Cathie and her team with Paul, Nancy, Ian, Dante, have won the STEM Research Project of the Year trophy at the Times Higher Education Awards 2022.

Speakers for the Symposium include: Alberto Naveira Garabato (National Oceanography Centre, Southampton), Marilena Oltmanns (National Oceanography Centre, Southampton), Wei Pan (ICL), Pauline Tedesco (ICL), Matthew Crowe (University College London) and Laura Cimoli (Scripps Institution of Oceanography, San Diego).

Please note: talks will take place at Room Huxley 340 at Imperial College and a welcome lunch at 170 Queen’s Gate. There will be also be a social BBQ dinner for MPE staff, students and invited guests after at 170 Queen’s Gate.

Please register for the event and lunch through this link here by 9th May.

Please download the flyer here.

Article on Paulina’s achievement can be found here.

• Event attribution

• Trend attribution

• Philosophical issues in attribution

• Causality

• Decision-making

The school is primarily aimed at MPE CDT graduate students but is open to a wider audience of scientists and scholars in mathematics and weather and climate science, including climate services. Scheduled activities are expected to occur between 10am and 5pm UK time, and will occur entirely online. The provisional schedule of lectures and practical sessions can be downloaded here. A more detailed syllabus, together with suggested background reading, will be provided later.

The lectures and practical sessions will be delivered by leading experts, including Aglaé Jézéquel (LMD/IPSL), Sebastian Sippel (ETH Zürich), Elisabeth Lloyd (Indiana), Marlene Kretschmer (Reading), Elena Saggioro (Reading), Dragana Bojovic (BSC), Raül Marcos (BSC), and Marta Terrado (BSC).

The school is co-sponsored by the World Climate Research Programme (WCRP) through its Lighthouse Activity ‘My Climate Risk’, and by the Walker Institute.

In order to benefit from the school, participants should have some basic background in weather and climate science, some knowledge of common statistical practices such as regression and uncertainty quantification, and some experience in scientific programming with either python or R.

We ask potential participants to indicate their interest by May 2nd, by completing this form. Successful applicants will be informed by Mid May and provided with joining information.

SYLLABUS FOR THE SCHOOL

MONDAY JUNE 6

Event attribution Aglaé Jézéquel 

Concepts of extreme event attribution :

• How do we define the event?
• Counterfactual and factual worlds
• Storyline and risk-based approach
• Attributing impacts related to extreme events
• Compound events attribution

Trend attribution I Sebastian Sippel 

Concepts of detection and attribution for long-term trends:

• Detection: Is there a change?
• Attribution: What is the reason for the change?
• Internal variability vs. forced response
• Separation of (forced) signal and noise
• Climate models as physics-based tools to identify an expected change signal (which observational changes will be tested against)  
• “Fingerprinting” of individual forcings (in spatial, temporal, vertical and/or multi-variate climate patterns) and traditional attribution “recipe”
• Key attribution results (e.g., from IPCC AR6 WG1)

Logic of model confirmation Lisa Lloyd

• Varieties of ways that models might be supported
• Fit
• Variety of Fit
• Independent support for aspects of the model
• Variety of independent support
• Robustness
  

Role of values in attribution Lisa Lloyd 

• Why values are always there
• Why that is not bad
• How to use values to help good science
• Tradeoffs of values in scientific investigation and practice

TUESDAY JUNE 7

Practical session on event attribution Aglaé Jézéquel 

Simple case study of EEA – calculation of a risk ratio

Causality I Marlene Kretschmer 

• Causal questions in climate science with a focus on teleconnections
• Association vs. causation
• The concept of causality in statistics: quantifying causal relationships from (observed) data
• Using causal networks to express scientific hypotheses
• Examples from climate science: common drivers, mediating pathways, direct and indirect effects, linear and non-linear causal effects

Trend attribution II Sebastian Sippel 

“New approaches” to detection and attribution (e.g., at regional scales):

• Dynamical adjustment: Separation of thermodynamical vs. dynamical trends as a tool to assess causes of regional-scale trends
• Statistical learning approaches to identify (forced) signals from climate patterns (e.g., pattern filtering, infilling, and/or pattern recognition, …)

WEDNESDAY JUNE 8

Practical session on causality Elena Saggioro 

Set of exercises on simple causal network structures and associated potential statistical pitfalls: common driver, mediators, indirect pathways, blocking pathways and non-linear causal effects.

THURSDAY JUNE 9

Causality II Marlene Kretschmer 

• Conditioning on a common effect: an example from climate science
• Particular challenges and opportunities in the use of causal networks in scientific practice
• Outlook: causal discovery
  

Decision-making Aglaé Jézéquel 

• How do we interact with stakeholders as climate scientists ?
• Interdisciplinarity with social sciences 
• Climate services : definitions, objectives and limits
• Case study of extreme event attribution as a climate service

Practical session on decision-making Raül Marcos, Marta Terrado and Dragana Bojovic

• A journey from climate information to decision-making: a tale of two worlds?

FRIDAY JUNE 10

Decision-making I: Knowledge coproduction  Dragana Bojovic

• Decision-making for climate change adaptation
• Involving stakeholders in decision-making
• Participatory methods 
• Climate services and need for transdisciplinary approaches
  

Decision-making II: Success story Marta Terrado 

• Coproduction of a climate service with a wine sector user 
• MED-GOLD Dashboard: a decision-making tool for the wine sector (different time scales: historical, seasonal, projections) 
• Translating skill to economic value, hit rate of the prediction: the weather roulette 
  

Lecturers and Support Specialists – Biographies

Dragana Bojovic is a senior researcher at Barcelona Supercomputing Center – Earth Science Department, where she co-leads the Knowledge integration team. She applies interdisciplinary and transdisciplinary approaches to improve climate knowledge and coproduce climate services. She has been collaborating with scientists, policymakers, and communities from different parts of the world, supporting knowledge exchange to enhance resilience to climate and other environmental changes. She holds a PhD in Science and Management of Climate Change (Ca’Foscari University of Venice) and a MSc in Environmental Change and Management (Oxford University). 

Aglaé Jézéquel is a researcher working at LMD in Paris. Her research interests include extreme event attribution, storylines, and climate services. Her work is interdisciplinary, between statistics of climate data and social sciences. She also teaches several classes at masters level, including a creative writing workshop on climate change in collaboration with a French writer.

Marlene Kretschmer is a postdoctoral researcher at the University of Reading. Before that she worked at the Potsdam Institute for Climate Impact Research in Germany where she received her PhD in climate physics. Her research focuses on identifying the large-scale atmospheric drivers of extreme weather and climate events, including dynamical stratosphere-troposphere coupling and its impacts for winter circulation and extremes. To address these issues, she is particularly interested in applying causal inference-based frameworks as well as novel statistical approaches from machine learning such as causal discovery algorithms.  Moreover, she is keen on applying these new techniques to evaluate teleconnection processes in climate models and to improve sub-seasonal to seasonal (S2S) forecasting of extreme events.

Elisabeth A. Lloyd, a philosopher of science working on models in science and their confirmation in evolutionary biology, started studying climate science and how it uses models in 2005. Since then, she has worked on a variety of key issues involving modeling questions, including model confirmation and model application in the context of attribution, and is especially interested in the development of attribution methods and the social values they hold. She has held faculty positions at University of California, Berkeley, and is now a Distinguished Professor at the specialty department in her field at Indiana University. A few weeks ago she found out she was elected to the American Academy of Arts and Sciences.

Raül Marcos-Matamoros is a Serra Húnter lecturer in Meteorology at the Barcelona University and a fellow researcher at the Barcelona Supercomputing Center (BSC) – Earth Science Department. In the BSC he collaborates with the Climate Services team where he is involved in the verification of seasonal prediction models and the development of climate services in different sectors (agriculture, forest fire prevention, water resources and renewable energies). He also works in the interaction with stakeholders to promote the integration of climate predictions in their decision-making workflows. He holds a PhD in Physics (University of Barcelona) and a MsC in Meteorology. 

Elena Saggioro is an interdisciplinary research fellow at the Walker Institute, and is completing her PhD in Mathematics of Planet Earth at the University of Reading. Her PhD research focused on applying a range of causal and Bayesian network methods to understanding the Southern Hemisphere stratosphere-troposphere coupling and its role in extending the predictability of large-scale tropospheric circulation variability. At the Walker Institute, she is exploring how causality and climate storylines can help in climate risk assessment and climate adaptation planning. Her work involves close engagement with a range of stakeholders, including national governments and public bodies in the UK and overseas.

Sebastian Sippel’s key research interests include improving understanding of climate variability, extremes, and their changes at global and regional scales, including land-atmosphere interactions and changes in ecosystem carbon cycling. Sebastian uses statistical learning techniques with a focus on identifying forced changes and internally generated variability. Sebastian works as a Senior Scientist (“Oberassistent”) and Lecturer in the Climate Physics group at ETH Zurich. Sebastian’s PhD research was on “Climate extremes and their impact on ecosystem-atmosphere interactions” at the Max Planck Institute for Biogeochemistry in Jena, Germany. Before that, he studied environmental sciences with a focus on environmental physics at Bayreuth University, Germany; with a year abroad in Oxford with studies on Environmental Change and Management.

Marta Terrado is a senior researcher and science communicator at the Barcelona Supercomputing Center – Earth Sciences department, where she co-leads the Knowledge Integration team. Applying various communication and dissemination tools and activities, she facilitates knowledge and technology exchange on climate services and air quality services to stakeholders from different sectors and governance levels to support their adaptation to climate change. She is an Environmental Scientist with a PhD in Earth Sciences (University of Barcelona), a MSc in Geographical Information Systems and a Postgraduate Degree in Science Communication.

Timetable

BSC Team = Dragana Bojovic, Raül Marcos, Marta Terrado

On behalf of  MPE CDT, we are pleased to announce the Mathematics of Planet Earth’ Exhibition 2022. We warmly invite you to visit the ‘Mathematics of Planet Earth’ Exhibition, which will be displayed in the main entrance of Imperial’s South Kensington Campus, between Saturday 28th May – Sunday 5th June 2022.

Free entry – all welcome!

About the exhibition

Mathematics of Planet Earth is an international exhibition displaying exhibits, videos and computer programs illustrating how mathematics plays a role in answering essential questions that concern our planet. Through a series of graphics, visualisations and hands-on experiments you will discover the contributions that mathematics makes to topics such as astronomy, fluid dynamics, seismology, glaciology and cartography. The MPE CDT will provide a number of activities and tours available for families and children of all ages.

Download the Exhibition booklet here.

Download the flyer here.

Please see below links for the MPE Exhibition 2020 and News Article :

• MPE Exhibition 2020
• News on MPE Exhibition 2020

Opening event

• Date: Saturday, 28 May 2022
• Time: 12:30 hrs
• Location: Imperial College Main Entrance, Exhibition Road, South Kensington Campus, SW7 2AZ

Guided tours

Hour-long guided tours run each day of the exhibition (apart from Saturday 28 May when the exhibition opens at 12.30). Guided tours are offered on the hour starting at 10.00am and are limited to 15 guests at a time; the last tour commences at 17.00. To book a tour, please email the MPE CDT.

Public lectures

The MPE Exhibition programme also includes a public lecture , featuring an outstanding and fascinating talk.

Date: Monday, 30th May 2022
Time: 16:00
Venue: ICCL402, EPSRC Centres for Doctoral Training Suite on Level 4N
Speaker: Prof Peter Lynch (An Irish meteorologist, mathematician, blogger and book author, University College Dublin)
Title: The Origins of Computer Weather and Climate Prediction: Fulfilment of Richardson’s Dream

Abstract:
Over the past few decades, weather forecasts have become impressively accurate. Forecast failures are rare and evoke strong reaction in the media. Climate prediction is still fraught with uncertainty, but great progress is under way.

The basic principles of numerical prediction and climate modelling were established long before electronic computers were available. Just one hundred years ago, Lewis Fry Richardson published his great work “Weather Prediction by Numerical Process”. The algorithm that he presented amounts in essence, to a general circulation model of the atmosphere, capable of describing both weather and climate.

In this lecture, we will review Richardson’s monumental study, and the triumphs and failures of his attempt to forecast the weather. His methods are at the core of atmospheric simulation and it may be reasonably claimed that his work is the basis of modern weather and climate forecasting. His dream has come true.

Interact online

• Follow the MPE CDT on Twitter
• You can interact with this event on Twitter using the hashtag #mathsonearth

Got a question? Email the MPE CDT >

Key Speaker and Guests:

Gonzalo Gonzalez De Diego (DPhil at Mathematical Institute, Oxford University))
Oana Lang (STUOD Research Associate at Mathematics Department, ICL)
Tobias Schwedes (Senior Data Scientist at Faculty)

Event Schedule:

 15:00- 15:10: Zoom call opens and Welcome

15:10 – 15:55: Gonzalo Gonzalez De Deigo on ‘Numerical approximation of viscous contact problems in glaciology’

15:55 – 16:05: Break

16:05 – 16:50 : Oana Lang on ‘On Data Assimilation, Particle Filters, and Stochastic Shallow Water Models’

16:50- 17:00 – Break

17:00 – 17:30: Tobias Schwedes on ‘Might a career in Data Science be up my alley?’

Gonzalo Gonzalez De Diego

Title: Numerical approximation of viscous contact problems in glaciology

Abstract: Viscous contact problems describe the time evolution of fluid flows in contact with a surface from which they can detach. These type of problems arise in glaciology when, for example, modelling the evolution of the grounding line of a marine ice sheet or the formation of a subglacial cavity. Such problems are generally modelled as a time dependent viscous Stokes flow with a free boundary and contact boundary conditions. Although these applications are of great importance in glaciology, a systematic study of the numerical approximation of viscous contact problems has not been carried out yet. In this talk, I will present some of the challenges that arise when approximating these problems and some of the ideas we have come up with for overcoming them.

Oana Lang

Title: On Data Assimilation, Particle Filters, and Stochastic Shallow Water Models

Abstract: I will start with a brief introduction on data assimilation, with particular focus on particle filters and the tempering procedure. I will then present a couple of applications for Lorenz ’63 and for a stochastic rotating shallow water model. Different types of stochasticity that can drive the underlying shallow water model, together with the corresponding theoretical justifications will be discussed. The aim of the talk is to introduce the audience to a particle filter based data assimilation methodology and to analyse the analytical properties of the models which can act as “signals” in this context.  

Prof Gavin Esler (Professor of Applied Mathematics, UCL)
Dr David Ham (Reader in Computational Mathematics, Mathematics Department, ICL)
Dr Zoe Goss (Renewable Technology Engineer at Frazer-Nash)
Dr Josephine Park (Principal Data Scientist at UK Health Security Agency)

Programme Schedule :

14.30 Prof Gavin Esler on ‘Can jet formation in beta-plane turbulence ever be described by a local closure theory?’

15.15 Refreshment Break

15.30 Dr David Ham on ‘How to not write your model: code generation for geoscientific simulation from a mathematical specification of the model’

16.20 Dr Zoe Goss on ‘Consultancy: a pathway from academia to industry’

16.40 Dr Josephine Park on ‘Data Science in the Public vs Private Sectors’

17.00 Q/A

Prof Gavin Esler
Title: Can jet formation in beta-plane turbulence ever be described by a local closure theory?
Abstract:
One of the oldest and best-studied models in geophysical fluid dynamics is that of stochastically forced turbulence governed by the barotropic vorticity equation on the beta-plane. In the right parameter regime zonal jets, resembling those seen in the extratropics of the giant planets, emerge spontaneously from the turbulence. Recently, new theories (by Srinivasan and Young, Woillez and Bouchet) have emerged for how the Reynolds stresses associated with the forced eddies in the system depend on the local flow conditions. Such theories hold the promise of a closure in which the resulting jet profiles could be explained by a single equation. The aim of the talk is to introduce these theories, reconcile their (apparent) contradictions, discuss their potential usefulness and their limitations, and explain why the answer to the titular question is, generally, no, but perhaps in some special situations, yes.

Dr David Ham
Title: How to not write your model: code generation for geoscientific simulation from a mathematical specification of the model
Abstract:
Solving forward and inverse models described by PDEs is at the core of much computational geoscience. A simulation is a complex combination of equations, discretisations, solvers, preconditioners, data assimilation and much more. The traditional approach to creating models has been to derive the computational algorithm with pen and paper, and then to write low-level code implementing that algorithm. This is a laborious, error-prone, and tedious process which makes model development a glacially slow progress. Having finally developed a forward model, much of the same work must then be re-done in order to create an adjoint model to solve inverse problems.

In this talk I will explain a totally different way of creating model code. Rather than implementing an algorithm, a mathematician can write the PDE they want to solve, and have the computer both derive the algorithm, write the code, and execute it in parallel. I will discuss the mathematical and computational abstractions that make this approach possible, and sketch how a specialised compiler can turn mathematics into code. I’ll also discuss the automation of inverse modelling and will touch on some of our very newest work in incorporating external data into models.

Dr Zoe Goss
Title: Consultancy: a pathway from academia to industry
Abstract:
Since graduating from the MPE CDT, I have joined the renewables technology team at Frazer Nash Consultancy. This has given me the opportunity to work in industry on exciting novel forms of renewable energy. At the same time, many of my colleagues still have strong links with academia; publishing journal papers, presenting at conferences and earning visiting professorship status. In the year I have been working for FNC I have supervised 3 masters students projects and a group undergraduate dissertation. I’ll be talking through my advice for choosing and applying for the right jobs for you, and why consultancy might be a great fit if you are caught choosing between worlds!

Dr Josephine Park
Title: Data Science in the Public vs Private Sectors
Abstract:
I’ll be walking through my career progression from PhD in Ocean Modelling, to Data Scientist at a Tech Consultancy to Principal Data Scientist (and armchair epidemiologist!) in the Civil Service. I’ll talk about some of the differences between Private and Public Sector employment, the types of projects I’ve worked on, and some of the experiences I’ve had working for the UK Health Security Agency during a global pandemic.

A new paper has been published in the SIAM Journal by MPE Reading Student, Louis Sharrock.

The paper can be viewed online here: Joint Online Parameter Estimation and Optimal Sensor Placement for the Partially Observed Stochastic Advection-Diffusion Equation | SIAM/ASA Journal on Uncertainty Quantification | Vol. 10, No. 1 | Society for Industrial and Applied Mathematics

Abstract

In this paper, we consider the problem of jointly performing online parameter estimation and optimal sensor placement for a partially observed infinite-dimensional linear diffusion process. We present a novel solution to this problem in the form of a continuous-time, two-timescale stochastic gradient descent algorithm, which recursively seeks to maximize the asymptotic log-likelihood of the observations with respect to the unknown model parameters and to minimize the expected mean squared error of the hidden state estimate with respect to the sensor locations. We also provide extensive numerical results illustrating the performance of the proposed approach in the case that the hidden signal is governed by the two-dimensional stochastic advection-diffusion equation.© 2022, Society for Industrial and Applied Mathematics Permalink: https://doi.org/10.1137/20M1375073

Read More: https://epubs.siam.org/doi/10.1137/20M1375073

The original press release can be found here.

Below is the project Mariana was working on:

Should I trust my neural network?

Assessing uncertainty in neural network predictions of ocean circulation under global heating.

In the coming decades, global heating will undoubtedly lead to changes in ocean circulation
potentially pushing the climate beyond key tipping points. However, how ocean circulation patterns
will change is still unclear and there is also a need for novel techniques which can deal with the
uncertainty associated with climate change prediction. In my project I worked with domain experts
at LOCEAN to use Bayesian Neural Networks (BNNs) to provide a probabilistic prediction of the
ocean regime. This probabilistic prediction provides researchers with a way of assessing how much
they should trust the result. Figure 1 shows the accuracy of my BNN as well as the entropy of the
results. Higher values of entropy indicate that the BNN has greater uncertainty about the prediction
there and therefore that the result should be treated with caution. Notably, comparing the two
panels shows that the highest uncertainty occurs where the BNN predicts the incorrect regime
which is a promising result.

Results gain considerable value to oceanographers and indeed the wider community if their source
and skill is physically interpretable. Techniques to address this issue are known as “explainable AI”,
but there has been very limited research into applying them to BNNs. Therefore, in this project I
explored how conventional methods such as SHAP can be adapted to make them suitable for BNNs.
Figure 2 shows an example of the SHAP values for a specific datapoint. The oceanographers at
LOCEAN were able to confirm that the BNN’s decision making process was physically realistic.

Professors Jennifer Scott and Dan Crisan welcomed everyone and then introduced a programme of talks from invited speakers:

Beatrice Pelloni, Professor of Mathematics, Heriot Watt University, Edinburgh and founding Director in Reading of the MPECDT presented “The mathematical description of large-scale atmospheric flows”.

Dr Sarah-Jane Lock, from ECMWF gave a talk titled “Are you sure I don’t need my umbrella?”

(Representing model uncertainty in weather forecasts).

We were then joined by three alumni speakers from cohorts Alpha, Beta and Charlie who gave short talks on the science they are now working on in their Postdoctoral Researcher positions:

Dr Lea Oljaca (University of Exeter): Attractors of nonautonomous dynamical systems

Dr Carlo Cafaro (University of Reading):My journey into probabilistic weather forecasts, from short to seasonal time scales

Dr Adrian Tsz Yan Leung (University of Reading): The impact of hybrid oceanic data assimilation in a coupled model: a case study of a tropical cyclone

Dan and Jennifer then closed the meeting and we had dinner in The Blandford’s Restaurant, Park House.