriate for WT. The Monte Carlo methods are also unsatisfactory in the context of nonlinear systems. A proper 14

statistical significance test for WT is an important research topic but beyond the main scope of this paper. Notwithstanding this, the WT is useful to help understand nonstationary influences. For the second area, there are a lot of studies using statistical analysis and modeling to investigate the mechanism(s) responsible for the temperature variability of the tropopause layer (Randel, et al., 2000; Seidel, et al., 2001; Santer et al., 2003; Fueglistaler et al., 2007). They emphasized the impacts of the well-mixed greenhouse gases and stratospheric ozone depletion on the temperature change in the tropopause layer. The current study shows that temperature response within tropopause layer to ENSO forcing is much stronger than the solar forcing, but the response explicitly depends on the location and the time scale. This result provided an alternative forcing coming from the interannual ENSO events and decadal solar forcing, which is not yet fully understood. Furthermore, the combined mechanism of solar cycle downward and coupled air-sea upward forcings in this study is different from the combined top-down and bottom up mechanism proposed by Meehl et al. (2009). The latter emphasized a combination of bottom up forcing by TSI radiation in the troposphere-ocean system, and top-down influence by UV radiation in the stratosphere could potentially explain the unexpectedly high amplitudes of observed climate responses in the tropospheric Pacific sector. But the exact relative contributions of the two mechanisms to tropospheric, and especially to tropopause climate forcing is not clear. In this study, based on the reconstruction of climate time series and lag correlation analysis, we emphasized that the interannual response mainly driven by ENSO forcing is confined within the tropical areas and the amplitude reaches a maximum 4-5 months later. The interdecadal response, 15

largely driven by solar forcing, displayed an out of phase response with the Arctic and all other latitudes. It is worth noting that the tropospause temperature anomalies during the recent 30-year dataset are dominated by the interannual variability, which is mainly from the coupled air-sea upward forcing.

Acknowledgements: The authors would like to thank Dr. C. Zou from NOAA/NESDIS/STAR for many excellent discussions and the MSU temperature data sets that were provided. We also thank the anonymous reviewers for their valuable comments and suggestions. This work was supported by the National Oceanic and Atmospheric Administration (NOAA), National Environmental Satellite, Data and Information Service (NESDIS), Center for Satellite Applications and Research (STAR). The views, opinions, and findings contained in this publication are those of the authors and should not be considered an official NOAA or U.S. Government position, policy, or decision.

References Baldwin, M.P., et al., 2001. The quasi biennial oscillation, Rev.Geophys., 39, 179229, doi:10.1029/1999RG000073. Fueglistaler, S., Dessler, A.E., Dunkerton, T.J., Folkins, I., Fu Q., Mote P.W., 2007. Tropical tropopause layer, Rev. Geophys., 47, RG1004. Gage, K.S., Reid, G.C., 1981. Solar variability and the secular variation in the tropical tropopause, Geophys. Res. Lett., 8(2), 187-190. Ge, Z., 2007. Annales Geophysicae Significance tests for the wavelet power and the wavelet power spectrum. Ann. Geophys., 25, 22592269. 16

Gray, L.J., et al., 2010. Solar influences on climate, Rev. Geophys., 48, RG4001, doi:10.1029/2009RG000282. Haigh, J. D., 1996. The impact of solar variability on climate, Science, 272: 981-984 Hatsushika, H., Yamazaki, K., 2001. Interannual variations of temperature and vertical motion at the tropical tropopause associated with ENSO, Geophys. Res. Lett., 28(15), 2891-2894. Holton, J.R., Tan, H.C., 1980. The influence of the equatorial quasi-biennial oscillation on the global circulation at 50 mb. J. Atmos. Sci., 37 (10), 2200-2208. Hood, L.L, Soukharev, B.E., 2010. Decadal variability of the tropical stratosphere: secondary influence of the El Nino southern Oscillation. J. Geophys.Res., 115, D11113, doi:10.1029/2009JD012291 Labitzke, K., 1987. Sunspots, the QBO and the stratospheric temperature in the north polar region. Geophys. Res. Lett., 14 (5), 535-537. Lau, K.M., Weng, H., 1995. Climate signal detection using wavelet transform: How to make a time series sing. Bull. Amer. Meteor. Soc., 76, 23912402. Kazuyoshi, K., Wang, B., 2010. Spatiotemporal Wavelet Transform and the Multiscale Behavior of the MaddenJulian Oscillation. J. Climate, 23, 3814-3834, doi: 10.1175/2010JCLI2693.1 Meehl, G.A, Arblaster, J.M., Branstator, G., van Loon, H., 2008. A coupled air-sea response mechanism to solar forcing in the Pacific region, J. Climate, 21: 28832897. Meehl, G.A., Arblaster, J.M., 2009. A Lagged Warm EventLike Response to Peaks in Solar Forcing in the Pacific Region. J. Climate, 22, 36473660. doi: 10.1175/2009JCLI2619.1 17

Meehl, G.A., et al., 2009. Amplifying the Pacific Climate System Response to a Small 11-Year Solar Cycle Forcing. Science 325, 1114. DOI: 10.1126/science.1172872. Powell, A, Xu, J., 2010. An investigation of the relationship between the equatorial quasi-biennial oscillation, the Arctic stratosphere and solar forcing. Journal of Atmospheric and Solar-Terrestrial Physics, 72, 13541363,

doi:10.1016/j.jastp.2010.09.024 Powell, A., Xu, J., 2011: Possible Solar Forcing of Interannual and Decadal Stratospheric Planetary Wave Variability in the Northern Hemisphere: An Observational Study, Journal of Atmospheric and Solar-Terrestrial Physics, 73, 825-838. DOI: 10.1016 /j.jastp.2011.02.001 Roy, I, Haigh, D.J., 2010. Solar cycle signals in sea level pressure and sea surface temperature, Atmos. Chem Phys., 10, 3147-3153. Randel, W.J., Wu, F., Gaffen, D.J., 2000. Interannual variability of the tropical tropopause derived from radiosonde data and NCEP reanalyses, J. Geophys. Res., 105, 15,509 15,524. Santer, B.D., et al., 2003. Behavior of tropopause height and atmospheric temperature in models, reanalyses, and observations: Decadal changes, J. Geophys. Res., 108(D1), 4002, doi:10.1029/2002JD002258. Seidel, D.J., Ross, R.J., Angell, J.K., Reid, G.C., 2001. Climatological characteristics of the tropical tropopause as revealed by radiosondes, J. Geophys. Res., 106, 7857 7878. Shindell, D., Rind, D., Balachandran, N., Lean, J., Lonergan, P., 1999. Solar cycle variability, ozone and climate, Science, 284: 305-308.

18

Torrence, C., Compo, G.P., 1998. A practical guide to wavelet analysis, Bull. Am. Meteorol. Soc., 79, 61-78. van Loon H., Meehl, G.A., 2008. The response in the Pacific to the Sun's decadal peaks and contrasts to Cold Events in the Southern Oscillation. J. Atmospheric and Solar-Terrestrial Physics, 70, 10461055. Varotsos, C., 1989. Connections between the 11-year solar-cycle, the QBO and total ozone - comments, Journal of Atmospheric and Terrestrial Physics, 51, 367-370. Varotsos, C., Kirk-Davidoff, D., 2006. Long-memory processes in ozone and temperature variations at the region 60S-60N. PHYSICS, 6 , 4093-4100. Varotsos, C., Efstathiou, M., Tzanis, C., 2009. Scaling behaviour of the global tropopause, Atmos. Chem Phys, 9 , 677-683. Weng, H.-Y., 2005. The influence of the 11 yr solar cycle on the interannualcentennial climate variability, Journal of Atmospheric and Solar-Terrestrial Physics. 67 793805. Weng, H.-Y., Lau, K.-M., 1994. Wavelet, period-doubling and time-frequency localization with application to satellite data analysis. Journal of Atmospheric Sciences 51, 25232541. White, W.B., Lean, J., Cayan, D.R., Dettinger, M.D., 1997. Response of global upper ocean temperature to changing solar irradiance. 3266 White, W.B., Liu Z., 2008. Non-linear alignment of El Nino to the 11-yr solar cycle. Geophys. Res. Lett., 35, L19607, doi:10.1029/2008GL034831. J. Geophys. Res., 102, 3255 ATMOSPHERIC CHEMISTRY AND

19

Ziemke, J.R., Chandra, S., Herman, J., Varotsos, C., 2000. Erythemally weighted UV trends over northern latitudes derived from Nimbus 7 TOMS measurements, J. Geophys. Res., 105, 7373-7382. Zou, C., Gao, M., Goldberg, M., 2009. Error structure and temperature trends in observations from the microwave sounding unit, J Climate, 22, 1661-1681. Zou, C., Wang, W., 2010. Stability of the MSU-Derived Atmospheric Temperature Trend, J. Atmos. Oceanic Technol., 27, 19601971.

Caption of Figures

Fig. 1 Global mean temperature retrieved from MSU CH3 measurement for the tropopause layer (300-100 hPa). (a) Time series of the temperature anomaly from 1981 through 2006; (b) Real coefficient of wavelet transform changing with time, (c) wavelet spectra. Fig. 2 Regression coefficient of MSU CH3 Temperature with the solar cycle and ENSO forcings in Winter (Dec-Feb) and Summer (Jul-Sep). Solar cycle: (a) Winter and (b) Summer; ENSO: (c) Winter and (d) Summer. The shaded areas indicate the statistical significance at the 95% level. Fig. 3 Lag correlation between F10.7cm solar flux and (a) MSU CH3 monthly temperature, (b) reconstructed interdecadal temperature; and the lag correlation between 20

the ENSO index and (c) MSU CH3 monthly temperature, (d) reconstructed interannual temperature. X coordinate indicates the lead time (month) by solar cycle or ENSO forcing. Dotted line indicates the 0.5 correlation coefficient at the 99% significant test.

21

Combined downward solar forcing and the upward ENSO forcing. Interdecadal temperature response to solar forcing was amplified lag 1-2 years. Interannual temperature response reaches a maximum lag 4-5 months to ENSO events.

22

Figure

Fig. 1 Global mean temperature retrieved from MSU CH3 measurement for the tropopause layer (300-100 hPa) (a) Time series of the temperature anomaly from 1981 through 2006; (b) Real coefficients of the wavelet transform with time, (c) Wavelet frequency spectra

Fig.2 Regression coefficient of MSU CH3 Temperature with the Solar cycle and ENSO forcings in winter (Dec-Feb) and summer (Jul-Sep) . Solar cycle: (a) Winter and (b) Summer; ENSO: (c) Winter and (d) Summer. The shaded areas indicate the statistical significance at the 95% level.

Corr. Solar F10.7cm Flux vs temperature Corr. ENSO vs temperature

a
60-90N 30-60N 0-30N 0-30S 30-60S 60-90S 95% 0 Corr. ENSO vs temperature (interannual ) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.1 0 1 2 3 4 5 6 7 8 Lag (month) 9 1 2 3 4 9 10 11 12 5 6 7 8 Lag (month)

1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1

12

24

36

48

60

72

84

96 108 120

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.1

Lag (month)

Corr. Solar F10.7cm Flux vs temperature ( interdecadal )

1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1

60-90N 30-60N 0-30N 0-30S 30-60S 60-90S 95% 10 11 12

12

24

36 48 60 72 Lag (month)

84

96 108 120

Fig. 3 Lag correlation between F10.7cm solar flux and (a) MSU CH3 monthly temperature, (b) reconstructed interdecadal temperature; and the lag correlation between the ENSO index and (c) raw monthly temperature , (d) reconstructed interannual temperature. X coordinate indicates the lead time (month) by solar cycle or ENSO forcing . Dotted line indicates the 0.5 correlation coefficient at the 99% significant test .