Global tropospheric ozone trends, attributions, and radiative impacts in 1995–2017: an integrated analysis using aircraft (IAGOS) observations, ozonesonde, and multi-decadal chemical model simulations

Abstract

Abstract. Quantification and attribution of long-term tropospheric ozone trends are critical for understanding the impact of human activity and climate change on atmospheric chemistry but are also challenged by the limited coverage of long-term ozone observations in the free troposphere where ozone has higher production efficiency and radiative potential compared to that at the surface. In this study, we examine observed tropospheric ozone trends, their attributions, and radiative impacts from 1995–2017 using aircraft observations from the In-service Aircraft for a Global Observing System database (IAGOS), ozonesondes, and a multi-decadal GEOS-Chem chemical model simulation. IAGOS observations above 11 regions in the Northern Hemisphere and 19 of 27 global ozonesonde sites have measured increases in tropospheric ozone (950–250 hPa) by 2.7 ± 1.7 and 1.9 ± 1.7 ppbv per decade on average, respectively, with particularly large increases in the lower troposphere (950–800 hPa) above East Asia, the Persian Gulf, India, northern South America, the Gulf of Guinea, and Malaysia/Indonesia by 2.8 to 10.6 ppbv per decade. The GEOS-Chem simulation driven by reanalysis meteorological fields and the most up-to-date year-specific anthropogenic emission inventory reproduces the overall pattern of observed tropospheric ozone trends, including the large ozone increases over the tropics of 2.1–2.9 ppbv per decade and above East Asia of 0.5–1.8 ppbv per decade and the weak tropospheric ozone trends above North America, Europe, and high latitudes in both hemispheres, but trends are underestimated compared to observations. GEOS-Chem estimates an increasing trend of 0.4 Tg yr−1 of the tropospheric ozone burden in 1995–2017. We suggest that uncertainties in the anthropogenic emission inventory in the early years of the simulation (e.g., 1995–1999) over developing regions may contribute to GEOS-Chem's underestimation of tropospheric ozone trends. GEOS-Chem sensitivity simulations show that changes in global anthropogenic emission patterns, including the equatorward redistribution of surface emissions and the rapid increases in aircraft emissions, are the dominant factors contributing to tropospheric ozone trends by 0.5 Tg yr−1. In particular, we highlight the disproportionately large, but previously underappreciated, contribution of aircraft emissions to tropospheric ozone trends by 0.3 Tg yr−1, mainly due to aircraft emitting NOx in the mid-troposphere and upper troposphere where ozone production efficiency is high. Decreases in lower-stratospheric ozone and the stratosphere–troposphere flux in 1995–2017 contribute to an ozone decrease at mid-latitudes and high latitudes. We estimate the change in tropospheric ozone radiative impacts from 1995–1999 to 2013–2017 is +18.5 mW m−2, with 43.5 mW m−2 contributed by anthropogenic emission changes (20.5 mW m−2 alone by aircraft emissions), highlighting that the equatorward redistribution of emissions to areas with strong convection and the increase in aircraft emissions are effective for increasing tropospheric ozone's greenhouse effect.

Understanding recent tropospheric ozone trends in the context of large internal variability: a new perspective from chemistry-climate model ensembles

“…We infer a role for increases in tropospheric ozone precursors in the tropical latitudes where ozone is efficiently produced throughout the year and, along with precursors, is convectively lofted to the free troposphere where the ozone lifetime is longer, as also pointed out by Zhang et al (2016). Increases in aircraft NO x emissions may play a particularly strong role in the upper tropospheric ozone increases (Wang et al 2022). From 1995 to 2014, upper tropospheric and lower tropospheric trends decouple both in sign, consistent with observations assessed by Gulev et al (2021).…”

“…We conclude that simulated boundary layer trends are too small and likely indicate an underestimate of emission growth in these regions. This conclusion is also supported by multi-decadal chemistry-transport model simulations driven by reanalysis ('observed') meteorology compared to not only the IAGOS data, but also to long-term ozonesonde measurements (Christiansen et al 2022, Wang et al 2022.…”

“…These findings, together with figure 1, imply a role for convective lofting of ozone and precursors to the free troposphere, where ozone production is more efficient and the lifetime is longer than near the surface, in driving the overall increase in the tropospheric ozone burden (Berntsen et al 1996, Naik et al 2005, Sauvage et al 2007. Increasing aircraft NO x emissions over this period may also play role in the overall burden increase (Wang et al 2022). The ozone increases in the lower troposphere are largest in the first two decades of the simulation, more in line with the overall trends in global anthropogenic emissions (table S2).…”

Multidecadal increases in global tropospheric ozone derived from ozonesonde and surface site observations: can models reproduce ozone trends?

“…In the Northern Hemisphere midlatitudes, chemistryclimate models were only able to reproduce ∼ 50 % of the observed ozone trend (Parrish et al, 2014), consistent with our current analysis using chemical transport models. Another analysis using GEOS-Chem from 1995-2017 found that the model underestimated global ozone trends compared to aircraft measurements and that aircraft emissions are a potential source of trend underestimation in the model (Wang et al, 2022). Tarasick et al (2019) also pointed out the role of data representativeness: uncertainty in estimated observational trends stems largely from data representativeness rather than the accuracy of historical data, pointing to the importance of increasing ozone monitoring station number and frequency, especially when the evaluation of model skill necessarily relies on comparison to sparse datasets.…”

“…Most estimates of ozone radiative forcing are calculated relative to the pre-industrial period (Skeie et al, 2020;Stevenson et al, 2013), and the small number of reliable measurements prior to the 20th century (Tarasick et al, 2019) provides challenges to constraining both long-and short-term radiative estimates. Even short-term changes in ozone can be difficult to reproduce; a recent chemical transport model simulation of global ozone trends over the past ∼ 20 years showed a consistent underestimate of observed ozone trends (Wang et al, 2022).…”