The effect of ENSO activity on lower stratospheric water vapor

F. Xie, W. Tian, J. Austin, J. Li, H. Tian, J. Shu, and C. Chen Key Laboratory for Semi-Arid Climate Change of the Ministry of Education, College of Atmospheric Sciences, Lanzhou University, China ICAS, School of Earth and Environment, University of Leeds, UK NOAA Geophysical Fluid Dynamics Laboratory, Princeton, NJ, USA State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

events are likely to dry the lower stratosphere over a narrow band of tropics (5 • S-5 • N) but have a moistening effect on the whole stratosphere when averaged over a broader region of tropics between 25 • S-25 • N. The moistening effect of La Niña events mainly occurs in lower stratosphere in the Southern Hemisphere tropics where a significant 20% increase in the tropical upwelling is caused by La Niña events. El Niño events 10 have a more significant effect on the tropical upwelling in the Northern Hemisphere extratropics than in Southern Hemisphere extratropics. The net effect of ENSO activities on the lower stratospheric water vapor is stronger in the Southern Hemisphere tropics than in the Northern Hemisphere tropics.

15
Due to its radiative and chemical significance the stratospheric water vapor has been widely studied in recent years in various aspects (e.g., Shindell, 2001;Forster and Shine, 2002;Stenke and Grewe, 2005;Tian et al., 2009). Sounding observations at Boulder indicate increasing lower stratospheric water vapor since 1981 (Oltmans et al., 2000). However, Boulder data and satellite observations from the Halogen Occultation 20 Experiment (HALOE), Stratospheric Aerosol and Gas Experiment II (SAGE II) as well as Polar Ozone and Aerosol Measurement III (POAM III) reveal that the lower stratospheric water vapor decreases after 2000 (e.g., Randel et al., 2006;Dhomse et al., 2008;Solomon et al., 2010;Hurst et al., 2011). To understand observed stratospheric water vapor trends, factors impacting or controlling stratospheric water vapor changes the stratospheric water vapor. It is well known that El Nino-Southern Oscillation (ENSO) has a large impact on the global circulation. Previous studies have pointed out that ENSO can modulate the stratospheric water vapor due to its adjust on temperature and transport processes within the tropopause region (e.g., Sassi et al., 2004). Observational and comprehensive modeling studies Geller et al., 2002;Hatsushika and Yamazaki 2003;Scaife et al., 2003;Fueglistaler and Haynes, 2005) found that strong El Niño events (warm phases of ENSO) have a moistening impact on the lower stratosphere while La Niña events (cold phases of ENSO) lead to a reverse change in the lower stratospheric water vapor. However, how and to what 15 an extent ENSO activities influence the stratospheric water vapor is still under debate. Scaife et al. (2003) simulated contributions of ENSO on increasing stratospheric water vapor, but they did not distinguish the different effect of El Niño activities and La Niña activities on the lower stratospheric water vapor. Using NCEP reanalysis and a general circulation model,  studied the effect of the El Niño and La Niña 20 activities on the tropical tropopause layer and they found that the thermal tropopause during a La Niña cold event was not an isentropic surface while El Niño events could change the location of the water vapor minimum in the upper troposphere through modifying the large scale circulation and convection in the central Pacific. Some other studies found convincing evidence that SST changes associated with ENSO can mod- 25 ify Brewer-Dobson (BD) circulation in the tropical stratosphere (e.g., Manzini et al., 2006;García-Herrera et al., 2006). It is apparent from those previous studies that transport changes caused by anomalous SSTs associated with ENSO can affect the lower stratospheric water vapor. However, contributions of anomalous BD circulation during ENSO events on stratospheric water vapor have not been discussed in detail in previous studies. Furthermore, the relative importance of ENSO warm and cold phases in affecting the lower stratospheric water vapor and the main entry locations of tropospheric air into lower stratosphere during ENSO warm and cold phases are remain unclear.

5
In this paper, the effect of El Niño activities on the lower stratospheric water vapor and temperature is further analyzed and then compared to the corresponding effect of La Niña activities. The BD circulation changes associated with ENSO activities are also examined and their effects on the lower stratospheric water vapor are discussed. Section 2 describes the data used in this study; Section 3 discusses lower strato-10 spheric temperature changes caused by ENSO activities and associated water vapor changes in the upper troposphere and lower stratosphere (UTLS). Section 4 analyses the changes in the BD circulation associated with ENSO activities and their effects on lower stratospheric water vapor changes. In Sect. 5 we discuss net effect of ENSO activities on lower stratospheric water vapor. Section 6 gives conclusions.

Data
The monthly mean European Center for Medium range Weather Forecasting (ECMWF) reanalysis data from 1961-2000 are mainly analyzed in this paper. The National Centers for Environmental Prediction (NCEP) reanalysis-2 data, which have longer time period extending to 2008, are employed as a supplement when ana-20 lyzing lower stratospheric water vapor trends. As the assimilated water vapor data from ECMWF/NCEP reanalysis may have systematic biases compared to real observations (Gettelman et al., 2010) ENSO (El Niño and La Niña) activities are decided by the ONI data which are defined as the three-month running-mean SST departures in the Niño 3.4 region (5 • N-5 • S, 170 • W-120 • W) (Smith and Reynolds, 2003). El Niño events are characterized by a positive ONI greater than or equal to +0.5 • C and La Niña events are characterized by a negative ONI less than or equal to −0.5 • C. The SST departures are based on a 5 set of improved homogeneous historical SST analyses (Extended Reconstructed SST -ERSST.v3) from NOAA (Smith et al., 2008). According to the ONI, long-term time series of ERA-40 data are divided into three groups with 122 months of data marked with El Niño events, 134 months of data marked with La Niña events and the other 224 months of data for normal conditions. The El Niño and La Niña anomalies are 10 calculated by compositing together the detrended and deseasonalized time series for El Niño and La Niña months, respectively.

Water vapor anomalies in the UTLS region associated with ENSO
Before analyzing water vapor anomalies associated with ENSO from ERA-40 reanalysis, it is necessary to check whether the assimilated water vapor data resembles 15 satellite measured water vapor in the tropical UTLS region. Figure 1 shows horizontal distributions of ERA-40 water vapor (averaged over 1961-2000) and MLS water vapor (averaged over 2005-2010) at 100 hPa. The latitude-height cross sections of zonal mean water vapor from ERA-40 reanalysis and MLS measurements are also given in Fig. 1 In this study, we mainly focused on the relative changes of the stratospheric water vapor caused by ENSO activities rather than analyzing absolute water vapor values. Therefore, an analysis of ERA-40 water vapor is still helpful since it has a longer time period than MLS and HALOE water vapor for the purpose of composite analysis. In the following, ERA-40 water vapor and cold point tropopause (or 100 hPa) temperatures, which directly control stratospheric water vapor values in ERA-40 data, will be analyzed combined with MLS and HALOE water vapor data in an attempt to cross check the robustness of results obtained. Figure 2 first shows 100 hPa temperature and water vapor anomalies associated with El Niño and La Niña events. It is apparent that ENSO in different phases have 10 different impacts on the 100 hPa temperature and water vapor at different tropical regions. An evident cooling can be noted in the middle and eastern Pacific at 100 hPa during El Niño events, while in the western Pacific region, where it is typically referred as the cold trap region (Newell and Gould-Stewart, 1981), the 100 hPa temperature shows a ∼0.6 K warming (Fig. 2a). The spatial pattern and magnitude of the tempera-15 ture anomalies exhibit in Fig. 2a are in accordance with those in previous studies (e.g., Randel et al., 2000;Scaife et al., 2003). The 100 hPa water vapor anomalies during El Niño events (Fig. 2b) are overall consistent with the 100 hPa temperature anomalies ( Fig. 2a), i.e., negative/positive temperature anomalies at 100 hPa are accompanied by negative/positive water vapor anomalies. As expected, La Niña events have an op-20 posite effect on the temperature and water vapor at 100 hPa as that of El Niño events ( Fig. 2c and d). Figure 2e, f, g and h shows the cold point tropopause temperature and water vapor anomalies associated with El Niño and La Niña events. We can see that the distributions and magnitudes of cold point tropopause temperature and water vapor anomalies are nearly the same as corresponding anomalies at 100 hPa, possibly due 25 to relatively coarse vertical resolution of ERA-40 data in the UTLS region. Figure 2 indicates that the ENSO signals in 100 hPa temperature are of opposite signs to those in tropical SSTs. Previous studies have also showed an opposite ENSO signal in the tropical lower stratosphere (Reid et al., 1989;Yulaeva et al., 1994). 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al.   et al. (2004) pointed out that the cooling of the tropical lower stratosphere over the eastern Pacific during warm phases of ENSO is related to internal equatorial waves forced by anomalous convection in the troposphere associated with SST changes. For the entry of water vapor into the lower stratosphere, it is mainly controlled by the tropopause temperature. Newell and Gould-Stewart (1981)  sarily imply a decease/increase in the lower stratospheric water vapor. The tropopause may lie above/below 100 hPa during the warm/cold phases of the ENSO due to intensified/weakened convection which acts to lift/suppress tropopause to a higher/lower level.  Fig. 3b and d, respectively. We can see that water vapor increases in the western Pacific and Indian Ocean from 360 K to 390 K isentropic surfaces but decreases in the middle and eastern Pacific above 370 K isentropic sur-25 face during El Niño events. As is pointed out in previous literatures, 360-390 K isentropic surfaces are overall lie in the lower stratosphere (e.g., Holton and Gettelman, 2001). Figure 3 suggests that El Niño events tend to moisten the lower stratosphere in broad regions of tropics except over the middle and eastern Pacific where the lower ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al. stratospheric water vapor decreases. La Niña events do an opposite effect as El Niño events, i.e, moistening the lower stratosphere over the middle and eastern Pacific but drying the lower stratosphere over the western Pacific and Indian Ocean. Also note that magnitudes of water vapor anomalies in the Southern Hemisphere tropics are overall larger than those in the Northern Hemisphere tropics. On the 380 K 5 isentropic surface, the maximum positive water vapor anomaly over the western Pacific and Indian Ocean during El Niño events is 0.10 ppmm (parts per million by mass) in the Northern Hemisphere tropics, but it reaches 0.14 ppmm in the Southern Hemisphere tropics. The minimum negative water vapor anomalies in the middle and eastern Pacific during El Niño are −0.19 ppmm in the Northern Hemisphere tropics and −0.31 ppmm 10 in the Southern Hemisphere tropics. The similar features can be noted during La Niña events. The result here suggests that the influence of ENSO activities on the lower stratospheric water vapor is stronger in the Southern Hemisphere tropics than in the Northern Hemisphere tropics.

ACPD
It is interesting that the effect of ENSO on the water vapor over the western Pacific 15 and Indian Ocean is most pronounced bellow 390 K isentropic surface while over the middle and eastern Pacific significant water vapor anomalies can be noted well above 390 K isentropic surface. Further check on the temperature anomalies on 370 K isentropic surface ( Fig. 3e) indicates that the temperature anomalies over the middle and eastern Pacific are much larger than the anomalies in the western Pacific and Indian 20 Ocean, consequently, the water vapor anomalies over the middle and eastern Pacific are much larger than those over the western Pacific and Indian Ocean on the same isentropic surface. The vertical velocity fields during ENSO events (Fig. 3f) show a strong upward motion around the middle and eastern Pacific, and a relatively weak upward motion over the western Pacific and Indian Ocean in ENSO situations suggesting 25 that vertical transport of water vapor is stronger over the middle and eastern Pacific than that over the western Pacific and Indian Ocean during ENSO events. This may be the possible reason that water vapor anomalies over the middle and eastern Pacific can be noted well above the 370 K isentropic surface. Note that the vertical velocity fields ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al. along the tropical longitudes look to be in accordance with the pattern of the Walker circulation in ENSO situations (Webster and Chang, 1988) suggesting that the Walker circulation also plays an role in modulating the tropical lower stratospheric water vapor. It is apparent from Figs. 2 and 3 that ENSO's effects on the lower stratospheric water vapor have large spatial variations. To estimate the net effect of El Niño/La Niña 5 events on the lower stratospheric water vapor, the zonal mean water vapor anomalies averaged over different latitude bands in the tropics are shown in Fig. 4. We can see that El Niño activities result in positive water vapor anomalies at the tropical lower stratosphere. The result is consistent with previous studies, i.e., El Niño events have a moistening impact on lower stratospheric water vapor (e.g., Fueglistaler and Haynes, 10 2005;Geller et al., 2002;Hatsushika and Yamazaki, 2003). However, Fig. 4a shows that El Niño activities tend to dry the tropical middle stratosphere. The zonal mean water vapor anomalies during La Niña events (Fig. 4b) show a quite different pattern from that during El Niño events. As expected, La Niña events result in a negative water vapor anomaly at the lower stratosphere over the equator be- in the whole stratosphere. The result suggests that La Niña events tend to dry the lower stratosphere over a narrow band of tropics (5 • S-5 • N) but overall tend to moisten the whole stratosphere over a broader region of tropics between 25 Although MLS water vapor time series span a short time period from 2005-2010, they can still be composited together with regard to El Niño and La Niña events for this 25 6-yr time period to provide more evidence on ENSO's effects on the lower stratospheric water vapor. Figure 5a and b shows 100 hPa MLS water vapor anomalies associated with El Niño and La Niña events. We can see that the spatial distributions of MLS water vapor anomalies at 100 hPa are generally in accordance with the corresponding ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al. ERA-40 water vapor anomalies ( Fig. 2b and d). An exception is that the MLS water vapor anomalies are much larger than ERA-40 water vapor anomalies over the western Pacific and Indian Ocean, possibly because MLS water vapor time series are too short. The longitude-height cross sections of MLS water vapor anomalies averaged between the 25 • N-25 • S latitude band composited for El Niño and La Niño events are shown 5 in Fig. 5c and d. Note that the vertical distributions of MLS water vapor anomalies exhibit a similar pattern as that of ERA-40 water vapor anomalies ( Fig. 3a and b and Fig. 3c and d), with ENSO anomalies over the western Pacific and Indian Ocean being most pronounced at the lower stratosphere while over the middle and eastern Pacific significant water vapor anomalies existing well in the middle stratosphere.

10
To further verify the results obtained from ERA-40 data, 12-yr HALOE water vapor time series, which cover the time period 1993-2004, are composited with respect to ENSO events. Figure 5e and f shows the zonal mean HALOE water vapor anomalies averaged over different latitude bands composited for El Niño and La Niña events, respectively. Consistent with Fig. 4, HALOE water vapor anomalies in Fig. 5 also show 15 that El Niño events tend to moisten the lower stratosphere but dry the middle stratosphere, while La Niña events have a moistening effect on the whole stratosphere when averaged over a broad region of tropics between 25 • S-25 • N. However, HALOE water vapor anomalies averaged over latitude band of 5 • S-5 • N indicate that the low stratosphere is not moistened during El Niño events and this is not in agreement with the 20 results in previous studies and in Fig. 4a. It is worth to point out there are many missing values in the HALOE water vapor data between 5 • S-5 • N, particularly bellow 100 hPa. Furthermore, HALOE water vapor retrievals bellow tropical tropopause is largely contaminated by clouds (e.g., Randel et al., 2001). Therefore, the HALOE water vapor anomalies averaged between 5 • S-5 • N may be lack of robustness bellow 25 100 hPa. However, we can see from above analysis that the results obtained from ERA-40 reanalysis are overall supported by MLS and HALOE water vapor observations. 4150 ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al.

BD circulation changes associated with ENSO
A noticeable feature in Fig. 4 is that the moistening effect of La Niña events mainly occurs in the lower stratosphere in the Southern Hemisphere tropics. To understand this asymmetric effect of La Niña events on stratospheric water vapor, we further analyze ENSO's effects on the BD circulation which is another important factor controlling the stratospheric water vapor change. It has been reported in previous studies that during warm phases of ENSO planetary wave activities can be significantly enhanced (Van Loon and Labitzke, 1987;Hamilton, 1993;Camp and Tung, 2007;Garfinkel and Hartmann, 2007;Free and Seidel, 2009;Sassi et al., 2004;Manzini et al., 2006;García-Herrera et al., 2006;Taguchi and Hart-10 mann, 2006) and hence, resulting in a stronger BD circulation in the stratosphere. Figure 6 shows ENSO-induced anomalies of vertical velocity of the BD circulation as well as corresponding E-P flux and zonal wind anomalies. Consistent with the previous studies (Manzini et al., 2006;Garfinkel and Hartmann, 2007;Free and Seidel, 2009), the vertical velocity of BD circulation between 15 • S-15 • N is significantly en-15 hanced/weakened during El Niño/La Niña events ( Fig. 6a and c). A more interesting feature in Fig. 6 is the hemispheric asymmetry of the E-P flux anomalies caused by ENSO events (Fig. 6b and d). El Niño events enhance the poleward/equatorward propagation of wave activities in the Northern/Southern Hemisphere. Consequently, the northern polar vortex is warmed and weakened with negative zonal wind anomalies  As mentioned in Sect. 3, La Niña events actually have a moistening effect in the southern hemispheric lower stratosphere. Figure 6a and c suggests that this moistening effect is closely related to changes in the tropical upwelling caused by ENSO events. We can see from Fig. 6a and c that the climate mean of tropical upwelling with an upward vertical velocity of the BD circulation (w * ) covers a wide area from 25 • S- in the Southern Hemisphere extratropics (Fig. 6c). Further analysis of the anomalous tropical upwelling w * at 100 hPa reveals that both El Niño and La Niña events cause a weakening of w * averaged over the Northern Hemisphere tropics between 2.5 • N-

25
• N, with a 3% and 10% decrease relative to normal conditions. However, the w * 15 averaged over the Southern Hemisphere tropics between 2.5 • S-25 • S is increased by 6% during El Niño events and increased by 20% during La Niña events. It is apparent that the effect of La Niña events on the tropical upwelling is more significant than that of El Niño events, particularly in the Southern Hemisphere tropics. The significant enhancement of upwelling in the Southern Hemisphere tropics between 2.5 • S-25 • S 20 will cause an increase of water vapor in the stratosphere despite that tropopause temperature changes caused by La Niña events may do an opposite effect, and this may be the main reason that La Niña events have a moistening effect on the stratosphere in the Southern Hemisphere tropics (Fig. 4b). The results here also imply that the effect of El Niño events on the lower stratospheric water vapor is more dominated by 25 tropopause temperature changes while the effect of La Niña effect is more related to the changes in tropical upwelling.
ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al.

Integrated effect of ENSO activities on lower stratospheric water vapor
It is evident from the above analysis that the lower stratospheric water vapor anomalies caused by El Niño and La Niña events have an over all opposite spatial distributions. It is necessary here to examine the integrated effect of ENSO on the stratospheric water vapor. Figure 7a shows the longitude-latitude cross section of temperature anoma-  Randel et al. (2000) analyzed tropical tropopause temperature changes associated with ENSO events and found that ENSO events cause a warming of the tropopause in the middle and eastern Pacific regions and a cooling in the western Pacific and maritime continent areas. The spatial pattern of tropopause temperature changes associated with ENSO showed by Randel et al. (2000) is quite similar as that 15 of 100 hPa temperature changes shown in Fig. 7a. Figure 7b and c shows zonal mean water vapor and temperature anomalies averaged between the 5 • S-5 • N and 25 • S-25 • N latitude bands, respectively. Again, these anomalies are composited from all El Niño and La Niño events during the period from 1961-2000. The integrated effect of ENSO has no significant impacts on equato- showed that zonal averaged ENSO temperature anomalies in the lower stratosphere are small but negative at the equator. When zonal mean water vapor and temperature anomalies are averaged between the 25 • S-25 • N latitude band (Fig. 7c) Figure 7b and c further confirms that the effects ENSO activities on the stratospheric water vapor and temperature have a latitudinal dependence 5 within the tropics. At this stage, it may be also instructive to examine whether ENSO events have an impact on long-term trends of the stratospheric water vapor. A qualitative analysis of linear trends of ERA-40 and NCEP reanalysis-2 cold point tropopause temperatures and water vapor reveals that those trends are quite different when the data recorder 10 marked with ENSO activities being either included or excluded implying that ENSO activities may have an ineligible impact on trends of cold point tropopause temperature and hence on stratospheric water vapor trends (not shown). However, the assimilated water vapor data from ECMWF/NCEP reanalysis have systematic biases compared to real observations (Gettelman et al., 2010) while the time series of observed strato-15 spheric water vapor are too short for trend analysis. Even the trends of cold point tropopause temperatures from ERA-40/NCEP reanalysis, which directly affect stratospheric water vapor, have uncertainties (Gettelman et al., 2009). This issue is worthy further investigation. 20 Through composite analysis of ERA-40 reanalysis, and NCEP reanalysis-2 data based on Oceanic Niño Index, the potential effects of ENSO events on the stratospheric water vapor are investigated. In agreement with earlier studies, El Niño events tend to moisten the lower stratosphere. However, it is found that El Niño events are likely to dry the middle stratosphere. La Niña events tend to dry the lower stratosphere over a ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al.  ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | spheric water vapor from balloonborne, frostpoint hygrometer measurements at Washington, D.C., and Boulder, Colorado, Geophys. Res. Lett., 27, 3453-3456, 2000. Randel, W. J., Wu, F., and Gaffen, D. J.: Interannual variability of the tropicsal tropopause derived from radiosonde data and NCEP reanalysis, J. Geophys. Res., 105, 15509-15523, 2000. ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Stenke, A. and Grewe, V.: Simulation of stratospheric water vapor trends: impact on stratospheric ozone chemistry, Atmos. Chem. Phys., 5, 1257Phys., 5, -1272Phys., 5, , doi:10.5194/acp-5-1257Phys., 5, -2005Phys., 5, , 2005. Taguchi   ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor  ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al.  ACPD 11,2011 The effect of ENSO activity on lower stratospheric water vapor F. Xie et al.