Geophysical Research Letters

Relating the strength of the tropospheric biennial oscillation (TBO) to the phase of the Interdecadal Pacific Oscillation (IPO)



[1] The phase of the Interdecadal Pacific Oscillation (IPO) can influence the Tropospheric Biennial Oscillation (TBO) such that a shift from a negative to positive phase of the IPO was associated with a weakening of the TBO after the mid-1970s. Here it is shown that the most recent transition in the late 1990s of the IPO from positive to negative was associated with a larger-amplitude TBO in the Indo-Pacific region, thus confirming and strengthening the previous results by extending them to the opposite phase of the IPO. A major contributor to this change in the TBO has been an ongoing increase of sea surface temperatures (SSTs) in the Indian Ocean that contributes to stronger trade winds in the Pacific, one of the processes previously identified with strengthening the TBO. Such modulation of interannual variability by decadal timescale processes has implications for understanding possible skill of decadal climate predictions.

1. Introduction

[2] A previous study [Meehl and Arblaster, 2011] noted that the mid-1970s climate shift [e.g.,Trenberth and Hurrell, 1994; Wang and An, 2001], when the Interdecadal Pacific Oscillation (IPO) transitioned from negative to positive (i.e. somewhat cooler to somewhat warmer tropical Pacific SSTs), was associated with interannual climate variability in the Indo-Pacific region becoming less biennial. In other words, the Tropospheric Biennial Oscillation (TBO) [e.g.,Meehl, 1987, 1997; Meehl et al., 2003; Li et al., 2001] became weaker. The IPO is the dominant feature of decadal variability of Pacific basin SSTs [Power et al., 1999], and is typically defined as the second empirical orthogonal function (EOF) of low-pass-filtered SSTs of near-global SSTs [e.g.,Parker et al., 2007], with the first EOF associated with the trend. In long climate model control runs, the IPO defined in this way is typically the first EOF, thus being the dominant mode of internally-generated decadal timescale variability in the Pacific region [Meehl et al., 2009; G. A. Meehl et al., Externally forced and internally generated decadal climate variability associated with the Interdecadal Pacific Oscillation, submitted to Journal of Climate, 2012]. The IPO is closely related to the Pacific Decadal Oscillation (PDO) [e.g., Mantua et al., 1997] and has a similar pattern, though the PDO is defined based on North Pacific SSTs, and consequently doesn't have as much power in the tropical and South Pacific as the IPO. The IPO has been connected to modulating the El Niño Southern Oscillation (ENSO) teleconnections from Australia to the tropical Pacific [Power et al., 1999; Arblaster et al., 2002], as well as to Indian monsoon precipitation [Meehl and Hu, 2006] and precipitation over the southwest U.S. [Meehl and Hu, 2006; Dai, 2012]. Though there have been studies making the case that the IPO is the low-frequency aspect of ENSO [e.g.,Jin, 1997], there have been papers that have outlined a distinct mechanism that can produce such decadal timescale variability in the Pacific [e.g., Meehl and Hu, 2006]. However, as with all internally generated variability of the climate system, the IPO operates in the context of changes in external forcing. In particular, the mid-1970s shift was shown to likely have involved both an internally generated component associated with the IPO, and an externally forced component related to increasing greenhouse gases in the atmosphere [Meehl et al., 2009].

[3] The mechanism of the TBO involves large-scale dynamically coupled interactions [Meehl, 1987, 1997; Meehl and Arblaster, 2002a, 2002b; Loschnigg et al., 2003]. It is based on the premise of air-sea coupling being particularly strong one season per year in the Indo-Pacific region during the passage of the “convective maximum”, a mass of convection and precipitation that moves with the seasonal cycle from the Indian monsoon in the northern summer southeastward to the Australian monsoon in southern summer and involving processes in the western tropical Pacific [Li et al., 2006; Annamalai and McCreary, 2005; Kug and Kang, 2006; Kug et al., 2006].

[4] The TBO has been noted to be the dynamical framework for regional processes related to meridional temperature gradients affecting the monsoon in the Indian sector [Meehl, 1994; Chang and Li, 2000], patterns of SST anomalies that involve El Niño and La Niña in the Pacific, and the Indian Ocean Dipole (IOD) [e.g., Webster et al., 1999; Saji et al., 1999]. Izumo et al. [2010] made use of the biennial transition processes in the TBO to document El Niño forecast skill by monitoring the state of the IOD in the northern fall.

[5] Given these seemingly self-sustaining processes,Meehl and Arblaster [2011]asked the question as to why the Indo-Pacific climate system isn't perfectly biennial. Among the possible causes, decadal-timescale internally-generated variability associated with the IPO was noted to have the potential to disrupt or modify biennial variability associated with the TBO in the Indo-Pacific region. For example, for the mid-1970s shift, a warmer tropical Pacific in relation to the tropical Indian Ocean weakened the trade winds, reduced coupling strength, and contributed to the Pacific SSTs and Indian monsoon becoming less biennial. After the late-1990s transition of the IPO from positive to negative [Lee and McPhaden, 2008; Dai, 2012; W. Han et al., Pacific decadal sea level change patterns associated with a warming Indo-Pacific Warm Pool, submitted toJournal of Climate, 2012], it has been noted that the equatorial eastern Pacific SSTs have returned to being more biennial [Barnston et al., 2012]. The contribution from the Indo-Pacific SST gradient to trade-wind strength has likely contributed to increasing the strength of the TBO as will be discussed below.

2. Data and Methods

[6] Observed SSTs used in this study are from Hurrell et al. [2008]which merges the Hadley Centre sea ice and SST dataset version 1 (HadISST1) and version 2 of the National Oceanic and Atmospheric Administration (NOAA) weekly optimum interpolation (OI) SST analysis. All-India rainfall data are described byParthasarathy et al. [1994], with more recent updates available from the Indian Institute of Tropical Meteorology ( The following caveat from the Indian Institute of Tropical Meteorology accompanies those data: “The data for the recent period 1991–2010 are preliminary estimates based on the subdivisional means supplied by the India Meteorological Department (IMD), which are in turn based on a variable network. However, the IMD data have been rescaled to conform to the long-term means of the respective subdivisions in the IITM-IMR data set. These data will be updated as and when the full set of data for 306 stations becomes available.”. The IPO is defined as the second EOF of 13-year low-pass–filtered seasonal near-global observed SSTs, which is well-separated from the third EOF according to the criteria ofNorth et al. [1982]. The Lanczos filter uses 73 weights, so 36 seasons are lost off the beginning and end of the time series. Power spectra are calculated using the Morlet wavenumber-6 wavelet analysis, following the methods ofWittenberg [2009].

3. The Late-1990s Climate Shift in the Pacific and the Effects on the TBO and Asia-Pacific Climate

[7] The late-1990s saw a shift of climate in the Indo-Pacific region, with the IPO transitioning from positive to negative, featuring tropical Pacific SSTs that went from anomalously warm to cool [e.g.,Lee and McPhaden, 2008; Dai, 2012]. This can be seen in Figure 1awhere pattern of the IPO and its PC time series illustrate the mid-1970s shift from negative to positive, and the late-1990s shift from positive to negative. Due the low-pass filter, the last date in the PC time series is DJF, 2003. Consequently the connections described here after the IPO shift around 1999 apply mainly to the early 2000s.

Figure 1.

(a) Second EOF of low-pass filtered seasonal SSTs that define the IPO and (b) PC time series of the second EOF of low-pass filtered seasonal SSTs. TheHurrell et al. [2008] SST dataset is a merge of the HadISST and NOAA OI.v2 SSTs, and the 1870–2010 period is used.

[8] The decadal timescale variability in Figure 1can also be seen in epoch differences of observed SSTs in the Indo-Pacific region inFigure 2. The mid-1970s shift is seen inFigure 2a as the difference of observed SSTs for the period 1976–1999 minus the period 1958–1975 (Figure 2a) that shows the warming of SSTs in the tropical Pacific and Indian Ocean (e.g., Wang and An, 2001). The late-1990s shift is shown inFigure 2b and is characterized by cooling of SSTs in the tropical eastern Pacific, while the Indian Ocean continues to warm.

Figure 2.

Epoch differences of SSTs from HadISST data (°C): (a) 1976–1999 minus 1958–1975 and (b) 2000–2010 minus 1976–1999.

[9] This transition from positive to negative IPO in the late-1990s, following the arguments presented byMeehl and Arblaster [2011], should produce a larger amplitude TBO in the early 2000s with the Indian monsoon and Nino3.4 SSTs becoming more biennial after the late-1990s. To investigate this possibility,Figure 3 shows wavelet spectra [Wittenberg, 2009] for All-India rainfall (Figure 3a) and Nino3.4 SSTs (Figure 3b). The wavelet spectra for four periods are depicted in Figure 3. The reference climatology from 1871–2006 is shown as the solid black line with positive (negative) IPO periods in red (blue). The 2-year wavelet spectrum band representing the TBO in the Indo-Pacific region (denoted by the vertical black lines inFigures 3a and 3b) shows that the pre-1970s shift and post-1990s shift periods, during times of negative IPO (Figure 1b), show larger amplitude TBO than times of positive IPO during 1895–1910 and 1976–1999. For both negative IPO periods, the wavelet spectra amplitudes for the TBO 2-year band are greater than the long-term climatology, and are more than double the amplitude of the 1976–1999 period of weak TBO for Indian monsoon rainfall (Figure 3a). For Nino3.4 SSTs the TBO amplitude in the wavelet spectra 2-year band is roughly twice that for the post-1990s shift period and nearly a factor of three larger for the pre-1970s shift period (Figure 3b). The other prominent positive IPO period in Figure 1 from 1895 to 1910 is consistent with the positive IPO period of 1976–1999, and shows a much lower TBO amplitude than the negative IPO periods.

Figure 3.

Global wavelet spectra, the climatology from 1871–2006 is shown as the solid black line with positive (negative) IPO periods in red (blue) as denoted in each panel for (a) monthly all-India rainfall. (b) Same as Figure 3a except for Nino3.4 SSTs. The TBO is represented by the 2 year band denoted by the vertical black line in each panel. A Monte Carlo analysis randomly re-sampled the data over the 1870–2010 period, recalculated the spectra, and was repeated 1000 times to form a distribution shown by the light gray lines, with thin black lines at the 5% and 95% levels.

[10] In order to test the significance of these various spectral peaks, a Monte Carlo analysis was performed by randomly re-sampling the data over the 1870–2010 period and recalculating the spectra. This was repeated 1000 times to form a distribution. The spectral peak at 2 years falls in the upper 95% of the distribution for both the 1958–1975 and 2000–2010, i.e., negative IPO, periods. For the positive IPO periods the power of the spectral peak at 2 years is either within the 5–95% of the distribution or even weaker.

[11] These results confirm what was posed by Meehl and Arblaster [2011] for the weakening of the TBO in Indian monsoon rainfall and Nino3.4 SSTs going from negative to positive IPO around the time of the 1970s shift. This also shows for the first time that a similar principle applies to the opposite situation when the IPO transitions from positive to negative in the late 1990s shift, and there is a stronger TBO evidenced in monsoon rainfall and eastern equatorial Pacific SSTs.

[12] Meehl and Arblaster [2011]summarize factors that could contribute to changes in climate base state on decadal timescales, such as those involved with the IPO, that could affect the amplitude of the TBO. All of these operate under the influence of the changing base state due to increasing greenhouse gases (GHGs) in the atmosphere. Just as the mid-1970s shift likely had a contribution from the IPO transitioning naturally from negative to positive, as well as a contribution from increasing GHGs, such that both these influences were acting to warm tropical Pacific SSTs, the IPO transition in the late-1990s also likely had contributions from natural and external factors. However, for a transition from positive to negative IPO, the internally generated processes that would drive the SSTs toward cooling must overcome the externally forced pattern in the tropical Pacific from increasing GHGs that would tend to warm those SSTs [Meehl et al., 2009].

[13] As noted in Figure 2, while decadal variability marked the epochs of positive and negative IPO in the tropical Pacific, the tropical Indian Ocean warmed throughout the entire period consistent with the response to ongoing increases of GHGs. This warming, along with the transition to negative IPO in the Pacific, produces an enhanced east-west SST gradient across the Indo-Pacific region prior to the mid-1970s and again early in the 21st century, and contributes to strengthened trade winds as well as higher sea levels in the western Pacific in the recent period (Han et al., submitted manuscript, 2012). Such stronger trades are one of the ingredients that contribute to a larger-magnitude TBO through increased coupling strength as documented in observations and model simulations [Yu et al., 2009; Meehl and Arblaster, 2011]. The relatively warmer conditions in the far western Pacific and tropical Indian Ocean also shift the ascending branch of the Walker Circulation westward, a condition identified by Meehl and Arblaster [2011] as a contributing factor to a stronger TBO.

4. Conclusion

[14] Analysis of decadal timescale variability of the Interdecadal Pacific Oscillation shows that a negative phase of the IPO (with lower than normal SSTs in the tropical Pacific on decadal timescales) is associated with a larger amplitude TBO (stronger biennial variability of the Indian monsoon and Nino3.4 SSTs), and vice versa for a positive phase of the IPO. Thus, decadal climate prediction efforts involving the evolution of the IPO [e.g., Meehl et al., 2010] could provide useful information regarding near-term seasonal-to-interannual variability associated with the TBO.


[15] Portions of this study were supported by the Office of Science (BER), U.S. Department of Energy, Cooperative Agreement DE-FC02-97ER62402, and the National Science Foundation. The National Center for Atmospheric Research is sponsored by the National Science Foundation. We thank the Indian Institute of Meteorology (IITM) for making their Indian regional/subdivisional Monthly Rainfall data set freely available and Dennis Shea for recent updates to theHurrell et al. [2008] SST dataset. Figures were created with the NCAR Command Language (Version 6.0.0) [software], (2012), Boulder, Colorado: UCAR/NCAR/CISL/VETS: Wasyl Drosdowsky, Robert Fawcett and two anonymous reviewers provided constructive comments on the manuscript.

[16] The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.