What is up with Dansgaard - Oeschger events? Answering the question using conceptual models. 

Speaker : Bryony Hobden 

Date : April 30th 2024 

Abstract : 

The last glacial period (110-10 Kya) was a time of extreme instability, marked by millennial-scale climate oscillations known as Dansgaard-Oeschger (DO) events. A DO event is characterised by a rapid increase in temperature (5-16°C in high northern latitudes) over decadal timescales, followed by cooling over extended periods of centuries to millennia. These DO events, unlike the anthropogenic climate change we are experiencing today, are truly unique. While the high northern latitudes rapidly warmed, there was simultaneous cooling over large portions of the Southern Hemisphere. Understanding how Earth’s climate was able to undergo these natural ‘tipping points’ is not just a scientific curiosity but a crucial step towards comprehending the stability of our current climate under the pressures of global warming. The precise cause of these abrupt shifts is still a subject of debate.

We approach these mysterious millennial-scale oscillations from a conceptual modelling and dynamical systems theory perspective. Building upon the low-dimensional model presented by Boers et al. (2018) we present a stochastic nonlinear model that can induce self-sustaining oscillations through feedbacks between sea ice and the Atlantic Meridional Overturning Circulation. Here, transitions between stadial (cold) and interstadial (warm) states result from bifurcations in an underlying fast subsystem connected to sea ice extent. Changes in the subsurface water temperature in the North Atlantic modulate the duration spent in stadial conditions, with canard trajectories offering explanations for interesting behaviour for sustained stadial periods and smaller transitions that do not trigger full DO events. The type of bifurcations we see in the system have implications for early warning signals and, therefore, predictability of these events. 

From Quantum Mechanics to Greenhouses

Speaker : Daniel Williams

Date : May 14th 2024 

Abstract : 

We all know that carbon dioxide and water vapour are greenhouse gases, but do you really understand why they are? What is it that makes nitrogen and oxygen—so abundant in our atmosphere—useless as greenhouse gases, whilst trace gases such as CFCs can have a global warming potential many times that of even CO2? In this talk I will try to provide, from first principles, an explanation of the interplay of factors that contribute to the absorption spectra and hence greenhouse gas potential of different atmospheric species. With the knowledge developed from this I will introduce the concept of the Runaway Greenhouse Effect, a fate that has already befallen one planet in our solar system (Venus) and is likely to be the ultimate destiny of the climate on Earth in the distant future.

Seasonal forecasting of the European North-West shelf seas: sources of, limits to and prospects for winter and summer ocean temperature predictability.

Speaker : Jamie Atkins 

Date : May 21st 2024 

Abstract : 

The European North-West shelf seas (NWS) support economic interests and provide environmental services to adjacent countries. Expansion of offshore activities, such as renewable energy infrastructure, aquaculture, and growth of international shipping, will place increasingly complex demands on the marine environment over the coming decades. Skilful forecasting of NWS properties on seasonal timescales will help to effectively manage these activities. Here we quantify the skill of an operational large-ensemble ocean-atmosphere coupled global forecasting system (GloSea), as well as benchmark persistence forecasts, for predictions of NWS sea surface temperature (SST) at 2-4 months lead time in winter and summer. We identify sources of and limits to SST predictability, considering what additional skill may be available in the future. We find that GloSea NWS SST skill is generally high in winter and low in summer. GloSea outperforms simple persistence forecasts by adding information about atmospheric variability, but only to a modest extent as persistence of anomalies in the initial conditions contributes substantially to predictability. Where persistence is low – for example in seasonally stratified regions – GloSea forecasts show lower skill. GloSea skill can be degraded by model deficiencies in the relatively coarse global ocean component, which lacks dynamic tides and subsequently fails to robustly represent local circulation and mixing. However, “near perfect atmosphere” tests show potential for improving predictability of currently low performing regions if atmospheric circulation forecasts can be improved. This underlines the importance of coupled atmosphere-ocean model development for NWS seasonal forecasting applications.

High Impact Weather in the Mid-Latitudes: A Neural Network Approach to Identifying Dry Intrusion Outflows

Speaker : Owain Harris 

Date : May 28th 2024 

Abstract : 

Dry intrusions are coherent airstreams that originate in the upper troposphere, or lower stratosphere, and descend towards the surface in the mid-latitude regions. Here, they have been linked with various types of high-impact weather including extratropical storms, atmospheric rivers, wildfires, and dust storms. Additionally, they can transport ozone-rich air into the troposphere, where it is harmful to human and environmental health. Our ability to study the future of these impacts is limited by the computationally expensive Lagrangian methods which are currently used to identify dry intrusion trajectories. Its demand for high resolution spatial and temporal data makes it incompatible with available climate projection products. By introducing the concepts of image segmentation and convolutional neural networks, this talk will discuss how machine learning may offer a new opportunity to uncover the response of dry intrusions to anthropogenic climate change. We will see if the same techniques used in facial recognition can be adapted to identify dry intrusions in the atmosphere. Meanwhile, asking what the limitations of this may be, and how we can interpret such a model.

Spatio-Temporal Chain Event Graphs - Translating Expert Judgement into Statistical Models

Speaker : Hollie Calley 

Date : June 4th 2024 

Abstract : 

In domains such as criminal justice, public health, and legislative development, statisticians play a vital role in converting verbal problem descriptions into statistical models. A domain expert can provide a statistician with a description of a process of events, such as which events might happen, why those might happen, and the possible outcomes generated by certain sequences of events. Translating these complex descriptions into a coherent model becomes a substantial task. Bayesian models encode these expert judgements in the form of prior distributions on specific variables within the model. To a non-statistician this process can often be hard to interpret, leading to a lack of understanding of how the model is affected by the inputs, how outputs are produced, or indeed if those outputs are reliable. Yet, in the age of AI and black box models such as Bayesian Neural Networks, the appeal for interpretable models for use in these situations is increasing. One such interpretable model is the Chain Event Graph (CEG), which produces a visual representation of the underlying statistical model along with statistical outputs, enhancing accessibility and comprehension. Currently, there is a trade off in terms of model complexity versus interpretability. Popular methods for Bayesian inference can handle much more complex interactions between variables than CEGs, but without statistical knowledge, it is hard to interpret their outputs. To address this, we must either enhance the interpretability of complex models for both statisticians and non-statisticians or extend simpler models to cover a broader range of situations. While temporal modelling using CEGs has seen progress, integrating spatial terms into CEG models remains unexplored. This talk will discuss the potential advancements in incorporating spatial elements into Chain Event Graphs and comparing the results to more popular models, aiming to bridge the gap between complexity and interpretability.

Exponential sum with additive coefficient.

Speaker : Madhuparna Das 

Date : June 11th 2024 

Abstract : 

Exponential sums are interesting objects to study, especially with the large moduli. In 1977 Montgomery and Vaughan gave a non-trivial estimate for the exponential sums with multiplicative coefficients. We implement their method to prove this result for the class of additive functions with large moduli.

Effects of stratified MHD on shear-driven instabilities in the tachocline with application to the

solar dynamo

Speaker : Velizar Kirkow 

Date : June 25th 2024 

Abstract : 

The solar tachocline is a thin, stably stratified layer in the solar interior located between the radiative and convective zones. Between the solid body rotation of the radiative zone and differentially rotating convective zone, the solar tachocline is a region of strong radial fluid shear which is believed to drive the turbulence in this layer. The dynamics herein prevent vertical momentum transport which would otherwise cause mixing between the radiative and convective zones over time. More precisely the turbulence is horizontally directed parallel to the east-west large-scale magnetic field the tachocline is permeated by. The mean fluid flow and magnetic field are in alignment with one another due to the Omega-effect which takes place in the tachocline in the Omega-alpha dynamo model. Hence, the tachocline is thought to be the seat of the solar dynamo which is used to explain the fast dynamics of the sun on an 11-year cycle. While it is understood that a shear-driven instability is most likely responsible for the dynamics of the tachocline, it remains unclear how small-scale events may contribute to the large-scale dynamics of the solar dynamo. We therefore build on current MHD (magnetohydrodynamic) literature by including the effects of stable stratification on an idealised two-dimensional interface model of the tachocline, so far not considered in relevant literature. Using this model, we study the effects of vertical stable stratification and horizontal magnetic field on the stability of the Kolmogorov shear flow. The instability species which destabilise the flow fall into two broad categories: oscillatory and non-oscillatory. The latter dominates the former should they coexist and are strongly damped by stratified MHD effects. However, oscillatory instabilities respond non-typically to stratified MHD effects and we find that they are in fact destabilised for certain specific stratified MHD couplings – this is of note as these instabilities persist into large stratified MHD couplings characteristic of the solar tachocline. As a result, our results show that stable stratification permits shear-driven instabilities to persist into higher magnetic field strengths than otherwise. We explore shear-driven instabilities under stratified MHD couplings at three different levels of approximation: linear stability, weakly nonlinear theory and full nonlinear regime. We present also some preliminary results using a reduced model for the oscillatory instabilities. 

A Precursor to Solar Prominence Eruptions: Detection and Analysis of EUV Prominence Oscillations

Speaker : William Beckwith-Chandler 

Date : July 2nd 2024 

Abstract : 

Solar prominences are large plasma structures, extending high into the solar atmosphere. They are cool and dense in comparison to the surrounding environment and supported against gravity by strong magnetic fields. The eruption of these structures can have a significant influence on the solar-terrestrial environment. However, accurately predicting the eruptions remains a challenge. We apply automated detection methods for Extreme Ultraviolet (EUV) prominences observed by the Solar Terrestrial Relations Observatory (STEREO) and Solar Dynamics Observatory (SDO). We study an event, during March 2011, when each STEREO spacecraft is in quadrature with respect to the Earth  and SDO. For two time ranges, we obtain longitudinal height profiles as a function of time. We also track the corresponding EUV filaments across the solar disk, which reveal the emergence of 8 - 16 hour oscillations in the EUV filament channels. Our analysis shows a correlation between the prominence's increasing height and the oscillation periods, suggesting a potential link to the subsequent eruption observed by the STEREO spacecraft off-limb. These findings offer new insights into prominence dynamics and may pave the way for improved eruption prediction, aiding future space weather forecasting. Furthermore, we explore how we can obtain prominence statistics over more than a full 11-year solar cycle, using artificial intelligence. Our results show that the prominences follow a pattern over the solar cycle, rather like the Wolf number for sunspots.