Whole Atmosphere Community Climate Model (WACCM) simulations are used to investigate the migrating and nonmigrating tidal variability in the mesosphere and lower thermosphere (MLT) due to the El Niño–Southern Oscillation (ENSO). The most notable changes occur in the equatorial region during Northern Hemisphere winter in the diurnal migrating tide (DW1), diurnal eastward propagating nonmigrating tides with zonal wavenumbers 2 and 3 (DE2 and DE3), and the semidiurnal westward propagating nonmigrating tide with zonal wavenumber 4 (SW4). The WACCM simulations indicate that the ENSO represents a source of interannual tidal variability of ∼10–30% in the MLT. The tidal changes are attributed to changes in tropical precipitation, altered tidal propagation due to changing zonal mean zonal winds, and changes in planetary wave activity associated with the ENSO. During the El Niño phase of the ENSO the DE2 and DE3 are decreased, and the DW1 and SW4 are enhanced. The opposite response occurs during the La Niña phase of the ENSO; however, the magnitude of the tidal changes due to El Niño and La Niña are different. This is especially notable for the DE2 and DE3 which are enhanced by ∼2 K during La Niña time periods, and only reduced by ∼1 K during El Niño time periods. The results demonstrate that changing sea surface temperatures associated with the ENSO significantly impact the overall dynamics of the MLT. Our results further suggest that the ENSO is a source of significant interannual variability in the low-latitude ionosphere and thermosphere.
 Upward propagating solar driven atmospheric tides contribute significantly to the overall dynamics of the mesosphere and lower thermosphere. These tides are forced in the troposphere due to a combination of infrared radiative heating of water vapor and latent heat release associated with tropical convection [e.g., Chapman and Lindzen, 1970; Lindzen, 1978]. Though generated in the troposphere, the tides propagate vertically and achieve large amplitudes in the mesosphere and lower thermosphere (MLT). The tides in the MLT are comprised of both migrating and nonmigrating tides. Migrating tides propagate at the same speed as the apparent motion of the Sun and are generated by longitudinally invariant tidal forcing in the troposphere and stratosphere. In contrast, a longitudinally varying source in the troposphere generates nonmigrating tides, which propagate either faster or slower than the apparent motion of the Sun. Although generally smaller in amplitude than the migrating tide, nonmigrating tides are considered to be particularly important due to the fact that they can introduce significant longitudinal variability [e.g., Zhang et al., 2006]. Nonmigrating tides are also the primary driver behind longitude variability in the low-latitude ionosphere [England, 2012], and thus serve as an important mechanism for the coupling between the lower atmosphere and ionosphere.
 One potentially significant source of interannual tidal variability is the El Niño–Southern Oscillation (ENSO). Large-scale changes in tropospheric convection occur in response to the ENSO [Trenberth et al., 2002; Lieberman et al., 2007]. These changes will alter the tidal forcing in the troposphere, and may therefore be expected to generate significant tidal variability in the MLT. The connection between diurnal tide variability and the ENSO was first recognized by Vial et al. . Vial et al.  observed a significant correlation between the diurnal tide in surface pressure measurements and the ENSO, and hypothesized that the interannual variability in the diurnal tide was due to changes in water vapor and/or latent heating associated with the ENSO. Observations have also indicated a connection between MLT temperatures and winds and the ENSO [Gurubaran et al., 2005; Lieberman et al., 2007]. The modeling results of Lieberman et al.  revealed that changes in tropospheric tidal heating during the 1997/1998 El Niño event were responsible for the observed changes in the diurnal tide amplitude in the MLT. The zonal mean zonal winds in the stratosphere and mesosphere are also known to be influenced by the ENSO [Sassi et al., 2004; Manzini et al., 2006]. In addition to the changes in tropospheric tidal forcing, the changes in zonal mean zonal winds may impact tides in the MLT by influencing the vertical propagation of the tides. Based on these prior studies one may surmise that the ENSO represents a potentially significant source of interannual variability in the MLT. However, only a few studies [Gurubaran et al., 2005; Lieberman et al., 2007; Oberheide et al., 2011b] have investigated the role of the ENSO on driving interannual tidal variability in the MLT, and a comprehensive view of the tidal changes that occur in response to the ENSO is therefore lacking. In the present study we use the Whole Atmosphere Community Climate Model (WACCM) to elucidate the tidal changes that occur in the MLT in response to the ENSO.
2. Whole Atmosphere Community Climate Model Simulations
 The WACCM is a model of the Earth’s atmosphere extending from the surface to the lower thermosphere (∼145 km). Specific details regarding the numerics and parameterizations in the WACCM can be found in Garcia et al.  and Richter et al. and are not repeated herein. The WACCM is part of the National Center for Atmospheric Research (NCAR) Community Earth System Model (CESM). WACCM simulations can use either specified ice and ocean at the model lower boundary or be coupled to ice and ocean models. When run coupled to another model, a coupler exchanges the fluxes and state information at the interface of the two models. Since we are interested in middle atmosphere variability due to the ENSO, the WACCM results presented herein are based on simulations with a fully coupled ocean model. To separate ENSO driven variability from internal model variability, we have performed a five-member ensemble simulation. Each ensemble member is run for 30 years, providing a total of 150 years of simulation results. The simulations are ’free-running’, and each ensemble member can be considered as an independent realization of the coupled atmosphere-ocean system. The simulations are therefore not representative of any specific time period. ENSO events are generated due to internal model dynamics, and this results in the ENSO events being different for each ensemble member. The solar flux is held constant at 138 sfu and geomagnetic activity is set to Kp of 3 for the duration of the simulations. We do not include a quasi-biennial oscillation (QBO) in our simulations in order to isolate the variability due to the ENSO, and prevent any nonlinear effects of the ENSO and QBO [Calvo et al., 2009] from influencing our investigation into the role of the ENSO on tidal variability.
 To investigate the tidal variability due to the ENSO, the model simulations are separated into periods of El Niño, La Niña, and neutral (i.e., neither El Niño or La Niña conditions). The El Niño time periods are considered to occur when the five-month smoothed Niño 3.4 index (average sea surface temperature between 5°N–5°S and 120°W–170°W) exceeds 1°C for a period of at least six months. Likewise, the La Niña periods are when the five-month smoothed Niño 3.4 index is below −1°C for at least six months. Based on this criteria, our simulations include 26 El Niño and 31 La Niña events. Since the ENSO signal is most prominent in the Northern Hemisphere winter months [e.g.,Sassi et al., 2004; Manzini et al., 2006], we restrict our focus to the time period from October to April. For each month, the composite average is calculated for the El Niño, La Niña, and neutral time periods. The composite average of the neutral time periods is subtracted from the El Niño and La Niña composite averages yielding the tidal anomalies that are presented in Section 3.
3. Results and Discussion
 The yearly Northern Hemisphere winter (December–February) average relative temperature amplitude of the migrating diurnal tide (DW1), nonmigrating eastward propagating diurnal tide with zonal wavenumbers 2 and 3 (DE2 and DE3), and nonmigrating westward propagating semidiurnal tide with zonal wavenumber 4 (SW4) at 100 km and the equator are shown in Figure 1. For clarity, each ensemble member is offset by 30 years in Figure 1. The tidal amplitudes in Figure 1 are relative to the 150 year average amplitude during December–February. The December–February average Niño 3.4 index is also shown in Figure 1. Note that the largest response to the ENSO is observed in the DW1, DE2, DE3, and SW4. We will thus limit our focus to these tides. The climatology of these tides is generally reproduced in the WACCM (e.g., H.-L. Liu, WACCM-X Simulation of Upper Atmosphere Wave Variability, submitted toGeophysical Monograph Series, 2012). The tidal behavior is therefore considered to be well represented, and the interannual variability due to the ENSO is thought to be realistically reproduced. From Figure 1 it is evident that interannual variability of ∼20–30% occurs in the amplitude of the tides during Northern Hemisphere winter. Furthermore, a clear relationship between the Niño 3.4 index and the tidal amplitudes is apparent, indicating that the interannual variability in the tides is due at least in part to the ENSO. The results in Figure 1 illustrate that the response of the tides to the ENSO is different for the different tides. In general, the DW1 and SW4 are correlated with the ENSO, while the DE2 and DE3 are anticorrelated with the ENSO. Based on all five ensemble members, the linear correlation coefficients between the December–February average Niño 3.4 index and the December–February average tidal amplitudes are 0.38, 0.44, −0.46, and −0.25 for the DW1, SW4, DE2, and DE3, respectively. As will be discussed later, the tidal response to El Niño and La Niña events are not equivalent, and this may result in the relatively weak correlation between the tides and Niño 3.4 index.
 The results in Figure 1 clearly demonstrate that the ENSO drives interannual variability in migrating and nonmigrating tides in the MLT, and we now turn our attention to the average tidal response to El Niño and La Niña events. The average tidal changes that occur during La Niña are presented in Figure 2. The tidal anomalies during El Niño events are shown in Figure 3. Note that the DW1 and DE2 are averaged over December to February, while the DE3 and SW4 are average over February to March. As will be discussed later, these time intervals correspond to the times when the greatest changes are observed in the individual tides (see Figure 4). During La Niña events (Figure 2), there are notable enhancements in the DE2 and DE3 in the equatorial region at MLT altitudes. The enhancements are statistically significant, and are on the order of 1–2 K, which corresponds to ∼30% for DE2 and ∼20% for DE3. The DW1 is only slightly reduced during La Niña time periods. A slight reduction of 1 K (∼10–15%) also occurs in the SW4. As can be seen in Figure 3, the opposite response occurs during El Niño time periods. That is, the DW1 and SW4 are enhanced, and the DE2 and DE3 are generally reduced (note that the enhancements in the MLT are not statistically significant). The response of the DE2 and DE3 to El Niño is considerably less clear than the response of these tides to the La Niña. The reductions in the DE2 and DE3 are also only ∼1 K compared to the ∼2 K enhancements that occur during La Niña time periods. The greater response of these nonmigrating tides to La Niña is consistent with TIMED/TIDI observations which similarly show a greater DE3 increase during La Niña compared to the DE3 decrease during El Niño [Oberheide et al., 2011b].
 The temporal evolution of the tidal anomalies at the equator during the La Niña phase of the ENSO are shown in Figure 4. Note that we only present the temporal evolution for the La Niña phase since all of the tidal anomalies are pronounced, and statistically significant, during this phase of the ENSO. One important feature to note is that all of the anomalies persist for several months during Northern Hemisphere winter. This indicates that during El Niño/La Niña winters the tidal behavior of the MLT is significantly altered for an extended time period. From Figure 4 it is apparent that the timing of the tidal response to the ENSO is not uniform among the different tides. The DW1 and DE2 anomalies are greatest during January–February, while the DE3 and SW4 anomalies are largest in March. The reason why the different tides exhibit a different temporal response is not presently known. One possibility is that the changes in tropospheric precipitation due to the ENSO change slowly over several months [e.g., Trenberth et al., 2002]. This may result in the enhancement or reduction of the tropospheric tidal forcing of different tidal components at different time periods. That is, anomalously large forcing of the DE2 may be greatest in January–February, while anomalously large forcing of the DE3 may be greatest in March. We note that although the DE3 amplitude is largest during July–September, only small changes (<1 K) in the DE3 occur in response to the ENSO during July–September. The ENSO tends to maximize during Northern Hemisphere winter months, and this may explain why the DE3 response is greatest during March–April.
 A complete analysis of the drivers of the ENSO driven tidal variability is beyond the scope of the present study. Nonetheless, it is worthwhile to briefly discuss the potential causes of the tidal variability. First, it is well known that the distribution of tropospheric convection changes due to the ENSO [e.g., Lieberman et al., 2007]. This will alter the tropospheric tidal forcing, and, depending on the tide, will result in either an enhancement or decrease in the tidal forcing and in-turn a change in the tidal amplitudes in the MLT [Lieberman et al., 2007; Oberheide et al., 2011b]. All of the tides considered in the present study can be generated in the troposphere by radiative and latent heating [Hagan and Forbes, 2002; Zhang et al., 2010]. It is therefore considered a strong possibility that the tidal changes that occur due to the ENSO are driven by changes in tropospheric forcing of the tides. Whether tropospheric changes result in an increase or decrease in tidal amplitude will depend on the specific tide. This may explain why certain tides, such as the DE2 and DE3, exhibit a comparatively stronger response to the ENSO. Furthermore, changes in the lower atmosphere are not symmetric between El Niño and La Niña events [e.g., Hoerling et al., 1997]. This asymmetry may provide an explanation for why the DE2 and DE3 exhibit a stronger response to La Niña compared to El Niño events. Another potential source of the tidal variability in the MLT are the changes in the zonal mean zonal winds that occur in response to the ENSO [Sassi et al., 2004]. Since the tidal propagation from the troposphere to the MLT is sensitive to the background zonal mean zonal winds [e.g., Ekanayake et al., 1997], the ENSO induced changes in the zonal mean zonal winds may alter the propagation of tides into the MLT. Another possible source of the tidal variability is changes in planetary waves due to the ENSO [Manzini et al., 2006]. Changes in the planetary waves may influence the tides in the MLT through wave-tide interactions. Altered planetary wave activity due to the ENSO can also influence the zonal mean winds, leading to an additional modulation of the tides. Additional studies are planned to gain a more comprehensive understanding of the source of the tidal variability in the MLT.
 Although the present study has focused on the tidal changes in the MLT, it is expected that the ENSO will also introduce interannual variability in the ionosphere and thermosphere. Nonmigrating tides in the MLT produce considerable longitude variability in the low-latitude ionosphere [England, 2012]. The interannual variability in the nonmigrating tides due to the ENSO is anticipated to generate interannual variability in the low-latitude ionosphere [e.g.,Pedatella and Forbes, 2009]. Nonmigrating tides can also propagate directly into the upper thermosphere [Forbes et al., 2009]. ENSO induced variability may therefore also be present throughout the thermosphere. Although the WACCM simulations have clearly illustrated the role of the ENSO on producing interannual tidal variability in the MLT, it is clear that additional research is necessary to understand the extent of the ENSO influence on the upper atmosphere. Last, we note that we have not included a QBO in the present simulations. The QBO and ENSO signals may interact nonlinearly [Calvo et al., 2009], and further study is necessary to understand the combined effects of the QBO and ENSO on tides in the MLT. Such research is required in order to better interpret the sources of interannual tidal variability in observations.
 The present study has demonstrated that significant interannual tidal variability in the MLT occurs due to the ENSO. During the El Niño phase of the ENSO, the WACCM simulations reveal an enhanced DW1 and SW4, and reduced DE2 and DE3. The opposite response occurs during the La Niña phase; however, the response of the tides in the MLT tends to be more pronounced for La Niña compared to El Niño. This is especially the case for the DE2 and DE3 which exhibit a significantly stronger response to La Niña events. The tidal response to the ENSO is ∼10–30%, and therefore represents a significant amplitude modulation of the tides during Northern Hemisphere winter months. In the context of other variability, the ENSO driven variability is smaller than the day-to-day and seasonal variability, but similar to the modulation due to the QBO (e.g., Liu, submitted manuscript, 2012). The changes in the migrating and nonmigrating tides in the MLT due to the ENSO are attributed to changes in the tropospheric tidal forcing, changes in zonal mean zonal winds which influence the tidal propagation, and changes in planetary wave activity. Additional research is planned to understand the role that each of these sources has in generating the tidal variability due to the ENSO.
 The National Center for Atmospheric Research is sponsored by the National Science Foundation. This work was supported in part by a NCAR Advanced Study Program Postdoctoral Fellowship (N. Pedatella). Additional support from National Science Foundation CEDAR grants ATM-0836386 and AGS-1138784 and NASA LWS Strategic Capability grant NNX09AJ83G is acknowledged. Resources supporting this work were partially provided by the NASA High-End Computing (HEC) Program through the NASA Advanced Supercomputing (NAS) Division at Ames Research Center.
 The Editor thanks S. Gurubaran and an anonymous reviewer for assisting in the evaluation of this paper.