Influence of El Niño-Southern Oscillation in the mesosphere
CAS Key Laboratory of Geospace Environment, Department of Geophysics and Planetary Science, University of Science and Technology of China, Hefei, China
Corresponding author: T. Li, CAS Key Laboratory of Geospace Environment, Department of Geophysics and Planetary Science, University of Science and Technology of China, Hefei, Anhui 230026, China. (firstname.lastname@example.org)
 Using the middle atmosphere temperature data set observed by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) satellite experiment between 2002 and 2012, and temperatures simulated by the Whole Atmospheric Community Climate Model version 3.5 (WACCM3.5) between 1953 and 2005, we studied the influence of El Niño-Southern Oscillation (ENSO) on middle atmosphere temperature during the Northern Hemisphere (NH) wintertime. For the first time, a significant winter temperature response to ENSO in the middle mesosphere has been observed, with an anomalous warming of ~1.0 K/MEI (Multivariate ENSO Index) in the tropics and an anomalous cooling of ~ −2.0 K/MEI in the NH middle latitudes. The observed temperature responses to ENSO in the mesosphere are opposite to those in the stratosphere, in agreement with previous modeling studies. Temperature responses to ENSO observed by SABER show similar patterns to those simulated by the WACCM3.5 model. Analysis of the WACCM3.5 residual mean meridional circulation response to ENSO reveals a significant downwelling in the tropical mesosphere and upwelling in the NH middle and high latitudes during warm ENSO events, which is mostly driven by anomalous eastward gravity wave forcing in the NH mesosphere.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 El Niño-Southern Oscillation (ENSO) is a coupled ocean-atmosphere phenomenon related to sea surface temperature variations in the tropical east-central Pacific Ocean. It dominates the interannual variability of the equatorial troposphere [Yulaeva and Wallace, 1994; Calvo-Fernandez et al., 2004] and also significantly affects stratospheric variability in both the tropics and extratropics. In the tropical lower stratosphere, Randel et al.  found significant zonal mean temperature and ozone responses to ENSO using radiosonde and satellite data sets together with the Whole Atmosphere Community Climate Model version 3.5 (WACCM3.5). Randel et al.  also showed that the zonal mean temperature and ozone responses to ENSO in the tropical lower stratosphere are in phase with each other, suggesting an ENSO modulation of the tropical upwelling in this region. In a follow-up study, Calvo et al.  corroborated the increased tropical upwelling in the lower stratosphere during warm ENSO events and showed that this was mainly due to changes in the filtering of orographic gravity waves by anomalies in the location and intensity of the subtropical zonal mean jets associated with ENSO.
 On the other hand, previous studies revealed enhanced planetary wave propagation and dissipation during warm ENSO events in the extratropics of the Northern Hemisphere (NH), which decelerate the stratospheric polar vortex and enhance the Brewer-Dobson circulation in the stratosphere. This leads to an anomalous cooling in the tropical upper stratosphere and warming in the polar region [e.g., Sassi et al., 2004; Bronniman et al. 2004; Garcia-Herrera et al., 2006; Manzini et al., 2006; Free and Seidel, 2009; Hood et al. 2010].
 Although the behavior of ENSO signal in the stratosphere is now well established, it has not been clear as to whether or not ENSO effects are significant above this layer. Sassi et al.  found a significant warming in the tropical middle mesosphere and cooling in the NH polar mesosphere during warm ENSO events in winter using an earlier version of WACCM (WACCM1b). The only sources of variability in the middle atmosphere in this version of WACCM were those from internal model dynamics and specified sea surface temperatures. They suggested that, since the stratospheric westerly zonal winds during warm ENSO events are weaker, the less eastward portion of NH upward-propagating gravity waves are then filtered via critical level absorption by the westerlies. As a result, the climatological gravity wave forcing in the NH winter mesosphere, which is strongly westward, becomes less so. Thus, the anomalous gravity wave forcing in the NH mesosphere associated with ENSO becomes eastward. This weakens the mesospheric residual mean meridional circulation and leads to cooling of the polar mesosphere. In addition to these model studies, using 13.5 years of Rayleigh lidar temperature data over Hawaii (19.5 N), Li et al.  showed a significant positive winter temperature response of ~2.5 K/MEI to ENSO near 70 km, consistent with WACCM1b model results in the middle mesosphere of NH subtropical region.
 Up to now, there have been no published global observations with sufficient duration to confirm the winter mesospheric temperature response to ENSO revealed in the general circulation model results. In our study, we use middle atmosphere temperature profiles between 2002 and 2012 observed by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) instrument onboard the Thermosphere Ionosphere and Mesosphere Electric Dynamics (TIMED) satellite, and a four-member ensemble of simulations from a more recent version of WACCM (WACCM3.5) between 1953 and 2005, to study ENSO influence on the middle atmosphere zonal mean temperature during the NH wintertime.
2 Data Sets and Analysis Method
 The SABER instrument onboard the TIMED satellite [Russell et al. 1999] has been providing temperature profiles in the middle atmosphere since January 2002 that are retrieved from atmospheric limb radiance measurements in the 15 µm CO2 band. The TIMED satellite is in a 74° inclined 625 km orbit gathering data during ~15 orbits a day with descending and ascending data points separated by ~10 h. The TIMED spacecraft yaws 180° every ~60 days to keep the SABER infrared detectors pointed away from the sun. This requirement, together with the viewing geometry, leads to alternating 53°S–83°N and 53°N–83°S sampling, such that only latitudes in the range of 53°S–53°N are sampled continuously during the mission. A number of SABER temperature validation studies have been conducted and reported by Mertens et al.  and Remsberg et al. . The absolute temperature measurement accuracies are ±0.5 K below 75 km and ±1–2 K from 75 to 95 km [Remsberg et al., 2008]. For this study, we use both ascending and descending data sets of SABER Level 2A version 1.07 data between 20 and 80 km from January 2002 to December 2012. To effectively eliminate solar tide aliasing of the zonal mean profiles, especially in the mesosphere, a data set covering a full diurnal cycle is needed. Since it takes ~60 days (i.e. ~2 months) for SABER to complete a ~24 h local time sampling interval with both descending and ascending orbits included, the effective time resolution of the SABER data, as processed for this study, is bimonthly. Our approach is justifiable since we are comparing SABER results to monthly-mean WACCM results. To obtain the SABER bimonthly mean profiles, we first separately calculate the daily ascending and descending zonal mean temperatures within 5° latitude bands centered at 45°S, 40°S, etc., up to 45°N, and then for each latitude, we average all available daily zonal mean profiles within a 60-day window centered at the middle of each month.
 WACCM3.5 is a fully interactive chemistry-climate model that ranges from the surface up to the thermosphere. A four-member ensemble of simulations analyzed here runs from 1953 to 2006 with a horizontal resolution of 1.9° latitude by 2.5° longitude. This simulation was part of the second Chemistry Climate Model Validation activity conducted as part of the Stratospheric Processes and Their Role in Climate project. It was also analyzed by Randel et al.  and Calvo et al.  to study ENSO effects in the lower stratosphere. The simulations were run with prescribed observed sea surface temperatures. Concentrations of greenhouse gases and halogen species are calculated explicitly in the model; observations of their mixing ratios at the surface are used as lower boundary conditions. A Quasi-Biennial Oscillation (QBO) was imposed by relaxing the winds to the observations in the tropics [Matthes et al., 2004], and heating and chemical effects of volcanic eruptions were included [Tilmes et al., 2009]. Detailed information of the base model WACCM3 is provided in Garcia et al. , and the main updates in WACCM3.5 that are different from WACCM3 are listed in Calvo et al. . In our study, we use monthly means below 0.01 hPa (~80 km) for the 1953–2005 period to compare with the SABER results. The variables analyzed here are zonal mean temperature, the transformed Eulerian mean meridional and vertical velocities, and (defined in Andrews and McIntyre ), the Eliassen-Palm (EP) flux divergence (accounts for resolved waves in the model), and parameterized gravity wave (GW) drag. Details of the GW parameterization implemented in WACCM3.5 can be found in Richter et al. .
 To derive the temperature response to ENSO, we applied a multivariate linear regression analysis at each altitude and latitude to the deseasonalized bimonthly (SABER) or monthly (WACCM) mean time series. The analysis includes terms to account for the long-term trend (assumed to be linear), the 11 year solar cycle (using solar F10.7 cm radio flux as a proxy), the QBO (two orthogonal QBO time series derived from the equatorial stratospheric zonal winds at seven different levels [Wallace et al., 1993]), ENSO (using the Multivariate ENSO Index (MEI) as a proxy [Randel et al., 2009]), and the effect of volcanic eruptions (using stratospheric aerosol optical depth as a proxy [Sato et al., 1993]). The evolution of each reference time series and detailed description of the linear regression analysis can be found in Li et al. . To fit the seasonal variability, the coefficient of each regression term (e.g., ε for ENSO) is assumed to be of the form ε = A1 + A2cosωt + A3sinωt + A4cos2ωt + A5sin2ωt with ω = 2π/(12 months), where A is a parameter to be determined in the fitting and t is the time in months. The response of ENSO in winter is the average of ENSO coefficient (e.g., ε) in December, January, and February. The uncertainties in the fitting coefficients are estimated according to the variances and covariances obtained from the linear-least square fit expressed as [Randel and Cobb, 1994]
where σ2(A1) and σ2(A1,A2) are the variance and covariance of the regression coefficients.
 The MEI is a Pacific-basin-wide index created from several atmospheric and oceanic variables such as the sea surface temperature, outgoing long-wave radiation (deep convection), and the pressure gradient between two remote locations across the Pacific, with atmospheric variables lagged by ~2 months [Randel et al., 2009]. The time series of the MEI index between 1953 and 2012 is shown in Figure 1 (the text file is available at http://www.esrl.noaa.gov/psd/enso/mei/mei.html). Positive MEI index values represent a warm event (El Niño), while negative values represent a cold event (La Niña). The horizontal dashed lines represent 1σ standard deviation, indicating the strength of warm or cold ENSO events. During the SABER observation period between January 2002 and December 2012, there were three strong El Niño events in winter 2002/2003, 2006/2007, and 2009/2010, respectively, and two strong La Niña events in winter 2007/2008 and 2010/2011, respectively.
3 Results and Discussion
 Figure 2 shows the meridional cross section of the zonal mean temperature response to ENSO during the NH winter derived from (a) the WACCM3.5 simulation and (b) the SABER observation. In the tropical stratosphere, WACCM3.5 (Figure 2a) shows a statistically significant cooling of about −0.6 K/MEI in the lowermost stratosphere, below 30 hPa, and about −0.2 K/MEI in the middle and upper stratosphere. A positive temperature response is simulated over the NH polar region with a maximum of ~1.0–1.2 K/MEI. These signals agree with those observed by SABER between 45°S and 45°N (Figure 2b), with the largest cooling of about −1.0 K/MEI occurring over the tropics in the lower and upper stratosphere and significant warming of ~1.4 K/MEI occurring over the midlatitudes (~45°N). In the lower stratosphere (below ~25 km), the SABER results show cooling responses of −0.4 to −0.8 K/MEI throughout the tropics, which is somewhat stronger than the previous radiosonde observations presented in Figure 2b of Randel et al.  but comparable to the radiosonde results shown in Figure 4 of Free and Seidel .
 In the mesosphere, WACCM3.5 shows a clear positive temperature response over the tropics with a maximum of 0.4 K/MEI near 65–75 km and a negative temperature response at the NH middle and high latitudes with a maximum of −1.6 K/MEI near the pole. Similar patterns are observed by SABER between 45°S and 45°N in the mesosphere, but their magnitudes are ~3 times stronger than those simulated by WACCM3.5. We also note here that ENSO events in the period examined with WACCM (1953–2005) are generally stronger than in the period covered by SABER measurements (2002–2012), and thus the different intensity of ENSO could not explain the larger signal obtained in SABER. This difference may be partially due to the different lengths of time used in the linear regression analysis. However, we also analyzed 11 years of WACCM data between 1995 and 2005, and the results are still ~1.5 times weaker than the SABER results. In addition, the out-of-phase behavior of the signals between the stratosphere and mesosphere is clear in both WACCM3.5 and SABER results. In the mesopause region (above 80 km), the SABER results show a tropical cooling which is still significant in WACCM3.5 although much weaker and at the higher altitudes (not shown). These differences could be related to the WACCM's gravity waves parameterization, although a deeper analysis needs to be done to evaluate this suggestion. In this paper, we focus on ENSO in the mesosphere where the agreement between SABER and model is the largest.
 As was discussed in section 1, during warm ENSO events in winter, the anomalously enhanced westward planetary waves at the NH middle latitudes can propagate upward and poleward through the stratosphere westerly wind [Garcia-Herrera et al., 2006]. Thus, enhanced dissipation of westward planetary waves in the upper stratosphere of NH middle and high latitudes can decelerate the westerly zonal mean wind. The resulting Brewer-Dobson circulation is enhanced in the stratosphere during warm ENSO events, causing anomalous warming in the polar region and cooling in the tropical stratosphere. In addition, the weakening of the zonal mean zonal winds in the NH stratosphere can also facilitate the propagation of eastward gravity waves into the mesosphere modulating the residual circulation there.
 Figure 3 shows the meridional cross section of the WACCM3.5 responses to ENSO in (a) the total wave forcing including both resolved and parameterized waves, (b) EP flux divergence of the resolved Rossby waves in the model, and (c) parameterized GW drag. The same linear regression fit method applied to temperature is used for wave forcing. In the upper stratosphere, the westward wave forcing primarily from resolved planetary wave EP flux divergence (compare Figures 3a and 3b) is clearly enhanced during warm ENSO events, corresponding to the decelerated westerly zonal mean zonal wind (not shown). This is consistent with the earlier WACCM1b simulation [Garcia-Herrera et al., 2006]. However, in the NH mesosphere, WACCM3.5 also reveals an anomalous enhancement of eastward wave forcing (Figure 3a), which mostly corresponds to anomalous eastward gravity wave drag (Figure 3c). A decomposition of the anomalous gravity wave forcing associated with ENSO into different origin components reveals that the anomalous wave forcing from the convection component dominates in the subtropics, while the anomalous wave forcing from the frontal component dominates at the middle and high latitudes (not shown).
 Thus, the dissipation and breaking of enhanced eastward propagating gravity waves in the NH middle and high latitude mesosphere weakens the mesospheric residual mean meridional circulation and leads to an anomalous cooling in the north polar mesosphere as shown by Sassi et al. . However, no explanation has been proposed for the strong positive temperature response to ENSO in the tropical mesosphere. The mesospheric warming in the tropics during warm ENSO events is likely the result of an anomalous residual mean meridional circulation induced by enhanced eastward wave forcing in the NH mesosphere (Figure 3a). This ENSO-induced anomalous circulation is in the opposite direction to the climatological residual mean meridional pole-to-pole (summer to winter) circulation. Since there is no significant complementary flow in the summer (southern) hemisphere, the continuity of mass produces an anomalous downwelling in the tropical mesosphere and leads to an anomalous warming in this region.
 For such an explanation to be firmly established, we further analyze the residual mean meridional circulation (, ) using the linear regression fit method. Its response to ENSO in winter is shown in Figure 4. The residual mean meridional circulation can approximately trace the net transport of air parcels in the middle atmosphere. We clearly see a significant anomalous residual mean meridional circulation in the mesosphere from the middle and high latitudes to the equatorial region with downwelling in the tropical mesosphere and upwelling in the NH middle and high latitudes during warm ENSO events. This indicates that adiabatic temperature change due to downwelling/upwelling primarily contributes to the tropical mesospheric temperature response to ENSO obtained from the model simulations and satellite observations. An opposite but weaker anomalous circulation is evident in the stratosphere in response to ENSO, with upwelling in the tropics and downwelling in the NH high latitudes, consistent with previous WACCM1b model results [Sassi et al., 2004; Garcia-Herrera et al., 2006].
 It is also interesting to note the anomalous upwelling in the mesosphere of Southern Hemisphere (SH) high latitudes (and an anomalous cooling in Figure 2a) and downwelling near 40°S–30°S in WACCM3.5. The tropical mesospheric warming during warm ENSO events increases the temperature gradient between low and high latitudes in the SH mesosphere [Karlsson et al., 2009] leading to an enhanced easterly zonal wind in the SH austral summer. The more westward part of the upward-propagating gravity waves spectrum are then filtered via critical level absorption by the easterlies in the SH mesosphere. As a result, the anomalous gravity wave forcing associated with ENSO in this region is eastward (as shown in Figure 3a), strengthening the SH branch of the mesospheric residual mean meridional circulation with anomalous upwelling in the SH high latitudes (as shown in Figure 4). Since this ENSO effect in the SH mesosphere is secondary, the mesospheric cooling in the SH high latitudes is much weaker than that in the NH (as shown in Figure 2).
 In this paper, we have derived the middle atmosphere temperature response to ENSO during the NH wintertime using the SABER data set (2002–2012) and a WACCM3.5 model simulation (1953–2005). SABER has provided, for the first time, observational evidence of the significant temperature response to ENSO in the mesosphere, and the data show similar patterns to those simulated by the WACCM3.5. The SABER-observed temperature response to ENSO is generally ~3 times stronger than the WACCM3.5 results, although this might be partially due to differences in the analyzed periods. Clear out-of-phase features between the stratosphere and mesosphere are also evident.
 Analysis of the WACCM3.5 residual mean meridional circulation (, ) response to ENSO reveals an anomalous circulation in the mesosphere with downwelling in the tropical mesosphere during warm ENSO events leading to warming in this region. This downwelling in the tropical mesosphere during warm ENSO events is driven by anomalous eastward wave forcing mostly contributed from anomalous eastward gravity wave drag in the NH mesosphere.
 This work was carried out at the University of Science and Technology of China, with support from the National Natural Science Foundation of China grants (41225017, 41074108, 41127901, 41025016, 41121003), the Chinese Academy of Sciences Key Research Program KZZD-EW-01, and the Fundamental Research Funds for the Central Universities. N.C.'s work was partly supported by the Spanish Ministry of Science through the project Supercomputing and e-Science, Consolider CSD 2007-00050. C.Y.S.'s work was supported by NSF/AGS grant 1136082. We are grateful to the SABER retrieval team for producing a high-quality data set that is made readily available to the scientific community. The SABER data were downloaded from http://saber.gats-inc.com/.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.