Seasonal Evolution of Titan's Stratosphere During the Cassini Mission

Titan's stratosphere exhibits significant seasonal changes, including breakup and formation of polar vortices. Here we present the first analysis of midinfrared mapping observations from Cassini's Composite InfraRed Spectrometer to cover the entire mission (Ls=293–93°, 2004–2017)—midnorthern winter to northern summer solstice. The north polar winter vortex persisted well after equinox, starting breakup around Ls∼60° and fully dissipating by Ls∼90°. Absence of enriched polar air spreading to lower latitudes suggests large‐scale circulation changes and photochemistry control chemical evolution during vortex breakup. South polar vortex formation commenced soon after equinox and by Ls∼60° was more enriched in trace gases than the northern middle‐winter vortex and had temperatures ∼20 K colder. This suggests that early‐winter and middle‐winter vortices are dominated by different processes—radiative cooling and subsidence‐induced adiabatic heating respectively. By the end of the mission (Ls=93°) south polar conditions were approaching those observed in the north at Ls=293°, implying seasonal symmetry in Titan's vortices.


Introduction
Saturn's largest moon Titan has a thick atmosphere comprising ∼98% nitrogen and ∼2% methane with ∼1.5 bar surface pressure (Fulchignoni et al., 2005). Titan has a rich C-N-H photochemistry originating from radicals formed by dissociation of N 2 and CH 4 by ultraviolet photons and magnetospheric electrons (Dobrijevic et al., 2014;Krasnopolsky, 2009;Lavvas et al., 2008;Loison et al., 2015;Vuitton et al., 2019;Wilson & Atreya, 2004). These radicals produce a wide range of higher-order hydrocarbon and nitrile trace gas species with lifetimes ranging from a few seconds to thousands of years (Vuitton et al., 2019;Wilson & Atreya, 2004). Oxygen species also contribute to atmospheric chemistry (Dobrijevic et al., 2014;Hörst et al., 2008), but only H 2 O, CO, and CO 2 have been detected so far. Trace gases are produced in the upper atmosphere (∼1,000 km) and condense in the cold lower stratosphere, giving rise to vertical abundance profiles that have increasing volume mixing ratio (VMR) with increasing altitude (Coustenis et al., 1991;Teanby et al., 2007;Vinatier et al., 2007). The vertical mixing timescale between source and sink regions controls the vertical gradient, with the shortest-lifetime gases having the steepest vertical gradients due to their more rapid loss away from the source region  10.1029/2018GL081401 . Cassini completed 294 Saturn orbits and 127 close Titan flybys between orbital insertion on 1 July 2004 (L s =293 • ) and its final plunge into Saturn's atmosphere on 15 September 2017 (L s =93 • ). Saturn/Titan's year lasts 29.46 Earth years, so Cassini's 13-year Saturn system tour covered almost half a Saturn/Titan year. Saturn's, and hence Titan's, obliquity is 26.7 • , so significant seasonal effects were observed during the mission, spanning northern winter to northern summer solstice. Terrestrial general circulation models (GCMs) adapted to Titan's atmosphere predict that middle atmosphere (stratosphere and mesosphere) meridional circulation is dominated by a single south-to-north circulation cell during northern winter, with upwelling in the southern hemisphere and subsidence at the north pole. This circulation reverses around equinox via a short-lived intermediate circulation with two hemispheric cells that upwell at the equator and subside at both poles (Hourdin et al., 1995;Lebonnois et al., 2012Lebonnois et al., , 2014Lora et al., 2015;Newman et al., 2011). Such changes have been investigated by observing how short-and intermediate-lifetime trace gas distributions vary over the mission. For example, subsidence advects photochemical species downward, which causes stratospheric abundances to increase. Therefore, gas abundance can be used as a tracer of vertical motion (Teanby et al., 2012). This is particularly obvious over Titan's winter poles, where subsidence can lead to very large trace gas enrichments (Coustenis et al., 2016(Coustenis et al., , 2018Teanby et al., 2012Teanby et al., , 2017Vinatier et al., , 2015.
Cassini's varied orbital tour provided a unique vantage point for observing Titan's winter pole, which has been known since the Voyager flybys to be the most enriched in trace gases . Cassini observations are particularly valuable as observing the winter pole is not possible from Earth due to viewing geometry. Subsets of the Cassini data have been analyzed previously and show that in northern winter the north pole was much more enriched in trace gases than other latitudes (Coustenis et al., 2007(Coustenis et al., , 2010(Coustenis et al., , 2016Flasar et al., 2005;Sylvestre et al., 2018;Teanby et al., 2006Teanby et al., , 2010b. Thermal winds derived from stratospheric temperature fields show that the northern winter pole was surrounded by a strong polar vortex, which isolated the polar air mass, allowing extreme gas enrichments to develop (Achterberg et al., 2008(Achterberg et al., , 2011Teanby et al., 2010b). However, there is also some evidence for cross-vortex mixing of intermediate-lifetime species such as HCN in the middle stratosphere . Short-lifetime gases are more enriched due to their steeper vertical gradient and resulting greater sensitivity to downward advection (Teanby et al., , 2010a. Later in the mission, after the 2009 northern spring equinox, the south pole began to become enriched in trace gases, indicating that the circulation had reversed and subsidence was now occurring at the southern pole (Teanby et al., 2012). There was also compositional evidence for the transitional two-cell circulation predicted by numerical models (Vinatier et al., 2015). Following equinox, enrichment at the south pole was greater than that observed in the north at the start of the mission (Coustenis et al., 2018;Teanby et al., 2017). The south polar stratosphere also achieved extremely cold temperatures, which created ice clouds of HCN (de Kok et al., 2014) and benzene  at ∼300 km. These cold temperatures were not observed in the north and could be caused by extreme trace gas enrichments acting as infrared coolers, combined with slow initial subsidence producing only modest levels of adiabatic heating (Teanby et al., 2017). This illustrates the importance of trace gases in Titan's overall atmospheric energy budget Teanby et al., 2017).
The meridional circulation also affects Titan's aerosols. Cassini's Visual and Infrared Mapping Spectrometer observations show winter subsidence is associated with thick lower stratosphere condensate clouds over the poles (Le Mouélic et al., 2018). Cassini's Imaging Science Subsystem observations of Titan's detached haze layer at 350-500 km imply upwelling speeds greater than haze particle free-fall velocity are required to dynamically clear the lower mesosphere of haze (West et al., 2018). These Visual and Infrared Mapping Spectrometer and Imaging Science Subsystem observations further confirm the meridional circulation inferred from Cassini CIRS and GCMs.
Here we present the first analysis of all CIRS midinfrared mapping sequences from the entire Cassini mission, spanning 2004-2017 (L s = 293-93 • , ΔL s = 160 • ), almost half a Titan year. This unique data set allows stratospheric temperature and composition to be determined along with seasonal variations. Previous CIRS studies have been limited to southern latitudes only (e.g., Teanby et al., 2017) or only included data taken up until 2016 (e.g., Coustenis et al., 2018). In addition, our analysis uses an improved methodology, incorporating limb observation-based temperature a priori, which gives more robust temperature and composition inversions. CIRS has the advantage of high spatial and temporal resolution, including the polar regions, from a single instrument, which we take full advantage of by simultaneously analyzing the entire data set with the same methodology. Previous studies have only analyzed partial data sets, which made direct comparison between seasons more difficult. We use our analysis to answer three key questions: (1) are thermal and chemical behaviors of Titan's north and south polar vortices comparable? (2) what are the main factors controlling chemistry during polar vortex breakup? (3) how long does the winter polar vortex breakup take? These questions are essential for understanding Titan's atmospheric chemistry and dynamics, in addition to constraining future GCMs and photochemical models.

Radiative Transfer Analysis
CIRS observation coverage and example spatial binning to improve signal-to-noise ratio are shown in Figure 1. Observations sequences and data preprocessing are summarized in supporting information Table  S1 and Text S1.
The inversion method is explained in detail in previous papers Teanby et al., 2006Teanby et al., , 2010b and is briefly summarized here. Spectra were inverted for temperature and composition using the NEMESIS retrieval code , which employs an optimal estimation technique. Partial derivatives of the forward modeled radiances were calculated analytically with respect to each model variable using radiative transfer theory. The temperature and composition were then iteratively adjusted to fit the data while remaining as close to the a priori profiles as possible . (c, e) Fits to the trace gas vibrational features in the FP3 spectrum and (d, f) corresponding trace gas volume mixing ratio (VMR) profiles. All gases are assumed to have a uniform abundance profile above the condensation level. This simplified parameterization is sufficient to fit these data. Note that C 2 H 4 does not condense under Titan's atmospheric conditions.
Here we use a two-stage inversion process similar to that in Teanby et al. (2010b). First, a continuous temperature profile was retrieved for each binned spectrum using the 4 methane band in FP4 over the spectral range 1,240-1,360 cm −1 . A latitude-and time-dependent temperature a priori (supporting information Text S2) was used as the starting point, with an a priori error varying as a function of latitude between eq = 2 K at the equator and pole = 5 K at the poles according to = pole − ( pole − eq ) cos . This accounted for the greater uncertainty in polar temperatures compared to the more stable and well-constrained equator, where we have the advantage of Huygens probe direct temperature measurements (Fulchignoni et al., 2005). A correlation length of 1.5 atmospheric scale heights was assumed for the off-diagonal elements of the temperature a priori covariance matrix to ensure a smooth temperature profile was retrieved . Second, the temperature was fixed and uniform a priori gas profiles (supporting information Text S3) were scaled to provide the best fit to the observed spectra. Example fits to the data are shown in Figure 2 along with the corresponding fitted temperature and composition profiles.
Nadir observation sensitivity is limited to midstratosphere to lower mesosphere (∼5-0.1 mbar), with peak information content at ∼1 mbar. Near the winter pole the range of sensitivity is 1-0.01 mbar for cases with a . The 1-mbar surfaces are fitted to irregularly spaced inversion results using a 2-D spline in tension (Smith & Wessel, 1990) with a grid spacing of 0.1 years (∼1 • of L s ) and 1 • latitude. Gray areas indicate data gaps. Units of composition are log 10 (VMR). hot stratopause and very cold lower to middle stratosphere. Contribution functions for both these cases are plotted in   Figure 3). Figure 3 shows the latitudinal variation of temperature and composition at the 1-mbar pressure level with season for the entire Cassini mission. Figure 4 shows the seasonal variation in time series form for latitudes 80 • S, 50 • S, 0 • N, 50 • N, and 80 • N. In this section we briefly discuss gross features of the temperature and composition results. Implications for Titan's atmosphere are considered in section 4.

Seasonal Temperature Variations
The midstratosphere 1-mbar temperatures shown in Figure 3 are warmest at the equator, with peak temperatures occurring at L s ∼ 30 • . The equatorial temperature maximum moves northward as the season progresses from northern winter to northern summer, from ∼10 • S at L s = 293 • to ∼0 • N from L s ∼ 30 • onward. Equatorial temperatures at 1 mbar are relatively stable until L s ∼ 30 • , after which they reduce by ∼5 K between L s = 30 • and 93 • . Temperatures at 1 mbar are coldest over the winter poles. The south pole has the coldest observed temperatures at this pressure level, occurring at L s ∼ 60 • in early southern winter. Figure 4 shows the temperature variation at 5, 1, and 0.1 mbar for comparison. The temperature at 0.1 mbar displays the most seasonal variation, especially near the poles with a ∼20 K preequinox cooling at the north pole and a ∼45 K postequinox cooling and subsequent recovery at the south pole. The 5-mbar temperature exhibits a steady postequinox decrease for the equator and southern hemisphere and a steady increase in the northern hemisphere.  (Teanby, 2007). VMR = volume mixing ratio.

Seasonal Composition Variations
Seasonal composition variations in Figure 3 are most appropriate for the 1-mbar level, based on contribution functions for the uniform profiles assumed here. Gases display a consistent behavior, with high concentrations over the poles during winter and approximately constant abundance at the equator, except for CO 2 , which is roughly constant at all latitudes. The degree of enrichment over the poles depends on the gas species; HC 3 N and C 4 H 2 show the most extreme polar enrichments (2-3 orders of magnitude), whereas HCN, C 2 H 2 , C 2 H 4 , C 2 H 6 , and C 3 H 4 have more modest enrichments (∼1 order of magnitude or less). HC 3 N and C 4 H 2 react faster to the seasons than the more modestly enriched species, with HC 3 N changing the fastest.
Note that absolute VMR values depend on profile assumptions, which can vary with latitude and season. For most gases, which have modest vertical gradients, this effect is small and the absolute VMRs can be 10.1029/2018GL081401 considered representative of the stratosphere. However, for short-lifetime gases like HC 3 N and C 4 H 2 that have very steep vertical gradients, the peak of the contribution function can shift to lower pressures, which leads to an overestimate of the stratospheric abundances (Teanby et al., 2006). Therefore, the uniform profile results presented here should only be used to inspect relative abundance variations as these are robust to profile assumptions. For example, relative abundance comparisons of HC 3 N for different latitudes and times are valid, but absolute abundance comparisons between HC 3 N and C 2 H 2 are not valid. This is a fundamental limitation of the CIRS nadir spectra, which are sensitive to a broad pressure range and contain limited vertical information. Where absolute abundances are critical (e.g., for photochemical model profile comparisons) limb viewing data is required (e.g., Teanby et al., 2012;Vinatier et al., 2015).

Polar Vortex General Characteristics
In general, Titan's stratospheric winter polar vortices are characterized by cold temperatures at 1 mbar, 20-50 K colder than at the equator (Figure 3). These cold temperatures cause a strong thermal gradient, which drives strong circumpolar winds, creates a potential vorticity gradient and mixing barrier, and effectively isolates the polar airmass from more equatorial latitudes (Achterberg et al., 2008(Achterberg et al., , 2011Flasar et al., 2005;. This allows subsiding air masses to enrich the polar stratosphere in trace gas species, which are advected down from the upper atmosphere photochemical production zone where they are more abundant . Subsidence causes significant adiabatic heating at low pressures and leads to a hot stratopause over the winter pole (Achterberg et al., 2008;Teanby et al., 2017). Trace gas enrichment is largest for short-lifetime gases (HC 3 N and C 4 H 2 ), which tend to have steeper vertical gradients and are more susceptible to enrichment by subsidence . The polar isolation is also most effective for short-lifetime species, whereas intermediate-lifetime species such as HCN and C 2 H 2 can be mixed across the vortex barrier and leach to lower latitudes . Longer-lifetime species are less enriched, with the very long lifetime species CO 2 showing virtually no variation, indicating a more uniform vertical profile consistent with a lifetime much longer than any dynamical timescale. This general picture is confirmed by our new results (Figures 3 and 4), but our longer time series covering the whole Cassini mission (L s = 293-93 • ) allows further insight into vortex breakup and formation processes by observing both north and south poles (sections 4.2 and 4.3).

North Polar Vortex Evolution
The northern winter vortex was already well established when Cassini arrived at L s = 293 • and extended from the north pole down to ∼45 • N. Initially, the 1-mbar temperatures were around 140 K, compared to 170 K at the equator, but had warmed to 160 K by the end of the mission at L s = 93 • (Figure 4). As the mission progressed the vortex shrank in extent, being limited to latitudes north of 60 • N from L s = 20 • onward ( Figure 3). Our results show the north polar vortex endured well beyond the northern spring equinox (L s = 0 • ), maintaining cold temperatures and enriched trace gas abundances until at least late spring (L s ∼ 60 • ). After this point vortex dissipation was first visible in temperature and short-lifetime gases (HC 3 N and C 4 H 2 ) as a warming and abundance reduction. Early stages of vortex breakup were also visible in a subset of observations taken at 70 • N before mid-2016 (L s < 80 • ) by Coustenis et al. (2018).
The hot stratopause at 0.1 mbar driven by adiabatic heating was already cooling at L s ∼ 320 • and had stabilized in temperature by L s = 0 • , indicating a weakening in north polar subsidence. The midstratosphere at 1 mbar was slower to respond because of the longer radiative time constant, with temperatures increasing more gradually by ∼20 K between L s = 0 • and 90 • , due to increased insolation. These observations indicate weakening of the vortex. A reduction in HC 3 N and C 4 H 2 abundance starting at L s ∼ 50 • is the first sign of vortex breakup. Longer-lifetime species dissipated over the proceeding L s ∼60-90 • period. There was a north polar data gap at L s ∼ 70 • , but the final orbits of Cassini around northern summer solstice (L s = 90 • ) showed that north polar gases had attained almost equatorial abundances and vortex breakup was complete.
Interestingly, our results show that after vortex breakup, spreading of air enriched in trace gases to lower latitudes did not occur (Figure 3). This shows that polar gas depletion by small-scale cross-latitude mixing, which would enhance trace gas abundances at subpolar latitudes postvortex breakup, is small compared to other effects. Furthermore, this suggests that the bulk of stratospheric trace gas loss must be due to a combination of large-scale dynamics and photochemistry. In this scenario, upwelling from the reversed meridional Hadley circulation would advect trace gas depleted air from lower latitudes into the north polar midstratosphere, leading to a reduction in abundances. Photochemical loss, enabled by increasing insolation after equinox, could also contribute to a reduction in polar stratospheric trace gas abundance during northern spring.

South Polar Vortex Evolution
South polar nadir observations have a data gap after equinox, but previous limb viewing observations showed the south polar winter vortex formed almost immediately after northern spring equinox (Teanby et al., 2012;Vinatier et al., 2015). Early southern winter vortex temperatures dramatically cooled to ∼120 K during L s = 0-60 • at 1 mbar. This is much colder than was observed in the middle-winter north polar vortex and can explain the high-altitude HCN and C 6 H 6 ice clouds observed by de Kok et al. (2014) and Vinatier et al. (2018). The extremely cold temperatures could be caused by enhanced radiative cooling from enriched trace gases, which are effective infrared coolers Teanby et al., 2017). In the early stages of vortex formation this cooling was not mitigated by significant adiabatic heating as the reversed meridional Hadley circulation was initially quite sluggish (Teanby et al., 2012(Teanby et al., , 2017. It is also possible that enhancement of infrared cooling gases may form a radiative feedback, contributing to establishing the reversed circulation (Teanby et al., 2017). Postequinox the south polar vortex continued to grow and by L s ∼ 90 • it was similar in extent to the middle-winter northern vortex, reaching 45 • S. It is likely that the early northern winter vortex also exhibited this behavior, but this phase would occur for L s = 180-270 • prior to Cassini's arrival, so remains unobserved.
For the shortest-lifetime gases (e.g., HC 3 N and C 4 H 2 ), there appeared to be some overshoot of enrichment during vortex formation, with south polar abundances being enhanced by up to an order of magnitude compared to those observed in the middle-winter north polar vortex. This can be seen in Figure 4, with abundances peaking around L s = 60-90 • at 80 • S. Therefore, the largest winter polar trace gas enrichments occurred between equinox and winter solstice. This effect was also visible in limb observations (Teanby et al., 2012(Teanby et al., , 2017Vinatier et al., 2015), but with peak abundances occurring earlier (L s < 60 • ) at high altitude (>300 km, <0.1 mbar) and later (L s > 60 • ) at low altitude (<300 km, >0.1 mbar). Near the end of the mission (L s = 93 • ), temperatures and abundances appeared to be trending toward those observed in northern midwinter (Figure 4). This suggests that north and south polar vortices have similar temperatures and chemistry at similar seasonal phases.
One potential explanation for the extreme south polar enrichments observed during L s = 0-90 • requires a combination of photochemistry, insolation, and dynamics. Close to equinox, upper atmosphere abundances of trace gases were highest because the southern hemisphere was just exiting summer-a period of relatively high insolation and photochemical production. Then, after equinox, as the circulation reversed and subsidence developed, the trace gas vertical profiles were advected downward, leading to large stratospheric enrichments. However, as the season progressed, overall insolation in the southern hemisphere steadily decreased, in turn reducing upper atmosphere photochemical production, which led to lower trace gas abundances at high altitude. This reduced the supply of trace gases into the vortex and led to less extreme polar enrichments. The latitude extent of the vortex also could play a role. Initially, the vortex had a small-latitude extent, so drew air from the upper atmosphere directly above the pole, which had constant illumination above 300-km altitude throughout winter, but as the vortex grew the air in the vortex was sourced from a wider region of the southern hemisphere, which had lower insolation and reduced upper atmosphere trace gas abundances. This conceptual explanation would need a coupled GCM and photochemical model to investigate further.

Equatorial Temperatures
The 1 mbar temperature maximum was skewed toward the subsolar point but stayed within 10 • of the equator. The temperature at 5 mbar exhibited a steady 5-K cooling from L s = 293-93 • , which was caused by the increase in Sun-Saturn distance over the mission (Figure 1a). This effect was also clearly observed in the lower stratosphere (10-30 mbar) in previous analysis of far-infrared FP1 spectra (Sylvestre et al., 2019). At lower pressures (∼1-0.1 mbar) dynamics had a larger effect on the temperature  and the effect of the increasing Sun-Saturn distance was less apparent.

Conclusions
Cassini CIRS nadir mapping observations of Titan were analyzed for the entire mission duration, which ran from northern midwinter (L s = 293 • ) until just after northern summer solstice (L s = 93 • ), to obtain seasonal 10.1029/2018GL081401 latitude variation maps at the 1-mbar pressure level. The equatorial composition was stable for all the species analyzed, whereas the midlatitudes and polar regions exhibited significant variability.
Winter vortices were observed at both poles with cold temperatures and isolated air masses enriched in trace gas species. The enrichments were largest for short-lifetime gases and more modest for those with longer lifetimes, in keeping with previous studies (Teanby et al., , 2010a. However, early-and middle-winter vortices evidently had very different properties. When newly formed, the south polar vortex was significantly more enriched in trace gases and achieved colder temperatures than its established northern counterpart. Trace gas abundances were up to an order of magnitude greater at the southern early-winter pole than the northern middle-winter pole. This could be due to enhanced radiative cooling from trace gases combined with a relatively weak subsidence and initially steeper photochemical profiles during vortex formation. This suggests early vortex temperature structure was dominated by radiative cooling, whereas middle-winter vortex temperatures were dominated by subsidence-driven adiabatic heating. Dissipation of the north polar vortex was a gradual process and was only complete ∼90 • of L s after equinox. Gas enrichments did not appear to spread from high to low latitudes during vortex breakup, suggesting that photochemistry and depletion due to large-scale dynamics related to the meridional circulation (upwelling) were more important loss mechanisms than small-scale cross-latitude mixing. Near the end of the mission the temperature and composition of the south polar vortex were trending toward those observed in the north at the start of the mission. However, as the Cassini data set covers ΔL s = 160 • , we are 20 • of L s short of a complete half Titan year record. Therefore, it is not possible to definitively confirm that north and south poles reach equivalent states for equivalent seasons, although a short extrapolation of the data suggests seasonal equivalence is very likely. This implies temperature and composition differences between the middle-winter north polar vortex and early-winter south polar vortex were primarily due to seasonal phase, not any hemispheric asymmetry.
Over the next few years it will be important to observe the north polar temperature and composition using ground-and space-based facilities such as ALMA and JWST. ALMA has already produced moderate resolution latitude maps, which have the potential to complete the CIRS record at high northern latitudes (Thelen et al., 2019(Thelen et al., , 2018. Unfortunately, with the loss of Cassini, the winter poles can no longer be observed until we next visit Titan. Acknowledgments N. A. T., M. S., J. S., and P. G. J. I. were funded by the UK Science and Technology Facilities Council (STFC). C. A. N. was supported by the NASA Cassini project and the NASA Astrobiology Institute. S. V. was funded by the French Centre National d'Etudes Spatiales (CNRS). The Cassini CIRS observations, summarized in Table S1, are publicly available from NASA's Planetary Data System (https://pds.nasa.gov). Temperature and composition results from this paper are available in supporting Data Sets S1 and S2, with a gridded NetCDF format product in Data set S3.