Quantifying the extent and intensity of marine heatwaves in a changing climate

Quantifying the extent and intensity of marine heatwaves in a changing climate

Research Question and Project Fit

This 2027 UNRISK project asks: how can extreme ocean temperatures be modelled across latitude, longitude and depth, rather than inferred only from the sea surface? The project description identifies Dr Ben Youngman as primary supervisor, Professor Theo Economou as academic supervisor, Exeter Mathematics and Statistics as host, and the Met Office as project partner. Regular Met Office visits are advertised, while the CASE Partner field is empty. That distinction prevents assuming CASE funding or a particular partner placement. Project description.

The explicit applicant requirements are a strong mathematics or statistics background, interest in environmental applications, and at least some programming experience. The scientific attraction is the combination of extreme-value methods, smooth modelling and physical interpretation. Its main uncertainties concern the exact scenario/data resources, how dependence and event duration will be handled, and whether the resulting maps can support decisions beyond temperature exposure.

The 2027 application guidance schedules the application system to open on 16 November 2026, with a deadline of 13:00 GMT, 13 January 2027. Its warning explains that early Leeds-system submissions lack the relevant codes and may not be retrievable. Prepare the application narrative and check the updated instructions before submitting; personal submission availability should be confirmed through the application route. Application timetable.

Research Problem and Doctoral Work

The advertised setting is the northwest European shelf seas. Surface temperature is comparatively observable, but organisms experience conditions throughout the water column and at the seabed. A surface anomaly can therefore give an incomplete description of exposure. The project proposes extreme value theory (EVT) and generalised additive models (GAMs) to represent smooth three-dimensional variation in extreme temperatures, using high-resolution marine climate projections. Atmospheric and oceanic drivers, machine learning and physical knowledge will help interpret changes under Shared Socioeconomic Pathways (SSPs). Outputs are intended to include risk maps and scenario estimates, with possible transfer to other shallow seas.

Several distinctions will determine doctoral quality. Three geographic coordinates do not automatically imply a three-variable joint-tail model. A flexible model for the marginal temperature distribution at every location is different from a model for temperatures becoming extreme simultaneously across locations or depths. Likewise, estimating an extreme temperature quantile is different from estimating an event’s duration or spatial footprint. The advertisement names EVT/GAM methods but does not settle the exact event definition, threshold baseline, declustering or dependence model.

A defensible study should identify which observations validate the projected surface and near-bed temperatures, which model biases affect the tail, and how uncertainty from the climate ensemble differs from statistical estimation uncertainty. Correlated grid cells cannot simply be treated as independent new information. Holding out periods, ensemble members or regions could help assess transferability, but these are suggested research approaches rather than assigned milestones.

Research decision	Useful evidence	Main limitation to address
Define exposure below the surface	Depth-resolved observations and clear heatwave/threshold definitions	Sparse validation at depth and changing baselines
Fit spatially varying extremes	Tail diagnostics, smoothness sensitivity and realistic uncertainty	Marginal GAMs do not themselves establish joint spatial risk
Explain drivers	Physically informed covariates and sensitivity to competing explanations	Predictive association is not necessarily a causal mechanism
Compare future scenarios	Exact forcing scenarios, ensembles and model provenance	One high-emissions ensemble is not all advertised SSPs
Produce decision-relevant maps	Defined habitat/fisheries use case and uncertainty communication	Physical exposure does not establish ecological damage

The potential contribution is a transparent framework connecting depth-resolved hazard estimates with uncertainty. A suitable thesis could develop and evaluate the statistical method, apply it to a bounded shelf-sea case, then test limited transfer. The exact dataset, ecological response information, computing allocation and transferable region remain unknown. The doctorate does not promise validated ecosystem forecasts or successful adaptation interventions.

School, Discipline, Degree and Training

The degree-awarding institution would be the University of Exeter, United Kingdom, with Mathematics and Statistics as the advertised host. UNRISK is a Leeds-managed, multi-university training centre; Leeds administration and the Met Office partnership do not make either the degree-awarding institution. Exeter’s Mathematics research-degree page provides the relevant programme context, but the precise UNRISK registration code and campus should be confirmed. Mathematics research degrees.

The department connects statistical forecasting, uncertainty quantification and climate dynamics, and reports a strong Met Office academic partnership. It cites mathematics at twentieth in the 2026 Times/Sunday Times guide and 2027 Complete University Guide, and a 2024 Leiden citation-based Mathematics/Computer Science indicator. These are dated, distinct metrics, not a ranking for this PhD or proof of personal supervision. Exeter’s shared current China page reports QS 136th overall for 2027, also distinct from subject quality. Institutional context.

UNRISK combines immediate doctoral research with early cohort training, residential workshops, peer learning and partner Challenge Weeks. Attending subject modules does not automatically make cohort activities part of the formal degree award. At least two academic supervisors and a three-month placement within the funded period are described; the advertised team supplies two names, but the placement host and calendar are not confirmed. Training structure.

Exeter Mathematics and Statistics guidance requires an upgrade report by month nine, a viva within six weeks afterwards and completion of the upgrade process by month twelve. This is dated host guidance, not a verified 2027 project-specific calendar. Confirm how cohort training, visits and placement fit these milestones. Host progression guidance.

Supervisors, Resources and Three Recent Originals

Youngman’s official profile identifies him as Senior Lecturer, working on environmental extremes, spatial statistics, natural-hazard modelling and recalibration using measurements and climate output. His maintained page explicitly connects software development with these methods and records the September 2026 evgam update. That supports a relevant methodological environment, not a requirement to arrive knowing every package or a promise of maintained project-specific ocean code. Official Youngman profile, maintained research/software page.

Economou’s page is headed “Professor Theo Economou” and lists the position “Associate Professor”. It describes Bayesian hierarchical modelling, environmental data blending, interpretable modelling and climate impacts. The page’s generic statement that no funding is currently available does not cancel this separately advertised competitive UNRISK route. Individual funding and capacity still need agreement. No marine-field supervisor beyond the advertised academic team is named. Official Economou profile.

What the selected originals contribute

Marginal extremes: Youngman, Journal of Statistical Software, 2022 — evgam

This original software/method paper puts GAM forms on extreme-value distribution parameters, with objective smoothing and return-level uncertainty estimation. Its examples include spatial rainfall and seasonal temperature; the Fort Collins GPD illustration estimates a 100-year return level of 39.1°C. Threshold choice and clustering matter, and one uncertainty example expressly does not propagate extremal-index uncertainty. Youngman is sole author and his maintained page identifies him as the package creator. This is direct supervisor evidence for EVT/GAM implementation, not a depth-resolved marine heatwave model or complete joint spatial dependence. Original paper.

Dependence: Murphy-Barltrop, Wadsworth, de Carvalho and Youngman, Extremes, 2026

The framework uses polar/radial geometry and two GAM steps for a high quantile and a tail gauge function. Simulations capture many smooth trends but miss abrupt dependence transitions and pointed shapes; the implemented framework is bivariate. The stock-market example lacks joint uncertainty quantification, and stationary diagnostics also look satisfactory, so trend significance is unresolved. Contributions identify Murphy-Barltrop as method/simulation/case-study lead; Youngman contributed computational implementation/development and manuscript feedback. It is a recent supervisor-coauthored method, not proof that a three-dimensional ocean field has already been solved. The publisher confirms 24 July 2026 publication despite the personal list’s “to appear” wording. Original article.

Projection context: Tinker and colleagues, Ocean Science, 2024

This project-linked external Met Office-led original downscales twelve HadGEM3-GC3.05 members with a 7-km NEMO shelf model under RCP8.5, providing projections and a present-day control dataset. Its 3.11°C ensemble-mean surface warming for 2079–2098 relative to 2000–2019 is a mean-climate result, not a heatwave return level or frequency. One parent-model ensemble and one forcing scenario do not span all structural uncertainty or the advert’s SSP programme. Tinker undertook processing, analysis and data release; other named authors developed the driving ensemble and regional model. Neither Youngman nor Economou is an author. Original article.

Together these studies provide two supervisor-linked methodological papers and one relevant partner/data paper. They do not establish three supervisor-authored marine studies. The project materials also cite DOI 10.5194/sp-4-osr8-7-2024; no quantitative claim here relies on it. The project must identify which newer scenario resources are actually available.

The project information lists B.Youngman@exeter.ac.uk for the principal supervisor and T.Economou@exeter.ac.uk for the academic co-supervisor; ENV-PGR@leeds.ac.uk handles UNRISK application queries. The Leeds administrative address does not change the Exeter host.

The practical resource questions concern data access, storage and computation, rather than an inferred wet laboratory. Ask what resolutions and ensemble sizes are manageable, which software is maintained, who provides oceanographic advice, and whether partner approvals restrict publication. Regular visits can strengthen domain learning, but their frequency, expenses and supervision arrangements require agreement.

Funding, Living Cost and International Conditions

If awarded, the UNRISK scholarship covers full tuition including the international fee difference, plus a UKRI-standard stipend for three years nine months (45 months). The expected three-month placement is included, not an additional funded extension. The £6,000 individual Research Training and Support Grant and £5,000 per-student cohort-training allocation are research/training budgets, not personal cash or annual stipend additions. Flexible-fund access is not an automatic individual payment. Funding and entry requirements.

The scheme confirms international participation but limits international awards; it does not provide Chinese admission odds. Winning an award and obtaining ordinary Exeter admission are separate. The FAQ excludes relocation-related accommodation/travel, visa and immigration health surcharge. The named Met Office partnership does not establish CASE status or extra CASE money. Require a personal award letter covering fees, maintenance, funded dates, placement expenses and any writing-up beyond funding. Current application resources.

The 2026–27 UKRI table gives £21,805 stipend for 2026–27, not a published 2027–28 amount. It also distinguishes international cohort limits by council. No council-specific international-award percentage is established. No flat multi-year total or individual net-income promise follows. UKRI rates and grant guidance.

Exeter’s April 2026 basic single-student estimate is £1,300–2,000 monthly, or £15,600–24,000 over twelve months. Using the same-period £21,805 stipend for illustration leaves £6,205 at the low end or a £2,195 deficit at the upper end, before exceptional expenses. This is a dated scenario, not a 2027 affordability guarantee. Private rent of £135–250 weekly excludes bills; inspect tenancy length, deposits and actual housing offers. University budget.

The cited outside-London visa maintenance threshold is £1,171 monthly for up to nine months (£10,539) plus applicable unpaid fees. It is different from annual living costs. Chinese applicants’ differential documentary treatment does not remove the underlying funds requirement, and supporting evidence can be requested. Immigration charges and the health surcharge should be refreshed before the future visa application. Visa money rules.

Applicant Fit, China-Specific Assessment and Preparation

The project’s mathematics/statistics strength and some programming are explicit prerequisites. Strong environmental motivation cannot replace quantitative preparation. EVT theory, GAMs, model checking, reproducible gridded-data analysis and uncertainty communication are useful preparation priorities, not additional mandatory software requirements in the advert.

UNRISK accepts relevant 2:1-equivalent qualifications and does not universally require a master’s. Exeter Mathematics itself permits an undergraduate or postgraduate degree with substantial mathematics, expecting a First or strong Upper Second equivalent. However, Exeter’s generic Chinese research-degree guidance normally expects a completed three-year Chinese master’s with both certificates. Its 75% undergraduate equivalence belongs to taught postgraduate admission, not an automatic bachelor-only PhD permission. The CDT and host must assess the individual qualification route; do not automatically reject or accept a Chinese bachelor-only applicant. China guidance, host eligibility responsibility.

The CDT’s Exeter link uses English Profile A, IELTS 6.0 overall and 6.0 each component. This aligns with Mathematics at Streatham; other campuses/programmes have different profiles. Verify the exact registration/offer rather than assuming another campus or programme profile applies. English need not be completed at initial application but must satisfy the host before registration. UNRISK excludes applicants already awarded a PhD or currently registered for one; a previous unfinished doctorate may be considered under its FAQ. Profile A.

China is not ATAS-exempt. Applicability depends on the precise course/research code and CAS, not simply the words “statistics” or “marine”. Confirm any requirement with Exeter; no nationality-wide prohibition or assured clearance is established. ATAS rules.

A convincing preparation example would explain a tail model’s assumptions, diagnostic failures and uncertainty, then connect the estimates to a defined decision. Clearly distinguish simulated projection data from measurements and hazard exposure from biological impact.

Career and Transferable Skills

Environmental statistics, ocean/climate analytics and risk modelling are plausible directions for this training. The transferable contribution is learning to explain tail-model assumptions, distinguish projection from measurement, and communicate uncertainty in a specific decision. These are inferences from the documented work, not salary forecasts, employment promises or measured group placement outcomes.

Application Stages, Risks and Next Actions

The 2027 timetable gives interview invitations 18–26 February 2027, an open day 1 March, interviews 3–5 March, and offers 10–17 March. October 2027 entry is programme context; the personal start day remains to confirm. All hosts initially use Leeds OLA plus a UNRISK assessment form, selecting NERC UNRISK CDT. Successful Exeter candidates subsequently complete a host application.

The detailed body remains labelled the 2026 process awaiting an autumn update. It presently permits up to two projects, expects contact with at least one supervisor per project, uses a structured statement instead of a traditional CV and does not require initial references. Transcripts and qualification information are requested; references may arise at the later Exeter stage. Check the updated 2027 instructions after opening, rather than imposing generic self-funded Mathematics proposal/deadline rules on the CDT.

Decision area	Assessment	Necessary verification
Institution and training	Relevant Exeter statistics/climate setting, cohort and placement	Exact registration/campus and 2027 progression calendar
Team and resources	Named statistical supervisors plus Met Office access	Domain mentoring, visit expenses, computing/data rights
Scientific direction	Strong depth-resolved extremes problem	Heatwave definition, dependence, scenario coverage, validation
Funding and living	Full international fees if awarded; 45 months	2027 stipend, individual award, housing and writing-up
Applicant feasibility	International quantitative route exists	Personal Chinese degree equivalence, English and ATAS
Timing and risk	Prepare for 16 November opening	Updated forms; January deadline and competition

Before applying, discuss the event definition, depth validation, available projection scenarios and who supports oceanographic interpretation with B.Youngman@exeter.ac.uk and T.Economou@exeter.ac.uk. Ask how uncertainty from marginal modelling, dependence and climate ensembles will be assessed together, and what useful outcome remains if a fully transferable framework proves too broad.

Before accepting, confirm personal award terms, travel/placement support, data licences and publication permissions, supervision frequency and compute allocation. This dossier supports preparing a focused application for the announced opening. It does not establish present submit readiness, a guaranteed award or successful ecological decision support.

Evidence Scope and Sources

The cited research has different scopes: the 2026 dependence method is bivariate, while the 2024 climate projection uses one parent-model ensemble and RCP8.5. Neither establishes the full advertised scenario programme. The application schedule gives dates but does not establish individual submission availability. Funding and cost figures apply only to their stated years and purposes.

Project, supervisors and original research: native project, Youngman, Economou, Youngman maintained page, evgam original, 2026 geometric original, 2024 marine original.

Programme and application: Mathematics doctorate, upgrade policy, UNRISK training, application timeline, entry/funding, resources/FAQ, China qualifications, English A.

Finance and eligibility: budget, UKRI stipend, visa finances, ATAS, UKRI host responsibilities.