ENSO and Southeast Asian biomass burning modulate subtropical trans-Pacific ozone transport

Abstract Trans-Pacific transport of enhanced ozone plumes has been mainly attributed to fossil fuel combustion in Asia in spring, but less attention has been paid to vegetation fires in Asia. Here we show that the El Niño-Southern Oscillation (ENSO)-modulated fires in Southeast Asia, rather than Asian fossil fuel plumes, dominate the interannual variability of springtime trans-Pacific transport of ozone across the entire North Pacific Ocean. During El Niño springs, the intensified fires from both the Indochinese Peninsula and Indonesia, together with large-scale circulation anomalies, result in enhanced ozone plumes that stretch over 15 000 km in both the lower-middle and upper troposphere. This enhancement is also observed in the in situ measurements of ozone concentration, with an almost 10% increase at Mauna Loa Observatory, Hawaii, a unique site to monitor the long-distance transport over the North Pacific. This study reports an unexpectedly strong influence of vegetation fires, linked with climate variability, on global tropospheric chemistry and proves once more how complex the interactions in the climate system are.


INTRODUCTION
Ozone (O 3 ), together with the hydroxyl radical (OH), dominates the atmospheric oxidizing capacity and cycling of reactive trace gases in the troposphere, and thereby plays a key role in atmospheric chemistry and climate change [1][2][3][4][5][6]. As a result of its lifetime of a few weeks in the free troposphere, O 3 can be transported on regional to intercontinental scales [2,[7][8][9][10][11]. Previous studies of long-range transport of O 3 and its precursors from Asia have mainly focused on anthropogenic fossil fuel (FF) combustion sources [7,[10][11][12][13][14]. Intercontinental transport from Asia to North America at mid-latitudes is particularly strong in spring because of frequent cold fronts related to a favorable position of the jet stream [5,8,9,15,16]. Here we instead focus on the subtropics and show that the springtime biomass burning (BB) emissions in Southeast Asia, including the Indochinese Peninsula and Indonesia, dominate the O 3 interannual variability across the subtropical North Pacific Ocean under the influence of the El Niño-Southern Oscillation (ENSO). Figure 1 shows how the springtime distribution of O 3 and its precursors across the subtropical North Pacific Ocean are linked to ENSO. The IASI satellite retrievals show that during El Niño, a positive anomaly of tropospheric O 3 (below 6 km) extends in spring from the South China Sea northeastward to southwestern North America and the Gulf of Mexico (Fig. 1a). This belt of O 3 anomalies, with an average value of 1.7 DU, has a width of about 1500 km and a length of over 15 000 km, far longer than the mid-latitude intercontinental transport pathways from East Asia to North America and from eastern North America to Europe [17,18]. As a precursor of O 3 and a tracer for combustion sources [19], carbon monoxide (CO) from MOPITT satellite retrievals also shows an anomalous distribution during El Niño springs (Fig. 1b).

Interannual variability of O 3 over the subtropical North Pacific Ocean
Simulations using the Community Earth System Model (CESM) CAM-Chem [20] in specified dynamics mode based on the MERRA reanalysis for the period 1981-2015 capture well the positive O 3 anomaly (Fig. 1c). By applying Empirical Orthogonal Function (EOF) analysis, we find that the first principal component (PC1) of the simulated springtime O 3 given in Fig. 1d is consistent with the O 3 distributions in Fig. 1a and c, and can explain 25% of the O 3 variability in the subtropical North Pacific ( Supplementary Fig. 1). The time series of O 3 and CO from satellite retrievals and model simulations in the marked region across the northern subtropical Pacific (defined in Fig. 1a), as well as the EOF results, all consistently show in-phase fluctuations with the pre-spring (December-February) ENSO index Niño 3.4 [21] (Fig. 1e and f, Supplementary Fig. 1).

Sea surface temperature (SST) anomalies shift atmospheric circulations and modulate spring O 3
ENSO is known to be one of the main processes that regulates the interannual variability of the atmospheric circulation [23][24][25][26], and has been found to exert a strong impact on tropospheric ozone and other chemical constituents in tropical regions and in the Northern Hemisphere mid-latitudes [27][28][29]. To further isolate the impact of ENSO on the O 3 enhancement, we conducted two sensitivity runs of CAM-Chem forced with composite typical ENSO SST with climatological emissions [21,30] (Supplementary Table 1). The O 3 differences between the two runs ( Fig. 3a) are consistent with the satellite measurements and the simulations nudged with the MERRA data ( Fig. 1a and c), indicating that the ENSO SST anomaly is the main driver of the shifting circulation patterns that influence the inter-annual variability of lower tropospheric O 3 across the subtropical North Pacific Ocean. The ENSO-induced wind-vector anomalies clearly indicate enhanced continental outflow from the Asian subtropics and mid-latitudes (20 • N-40 • N) to the central Pacific (Fig. 3). Figure 3a also suggests an ENSO-induced enhancement of anti-cyclonic circulation in the subtropical western Pacific, which may cause more descending motion and stronger solar radiation and hence enhanced chemical production of O 3 , despite a relatively weak CO anomaly in the western Pacific (Fig. 3a, Supplementary Fig. 2a).
The vertical cross-section of the differences in O 3 fluxes along 120 • E between the two runs clearly shows two isolated centers of outflow anomalies from Asia ( Fig. 3b and Supplementary Fig. 2b). In the lower troposphere, the dominant O 3 and CO fluxes in the western Pacific exist around 26 • N (Fig. 3b), corresponding to the anomaly of northeastward transport in the coastal region of southern China (Fig. 3a, Supplementary Fig. 2a). Further south, near 10 • N-15 • N, the O 3 enhancement in El Niño spring is associated with stronger westward transport (i.e. negative flux) of the recirculated aged plume. In the central-eastern Pacific (i.e. along 155 • W), the ENSO induced tropospheric O 3 enhancement has a maximum value around 700 hPa between 10 • N and 25 • N, associated with stronger westward transport of aged continental flows recirculated from Asia ( Fig. 3a and c). Interestingly, MLO is located at the key location where the flow bifurcates to the west and the east horizontally, and where the enhancement reaches its maximum vertically, making it a key site for monitoring the ENSO-induced long-range transport of Asian pollutants.

Stronger biomass burning in ENSO years dominates spring O 3 enhancement
Previous studies [7,12,31,32] revealed transport pathways of Asian FF pollutants across the North Pacific Ocean. In our study, the enhanced transport flux of O 3 and CO plumes from tropical and subtropical Asia are consistent with the previously reported pathways of Southeast Asia BB plumes [8,9], which are particularly strong in spring [8]. In addition, the BB emission intensity in this region is also linked with ENSO because of reduced precipitation during El Niño years [27,33]. As shown in Supplementary Fig. 3 and Supplementary Table 2, the BB emission intensity can be up to 57% higher in El Niño years. Lagrangian backward dispersion modelling indicates that during the El Niño springs, air masses at MLO experience a greater residence time above South and Southeast Asia (Supplementary Fig. 4). Accordingly, measurements at MLO in spring show a good correlation between O 3 and CH 3 Cl, an effective BB tracer, and the Lagrangian modelling results clearly establish a link to the main BB source regions in tropical and subtropical Asia ( Supplementary  Fig. 5). The springtime maxima of O 3 and CO at MLO (Fig. 2b and c) also indicate a O 3 / CO ratio of about 0.3, which is quite typical of aged BB plumes [34,35].
We further conducted a series of CAM-Chem simulations (Supplementary Table 1) to quantify the relative contributions from major source regions in Asia. The FF emissions in China mainly cause a positive O 3 anomaly in the eastern Pacific (Fig. 4a), consistent with the wind anomaly shown in Fig. 3a, while those from India cause a weak anomaly in the eastern Pacific, but a stronger anomaly over southern China and southern India (Fig. 4b). However, for BB emissions in Southeast Asia, both the ENSOinduced anomalies in circulation and enhanced BB emissions make stronger contributions to the O 3 and CO anomalies stretching from subtropical Asia to North America ( Fig. 4c and d, Supplementary  Figs 6 and 7). The ENSO-induced circulation  anomaly causes enhancements of O 3 and CO in the lower-middle troposphere (between 600 and 800 hPa), with a bifurcated wind flow anomaly in the central Pacific (Fig. 4e, Supplementary  Fig. 6e). However, the difference in BB emissions produces a stronger impact over the subtropical Pacific, with two isolated strong vertical anomalies in the lower troposphere and the upper troposphere, for both O 3 and CO ( Fig. 4f and Supplementary Fig. 6f). The enhanced BB emissions from the Indochinese Peninsula and Indonesia in El Niño springs control the O 3 anomalies in lowermiddle troposphere and the upper troposphere, respectively, over the entire subtropical North Pacific Ocean (Fig. 5). These results are consistent with the vertical O 3 anomalies revealed by ozonesonde profiles from Hilo, Hawaii (close to MLO) ( Supplementary Fig. 8), located between the two belts of tropospheric O 3 column enhancements from Indonesia and the central-eastern Pacific, respectively ( Fig. 5a and b). The detailed transport mechanism and key processes of the BB smoke transport from the Indochinese Peninsula and Indonesia that influence the O 3 enhancement over the subtropical Pacific are illustrated in Fig. 6.

CONCLUSION
Our results have expanded understanding of the intercontinental-scale transport of Asian pollutants to North America [7,12,31,32] and of the roles of interannual variability in emissions and meteorology on tropospheric O 3 [14,27,36]. We find that ENSO plays a greater role than previously thought as a driver of the interannual variability of both lower-tropospheric and upper-tropospheric O 3 over the subtropical North Pacific Ocean and its subsequent transport to North America. In particular, our work shows that the year-to-year difference in BB emissions modulated by ENSO makes a stronger contribution than the anomalies because of meteorological variability alone. It proves once more how complex the interactions in the climate system are. This study highlights the importance of continuous measurements in the remote North Pacific Ocean for characterizing the impacts from both natural climate variability and human activity, and also suggests that better ENSO forecasting could improve modelling of continental-scale long-range transport of air pollutants.

Observations, data and ENSO events definition
Ambient ozone mixing ratios have been measured at MLO since 1973 [37] and the MLO CO record covers 1989 to the present [38]. To exclude possible influences from daytime upslope winds [22], our analysis is restricted to nighttime (0:00-8:00 local time) observations. We use the monthly 0-6 km column O 3 dataset retrieved by Infrared Atmospheric Sounding Interferometer (IASI) [39] from 2008 to 2017 to compute the springtime column O 3 composites in different ENSO phases. Daytime CO profiles from the Measurements of Pollution in the Troposphere (MOPITT) [8,40]   Era-Retrospective Analysis for Research and Applications (MERRA) reanalysis dataset [42], have been conducted in this study. Detailed information of model description, experiments configurations and emission scenarios of these experiments are described in the Supplementary data.

Lagrangian particle dispersion modeling
Lagrangian particle dispersion modeling is applied to demonstrate the transport pathways and track the potential sources of air masses recorded at MLO based on the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model [43]. Three-thousand particles are released every hour at 3400 m above sea level over MLO, which are then tracked backward in time for 20 days. The footprint of the retroplume, which is represented by the residence time of particles below 100 m [44], is used to identify potential emissions source regions and to gauge the relative impact of the emissions at MLO.

Data and materials availability
The observations, met data and modeling codes are freely available. Measurements from MLO are available from the NOAA Global Monitoring Division (GMD). Methyl chloride data from MLO were provided by S. Montzka, NOAA.