Enhanced Aerosol Mass in the Tropical Tropopause Layer Linked to Ozone Abundance

Aerosol particles play a critical role in the tropical tropopause layer (TTL) through cloud formation and heterogeneous chemistry, influencing the radiative and chemical balance of the stratosphere. However, aerosol measurements in the TTL are sparse, resulting in poor knowledge of aerosol abundance and distribution in this important region. Here, we present in situ aircraft measurements over the western tropical Pacific, revealing a persistent and altitude‐dependent enhancement of aerosol mass in the TTL compared to the convectively influenced troposphere below. Notably, our data demonstrate a striking positive correlation between aerosol mass and ozone. Model simulations suggest that organic materials constitute a substantial fraction of the total aerosol mass within the TTL. We further derived an empirical parameterization of TTL aerosol mass as a function of ozone based on their linear relationship. This framework holds potential for estimating the TTL aerosol abundance but requires further validation and refinement through future measurements.


Introduction
As the main pathway for the transport of tropospheric air into the stratosphere, the TTL largely determines the entry values for the materials entering the stratosphere (Fueglistaler et al., 2009).The properties of the TTL air and the processes occurring in the TTL thus affect the global stratosphere and climate (Randel & Jensen, 2013;SPARC, 2006).
Aerosol is an important component in the TTL.The composition of the TTL aerosol is influenced by tropical dynamics (e.g., transport) and regional continental air sources (Froyd et al., 2009).Field observations have suggested that new particle formation events frequently occur in the TTL, in particular the lower TTL just below the tropopause (Brock et al., 1995;Weigel et al., 2011Weigel et al., , 2021)).Recent advancements in in-situ SO 2 measurement in the TTL suggest little contribution of SO 2 to stratospheric aerosols, revealing a significant gap in the stratospheric aerosol budget (Rollins et al., 2017).Aerosol particles in the TTL affect the stratospheric water vapor budget through TTL dehydration processes by serving as nuclei for the formation of cirrus clouds (Penner et al., 2009).TTL circus clouds have substantial impacts on the earth's radiative balance (Hong et al., 2016).By providing condensed surface areas, TTL aerosol can also facilitate condensation of low vapor pressure gases such as sulfuric acid (Brock et al., 1995) and promote heterogeneous chemistry that depletes ozone once they are transported to the stratosphere (Tolbert et al., 1988).Despite the importance, the abundance and properties of the TTL aerosol remain poorly characterized.The complex dynamic and chemical processes in the TTL make it difficult to elucidate the formation mechanisms of the TTL aerosol.In-situ measurements of TTL aerosol are limited, hindering our understanding on the climate impacts of the TTL and stratospheric aerosols.
To gain a deeper insight into the abundance, spatial distribution, and formation mechanisms of TTL aerosols, we carried out aircraft measurements of aerosols over the western Pacific warm pool during the Pacific Oxidants, Sulfur, Ice, Dehydration, and cONvection (POSIDON) campaign in October 2016.The western Pacific warm pool plays a leading role for transport of air into the TTL (Fueglistaler et al., 2005).Our measurements were carried out during nine flights aboard a NASA WB-57F high-altitude aircraft stationed in Guam.Throughout these flights, we extensively characterized TTL aerosols and trace gases (flight tracks are shown in Figure 1).To complement the measurements, we employed modeling techniques to obtain insights into the chemical composition of the aerosols.The combined approach of aerosol measurements, tracer analysis, and modeling offers insight into aerosol abundance and formation mechanisms within the TTL.

Measurements
The POSIDON campaign, took place during October 2016, aimed at improving our understanding of the physical and chemical processes occurring in the TTL.A total of nine flights were carried out on board the NASA WB-57 high altitude aircraft from Guam (13.5°N, 144.8°E).These flights covered the region from 0 to 15°N and from 130 to 160°E (Figure 1a), with vertical coverage spanning 0-19 km (Figure 1b).Each flight path consisted of several upward and downward segments between 14 and 18 km in altitude, providing extensive sampling of the TTL.The measurement region lies over the tropical warm pool, where high sea surface temperatures lead to widespread deep atmospheric convection (Yan et al., 1992).
The aerosol particles were sampled through a near-isokinetic inlet that reduces the ambient sample speed from the aircraft speed of ∼90-200 to ∼3 m s 1 while maintaining ambient aerosol mixing ratios (Jonsson et al., 1995).This inlet has been employed in various aircraft campaigns for aerosol sampling (e.g., Schwarz et al., 2006).Aerosol number size distribution was measured in situ by a custom-built optical particle counter, the portable optical particle spectrometer (POPS), which was mounted in the fuselage bay of the aircraft.The POPS uses a 405 nm laser to count and size individual aerosol particles with diameters from 140 to 3,000 nm (Gao et al., 2016).The scientific application of the POPS has been demonstrated by recent field campaigns (Cui et al., 2018;Liu et al., 2021;Yu et al., 2017).In the POPS instrument, each particle passing across the laser beam produces a pulse by scattering the laser light.The particle number is determined by the number of the pulses.The particle size is calculated from the intensity of the pulse, which was calibrated using a series of differential mobility analyzer (DMA) size-selected dioctyl sebacate (DOS) particles prior to the campaign.The aerosol mass was calculated from the aerosol number and size assuming that the particles are spherical and have a constant density of 1.6 g cm 3 .The mass mixing ratio (MMR) of aerosols was determined as the ratio of aerosol mass to the density of ambient air, which was calculated from the measured air pressure and temperature.A lognormal fit to the measured mass size distribution derived from the POPS measurements in the TTL (Figure S1 in Supporting Information S1) suggests that the POPS measurements captured approximately 50% of the total aerosol mass.Laboratory tests suggest that the POPS instrument used during POSIDON can provide reliable measurements under pressures as low as 70 hPa, which corresponds to an altitude of 18.9 km in this study.Therefore the POPS data acquired above 18.9 km are not used in the analysis.
The size distribution of particles with diameter larger than 3 μm was measured with a fast cloud droplet probe (FCDP).The FCDP detects particle forward scattering to determine the number and size of particles (Lance et al., 2010;McFarquhar et al., 2007).We assume that all particles greater than 3 μm are ice crystals as the number of large aerosol particles in the upper troposphere is negligible (Jensen et al., 2013).We found that ice crystals could likely abrade the inlet materials, resulting in artifacts in POPS-measured aerosol size distribution.Such interference has been observed in previous aircraft measurements (Murphy et al., 2004).For this reason, we excluded the POPS data when ice crystals were present from the analysis.
Additional real-time measurements included: (a) ozone mixing ratio obtained using a custom UV spectrophotometer designed for high altitude airborne deployment with high accuracy and precision (Gao et al., 2012), (b) water vapor mixing ratio measured by a two-channel tunable diode laser-based hygrometer, which is capable of accurately measuring low-concentration (below 1 ppm) water vapor in the upper troposphere and lower stratosphere (Thornberry et al., 2015), (c) N 2 O mixing ratio was measured by the PAN and other Trace Hydrohalocarbon ExpeRiment (PANTHER) instrument, which uses gas chromatographs with electron capture and mass spectroscopic detection (Elkins et al., 2002;Wofsy, 2011), and (d) ambient air pressure and temperature measured by the Meteorological Measurement System (MMS; Chan et al., 1989;Scott et al., 1990), which also records aircraft position with 1-s time resolution.

Modeling
We employed the Community Aerosol and Radiation Model for Atmospheres (CARMA), an advanced sectional aerosol model (Toon et al., 1988;Yu et al., 2015).CARMA is coupled with the Community Earth System Model (CESM), allowing for comprehensive analysis of aerosols.The model operates at a spatial resolution of 1.9°× 2.5°and employs a time step of 30 min.The domain extends from 4.74°N to 23.68°N in latitude and from 137.5°E to 162.5°E in longitude.The model incorporates 35 vertical pressure levels, spanning from the Earth's surface up to 200 hPa, and an additional 21 vertical pressure levels from 200 to 2 hPa.Simulations are nudged to meteorology from Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2; Gelaro et al., 2017).
CARMA tracks two groups of aerosols, with each group containing 20 size bins.The first group consists of pure sulfate particles with aerosol diameter ranging from 0.4 nm to 2.6 μm.These particles form through nucleation and condensation of water and sulfuric acid vapor (Zhao & Turco, 1995).The second group comprises internally mixed aerosols with the diameters varying from 100 nm to 17 μm.These mixed aerosols consist of particles that contain organic compounds, black carbon (BC), sea salt, dust, and condensed sulfate.
In addition, we employed a straightforward, observationally constrained chemical model to characterize the vertical distribution of O 3 within the TTL.This model is a one-dimensional column model that allows updrafts and vertical mixing but assumes no horizontal mixing.The O 3 formation process is represented using the Chapman mechanism (Seinfeld & Pandis, 2016).The model calculation is performed in the aircraft measurement region for the altitude range of 14.5-18.9km, with an input O 3 concentration of 28.8 ppb at 14.5 km, which represents the average observed O 3 concentration at that altitude.A detailed description of the model is provided in the Supporting Information.

Determination of TTL
We adopt the upper boundary of the TTL to be at 19 km, following SPARC (2006).This altitude is 1.5 km higher than the cold point tropopause (CPT) as shown in Figure S2 in Supporting Information S1.The CPT corresponds to an altitude of 17.5 km and air temperature of approximately 190 K.These values are in line with previous measurements of TTL (Fueglistaler et al., 2009;Gettelman et al., 2004).
The lower boundary of the TTL is determined by examining the vertical profile of potential temperature (θ; Figure S3 in Supporting Information S1).The curvature of the altitude-θ profile changes at 14.5 km, which is mathematically characterized as the lapse rate minimum (LRM) of θ.This change reflects the transition of stability regimes, that is, deep convection dominates air stability below the LRM and radiation starts to influence air temperature above the LRM.Accordingly, we identify the LRM level at 14.5 km (355 K, 140 hPa) as the base of the TTL (Gettelman & Forster, 2002), which is consistent with previous studies (Fueglistaler et al., 2009;Sunilkumar et al., 2017).As a result, the TTL spans from 14.5 to 19 km, and the majority of our measurements were conducted within the TTL (Figure 1).

Vertical Profile of Aerosols and Ozone
The vertical profile of aerosol mass from the surface to the top of the TTL can be segmented in three layers with distinct characteristics (Figure 2a).Layer I spans from the surface to 5 km altitude and represents the lower troposphere, in which the MMR of aerosols decreased logarithmically with altitude from 2 to 0.08 μg kg 1 likely because of the influence of surface emissions from Guam.Layer II ranges from 5 to 14.5 km altitude.In this layer, the aerosol MMR remained at approximately 0.08 μg kg 1 with small variability.Such low-concentration aerosol layers immediately below the convection outflow have been observed previously over the Northern Indian Ocean (de Reus et al., 2001) and the rain forests in South America (Andreae et al., 2018;Krejci et al., 2003).The observation indicates that deep convection serves as an effective sink for aerosol particles (Yu et al., 2019).Layer III lies in the TTL, extending from 14.5 to 18.9 km, with the upper boundary 1.4 km higher than the CPT.In this layer the aerosol MMR increased rapidly from 0.08 μg kg 1 at 14.5 km to 1.5 μg kg 1 at 18.9 km.Our observations show a sustained increase of aerosol MMR from the upper troposphere to lower stratosphere across the tropopause, rather than abrupt transitions.This suggests that the CPT has no unique role for the transport of aerosols.Accounting for the mass of <140 nm particles that were not measured by the POPS instrument, the aerosol MMR at 18.9 km would be approximately 3 μg kg 1 .This concentration is close to the measurement during the POLARIS mission in 1997 (Mclinden et al., 1999), in which an aerosol MMR of approximately 3.5 μg kg 1 was observed for aerosols with size range of 0.07-1 μm at 18.9 km in eastern Pacific (17.5°N, 159.3°W).
The ozone concentration remained below 35 ppb in layers I and II and began to increase at 14.5 km (Figure 2b).This observation is consistent with previous ozonesonde measurements in the tropical Pacific (Folkins  , 1999;Folkins & Martin, 2005).The concurrence of ozone minimum and LRM is in line with the analysis by Gettelman and Forster (2002) and Folkins et al. (2002).

Mechanisms for Aerosol Enhancement in the TTL
We find that the aerosol MMR was tightly correlated with O 3 in the TTL during all flights with small variation between flights.This can be clearly seen in the example time series of O 3 and aerosol MMR for the flight on October 18 shown in Figure 3a.The Pearson's correlation coefficient (r) for the campaign-average O 3 and aerosol MMR (averaged into 2-K θ intervals) was 0.98 in layer III.In contrast, the aerosol MMR and O 3 were anticorrelated with an r value of 0.35 in layer I (Figure 3b), and no correlation was observed between aerosol MMR and O 3 in layer II.
The main processes contributing to the increase in TTL O 3 from the bottom to the top of the TTL include chemical production via photolytic dissociation of molecular oxygen (O 2 ) (Crutzen et al., 1999;Prather, 2009) and isentropic in-mixing of stratospheric air from the extratropical lower stratosphere (Konopka et al., 2010;Ploeger et al., 2012).The relative contribution of these processes to O 3 remains unclear.While some studies argue that in situ chemical production dominates (Avallone & Prather, 1996), others suggest that isentropic stratospheric inmixing can contribute to O 3 by as much as 40%-60% (Abalos, Randel, et al., 2013, Abalos, Randel, et al., 2013;Konopka et al., 2009Konopka et al., , 2010;;Ploeger et al., 2011Ploeger et al., , 2012;;Sargent et al., 2014).This contribution is most significant during the boreal summer months (June-August), primarily driven by the Asian summer monsoon, and gradually decreases during transition into the fall and winter months.In particular, these studies suggest that the contribution from in-mixing falls within the range of 0%-20% in October, averaged over a ±10°N latitude range, with the contribution increasing with altitude.
We used a one-dimensional column model (over the flight region) and tracer analysis to examine these processes.The model calculates the vertical distribution of O 3 within the TTL, considering a constant slow vertical ascent rate of 0.25 mm s 1 (Avallone & Prather, 1996;Park et al., 2010) of air in the TTL and assuming it evolves in isolation (the model description section in the Supporting Information).Remarkably, the calculated vertical profile of O 3 closely aligns with the observed O 3 profile (Figure 4a).This result is in line with earlier investigations that yielded similar results using a column model (Avallone & Prather, 1996) and with research that assumed the tropics are isolated from extra-tropics when explaining the annual cycles of ozone above the tropical tropopause (Randel et al., 2007;Schoeberl et al., 2008).However, we acknowledge the simplicity of the model.For example, the model does not consider the chemistry of nitrogen oxides (NO and NO 2 ; not measured during POSIDON) and O 3 , which could be important in the TTL O 3 abundance (Nussbaumer et al., 2003), and uncertainties exist in the assumed ascent rate.On the other hand, the tracer analysis shows evidence of stratospheric in-mixing.Figure S4 in Supporting Information S1 shows that N 2 O was anticorrelated with O 3 in the TTL during POSIDON.The observed negative correlation indicates that the sampled TTL air included a contribution from mixing of stratospheric origin (Folkins et al., 1999).This is because the strong UV radiation at higher altitudes in the stratosphere photolyzes N 2 O while leading to O 3 production, resulting in an anticorrelation of N 2 O and O 3 (Assonov et al., 2013).In contrast, N 2 O in the troposphere is inert and uniformly distributed, thus anticorrelation of N 2 O with O 3 is not expected and not observed below the TTL during our measurements.The synthesis of modeling and tracer analysis indicates contributions from both chemical production and stratospheric in-mixing to TTL O 3 .Given the strong correlation between O 3 and aerosols in the TTL, these results suggest it is possible that TTL aerosols might also be influenced by these chemical and physical processes.
To investigate the chemical composition of TTL aerosol, we employed the CARMA model embedded in CESM to simulate the aerosol formation and growth processes.The model reproduced the O 3 profile reasonably well (Figure 4a).Our model results suggest that sulfate aerosol exhibits a consistent increase with altitude, whereas organic aerosol displays an ascending trend below the tropopause but declines above it.Simulated contributions of black carbon, dust, and sea salt are negligible partly because these aerosols are removed through the secondary activation above the cloud base via deep convection (Yu et al., 2019).Moreover, our modeled vertical profile of the total aerosol MMR demonstrates good agreement with the in situ measurements (Figure 4b) below the tropopause, with organic aerosol constituting a significant fraction of the aerosol mass.These modeling results imply that sulfur alone is insufficient to explain the observed aerosol mass, and organic precursors may play a pivotal role in the formation and/or growth of TTL aerosols under low-temperature conditions.This finding aligns with previous findings from TTL measurements during the Pre-AVE and CR-AVE campaigns over Southwest Central America in boreal winter (Froyd et al., 2009), which highlighted the prevalence of organicsulfate particles as the most abundant particle type in the TTL and lower stratosphere.We note that the model underestimates the aerosol mass above the tropopause, with the difference increasing with altitude within the TTL, reaching 60% at an altitude of 1 km above the CPT.Additional research is needed to understand this discrepancy.
We postulate that TTL aerosols are likely generated through the processes of new particle formation and subsequent growth following convective outflow in the upper troposphere (Williamson et al., 2019).The observed aerosol number size distribution in the TTL suggests particle growth with increasing altitude (Figure S5 in Supporting Information S1), although obtaining measurements of smaller particles would enhance our understanding.During convective transport, soluble species are effectively removed, whereas the insoluble and weakly soluble species can endure washout and gradually ascend upward, with minimal loss, to reach the stratosphere (Bechara et al., 2010).The TTL, characterized by low temperature, low particle surface area density, and high relative humidity, provides ideal conditions for new particle formation and growth.Multiple studies have proposed the upper troposphere as the primary nucleation region (Brock et al., 1995;Weigel et al., 2011), and investigations have revealed the involvement of organics in new particle formation and initial growth in the remote tropical upper troposphere (Kupc et al., 2020).Froyd et al. (2009) observed an increase in the signal of particulate oxidized organics with altitude in the TTL up to 19 km via aircraft measurements.They hypothesized that this increase could be attributed to particle growth through aging processes.Future measurements analyzing the composition of oxidized organic species will enhance our understanding of particle formation and growth pathways.Exceptional overshooting that crosses the tropical tropopause could also affect the aerosol abundance in the TTL (Vernier et al., 2011).This possibility is examined using H 2 O as a tracer.The θ-H 2 O profile displayed no spikes above the tropopause (Figure S6 in Supporting Information S1), suggesting the absence of overshooting during the measurements.
Building upon the strong correlation between TTL aerosol mass and O 3 in our measurements, we further derived an empirical parameterization of aerosol MMR as a function of O 3 in the TTL using linear regression.The derived relationship is expressed as MMR (μg kg 1 ) = 0.0074 (±2.2 × 10 4 ) × O 3 (ppb) + 0.23 (±0.049), accounting for the 50% mass that was not measured by the POPS instrument.This parameterization may be used to estimate TTL aerosol abundance and for validation of modeled results within the measurement region during the measurement months.

Conclusions
Our aircraft measurements conducted over the western Pacific revealed a vertically stratified aerosol distribution comprised of three distinct layers.The lower-troposphere layer, ranging from 0 to 5 km in altitude, exhibited a decrease in aerosol mass with height, primarily influenced by surface-level emissions.The layer between 5 and 14.5 km displayed consistently low aerosol concentrations with minimal variability, indicating effective aerosol removal through deep convection processes.In the TTL, we observed an enhanced aerosol mass that exhibited a strong correlation with O 3 .The modeling and tracer analysis suggest that TTL O 3 and aerosols likely originate from a combination of chemical production and stratospheric in-mixing processes.Furthermore, based on the linear relationship observed between aerosol MMR and O 3 in the TTL, we derived an empirical parameterization that may allow for the estimation of aerosol MMR as a function of O 3 within the observation region and during the observation months.
While the parameterization shows promise for estimating TTL aerosol abundance using ozone, which is relatively easier to measure and model, its application is constrained by significant limitations as measurements were conducted in a specific region of the TTL under background conditions during a particular time period, and the POPS instrument had limited measurement capabilities, including assumptions about aerosol density and a limited particle measurement range that excludes smaller particles.Therefore this framework requires extensive future research for verification and improvement.
Our study also highlights the challenge of accurately predicting aerosol mass, particularly above the tropopause, even for the most advanced models.This underscores the existence of significant gaps in our understanding of the origin and formation mechanisms of aerosols in the TTL.To achieve a more comprehensive understanding of TTL aerosol, it is crucial to conduct future measurements that encompass detailed aerosol chemical composition and gas-phase precursors across different locations and seasons.These endeavors will help diagnose potential deficiencies in the models and validate the modeling results.Consequently, they will lead to an improved comprehension of TTL aerosols and facilitate a predictive understanding of their effects on stratospheric chemistry, clouds, and climate.

Figure 2 .
Figure 2. Potential temperature (θ) versus (a) MMR and (b) O 3 for data acquired from 0 to 5, 5-14.5, and 14.5-18.9km measurements for all flights.Individual data points (5-s averages) are shown by the dots and the averages (by θ of 2 K) are shown by the solid squares.

Figure 3 .
Figure 3. (a) Example time series of aerosol MMR and O 3 for the October 18 measurements.MMR is represented by solid circles and colored by flight altitude.O 3 is represented by solid squares (not colored by altitude change).The gaps indicate ice-influenced data that were excluded from the analysis.(b) Scatter plot of MMR versus O 3 for the 0-5 km (layer I), 5-14.5 km (layer II), and 14.5-18.9km (layer III) measurements.Individual data points are shown by the dots and the averages are shown by the solid squares.Logarithmic scales are used to make the small values visible.

Figure 4 .
Figure 4. Comparison of modeling results to measurements for (a) O 3 and (b) aerosol MMR.Individual measurement points are shown by the gray dots.The median values and the 25th and 75th percentiles are shown by the blue line-square symbols and the error bars.The y-axis represents altitude relative to the tropopause.