Geophysical Research Letters

Influence of El Niño Southern Oscillation on stratosphere/troposphere exchange and the global tropospheric ozone budget

Authors

  • Guang Zeng,

    1. Atmospheric Chemistry Modelling Support Unit, Natural Environmental Research Council Centres for Atmospheric Science, and Centre for Atmospheric Science, Department of Chemistry, University of Cambridge, Cambridge, UK
    Search for more papers by this author
  • John A. Pyle

    1. Atmospheric Chemistry Modelling Support Unit, Natural Environmental Research Council Centres for Atmospheric Science, and Centre for Atmospheric Science, Department of Chemistry, University of Cambridge, Cambridge, UK
    Search for more papers by this author

Abstract

[1] We use a combined climate/chemistry model to assess the interannual variability of tropospheric ozone associated with the El Niño Southern Oscillation (ENSO). Simulations cover the years 1990 to 2001 with prescribed SSTs. El Niño and La Niña events are reproduced by the model. Previous studies have emphasized the impact of ENSO on tropospheric composition through changing emissions or convection. Here we show that there is a close relationship between ENSO and stratosphere-troposphere exchange (STE); different patterns of circulation and meteorology in El Niño and La Niña years result in significant differences in transport of ozone-rich air from the stratosphere to the troposphere. We calculate an anomalously large increase of STE following a typical El Niño year, most evident after the 1997–1998 El Niño event. La Niña events result in a decrease of STE. STE is one factor affecting tropospheric ozone; we find highest ozone after the 1997–1998 El Niño.

1. Introduction

[2] The El Niño Southern Oscillation (ENSO) has a strong impact on the interannual variability and distribution of tropical tropospheric ozone. Observed anomalies of tropospheric column ozone and upper tropospheric water vapor in the equatorial Pacific region strongly correlate with the ENSO signal [see, e.g., Chandra et al., 1998; Ziemke and Chandra, 2003]. During El Niño, the usually warm surface waters in the western Pacific and Indonesian region shift eastwards to the eastern Pacific and South America, accompanied by an eastward shift of the pattern of convection. During La Niña the opposite occurs: abnormally warm ocean temperatures and stronger convection occur over the western Pacific but with lower sea surface temperatures and weakened tropical convection in the eastern Pacific. Previous studies of the impact of ENSO on composition (O3 and its precursors) in the tropical region found an increase of tropospheric column O3 in the western Pacific over Indonesia in El Niño years associated with dryness, downward motion and suppressed convection [see, e.g., Chandra et al., 1998; Fujiwara et al., 2000; Hauglustaine et al., 1999; Thompson et al., 2001; Sudo and Takahashi, 2001]. The suppressed convection in the western Pacific leads to less efficient vertical transport of low surface concentrations of ozone from the surface. In contrast, a negative ozone anomaly in the eastern Pacific arises from increased humidity and enhanced upward transport.

[3] Large-scale circulation patterns vary greatly with ENSO. For example, Shapiro et al. [2001] have shown that ENSO has an important impact on dynamical features in the extra-tropical Pacific upper troposphere including the strength of the sub-tropical jet, expected to be an influence on stratosphere-troposphere exchange. Furthermore, James et al. [2003] also suggest that ENSO has an important influence on the interannual modulation of global STE. Due to the complexity of physical and dynamical processes determining STE (which have a wide range of temporal and spatial variabilities), STE is still inadequately quantified and model calculations of the ozone cross-tropopause flux differ widely (see, e.g., Table 4.12 of IPCC [Houghton et al., 2001]).

[4] Given the link between STE and the ENSO signal we suggest that there may be mechanisms other than those discussed so far in the literature (e.g., convection, emissions) by which chemical composition in the troposphere can be affected by ENSO. In particular, this paper aims to explore the large-scale features of ozone transport associated with ENSO. We aim to quantify the effect of ENSO on the ozone budget on a global scale, using a detailed chemistry-climate model.

2. Model and Experiments

[5] We use the UK Met Office Unified Model version 4.4 combined with a tropospheric chemistry module (UM/CHEM) [Zeng and Pyle, 2003, and references therein]. The model has 19 levels from the surface to 4.6 hPa at a horizontal resolution of 3.75° by 2.5°. The model is forced by prescribed sea surface temperatures (SSTs). The tracer advection scheme is based on work by Leonard et al. [1995]. The chemistry module, including emissions and wet/dry deposition of chemical species, is incorporated into the GCM. The chemical integration uses an implicit time integration scheme [Carver and Stott, 2000] with a 15 minute time step. The chemical mechanism used here includes a NOx/CO/CH4/NMHC oxidation scheme. The reaction rate coefficients are from recent IUPAC and JPL databases. We use pre-calculated photolysis rates from a 2-D model [Law and Pyle, 1993]. Daily concentrations of the stratospheric species O3 and NOy are specified at the top three model levels (above 30 hPa) using output from the 2-D model, to produce a realistic annual cycle of these species in the stratosphere. The model uses present day climatological O3 fields [Li and Shine, 1995] in the radiation calculations and does not allow for feedback of calculated O3 into the GCM.

[6] We perform simulations from 1990 to 2001 forced by observed SSTs covering that period. The emission scenarios are the same as those used by Zeng and Pyle [2003], which are based on IPCC SRES A2 scenario for the year 2000 [Nakićenović et al., 2000]. They vary seasonally but not annually. Hence our simulation does not specifically account for emissions by, for example, the 1997 Indonesian forest fires. Indeed, any modeled correlation with ENSO cannot be due to the emissions. The initial concentrations of longer lived species in the model are from a previous multiyear run using climatological present-day SSTs.

3. Results and Discussion

[7] The ENSO index is defined, following Ziemke and Chandra [2003], as the deseasonalized monthly mean tropical Tahiti (18S,150W) minus tropical Darwin (13S,131E) surface pressure to indicate El Niño and La Niña events in the Pacific. Figure 1a shows the smoothed ENSO index (also called the Southern Oscillation Index, SOI) calculated from the model for 1990–2001. El Niño and La Niña are indicated by negative and positive pressure anomalies respectively; high surface pressure is associated with suppressed convection and vice versa. The model, forced with observed SSTs, captures the major features of the observed ENSO index shown by Ziemke and Chandra [2003]. El Niño events dominate the years from 1991 to 1994 and from 1997 to 1998. 1996 and from late 1998 into 2000 are typical La Niña years.

Figure 1.

(a) Calculated time series of ENSO index for 1990–2001 smoothed using an FFT filter with a cut off timescale of 12 months, in hPa; (b) Solid line: deseasonalized monthly mean STE anomalies for 1990–2001, in Tg; Broken line: SOI from 6 month earlier.

[8] ENSO events have a major impact on the modelled distributions of trace gases in the tropics. Figure 2 shows the difference in modelled O3 between October 1997 (El Niño) and October 1996 (unperturbed) along the Equator. O3 increases by over 10 ppbv in the western Pacific and decreases by over 10 ppbv in the eastern Pacific as the result of the 1997–1998 El Niño event. This is in line with the simulation of Sudo and Takahashi [2001] and the observed anomalies that are derived from the Total Ozone Mapping Spectrometer (TOMS) [Chandra et al., 1998]. The model clearly captures the sense of the observed ENSO-related variations in tropical O3.

Figure 2.

Differences of simulated O3 (ppbv) at Equator between October 1997 and October 1996.

[9] To understand the impact of ENSO on a global scale, we have calculated the model global ozone budgets for 1990–2001. Figure 3a shows the annual total net influx of O3 from the stratosphere to the troposphere for 1990–2001, as diagnosed from our simulation. Note that here we mean by “STE” the net flux of stratospheric O3 into the troposphere. We adopt the 150 ppbv O3 isopleth to define the tropopause (e.g., also used by Shindell et al. [2001]). Figures 1a and 3a suggest there is a strong anti-correlation between the ENSO and STE signal; strong El Niño years in e.g., 1992 and 1998 coincide, with a few months phase delay, with large STE and La Niña years (e.g., 1996 and 1999) with small STE. The monthly mean deseasonalized STE is shown in Figure 1b. It anti-correlates well with the SOI from 6 months earlier (with correlation coefficients r = −0.6 over 1990–2001 and r = −0.78 over 1995–2001). This is consistent with the findings of Langford et al. [1998] who showed that 6-month lagged observed free tropospheric ozone at a northern midlatitude site anti-correlates with the SOI over the period 1993–1999. Langford [1999] attributes this anti-correlation to the variation of STE associated with the mean transverse circulation at the subtropical jet exit over the eastern Pacific. To further demonstrate the anti-correlation, Figure 3b shows the mean SOI calculated for July–December of the year previous to that used to calculate the annual mean STE in Figure 3a. The anti-correlation is strong (r = −0.7).

Figure 3.

Simulated variabilities covering 1990–2001: (a) Annual total STE in Tg; (b) mean SOI over July–December from previous year in hPa; (c) Annual total net chemical production of O3 in Tg; (d) Annual mean O3 abundance in the lower stratosphere (tropopause–100 hPa), in Tg; (e) Annual mean O3 abundance in the troposphere, in Tg; and (f) Annual mean OH abundance in the troposphere, in Mg.

[10] Figure 3c shows that the globally integrated O3 mass between the O3-defined tropopause and 100 hPa closely follows the variability of STE. The largest positive anomaly occurs in 1998 when the STE is greatest and the largest negative anomaly occurs in 1999 when the STE drops lowest. The enhanced net downward transport of ozone in a typical El Niño year coincides with a buildup of ozone in the extratropical lower stratosphere. In the troposphere, the O3 abundance is determined not just by STE, but also by the total net chemical production (NCP) and deposition at the Earth's surface. There is a buffering between these terms. For example, increased net downward transport of O3 from the stratosphere can lead to increased chemical destruction of O3 (see Figures 3a and 3d) so we do not expect a simple relationship between STE and global tropospheric ozone. The calculated global total tropospheric O3 masses for 1990–2001 are shown in Figure 3e. While ozone is relatively constant in the early model years (see discussion below) the most significant feature is the sharp increase of O3 abundance in 1998 which coincides with the large STE earlier in that year. This is consistent with the results of Langford et al. [1998]. We note also that a climatological analysis of the first four years of annually averaged MOZAIC ozone data [Köhler, 2003] on the 340 K surface also shows highest ozone following the 1997–1998 El Niño.

[11] We have also examined the response of OH to the ENSO signal (Figure 3f). OH abundance in the troposphere is primarily controlled by O3 and water vapor; hence the variability of tropospheric O3 inevitably leads to changes in OH abundance. Note again that a large increase in global OH, and oxidizing capacity, after the 1997–1998 El Niño event is a dominant feature. (OH decreases sharply at the beginning of the simulation. This is mainly due to low initial CO concentrations in the model (not shown) taken from a previous run which was forced by climatological SSTs. CO then stabilizes after two years of simulation. The effect of the low initial CO concentrations on O3 is not so evident, although the chemical-dynamical correlations we discuss are weakest during the first few years when the model is still settling down. These first years of the integration appear to be dominated by the response to the initialisation; any chemical-dynamical effects could be masked during this period.)

[12] Figure 3 featured global diagnostics. Chemical budgets in particular regions may respond in different ways to the ENSO signal and Figure 4 shows regional differences in the response of the modeled tropospheric ozone. In the extratropics (20–90°) ozone correlates with STE (Figure 3a) with, for example, high STE and a large O3 abundance in 1998. In the tropics, ozone and STE are less clearly correlated than in the extratropics. However, there is a strong correlation between extratropical O3 and the tropical O3 abundance one year later (r = 0.84) which could be caused by a time-delay related to hemispheric transport and mixing. Some of the tropical tropospheric variability will also be due to changes in convection (see Figure 2), not associated with STE; in the atmosphere emission changes will certainly also be important.

Figure 4.

(top) Annual mean O3 burdens (Tg) in the tropics (20°N–20°S) (solid line). Broken line is the O3 burden one year later. (bottom) O3 burdens (Tg) in the extratropics (90–20°N and 20–90°S).

4. Conclusions

[13] The El Niño Southern Oscillation significantly affects the interannual variability of the tropospheric ozone budget simulated by the UK Met Office Unified Model combined with a detailed tropospheric chemistry module. In particular, the 1997–1998 ENSO event coincides with strong signals in O3. We have calculated the modeled global budget of O3 to determine the relevant processes. During El Niño and La Niña events, shifts in circulation and meteorological patterns not only affect photochemistry in the tropics but also the transport of O3-rich air from the stratosphere to the troposphere. There is a strong correlation between STE and ENSO; the modelled annual total STE maximizes following El Niño years and minimizes following La Niña years. As a direct consequence of changes in STE, photochemical production of O3 in the troposphere is modified. The net effect on O3 is most pronounced in the extratropics where large STE coincides with an increase in O3. Of course, many other features will also contribute to observed changes in ozone. Nevertheless, in our model simulation, the O3 abundance in the tropics also shows some response to changes in STE, with a roughly one-year delay.

[14] ENSO affects global total tropospheric O3 not only via its effects on chemical processes (temperature-dependent chemistry, water vapour concentrations, emissions, etc.) but also via its profound effect on STE in the extratropics. Consequently, the oxidizing capacity of the troposphere is influenced by a number of processes related to ENSO, including an important contribution from STE.

Acknowledgments

[15] This work was supported by the NERC Centres for Atmospheric Sciences (NCAS) and the NERC UTLS thematic programme. The UK Met Office Hadley Centre is thanked for the use of the UM.

Ancillary