Why do we get the composition of the atmosphere over the Southern Ocean so wrong?  

Background and motivation 

The atmosphere over the Southern Ocean is the most pristine remote marine environment on Earth. Far from continental pollution sources, its composition is set by the interplay of natural ocean–atmosphere exchange (dimethyl sulfide, sea salt, halogens, etc.), long-range transport, stratospheric input and, around the continent itself, emissions from the sea-ice zone. This makes the region a unique natural laboratory. It is the closest modern analogue to the pre-industrial atmosphere, and therefore anchors our estimates of how much humans have perturbed oxidants and aerosol since industrialisation. It is also a region of first-order importance for climate, with aerosol–cloud interactions over the Southern Ocean remaining one of the largest sources of uncertainty in Earth-system models. 

Global chemistry models perform notably poorly here, and they fail in inconsistent ways. Multi-model assessments (the Tropospheric Ozone Assessment Report and CMIP6 evaluations) show systematic biases in Southern Hemisphere tropospheric ozone, with many models overestimating its abundance and misrepresenting its seasonal cycle at surface sites such as Cape Grim or at the Antarctic sites. A likely contributor is that many of these models do not represent the halogen chemistry (Cl, Br, I) occurring in these locations.  GEOS-Chem (the model at the centre of this project and one of the most widely used community chemistry-transport models) has the opposite problem around Antarctica: benchmark comparisons show recent versions of the model underestimate ozone at high southern latitudes, a bias that has grown as more comprehensive halogen chemistry has been introduced. Halogen chemistry reduces the global tropospheric ozone burden by roughly 10–15% in current model formulations, with the largest relative effects in the pristine Southern Hemisphere, yet the underlying processes (sea-salt debromination, dechlorination, iodine recycling, multiphase HOBr chemistry, etc.) remain poorly constrained — leaving open whether the model now destroys ozone too efficiently in this region, does so for the wrong reasons, places the loss in the wrong region or season (e.g. over the open ocean rather than in the springtime sea-ice zone, where blowing-snow and sea-ice bromine sources dominate), or is missing compensating sources such as stratospheric input.  

Over the last decade the discovery of new DMS oxidation intermediates such as HPMTF, and the importance of their processing by clouds, has changed our understanding of ocean–atmosphere sulfur chemistry. Given the large DMS emissions in the region, and the potential for halogen (BrO, Cl) oxidation of DMS to be a significant sink here, re-evaluating the simulated sulfur cycle in the Southern Ocean is a natural companion to the halogen work.   

Because the Southern Ocean atmosphere is so chemically clean, it is unforgiving: processes that might be masked by anthropogenic emissions elsewhere dominate the budgets here. A model that gets the Southern Ocean right for the right reasons has passed a demanding test of its underlying process descriptions — with direct consequences for estimates of the global ozone budget, the oxidising capacity of the atmosphere and the pre-industrial natural sulfate baseline against which aerosol radiative forcing is calculated. 

Hypotheses and research questions 

The project will test three linked hypotheses: 

• H1. The model’s ozone underestimate at high southern latitudes arises primarily because multiphase halogen chemistry destroys ozone too efficiently over the open Southern Ocean — through excessive sea-salt debromination (linked to the treatment of aerosol acidity) and/or iodine recycling — rather than from errors in stratospheric input or deposition. 
• H2. Reproducing the observed seasonal cycles of MSA and non-sea-salt sulfate across the Southern Ocean requires both the expanded DMS oxidation mechanism (HPMTF and its cloud loss, MSA formation, methanethiol) and halogen (BrO, Cl) oxidation of DMS; together these substantially alter the simulated natural sulfate burden. 
• H3. The NOx that sustains ozone production over the Southern Ocean is supplied mainly by long-range transport and thermal decomposition of PAN, with a contribution from oceanic alkyl nitrates, and errors in this supply are a secondary contributor to the ozone bias compared with halogen chemistry. 

Overarching question: can a single, self-consistent model configuration simultaneously reproduce the observed Ox, NOx, SOx and halogen environment of the Southern Ocean and Antarctic coast — and if not, what is missing? 

Project description and work packages 

WP1 — Observational synthesis and baseline model evaluation (Months 1–14). The student will assemble the pre-existing observational record for the region (south of ~40°S) into a single harmonised evaluation database, drawing on the platforms in Table 1. Priority will be given to a core set of species (O₃, BrO, IO, DMS, MSA, nssSO₄ and PAN) and platforms (ATom, SOCRATES, ACE, Cape Grim and the Antarctic stations), with other datasets added as time allows. A baseline simulation with the current standard version of GEOS-Chem (Classic and/or GCHP configurations as appropriate) will be sampled along flight tracks, cruise tracks and at station locations, and evaluated species-by-species (O₃, NO, NO₂, PAN, HNO₃, alkyl nitrates, DMS, SO₂, MSA, sulfate, BrO, IO, plus supporting CO/hydrocarbons). This delivers the first systematic multi-species benchmark of GEOS-Chem for the Southern Ocean and a quantified, prioritised list of model failings. 

WP2 — Oxidants, halogens and reactive nitrogen (Months 14–30). Building on WP1, the student will run targeted sensitivity studies of the processes thought to control the ozone bias: sea-salt aerosol debromination and its dependence on aerosol acidity; iodine emission and recycling; blowing-snow and sea-ice bromine sources at the Antarctic coast; heterogeneous halogen recycling on aerosol and cloud; dry deposition of ozone to the ocean and to sea ice; and the stratospheric contribution to the high-latitude ozone budget. An ozone budget analysis (production, loss, transport, deposition) for the Southern Ocean troposphere will quantify which terms drive the underestimate and whether revised process descriptions close the gap with observations for the right mechanistic reasons. Because the ozone budget also depends on the NOx supply, the reactive nitrogen budget (H3) is addressed within this package: simulated NO, NO₂, PAN, HNO₃ and alkyl nitrates will be evaluated against the aircraft and station observations, with sensitivity studies of PAN transport and thermal decomposition, oceanic alkyl nitrate emissions, Southern Hemisphere lightning NOx, and the halogen nitrate (BrNO₃, INO₃) hydrolysis that couples the nitrogen and halogen cycles. As NO and NO₂ observations here are sparse and often close to detection limits, PAN and alkyl nitrates will provide the main observational constraint. 

WP3 — The sulfur cycle (Months 24–36). The student will evaluate and, where needed, extend the expanded marine sulfur mechanism — DMS oxidation via both abstraction and addition channels, HPMTF formation and its cloud uptake, MSA production and loss, and marine methanethiol emissions — and test it against the DMS, SO₂, MSA and sulfate observations from the ship campaigns and station records (including the long MSA/nssSO₄ records at Cape Grim, Amsterdam Island and Antarctic sites). The interaction with WP2 is a key novelty: halogen (BrO, Cl) oxidation of DMS and multiphase HOBr + S(IV) chemistry couple the sulfur and halogen budgets, and these couplings will be quantified. 

WP4 — Synthesis (Months 36–42). The project culminates in an integrated ‘best estimate’ simulation combining the WP2–3 developments, re-evaluated against the full WP1 database, with revised regional budgets of Ox, NOx, SOx and reactive halogens, and paired present-day and pre-industrial simulations to assess the implications for the pre-industrial baseline, the global ozone budget and oxidising capacity. Thesis writing runs alongside this package. 

Each work package is designed to lead to a publication, and together they form a coherent thesis. The halogen–ozone work (WP2) forms the spine of the thesis; the depth of the sulfur work (WP3), and of the reactive nitrogen component of WP2, can be adjusted to the student’s interests and to what WP1 reveals to be most important. 

Risks and mitigation. The main risks are that harmonising the observational database takes longer than planned, and that model development in WP3 proves slower than expected. The first is mitigated by the core species and platform subset defined in WP1; the second by WP2 being built largely on sensitivity studies with existing model options, so that a complete thesis on halogens and ozone is achievable even if the sulfur work is reduced in scope.

Table 1. Pre-existing observational datasets underpinning the project. 

Platform	Datasets	Relevance
Aircraft	ATom-1 to -4 (2016–2018); ORCAS (2016); SOCRATES (2018)	Vertically resolved O₃, NOx, HOx and its precursors, sulfur species, halogenated species and aerosol through the remote Southern Ocean troposphere; the primary constraint on free-tropospheric composition.
Ships	ACE / ACE-SPACE circumnavigation (2016–17);  CAPRICORN I/II and MARCUS voyages; RV  Investigator and other SO cruises	Boundary-layer O₃, DMS, aerosol composition, CCN and halogen precursors across the full latitudinal gradient of the Southern Ocean (e.g. Humphries et al., 2021).
Surface stations	Cape Grim (GAW baseline); Halley; Neumayer; Amsterdam Island; Arrival Heights; Syowa	Multi-decadal records of O₃, DMS oxidation products (MSA, nssSO₄) and halogen species (BrO, IO), plus campaign measurements of NOx and other reactive species; constrain seasonal cycles and long-term behaviour the campaigns cannot.

 

Methods, tools and data 

This is a computational project. The student will run GEOS-Chem (Classic and, where high resolution or its cubed-sphere grid (which avoids the polar convergence of latitude–longitude grids around Antarctica) is needed, GCHP) on the University of York’s Viking HPC cluster and on national facilities (e.g. ARCHER2/JASMIN), and will build an analysis toolkit in Python (xarray, pandas, cartopy and related libraries) for model–observation comparison. All observations are pre-existing and publicly available (NASA campaign archives, NCAR/EOL, AAD/CSIRO voyage repositories, WMO GAW / EBAS, BAS and partner station archives); no fieldwork is required, though opportunities to engage with observational collaborators both within York and more widely will be actively encouraged. 

Timeline 

Period	Activity
Months 1–6	Literature; GEOS-Chem and HPC training; assemble and harmonise the observational database with a common sampling/averaging protocol.
Months 6–14	WP1: baseline GEOS-Chem evaluation across all platforms; paper on the multi-species model– observation comparison.
Months 14–30	WP2: halogen, oxidant and reactive nitrogen sensitivity studies; O₃ budget analysis; paper on the Southern Ocean ozone budget
Months 24–36	WP3: sulfur cycle evaluation and development, including halogen–sulfur coupling; paper on the Southern Ocean sulfur budget
Months 30–36	Completion of WP3; international conference presentation; start of synthesis simulations
Months 36–42	WP4: synthesis and pre-industrial simulations; thesis writing

 

Supervision

The project will be supervised by Prof. Mat Evans (global chemistry-transport modelling, halogen chemistry and GEOS-Chem development) and Prof. Pete Edwards (atmospheric measurement and instrument development for OH, NOx, SO₂ and ozone), both based in WACL. This pairs the model development with first-hand expertise in the measurements against which the model is evaluated, including their uncertainties and detection limits.

Training and student profile 

The student will join the Wolfson Atmospheric Chemistry Laboratories (WACL) at the University of York, one of the UK’s largest atmospheric chemistry groups, and will benefit from its links with the National Centre for Atmospheric Science (NCAS). They will receive training in atmospheric chemistry, global modelling, high-performance computing, scientific Python, statistics and data analysis, alongside transferable skills training (scientific writing, presentation, conference attendance — with the expectation of presenting at national and international meetings) through the NCAS training programme (e.g. the Introduction to Atmospheric Science and scientific computing courses). The project would suit a student with a strong background in chemistry, physics, mathematics, environmental science or a related quantitative discipline. Prior programming experience is helpful but not essential, as full training in GEOS-Chem and Python will be provided; enthusiasm for computational science is. 

Project summary

The atmosphere over the Southern Ocean is the most pristine remote marine environment on Earth and the closest modern analogue to the pre-industrial atmosphere. Its composition is set by natural ocean–atmosphere exchange (dimethyl sulfide, sea salt and halogens), long-range transport, stratospheric input and emissions from the sea-ice zone. Because anthropogenic influence is small, processes that are masked elsewhere dominate the budgets here, making the region a demanding test of chemistry models and a key constraint on estimates of how much humans have perturbed oxidants and aerosol.

Global models perform poorly in this region. Multi-model assessments (TOAR, CMIP6) show Southern Hemisphere ozone is commonly overestimated, probably because many models omit halogen chemistry. GEOS-Chem, one of the most widely used chemistry-transport models, has the opposite problem: it underestimates ozone at high southern latitudes, and the bias has grown as more comprehensive halogen chemistry has been added. At the same time, the discovery of new DMS oxidation intermediates such as HPMTF, and their removal by clouds, has changed our understanding of marine sulfur chemistry, with consequences for natural sulfate aerosol.

This project asks whether a single, self-consistent model configuration can simultaneously reproduce the observed ozone, reactive nitrogen, sulfur and halogen environment of the Southern Ocean and Antarctic coast, and if not, what is missing. It tests three hypotheses: (H1) the ozone underestimate arises mainly because multiphase halogen chemistry destroys ozone too efficiently over the open ocean, rather than from errors in stratospheric input or deposition; (H2) reproducing the observed seasonal cycles of MSA and non-sea-salt sulfate requires both the expanded DMS oxidation mechanism and halogen oxidation of DMS; and (H3) the NOx that sustains ozone production is supplied mainly by long-range transport and decomposition of PAN, and errors in this supply are secondary to halogen chemistry in explaining the ozone bias.

The project is computational and uses only existing, publicly available observations. WP1 (months 1–14) assembles aircraft (ATom, ORCAS, SOCRATES), ship (ACE, CAPRICORN, MARCUS) and station records (Cape Grim, Amsterdam Island and Antarctic sites) south of 40°S into a harmonised database, and delivers the first systematic multi-species benchmark of GEOS-Chem for the region. WP2 (months 14–30), the core of the thesis, uses targeted sensitivity studies and an ozone budget analysis to identify which halogen, deposition, stratospheric and reactive nitrogen processes drive the ozone bias. WP3 (months 24–36) evaluates and extends the marine sulfur mechanism, including its coupling to halogen chemistry. WP4 (months 36–42) combines these developments into a best-estimate simulation, with paired present-day and pre-industrial runs to assess the implications for the global ozone budget, oxidising capacity and the pre-industrial baseline. Each work package is designed to lead to a publication, and the depth of the sulfur and nitrogen components can be adjusted to the student’s interests and to what WP1 reveals.

The student will be supervised by Prof. Mat Evans and Prof. Pete Edwards in the Wolfson Atmospheric Chemistry Laboratories at the University of York, combining expertise in global modelling and atmospheric measurement. They will receive training in atmospheric chemistry, GEOS-Chem, high-performance computing and scientific Python, and will present their work at national and international meetings. The project would suit a student with a background in chemistry, physics, mathematics, environmental science or a related quantitative discipline.