The impact of the Madden-Julian Oscillation trend on the Arctic amplification of surface air temperature during the boreal winter

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 38,, doi: /2011gl049881, 2011 The impact of the Madden-Julian Oscillation trend on the Arctic amplification of surface air temperature during the boreal winter Changhyun Yoo, 1 Steven Feldstein, 1 and Sukyoung Lee 1 Received 5 October 2011; revised 8 November 2011; accepted 8 November 2011; published 20 December [1] One of the most prominent and important features of climate change is that surface air temperature (SAT) change is greatest at high latitudes. The cause for this Arctic amplification of SAT is uncertain. Using ERA-Interim reanalysis data, we show that Arctic amplification during the past 30 years (1979 to 2008) is linked to the Madden-Julian Oscillation (MJO), the primary mode of intraseasonal variability in the tropics. Specifically, it is shown that interdecadal changes in the frequency of occurrence of individual MJO phases have had considerable influence on the Arctic warming during the boreal winter. During that time period, MJO phases 4 6 exhibited a large increase and phases 1 2 amod- erate decrease in their frequency of occurrence. Time lagged composites of the SAT show that MJO phases 4 6, which correspond to enhanced localized tropical heating, are followed 1 2 weeks later by Arctic warming. Similarly, MJO phases 1 2, which are associated with more zonally uniform tropical heating, are followed by Arctic cooling. These relationships between the Arctic SAT and the spatial structure of the tropical heating are consistent with the poleward propagation mechanism of Lee et al. (2011a, 2011b). By incorporating both the trend in MJO phase and the intraseasonal SAT anomaly associated with the MJO, it was found that the MJO-induced SAT trend accounts for 10 20% of the observed Arctic amplification over the Arctic Ocean. Citation: Yoo, C., S. Feldstein, and S. Lee (2011), The impact of the Madden-Julian Oscillation trend on the Arctic amplification of surface air temperature during the boreal winter, Geophys. Res. Lett., 38,, doi: /2011gl Introduction [2] Arctic amplification, i.e., the interdecadal trend in surface air temperature (SAT) being larger at high latitudes than at other latitudes, is one of the most prominent and robust features of climate change over the last century [e.g., Johannessen et al., 2004; Serreze and Francis, 2006]. Despite its urgent implications, including an increased melting rate of the Greenland ice sheet and reduction in Arctic biodiversity, as well as alterations of the atmospheric circulation [Arctic Climate Impact Assessment, 2005], its cause is not well understood. Positive surface albedo feedback (SAF) [Budyko, 1969; Sellers, 1969], which is associated with seaice and snow-cover retreat, has been supported by some climate model simulations [Holland and Bitz, 2003; Hall, 1 Department of Meteorology, Pennsylvania State University, University Park, Pennsylvania, USA. Copyright 2011 by the American Geophysical Union /11/2011GL ]. However, there are model simulations which can successfully produce Arctic amplification without SAF, leading to the interpretation that the atmospheric circulation is in part responsible for the Arctic amplification [e.g., Alexeev et al., 2005; Cai and Lu, 2007]. [3] Observational datasets also show contradictory evidence as to whether Arctic amplification is caused by SAF or by changes in the atmospheric circulation. In their analysis of ERA-40 reanalysis data, Graversen et al. [2008] found that Arctic warming occurs well above the surface, hence they concluded that SAF cannot account for the warming and instead attributed the trend to an accelerated atmospheric moist energy transport from lower latitudes [Graversen, 2006]. However, this claim was questioned by Screen and Simmonds [2010], who concluded that reduction in sea ice cover played the major role in the trend, because the warming is strongest at the surface (in ERA-Interim reanalysis data) where fluxes have their greatest impact, and because of the large linear congruence between the declining sea ice and air temperature. On the other hand, the results from recent observational and modeling studies suggest that poleward heat transport associated with Rossby waves excited by tropical convection also plays an important role for the enhanced high latitude warming [Ding et al., 2011; Lee et al., 2011a, 2011b; Schneider et al., 2011]. In this paper, we present further evidence of this linkage between tropical convection and Arctic amplification by showing that interdecadal variability of the Madden-Julian Oscillation (MJO), the most prominent mode of intraseasonal variability in the tropics [Madden and Julian, 1994], makes an important contribution to Arctic amplification. [4] The recent study of Lee et al. [2011a] found evidence that Arctic amplification arises from changes in the frequency of occurrence of a few intraseasonal time-scale teleconnection patterns associated with increased convective precipitation over the Indian and western Pacific Oceans. Using a coupled self-organizing map analysis of 250-hPa streamfunction and tropical convective precipitation, they showed that the interdecadal trend in the Northern Hemisphere (NH) winter streamfunction field during 1959 to 2001 is associated with an increase in the frequency of patterns that resemble the positive Pacific/North American (PNA) and circumglobal streamfunction patterns [Branstator, 2002], and a corresponding decrease in the frequency of PNA and circumglobal streamfunction patterns of opposite sign. The high-latitude warming takes place through an increased adiabatic warming, poleward stationary eddy heat flux, and downward infrared radiative flux. This mechanism of tropical heating-driven Arctic amplification was first proposed by Lee et al. [2011b]. In that study, the mechanism was tested 1of6

2 Figure 1. The P2-P1 SAT, where P1 and P2 correspond to the boreal winter (November through March) for and , respectively. The contour interval is 0.5 C. Zero contours are omitted. by imposing zonally localized convective heating in an atmosphere-mixed layer ocean model of the Cretaceousearly Cenozoic equable climate. [5] Because the MJO is associated with zonally localized convective heating, the findings of Lee et al. [2011a] allude to the possibility that Arctic amplification may be related to interdecadal variability of the MJO. Indeed, the MJO has considerable influence on the extratropical atmospheric circulation, possibly through Rossby wave propagation [e.g., Matthews et al., 2004]. Recent studies have shown that the positive (negative) phase of the North Atlantic Oscillation (NAO) occurs with increased likelihood 1 2 weeks after MJO phases 3 4 (7 8), which corresponds to enhanced convection over the Indian (western Pacific) Ocean [Cassou, 2008; Lin et al., 2009] (these studies use the index of Wheeler and Hendon [2004] to define the MJO). In addition, the positive (negative) PNA has been observed to occur more frequently 1 2 weeks after MJO phases 5 8 (1 4) [Johnson and Feldstein, 2010]. Moreover, two independent studies of daily SAT station data over North America for the winter season revealed systematic variation of the SAT spatial pattern contingent upon the MJO phase [Vecchi and Bond, 2004; Lin and Brunet, 2009]. [6] Motivated by these previous studies, we investigate whether changes in the frequency of different MJO phases also contribute toward Arctic amplification. To address this problem, in this study, we will compare the interdecadal SAT trend associated with these changes in the MJO to the full interdecadal SAT trend. 2. Data and Methods [7] To study the MJO and its relationship with the SAT, we use 30 boreal winters (November through March during 1979 to 2008) of the daily multivariate MJO index [Wheeler and Hendon, 2004] ( The MJO index is defined by the principal components of the two leading combined EOFs of the intraseasonal 200- and 850-hPa zonal winds and outgoing longwave radiation (OLR), averaged over the band from 15 S to 15 N. The MJO has eight phases, following Wheeler and Hendon [2004, see Figure 8], with phases 2 4 corresponding to enhanced convection over the Indian Ocean and the western Maritime Continent, and phases 5 7 to strengthened convection over the eastern Maritime Continent and the western Pacific Ocean. [8] For daily SAT, we use the European Center for Medium-Range Weather Forecasts ERA-Interim (1979 to 2008) reanalysis [Dee and Uppala, 2009]. It will be shown that our results are insensitive to the choice of reanalysis dataset (see the auxiliary material). 1 The interdecadal trend in SAT is defined as the difference between the time mean for two different 15-year winter periods: 1979 to 1993 (P1) and 1994 to 2008 (P2) (see Figure 1). To examine the intraseasonal time scale SAT changes associated with the MJO, the seasonal cycle is removed at each grid point by subtracting the first three harmonics of the calendar mean for each day. This is followed by the application of a 101-point, day, band-pass digital filter. 3. Results [9] We start by probing the P2-P1 SAT change, which shows widespread warming over the Barents and Kara seas and much of the North American and Eurasian landmasses (about 1 4 C) (Figure 1). A comparison of the frequency of occurrence of those days that the MJO amplitude exceeds a value of 1.5 (our definition of an active MJO, which corresponds to 30% of the days within the dataset) for P2 (blue bars in Figure 2) and P1 (red bars in Figure 2) shows a notable increase for phases 4 6, a moderate decrease for phases 1 2, and minimal change for the other phases. The changes in the frequency of occurrence for phases 1 and 4 6 in Figure 2 exceed the 95% confidence level for a Monte Carlo test based on 1000 random samples. This trend in MJO frequency of occurrence is insensitive to the threshold value. Because the average MJO amplitude over the entire period is 1.25, we have also examined the frequency of occurrence with threshold values of 1.0 and 1.25, which include about 70% and 45% of the total number of days, respectively. For both cases, it is found that the changes in the frequency of occurrence are similar to those shown in Figure 2. [10] It is important to note that the change in the frequency of occurrence of the MJO phase corresponds to an interdecadal 1 Auxiliary materials are available in the HTML. doi: / 2011GL of6

3 Figure 2. Frequency of occurrence (%) for each MJO phase. The days are included when the MJO index exceeds a value of 1.5. The red bars are for P1, and the blue bars for P2. The total numbers of days are 712 (P1) and 771 (P2). The numbers on top of the bars indicate the average amplitude for each phase of P1 (red) and P2 (blue). trend in tropical convection. Both the increased frequency of phases 4 6 and decreased frequency of phases 1 2 indicate that tropical convection, hence precipitation, at interdecadal time scales is reduced over the Indian Ocean and increased over the Maritime Continent and western Pacific Ocean. Interestingly, these interdecadal time scale changes in the intraseasonal MJO resemble the interdecadal trend obtained from other independent precipitation datasets, such as the Global Precipitation Climatology Project, and Climate- Prediction-Center Merged Analysis of Precipitation [see Lee et al., 2011a, Figures 1b and 1c]. [11] Lagged composites of the SAT associated with each of the MJO phases show systematic variation. We focus on MJO phases 1 and 5 (Figure 3), because (i) those phases show large changes in their frequency of occurrence, and (ii) in general, phases 1 4 (5 8) indicate a similar Arctic cooling (warming) 1 2 weeks after the MJO passes through that phase (not shown). As can be seen from Figure 3, for phase 1, Arctic cooling is evident on lag +5 through lag +15 days (Figure 3, left), while phase 5 shows Arctic warming on lag +10 and lag +15 days (Figure 3, right). These changes in Arctic SAT are consistent with the findings of Lee et al. [2011a, 2011b], who showed that enhanced, zonally localized convection is associated with an elevated high latitude SAT. Similar changes are observed in tropical convection associated with the MJO. For example, in the first panels of Figure 3, which show the total OLR composites on lag day 0, where total is defined as the sum of the anomalous OLR associated with the MJO plus the climatological OLR, it can be seen that phase 5 is associated with more intense and zonally localized tropical convection than phase 1. Furthermore, the time scale of the high latitude SAT response to tropical convection is in agreement with that obtained from observational and idealized model studies of tropically-forced, poleward propagating Rossby wave trains [Hoskins and Karoly, 1981; Kiladis and Weickmann, 1992]. [12] To examine the effect of the MJO changes on the interdecadal SAT trend, we incorporate the changes in the MJO frequency of occurrence (Figure 2) with the intraseasonal SAT anomaly associated with the MJO (Figure 3), and retrieve an interdecadal MJO-induced SAT trend (hereafter, for brevity, the MJO-induced SAT trend). In mathematical form, ðt 2 ðl; qþ T 1 ðl; qþþ MJO ðtþ ¼ P 8 i¼1 DT 2;i ðl; q; tþn 2;i P 8 i¼1 N 2 DT 1;i ðl; q; tþn 1;i N 1 ; ð1þ where subscripts 1 and 2 denote P1 and P2, respectively, l and q are latitude and longitude, respectively, an overbar indicates the time mean, and t is the lag day. On the righthand-side, DT 1,i and DT 2,i are the MJO SAT anomalies for P1 and P2 (as in Figure 3), respectively, and N 1,i and N 2,i are the number of days in P1 and P2, respectively, for MJO phase i, that the threshold value is exceeded (as in Figure 2). Note that because of the application of the band-pass filter (section 2), DT i does not contain information of any longterm trend. The numbers in the denominator, N 1 and N 2, correspond to the numbers of days for the entire P1 and P2, respectively. [13] It is striking that the MJO-induced SAT trend patterns over the Arctic for lag +5 days (Figure 4b) and lag +10 days (Figure 4c), along with the time average (Figure 4d), are similar to the observed P2-P1 high latitude SAT trend 3of6

4 Figure 3. Total OLR composite on lag day 0 (first panels), with lagged composites of SAT on lag days 0, 5, 10, and 15 for MJO phases (left) 1 and (right) 5. Solid contours are positive, dashed contours negative, and the zero contours are omitted. pattern (Figure 1). It is also seen that the MJO-induced SAT trend shows cooling, primarily over the continents, overlapping somewhat with those regions where the P2-P1 SAT trend (Figure 1) is small. The statistical significance of individual grid points is also examined with a Monte Carlo test. For this test, the same calculation as in (1) was performed 1000 times, using randomly selected days, with the same total number of days in each time period. As can be seen in Figures 4a 4d, at high latitudes, the majority of grid points show statistical significance above the 90% level. [14] The contribution of the MJO-induced SAT trend to the total SAT trend, expressed as a percentage of the total trend, is 10 to 20% at high latitudes (Figure 4e). (The region centered at 110 W, 80 N with greater than 40% contribution from the MJO-induced warming has a total trend that is close to zero.) In addition, the 300-hPa geopotential trend associated with the MJO, which is calculated using the same methodology as that for the MJO-induced SAT, shows a clear signal of poleward Rossby wave propagation (Figure 4f ). Based on the findings of Lee et al. [2011a, 2011b], the results in Figure 4f suggest that the Arctic amplification associated with the MJO arises from changes to poleward propagating wave trains, which warm high latitudes via an increased poleward eddy heat flux and eddy-induced adiabatic warming. [15] This P2-P1 and MJO-induced SAT trends are found to be insensitive to the choice of reanalysis dataset: we repeated the calculations for Figures 1 and 4 with NCEP/ DOE reanalysis data [Kanamitsu et al., 2002], and found that the MJO-induced SAT trends are very similar to those shown in Figures 1 and 4 (see Figure S1 in the auxiliary material). Also, it is found that the SAT changes associated with the MJO (Figure 3) and MJO-induced SAT trend (Figure 4) are insensitive to the choice of the reanalysis dataset used to calculate the MJO index, as very similar results are obtained with an MJO index formulated with ECMWF data via the same methodology as that of Wheeler and Hendon [2004]. 4. Discussion and Conclusion [16] Motivated by the growing concern over Arctic amplification, perhaps the most prominent signal of climate change today, we have investigated the extent to which interdecadal changes in the MJO, the leading mode of variability within the tropics, contribute to Arctic amplification. An MJO-induced SAT trend is calculated which takes into account both the change in frequency and the intraseasonal SAT anomaly associated with each MJO phase. [17] The two main findings of this study are: (1) The MJO-induced SAT trend explains 10-20% of the total high latitude SAT trend. Other tropical convective processes, such as non-mjo intraseasonal tropical convection and El Niño/ Southern Oscillation are also likely to make an important contribution toward Arctic amplification. (2) Changes in the frequency of occurrence of an intraseasonal time scale process such as the MJO can influence interdecadal Arctic amplification. An examination of the MJO-induced 300-hPa geopotential trend provides insight into a plausible physical 4of6

5 Figure 4. The MJO-induced SAT trend patterns obtained by the method described in the text. (a d) Lag days 0, 5, and 10, along with the time average, are shown, respectively. The contour interval is 0.03 C. The statistical significance level is indicated by dots (90% level). (e) The local ratio between the MJO-induced SAT trend (Figure 4d) and the P2-P1 SAT trend (Figure 1), expressed as a percentage, is shown. Values are not displayed when negative, and the ratio is multiplied by the sign of the MJO-induced SAT. The contour interval is 5%. (f ) The MJO-induced 300-hPa geopotential trend is shown. The contour interval is 10 m 2 s -2. mechanism, leading to the suggestion that changes in poleward propagating Rossby wave trains warm the Arctic through an enhanced poleward heat flux and adiabatic warming. [18] In this study, we have not assessed the impact of the extratropics on the MJO. The recent study of Lin and Brunet [2011], who showed that MJO forecast skill increases with large NAO amplitude, together with the observed MJO excitation of the NAO [Cassou, 2008; Lin et al., 2008], alludes to a possible MJO-NAO cyclic relationship. This leads to the question of whether the trends in Figure 2 may be driven from the extratropics. However, since the MJO and NAO take place on different time scales, with the former process being quasi-periodic and the latter process resembling a first order autoregressive process [Feldstein, 2000], it is unlikely that the MJO-induced Arctic amplification has its origin in the extratropics via the above cyclic relationship. 5of6

6 [19] To discern in greater depth the process through which the MJO warms high latitudes, future study with numerical model simulations that use MJO-like convective heating is required. The results from model simulations of this type will be presented in an upcoming paper. [20] Acknowledgments. This study is supported by National Science Foundation grants ATM , ATM , and AGS , and by National Oceanic and Atmospheric Administration grant NA100AR We thank two anonymous reviewers for their constructive comments and suggestions. [21] The Editor thanks the anonymous reviewers for their assistance in evaluating this paper. References Alexeev, V., P. Langen, and J. Bates (2005), Polar amplification of surface warming on an aquaplanet in ghost forcing experiments without sea ice feedbacks, Clim. Dyn., 24(7 8), , doi: /s Arctic Climate Impact Assessment (2005), Impacts of a Warming Climate: Arctic Climate Impact Assessment, Cambridge Univ. Press, New York. Branstator, G. (2002), Circumglobal teleconnections, the jet stream waveguide, and the North Atlantic Oscillation, J. Clim., 15(14), , doi: / (2002)015<1893:cttjsw>2.0.co;2. Budyko, M. I. (1969), The effect of solar radiation variations on the climate of the Earth, Tellus, 21, , doi: /j tb00466.x. Cai, M., and J. Lu (2007), Dynamical greenhouse-plus feedback and polar warming amplification. Part II: Meridional and vertical asymmetries of the global warming, Clim. Dyn., 29(4), , doi: /s Cassou, C. (2008), Intraseasonal interaction between the Madden-Julian Oscillation and the North Atlantic Oscillation, Nature, 455(7212), , doi: /nature Dee, D. P., and S. Uppala (2009), Variational bias correction of satellite radiance data in the ERA-Interim reanalysis, Q. J. R. Meteorol. Soc., 135(644), , doi: /qj.493. Ding, Q., E. J. Steig, D. S. Battisti, and M. Kuttel (2011), Winter warming in West Antarctica caused by central tropical Pacific warming, Nat. Geosci., 4(6), , doi: /ngeo1129. Feldstein, S. B. (2000), The timescale, power spectra, and climate noise properties of teleconnection patterns, J. Clim., 13(24), , doi: / (2000)013<4430:ttpsac>2.0.co;2. Graversen, R. G. (2006), Do changes in the midlatitude circulation have any impact on the Arctic surface air temperature trend?, J. Clim., 19(20), , doi: /jcli Graversen, R. G., T. Mauritsen, M. Tjernstrom, E. Kallen, and G. Svensson (2008), Vertical structure of recent Arctic warming, Nature, 451(7174), 53 56, doi: /nature Hall, A. (2004), The role of surface albedo feedback in climate, J. Clim., 17(7), , doi: / (2004)017<1550:trosaf>2.0.co;2. Holland, M. M., and C. M. Bitz (2003), Polar amplification of climate change in coupled models, Clim. Dyn., 21(3 4), , doi: / s Hoskins, B. J., and D. J. Karoly (1981), The steady-state linear response of a spherical atmosphere to thermal and orographic forcing, J. Atmos. Sci., 38, , doi: / (1981)038<1179:tslroa>2.0. CO;2. Johannessen, O. M., et al. (2004), Arctic climate change: Observed and modelled temperature and sea-ice variability, Tellus, Ser. A, 56(4), , doi: /j x. Johnson, N. C., and S. B. Feldstein (2010), The continuum of North Pacific sea level pressure patterns: Intraseasonal, interannual, and interdecadal variability, J. Clim., 23(4), , doi: /2009jcli Kanamitsu, M., W. Ebisuzaki, J. Woollen, S.-K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter (2002), NCEP-DOE AMIP-II Reanalysis (R-2), Bull. Am. Meteorol. Soc., 83(11), , doi: /bams Kiladis, G. N., and K. M. Weickmann (1992), Circulation Anomalies associated with tropical convection during northern winter, Mon. Weather Rev., 120(9), , doi: / (1992)120<1900:caawtc> 2.0.CO;2. Lee, S., T. Gong, N. Johnson, S. Feldstein, and D. Pollard (2011a), On the possible link between tropical convection and the Northern Hemisphere Arctic surface air temperature change between , J. Clim., 24(16), , doi: /2011jcli Lee, S., S. Feldstein, D. Pollard, and T. White (2011b), Do planetary wave dynamics contribute to equable climates?, J. Clim., 24(9), , doi: /2011jcli Lin, H., and G. Brunet (2009), The influence of the Madden-Julian Oscillation on Canadian wintertime surface air temperature, Mon. Weather Rev., 137(7), , doi: /2009mwr Lin, H., and G. Brunet (2011), Impact of the North Atlantic Oscillation on the forecast skill of the Madden-Julian Oscillation, Geophys. Res. Lett., 38(2), L02802, doi: /2010gl Lin, H., G. Brunet, and J. Derome (2008), Forecast skill of the Madden- Julian Oscillation in two Canadian atmospheric models, Mon. Weather Rev., 136(11), , doi: /2008mwr Lin, H., G. Brunet, and J. Derome (2009), An observed connection between the North Atlantic Oscillation and the Madden-Julian Oscillation, J. Clim., 22(2), , doi: /2008jcli Madden, R. A., and P. R. Julian (1994), Observations of the day tropical oscillation A review, Mon. Weather Rev., 122(5), , doi: / (1994)122<0814:ootdto>2.0.co;2. Matthews, A. J., B. J. Hoskins, and M. Masutani (2004), The global response to tropical heating in the Madden Julian Oscillation during the northern winter, Q. J. R. Meteorol. Soc., 130(601), , doi: /qj Schneider, D., C. Deser, and Y. Okumura (2011), An assessment and interpretation of the observed warming of West Antarctica in the austral spring, Clim. Dyn., doi: /s x. Screen, J. A., and I. Simmonds (2010), The central role of diminishing sea ice in recent Arctic temperature amplification, Nature, 464(7293), , doi: /nature Sellers, W. D. (1969), A global climate model based on the energy balance of the Earth-atmospheric system, J. Appl. Meteorol., 8, , doi: / (1969)008<0392:agcmbo>2.0.co;2. Serreze, M., and J. Francis (2006), The Arctic amplification debate, Clim. Change, 76(3 4), , doi: /s y. Vecchi, G. A., and N. A. Bond (2004), The Madden-Julian Oscillation (MJO) and northern high latitude wintertime surface air temperatures, Geophys. Res. Lett., 31, L04104, doi: /2003gl Wheeler, M. C., and H. H. Hendon (2004), An all-season real-time multivariate MJO Index: Development of an index for monitoring and prediction, Mon. Weather Rev., 132(8), , doi: / (2004)132<1917:AARMMI>2.0.CO;2. S. Feldstein, S. Lee, and C. Yoo, Department of Meteorology, Pennsylvania State University, 503 Walker Building, University Park, PA 16802, USA. (sbf1@meteo.psu.edu; sl@meteo.psu.edu; cyoo@psu.edu) 6of6