Among the findings reported in the Summary for Policymakers of the 2001 Third Assessment Report of the Intergovernmental panel on Climate Change is that "Warm episodes of the El Niqo-Southern Oscillation (ENSO) phenomenon ... have been more frequent, persistent and intense since the mid-1970's compared to the previous 100-years." A key piece of evidence in support of this finding is Trenberth and Hoar's (1996; hereafter referred to as TH) analysis of an extended record of barometric pressure at Darwin, Australia. Pressure records for Tahiti (18oS, 150oW) and Darwin (12oS, 131oE) are representative of the opposing poles of the 'Southern Oscillation' (Walker 1924; Walker and Bliss 1932; Troup 1965; Trenberth 1976, 1984; Wright 1985; Trenberth and Shea 1987; Trenberth and Caron 2000). When smoothed as in Trenberth (1984), these time series are negatively correlated with one another at a level of -0.71 for 1935-2002 and their difference (standardized pressure at Tahiti, denoted TN, minus standardized pressure at Darwin, denoted DN, after Trenberth (1984)) is negatively correlated with sea surface temperature in the equatorial Pacific (5oN-5oS,170-120oW) at a level of -0.92 for 1950-2002 (both series smoothed as in Trenberth (1984)). This so called "Southern Oscillation Index (SOI)" has been used in numerous studies as an indicator of the status of the ENSO phenomenon as viewed from an atmospheric perspective. Mindful of the high degree of redundancy between the smoothed Darwin and Tahiti pressure records, TH elected to use the longer and more reliable Darwin record as a basis for investigating long term trends in the ENSO cycle. TH reported that Darwin pressure has tended to be above its long term normal since 1977, indicative of a prevalence of the warm (El Niqo) phase of the ENSO cycle. This tendency is clearly evident in the updated Darwin pressure time series shown in Fig. 1. Harrison and Larkin (1997), Rajagopalan et al. (1997, 1999), Wunsch (1999a,b), and Solow and Huppert (2003) have questioned the statistical significance of the prevalence of positive sea-level pressure (SLP) anomalies toward the end of the Darwin record, and Trenberth and Hoar (1997), Allan and D'Arrigo (1999), and Trenberth and Hurrell (1999a,b) have provided additional evidence in support of the TH findings. In this note we will address two different issues concerning the significance and interpretation of these findings: (1) the choice of metric that is used to describe and quantify changes in the Darwin sea-level pressure (SLP) record and (2) the interpretation of long term changes in the Darwin SLP record. TH described and tested the apparent nonstationarity of the Darwin SLP time series in terms of the occurrence of extended runs of high values, indicative of El Nino-like conditions, toward the end of the record. In particular, they noted that it remained above normal )(defined as the xxx mean) for 22 consecutive 3-month seasonal means starting in xxx and that the means for the perids extending from xxx and xxx until the end of the record were significantly above normal. What TH described as an increasing frequency of occurrence and persistence of anomalously high SLP toward the end of the Darwin record can equally well be interpreted as an upward trend, starting around the middle of the 20th century (Fig. 1). Since there is no evidence of bimodality in the Darwin SLP record, the more strightforward interpretation captures the essence of the changes that have taken place, and it is less subject to the criticisms raised by xxx. The notion of a trend also seems more appropriate for testing the hypothesis that greenhouse warming associated with the buildup of greenhouse gases in the earth's atmosphere could be a ffecting ENSO. It is easier to imagine a gradual increase in greenhouse gases might cause a gradual change in the mean state about which the ENSO cycle fluctuates than how it might affect the characteristics of warm events in particular. On the basis of the significance tests described in the Appendix, the upward trend in Darwin SLP since the mid-20th century is highly significant. Is the upward trend in Darwin SLP since the mid-20th Century indicative of a drift toward the warm polarity of the ENSO cycle? Darwin (12S,131E) is one of the opposing poles of SLP in the Southern Oscillation (Walker xxx). The station most frequently used to represent the other pole is Tahiti (18S,150W). SLP time series at thsese two stations are strongly anticorrelated and the SLP difference between them is highly correlated with equatorial Pacific sea surface temperature. The standardized difference between standardized SLP at the two stations (Tahiti minus Darin: T-D) widely refered to as the Southern Oscillation index, is thus an indicator of the status of ENSO: positive values are indicative of the cold polarity and negative values of the warm polatity. By analyzing the Tahiti SLP record in combination with the Darwin record, it is possible to distinguish between the SLP trends that are ENSO related and those associated with other phenomena. The former are well represented by time series of (T-D) and the latter by (T+D), as first proposed by Trenberth (19xx). If the upward trend in Darwin SLP since the mid-20th century were solely attributable to a drift in the ENSO cycle, it should have been adfcompanied by an equal andopposite trend in Tahiti SLP. It is evident from Fig. 1 that Tahiti has exhibited little, if any SLP trend since the mid-20th century. Hence, the trend in the SOI (T-D) is only about half as large as the trend in Darwin SLP and , based on the tests described in the Appendix, it is not statistically significant. The (T + D) time series (Fig. 1)) exhibits a pronounced upward trend which, based on the tests described in the Appendix, is even more statistically significant than the trend at Darwin. When global SLP is regressed on this time series, the resulting pattern, shown in Fig. 2, projects strongly on the Northern and Southern annular modes (e.g., see Thompson and Wallace, 2000). Darwin and Tahiti SLP both tend to be above normal when the annular modes are in their "high index" polarity, with anomalously low pressure over the polar cap regions and strong subpolar westerlies. The close agreement between the maps constructed from different data sets lends credence to this global pattern. Consistent with this result, the global patterns formed by regressing SLP upon the indices of the annular modes are characterized by positive values throughout the tropics (Baldwin 2001). The correlation coefficient between unfiltered time series of (TN + DN) and the NAM, as defined in Thompson and Wallace (2000), is 0.24 for all calendar months in 1950-2002, and the correlation between (TN + DN) and the SAM, derived from Antarctic 500 hPa geopotential heights by Thompson and Solomon (2002), is 0.20 for all calendar months in 1969-98. For reference, the correlations corresponding to a two-tailed p-value of 0.001, accounting for the month-to-month autocorrelation in the (TN + DN), NAM, and SAM time series in the manner suggested by Bretherton et al. (1999), are 0.15 and 0.18, respectively. Both Northern and Southern Hemisphere annular modes have exhibited pronounced trends toward their high-index polarity, as indicated in Table 2. The trend in the NAM has been documented by Hurrell (1995) and Thompson et al. (2000). The NCEP-NCAR reanalyses shows indications of a trend in the SAM, which is confirmed by Thompson and Solomon's (2002) analysis of station records from 1969 onward. Possible causes of these trends include stratospheric ozone depletion (Thompson and Solomon 2002; Gillett and Thompson 2003), the buildup of greenhouse gases (Shindell et al. 1999), and the trend in tropical sea surface temperatures, particularly over the Indian Ocean sector (Hoerling et al. 2001). The contribution of the trend in the annular modes to the trend in the (TN + DN) time series can be estimated as the product of the trend in the annular modes (Table 2) and the regression coefficient of (TN + DN) upon the annular mode time series. For standardized indices, the regression coefficients are the same as the correlation coefficients, listed in Table 3. For example, the contribution of the NAM to the (TN + DN) trend is 1.43 std.dev./53 years x 0.51 = 0.73 std.dev. / 53 years or 44%. Based on a similar analysis for the SAM, it appears that the annular modes account for about half of the positive trend in (TN + DN) from 1950 onward. Has the anomalous behavior of the ENSO cycle reported by TH persisted? Of the xxx 3-month seasons since the publication of TH (xxx, xxx, xxx, xxx, 2004) only xxx can be unambiguously classified as falling within warm episodes of the ENSO cycle [footnote defining what we mean by warm episode.] Official definitioin ? Of those xxx seasons, xx fell within the very strong 1997-98 El Nino event and the others within a relatively warm inteval in 2002-03, which barely qualifies as an El Nino event (e.g., see xxx). Despite the prevalence of neutral to cold conditions during this 8-year interval, Darwin SOP has averaged xxx standard deviations above its 1950-xxx normal. This seemingly contradictory behavior, reflects the anomalously high values of T-D SLP during this period. In contast to Darwin SLP, the SOI has averaged on the high (cold) side of normal, consistent with the SST data. Mike: where does the following fragment fit in: "... consistent with the equatorial Pacific SST data." Differences in analysis techniques notwithstanding, our analysis substantiates the findings of Trenberth and Hoar, that sea-level pressure at Darwin has been tended to be higher in recent decades than in the earlier part of the record. Although the prevalence of lower values of the Southern Oscillation Index from the mid 1970's through the mid '90's has contributed to this tendency, we are inclined to attribute most of it to phenomena other than ENSO: in particular, the trend in the annular modes toward their high index polarity and...... Acknowledgments TPM is funded by a grant from the NOAA CDEP program to the Center for Science in the Earth System under JISAO Cooperative agreement No. NA17RJ1232, and JMW is funded by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement No. NA17RJ1232. JISAO contribution number 1038. Appendix. Table 1 shows trends in Darwin and Tahiti sea-level pressure for the period of record 1950-2002. Specifics of how it was computed. In accordance with Troup (19xx) and Trenberth, the Darwin and Tahiti SLP time series are standardized by removing their time means and dividing them by the standard deviations.... To assess the robustness of the trends in Table 1, they were recomputed for the periods of record 1935-2002, and for the corresponding intervals ending with the publication of the TH paper (1950-1995 and 1939-1995). Trends for the interval 1935-2002 were also recomputed using cosine tapering: i.e., the time series x(t) were multiplied by the function cos (pi t / L) where time t is measured relative to the middle of the record and L is the length of the record. In all cases, the results were qualitatively consistent with Table 1, - p-values for the trend in Darwin SLP were near or below 0.05 - the Tahiti trend very small - the trend in (T + D) highly significant with p-values below 0.xx, and - the trend in (T - D) (the SOI) was not significant: the lowest p-value was 0.xx, for the 1950-96 period of record. REFERENCES Allan, R. J., and R. D. D'Arrigo, 1999: "Persistent" ENSO sequences: How unusual was the 1990-1995 El Niño? Holocene, 9, 101-118. Baldwin, M. P., 2001: Annular modes in global daily surface pressure. Geophys. Res. Lett., 28, 4114-4118. Bretherton, C. S., M. Widmann, V. P. Dymnikov, J. M. Wallace, and I. Bladé, 1999: The effective number of spatial degrees of freedom of a time-varying field. J. Climate, 12, 1990-2009. Compagnucci, R. H., M. A. Salles, P. O. Canziani, 2001: The spatial and temporal behaviour of the lower stratospheric temperature over the Southern Hemisphere: the MSU view. Part I: data, methodology and temporal behaviour. Int. J. Climatol., 21, 419-437. Gillett, N. P., and D. W. J. Thompson, 2003: Simulation of recent Southern Hemisphere climate change. Science, 302, 273-275. Harrison, D. E., and N. K. Larkin, 1997: Darwin sea level pressure, 1876-1996: Evidence for climate change? Geophys. Res. Lett., 24, 1779-1782. Hartmann, D. L., J. M. Wallace, V. Limpasuvan, D. W. J. Thompson, and J. R. Holton, 2000: Can ozone depletion and greenhouse warming interact to produce rapid climate change? Proc. Nat. Acad. Sci., 97, 1412-1417. Hoerling, M. P., J. W. Hurrell, and T. Xu, 2001: Tropical origins for recent North Atlantic climate change. Science, 292, 90-92. Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation regional temperatures and precipitation. Science, 269, 676-679. Intergovernmental Panel on Climate Change, 2001. Climate change 2001. The scientific basis. Cambridge Univ. Press. 881 pp. Jones, R. H., 1975: Estimating the variance of time averages. J. Appl. Meteor., 14, 159-163. Kidson, J. W., 1988: Interannual variations in the Southern Hemisphere circulation. J. Climate, 1, 1177-1198. _____, 1999: Principal modes of Southern Hemisphere low frequency variability obtained from NCEP/NCAR reanalysis. J. Climate, 12, 2808-2830. Kikkawa, S., and M. Ishida, 1988: Number of degrees of freedom, correlation times, and equivalent bandwidths of a random process. IEEE Trans. Inf. Theory, 34, 151-155. Kistler, R., W. Collins, S. Saha, G. White, J. Woollen, E. Kalnay, M. Chelliah, W. Ebisuzaki, M. Kanamitsu, V. Kousky, H. van den Dool, R. Jenne, and M. Fiorino, 2001: The NCEP-NCAR 50-year reanalysis: Monthly means cd-rom and documentation. Bull. Amer. Meteor. Soc., 82, 247-268. Rajagopalan, B., U. Lall, and M. A. Cane, 1997: Anomalous ENSO occurrences: An alternative view. J. Climate, 10, 2351-2357. _____, U. Lall, and M. A. Cane, 1999: Comment on "Reply to the comments of Trenberth and Hurrell." Bull. Amer. Meteor. Soc., 80, 2726. Santer, B. D., T. M. L. Wigley, J. S. Boyle, D. J. Gaffen, J. J. Hnilo, D. Nychka, D. E. Parker, and K. E. Taylor, 2000: Statistical significance of trends and trend differences in layer-average atmospheric temperature time series. J. Geophys. Res., 105, D6, 7337-7356. Shindell, D. T., R. L. Miller, G. Schmidt, and L. Pandolfo, 1999: Simulation of recent northern winter climate trends by greenhouse-gas forcing. Nature, 399, 452-455. Solow, A. R., and A. Huppert, 2003: On non-stationarity of ENSO. Geophys. Res. Lett., 30, 1910, doi:10.1029/2003GLO18225. Thompson, D. W. J., and J. M. Wallace, 1998: The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 25, 1297-1300. Thompson, D. W. J., and J. M. Wallace, 2000: Annular modes in the extratropical circulation. Part I: month-to-month variability. J. Climate, 13, 1000-1016. _____, and S. Solomon, 2002: Interpretation of recent Southern Hemisphere climate change. Science, 296, 895-899. _____, J. M. Wallace, and G. C. Hegerl, 2000: Annular modes in the extratropical circulation. Part II: Trends. J. Climate, 13, 1018-1036. Trenberth, K. E., 1976: Spatial and temporal variations of the Southern Oscillation. Quart. J. Roy. Meteor. Soc., 102, 639-653. _____, 1984: Signal versus noise in the Southern Oscillation. Mon. Wea. Rev., 112, 326-332. _____, and D. J. Shea, 1987: On the evolution of the Southern Oscillation. Mon. Wea. Rev., 115, 3078-3096. _____, and T. J. Hoar, 1996: The 1990-1995 El Niño - Southern Oscillation event: Longest on record. Geophys. Res. Lett., 23, 57-60. _____, and _____, 1997: El Niño and climate change. Geophy. Res. Lett., 24, 3057-3060. Trenberth, K. E., and J. W. Hurrell, 1999a: Comments on "The interpretation of short climate records with comments on the North Atlantic and Southern Oscillations." Bull. Amer. Meteor. Soc., 80, 2721-2722. _____, and _____, 1999b: Reply to Rajagopalan, Lall, and Cane's comment about "The interpretation of short climate records with comments on the North Atlantic and Southern Oscillations." Bull. Amer. Meteor. Soc., 80, 2726-2728. _____, and J. M. Caron, 2000: The Southern Oscillation revisited: Sea level pressures, surface temperatures and precipitation. J. Climate, 13, 4358-4365. Troup, A. J., 1965: The Southern Oscillation. Quart. J. Roy. Meteor. Soc., 91, 490-506. Walker, G. T., 1924: Correlation in seasonal variations of weather. IX: A further study of world weather (World Weather II). Mem. Indian Meteor. Dept., 24, 275-332. _____, and E. W. Bliss, 1932: World Weather V. Mem. Roy. Meteor. Soc., 4, 53-84. Woodruff, S. D., H. F. Diaz, J. D. Elms, and S. J. Worley, 1998: COADS Release 2 data and metadata enhancements for improvements of marine surface flux fields. Phys. Chem. Earth, 23, No. 5-6, 517-526. Wright, P. B., 1985: The Southern Oscillation: An ocean-atmosphere feedback system? Bull. Amer. Meteor. Soc., 66, 398-412. Wunsch, C., 1999a: The interpretation of short climate records with comments on the North Atlantic and Southern Oscillations. Bull. Amer. Meteor. Soc., 80, 245-256. _____, 1999b: Reply to comments on "The interpretation of short climate records with comments on the North Atlantic and Southern Oscillations." Bull. Amer. Meteor. Soc., 80, 2723. Zhang, Y., J. M. Wallace, and D. S. Battisti, 1997: ENSO-like interdecadal variability: 1900-93. J. Climate, 10, 1004-1020. Figure Captions FIG. 1. Standardized monthly Darwin pressure anomalies for 1882-2002. The series is filtered and standardized as for the shaded curve of Fig. 2 of TH (1882-1981 base period). The first and last 5 months are omitted to remove possible endpoint effects. FIG. 2. Monthly Tahiti and Darwin sea-level pressure time series (hPa) and the SOI (non-dimensional) for 1935-2002. The Tahiti and Darwin traces are obtained by smoothing each series with a 13-month running mean to remove much of the annual cycle, and the resultant Gibbs oscillations (Trenberth 1984) are greatly diminished with a subsequent 9-month running mean. The SOI is defined and filtered as in Trenberth (1984), with the normalization based on the standard deviation of all calendar months combined. The first and last 10 (5) months of Tahiti and Darwin (SOI) series are omitted to remove possible endpoint effects. The lightly shaded vertical bars are intended to highlight the large swings of the ENSO cycle. FIG. 3. Time series of monthly SOI and (TN + DN) (non-dimensional) for 1935-2002, and marine SLP anomalies (hPa) averaged over 20°N-20°S for 1950-2002, all smoothed as in Trenberth (1984). Anomalies and normalization are with respect to 1950-2002 for all series, and the first and last 5 months of the smoothed series are omitted to remove possible endpoint effects. A plot of smoothed SOI and (TN + DN) for 1935-82 is presented in Trenberth (1984). The marine observations are the Comprehensive Ocean-Atmosphere Data Set (COADS, Woodruff et al. 1998) for 1950-97 and the National Centers for Environmental Prediction (NCEP) marine real-time data (http://www.cdc.noaa.gov/cdc/data.nmc.marine.html) for 1998-2002. FIG. 4. NCEP-NCAR reanalysis (contours) and COADS/NCEP marine real-time (shading) SLP anomalies regressed onto standardized (TN + DN) (contour and shading intervals of 0.25 hPa per one standard deviation of the index), based on unfiltered monthly data for 1950-2002. The reanalysis data are 2.5° latitude-longitude resolution. The COADS (NCEP marine real-time) data span 1950-97 (1998-2002), and the original anomalies at 2° latitude-longitude resolution have been averaged into 4 by 6° latitude-longitude regions to reduce the noisiness. Tables Table 1. Trends and associated significance p-values for normalized, annual-mean Darwin, Tahiti, and SOI pressure anomalies for 1935-2002. Yearly values of Darwin and Tahiti are obtained as January through December averages, and they are then normalized by their respective standard deviations (0.67 and 0.51 hPa for Darwin and Tahiti, respectively). These indices are also used to construct an SOI, which is then normalized by the standard deviation of its values (see also Trenberth (1984) and TH). The resultant Darwin, Tahiti, and SOI series have means 0 and standard deviations 1. The trend is estimated with the method of least squares, the trend standard error includes the variance of the trend residual (following Santer et al. 2000), the degrees of freedom are for an effective sample size of order 1 (Jones 1975, Kikkawa and Ishida 1988, Bretherton et al. 1999), and the p-values are for two-tailed tests. Trend Degrees of Variable (std.dev./68 years) Freedom t-value p-value Darwin 0.80 41 1.51 0.14 Tahiti 0.04 46 0.08 SOI -0.41 45 -0.78 0.44 Table 2. As in Table 1, but for 1950-2002 and also including (TN + DN), 20°N-20°S average marine SLP anomaly, NAM, and SAM (1969-98 only) indices. Trend Degrees of Variable (std.dev./53 years) Freedom t-value p-value Darwin 1.30 34 2.31 0.03 Tahiti -0.04 32 -0.07 SOI -0.72 34 -1.22 0.23 (TN+DN) 1.67 27 2.79 0.01 20°N-20°S 0.94 33 1.59 0.12 NAM 1.43 31 2.46 0.02 SAM 1.67 30 1.52 0.14 Table 3. Correlation coefficients x 100 between Darwin, Tahiti, SOI, (TN + DN), 20°N-20°S average SLP, NAM, and SAM (1969-98 only) for a) unfiltered monthly- and b) annual-mean pressure indices for 1950-2002. a) Monthly-means Tahiti SOI TN+DN 20°N-20°S NAM SAM Darwin -34 -82 58 55 9 12 Tahiti 82 58 12 19 10 SOI 0 -26 6 -2 (TN+DN) 58 24 20 20°N-20°S 21 30 NAM -2 b) Annual-means Tahiti SOI TN+DN 20°N-20°S NAM SAM Darwin -72 -93 38 55 18 28 Tahiti 93 38 -29 21 -20 SOI 0 -46 2 -26 (TN+DN) 34 51 12 20°N-20°S 28 28 NAM 4 Table 4. Darwin SLP and SOI t-values, and two-tailed significance p-values for the difference of 1977-2002 and 1950-1976 means, and for the linear trend for 1950-2002. The degrees of freedom estimates are the same for both metrics, are derived as in Table 1, and are 34 for both Darwin and the SOI. Trend statistics are from Table 2. 1977-2002 minus 1950-1976 t-value p-value Darwin 2.71 0.01 SOI -2.05 0.05 1950-2002 trend Darwin 2.31 0.03 SOI -1.22 0.23