Past Seminars

April 8th (9am London)

Baoxiang Pan

CAS Institute of Atmospheric Physics

Generative modelling for climate simulation

Climate science aims to better understand various processes and statistical behavior of the climate system under natural, historical, and possible future forcing scenarios. We discuss how generative machine learning can inform better process representation and systematical understanding, using case examples of parameterization, state inference, and probabilistic forecast.

March 4 (9am London)

Lei Bai

Shanghai AI Laboratory

Learning High-resolution Global Data Assimilation and Weather Forecasting Systems 

Weather forecasts, spanning days to weeks, hold significant importance for planning and decision-making across various sectors.  Given the intricate nature of climate dynamics, the integration of artificial intelligence (AI) has become imperative in mid-term weather prediction. However, existing attempts are limited in two aspects: 1) relatively lower resolution than the most advanced operational forecasting system (e.g., 0.25-degree vs 0.09 degree), and 2) lack of the data assimilation module. This talk will present the recent works at the OpenEarthLab, aiming at learning a unified high-resolution global data assimilation and weather forecasting systems. 

February 6 (9am LA)

Pierre Gentine

Columbia University

Machine learning across scales to improve climate models: from emulation to discoveries

December 5 (9am London)

Martin Rypdal

UiT The Arctic University of Norway

From Climate Sensitivity to Early-Warning Signals

In the last decades, several statistical methods have been developed to estimate parameters in simple linear energy-balance models describing the mean surface temperature’s response to radiative forcing. Part of the motivation for the development of these methods has been to estimate or constrain climate sensitivity. In this presentation I will discuss how these ideas can be adapted to estimate the sensitivity of climate subsystems to changing conditions. If a system’s sensitivity can be tracked over time, we may use it as a possible early-warning indicator for tipping points. 

November 7 (9am London)

Bedartha Goswami

University of Tübingen

Understanding climate variability with statistical machine learning

Recent years have seen a boom in the application of statistical machine learning and deep learning methods to tackle problems in meteorology and climate science. In this talk, I will present two projects ongoing in my group that use machine learning to understand climate variability. I will show how we can use similarity-based networks of climate time series data to reveal new features of intraseasonal variability of summertime extreme rainfall propagation over South Asia that might potentially be useful for early warning systems. I will next present how we can use principal component analysis in combination with Gaussian Mixture Models to categorize extreme phases of the El Niño Southern Oscillation (ENSO) and show how the Eastern Pacific El Niño is better modeled as two separate categories: a weaker 'canonical' form and an extreme form, both of which show distinct idiosyncratic onset and development.

October 3 (9am LA)

Paul O’Gorman 

MIT

A new theory for extratropical storms in the limit of purely moist dynamics

Latent heating due to phase changes of water affects extratropical storms in the current climate and will do so to a greater extent in warmer climates. Previous work showed that as the climate warms the most unstable mode of moist baroclinic instability transitions from a periodic wave to an isolated vortex dominated by latent heating. This vortex mode is an example of a diabatic Rossby vortex (DRV) which is a moist storm that forms in the extratropics in the current climate, most famously in the case of winter storm Lothar. In this talk, I will describe a new theory for the DRV that makes analytical predictions in the limit of a convectively neutral stratification. I will also discuss the implications of the theory for the wave-vortex transition that occurs when the atmosphere is sufficiently moist, as well as new "DRV world" simulations that give insights into the finite amplitude behavior.

September 5 (9am LA)

Marina Hirota

Federal University of Santa Catarina

Disturbances, impacts and resilience of Amazonian ecosystems

Terrestrial ecosystems have been undergoing a remarkable and unprecedented combination of climatic and human-driven disturbances since the beginning of the 20th century. As a result, energy balance and biogeochemical and hydrological cycles are affected, which feed back into the climate system, dampening or amplifying vegetation cover changes. In the Amazon basin, disturbances such as anomalous droughts, wildfires and land use change are concurrently occurring, already leading to changes in ecosystem structure and functioning. Grounded on the theory of dynamical systems, I will talk about how disturbances are heterogeneously triggering feedback mechanisms at different spatial and temporal scales and decreasing resilience, potentially causing further changes in vegetation cover within the Amazonian ecosystems. 

June 6 (9am LA)

Rupert Klein

Freie Universität Berlin

How Mathematics helps structuring climate discussions

Mathematics in climate research is often thought to be mainly a provider of techniques for solving the continuum mechanical equations for the flows of the atmosphere and oceans, for the motion and evolution of Earth’s ice masses, and the like. Three examples will elucidate that there is a much wider range of opportunities.

Climate modellers often employ reduced forms of “the continuum mechanical equations” to efficiently address their research questions of interest. The first example discusses how mathematical analysis provides systematic guidelines for deriving and rigorously characterizing such reduced model equations.

The second example describes recent developments of data analysis techniques that are relevant for "small data", rather than "big data", problems in the context of seasonal weather (condition) prediction and climate research. As an example we will discuss the data-based one- to two-year prediction of El Niño periods.

Modern climate research has joined forces with economy and the social sciences to generate a scientific basis for informed political decisions in the face of global climate change. One major type of problems hampering progress of the related interdisciplinary research arise from often subtle language barriers. Accordingly, the third example describes how a mathematical formalization of the notion of “vulnerability w.r.t. climate change” has helped structuring the discourse in a related interdisciplinary research project.

May 2nd (9am LA)

Valerio Lucarini

University of Reading

Understanding Optimal Fingerprinting for Detection and Attribution via Response Theory    

Detection and attribution studies have played a major role in shaping contemporary climate science and have provided key motivations supporting global climate policy negotiations. The goal of such studies is to associate observed climatic patterns of climate change with acting forcings - both anthropogenic and natural ones - with the goal of making statements on the acting drivers of climate change. The statistical inference is usually performed using regression methods referred to as optimal fingerprinting. We show here how a fairly general formulation of linear response theory relevant for nonequilibrium systems provides the physical and mathematical foundations behind the optimal fingerprinting approach for the climate change detection and attribution problem. Our angle allows one to clearly frame assumptions, strengths and potential pitfalls of the method.

References:

V. Lucarini and M. Chekroun, arxiv:2212.02628 [physics.ao-ph] (2022)

V. Lucarini and M. Chekroun, arXiv:2303.12009 [physics.ao-ph] (2023)

April 4 (9am London)

Paul Ritchie 

University of Exeter

Overshoots and Rate-Induced Tipping in Conceptual Climate Models

Previous studies report low global warming thresholds above pre-industrial conditions for key tipping elements such as ice-sheet melt. If so, high contemporary rates of warming imply that exceeding these thresholds is almost inevitable, which is widely assumed to mean that we are now committed to suffering these so called bifurcation-induced tipping events. Here we show that this assumption may be flawed, especially for slow-onset tipping elements (such as the collapse of the Atlantic Meridional Overturning Circulation (AMOC)) in our rapidly changing climate. We demonstrate, using conceptual climate models, that a threshold may be temporarily exceeded without prompting a change of system state, if the overshoot time is short compared to the effective timescale of the tipping element. On the other hand, systems may exhibit rate-induced tipping points instead of (or as well as) bifurcation-induced tipping, where a system fails to adapt to rapidly changing external forcing. Such tipping points are much less widely known, and yet are arguably even more relevant to contemporary issues such as climate change. We illustrate this phenomenon using a model for the AMOC and the possibility of avoiding tipping by reversing the forcing. This has the potential to lead to multiple critical rates for the same peak change as the low rates required to avoid rate-induced tipping compete against the fast rates required for safe overshoots.

March 7th (9am Beijing)

Courtney Quinn

University of Tasmania

Dynamical Systems and Climate: Combining Theory with Data-Driven Methods

Working in climate mathematics brings many challenges, not least being the high dimensionality of the system state and the range of timescales involved.  Model behaviour often is dominated by fast growing errors on the shortest timescales.  When working with general circulation models (GCMs) one wants to be able to target the timescales of interest while still capturing the necessary complexity of the system.  Our work attempts to use techniques from dynamical systems theory to help inform data-driven processes, as well as use data-driven processes to help undertake studies of the system’s dynamics.

The first part of the talk will discuss the problem of state-estimation in coupled, multiscale systems.  We explore various methods of strongly coupled ensemble data assimilation and utilise the unstable subspace of the system to help determine the necessary ensemble size.  We also show that the selection of observation variables can impact the calculation of the unstable subspace and subsequently the state estimation. While the model we use is a paradigm example of atmosphere-ocean interconnectivity, we draw parallels to methods for more complex GCMs.

The second part of the talk focuses on regime detection and analysis of climate modes. Using a machine learning method which prescribes an underlying model for the data (namely FEM-BV-VAR), we reduce the problem to analysing the dynamics on the resulting regime-switching model.  We calculate the underlying stability of each model state as well as the growth rates and alignment of covariant Lyapunov vectors (previously identified as possible indicators of transitions).  On short timescales we denote the covariant Lyapunov vectors to be mixed singular vectors (MSVs) based on their calculation and lack of convergence.  Our application on the North Atlantic Oscillation shows agreement with other stability studies of the atmospheric mode, and our application to the 2010 Russian heatwaves connects MSVs to the initial perturbations empirically produced by the European Centre for Medium-Range Weather Forecasts.

February 7th (9am Los Angeles)

Davide Faranda

Université Paris Saclay

Analogues Methods for attribution of extreme events to climate change, application to the exceptional 2022 European-Mediterranean drought

A prolonged drought affected Western Europe and the Mediterranean region in the first nine months of 2022 producing large socio-ecological impacts. The role of anthropogenic climate change (ACC) in exacerbating this drought has been often invoked in the public debate, but the link between atmospheric circulation and ACC has not received much attention so far. Here we address this question by applying the method of circulation analogs, which allows us to identify atmospheric patterns in the period 1836-2021 very similar to those occurred in 2022. By comparing the circulation analogs when global warming was absent (1836-1915) with those occurred recently (1942-2021), and by excluding interannual and interdecadal variability as possible drivers, we identify the contribution of ACC. The 2022 drought was associated with an anticyclonic anomaly over Western Europe persistent over December 2021-August 2022. Circulation analogs of this atmospheric pattern in 1941-2021 feature 500 hPa geopotential height anomalies larger in both extent and magnitude, and higher temperatures at the surface, relative to those in 1836-1915. Both factors exacerbated the drought, by increasing the area affected and enhancing soil drying through evapotranspiration. While the occurrence of the atmospheric circulation associated with the 2022 drought has not become more frequent in recent decades, there is an increase of its interdecadal variability for which the influence of the Atlantic Multidecadal oscillation cannot be ruled-out.

December 6th (9am London)

Robbin Bastiaansen

Utrecht University

Tipping in spatially extended systems

In the current Anthropocene, there is a need to better understand the catastrophic effects that climate and land-use change may have on ecosystems, earth system components and the whole Earth system. The concept of tipping points and critical transitions contributes to this understanding. Tipping occurs in a system when it is forced outside the basin of attraction of the original equilibrium, resulting in a critical transition to an alternative, often less-desirable, stable state.

The general belief and intuition, based on simple conceptual models of tipping elements (i.e. ordinary differential equations), is that tipping leads to reorganization of the full (sub)system. In this talk, I will review and explore tipping in conceptual, but spatially extended, and potentially spatially heterogeneous, models (i.e. partial differential equations). In these spatially explicit models, additional stable states can emerge that are not uniform in space, such as Turing patterns and coexistence states (part of the domain in one state, the rest in another state with a spatial interface or front between these regions), which can lead to a different tipping behaviour. In particular, in these systems a tipping point might lead only to a slight restructuring of the system or to a tipping event in which only part of the spatial domain undergoes reorganization, limiting the impact of these events on the system’s functioning.

October 4th (9am London)

Niklas Boers

Potsdam Institute for Climate Impact Research

Critical Transitions in the Earth System

In response to anthropogenic release of greenhouse gases, the Earth is warming at unprecedented rates. It has been suggested that several components of the Earth’s climate system may respond with abrupt transitions between alternative stable states in response to gradual changes in forcing. Based on the theory of stochastically forced dynamical systems and their bifurcations, a methodology is presented to measure changes in the stability of a given equilibrium state from observational time series. The method is applied to observation-based data to investigate if and how the stability of the Greenland Ice Sheets, the Atlantic Meridional Overturning Circulation, the Amazon rainforest, and the South American monsoon, has changed in the course of the last century.

September 6th (9am Los Angeles)

Sam Stechmann

University of Wisconsin-Madison

Toward the Addition of Clouds to Fundamentals of Geophysical Fluid Dynamics

Clouds and rainfall are among the most challenging aspects of predicting weather and climate. In addition to difficulties in observing and simulating clouds, there is a large gap in our theoretical understanding of moist fluid dynamics versus dry fluid dynamics. Here, I will present some recent work toward closing this gap by defining basic principles of geophysical fluid dynamics (GFD) that hold even in the presence of phase changes and clouds. In particular, I will present cloudy versions of the conservation principles of energy and potential vorticity, and the quasi-geostrophic approximation. This work is part of a broader community effort that is helping to lay the theoretical foundations of moist atmospheric dynamics.

June 7th (9am Beijing) 

Nan Chen

University of Wisconsin-Madison

A Physics-Informed Data-Driven Algorithm for Ensemble Forecast of Complex Turbulent Systems

A new ensemble forecast algorithm, named as the physics-informed data-driven algorithm with conditional Gaussian statistics (PIDD-CG), is developed to predict the probability density functions (PDFs) of complex turbulent systems with partial observations. The PIDD-CG algorithm integrates a unique multiscale statistical closure modeling strategy with an extremely efficient nonlinear data assimilation scheme to create a mixture of conditional statistics. These conditional statistics serve as the forecast ensemble members to mitigate the curse of dimensionality in recovering high-dimensional PDF. The multiscale features in the time evolution of these conditional statistics ensembles are effectively predicted by an appropriate combination of physics-informed analytic formulae and recurrent neural networks. An information metric is adopted as the loss function for the latter to capture the desirable turbulent features more accurately. The proposed algorithm succeeds in efficiently forecasting both the transient and statistical equilibrium non-Gaussian PDFs of strongly turbulent systems with intermittency, regime switching and extreme events.

May 3rd (9am Los Angeles) 

Annalisa Bracco

Georgia Tech

Analyzing changes in the tropical Indo-Pacific in the last 6,000 years.

A 6000-year long simulation with a coupled climate model is analyzed to find how slow changes in the mean climate induced by the variations in the Earth’s orbit from mid- to late Holocene feedback on the main modes of climate variability in the tropical Indo-Pacific. We find that mean state changes induced a gradual shift in the dominant mode of variability of the Indian Ocean and a slow, steady strengthening of ENSO. Using a complex network methodology and principal component analysis, we investigate the climate modes evolution and their connectivity changes over 6000 years . To uncover the causal link between a constantly evolving mean state and such changes in variability we consider two independent analyses. First, we explore changes in the mean state by quantifying the departure of 500-year annual mean snapshots from the annual mean conditions averaged over the mid-Holocene. Second, we leverage ideas from dynamical system theory and characterize the nature of the Indo-Pacifc transition, in a state space representation, by accounting for its spatiotemporal and multivariable dependency. The analysis reveals that a strengthening of the Walker circulation, driven by enhanced convection over the central Pacific and by the weakening of the Asian monsoon, set the stage for a shift in modes in both basins.

April 5th (9am Sydney)

Geoff Stanley

UNSW, Syndey 

Topology and Optimization for Neutral Surfaces in the Ocean

Flow in the dark ocean interior is like a fine Canadian maple syrup poured over a stack of Dutch pancakes: the syrup runs quickly off the sides of the pancakes but diffuses slowly down through the stack, just as seawater mixes rapidly along a stack of stably stratified "neutral surfaces" while diffusing ~100 million times more slowly across neutral surfaces.  Unfortunately, these neutral surfaces are mathematically ill-defined, so oceanographers must use "approximately neutral surfaces", which are well-defined but only approximately aligned with the neutral plane of preferential mixing.  Potential density surfaces are the classic examples of approximately neutral surfaces, but their misalignment from the neutral plane is rather severe and causes strong but spurious vertical mixing across the neutral plane.

In this talk, I will show two types of surfaces that are vast improvements over potential density surfaces.  The first surface centres around an old topological tool called the Reeb graph that enables the study of multivalued functional relationships, in this case between pressure and in-situ density on the surface.  The second surface uses optimization and the calculus of variations to minimize a measure of error from perfect neutrality.  I will also derive the exact geostrophic streamfunction on a neutral surface (which was known to exist for 30 years) and apply this theoretical result to calculate new geostrophic streamfunctions that are ~10x more accurate on approximately neutral surfaces than previously existing ones. 

Open source code is available to facilitate the use of accurate neutral surfaces such as these. 

February 1st (9am Los Angeles)

Ted Shepherd

Department of Meteorology, University of Reading

Bringing physical reasoning into statistical practice in climate-change science

The treatment of uncertainty in climate-change science is dominated by the far-reaching influence of the ‘frequentist’ tradition in statistics, which interprets uncertainty in terms of sampling statistics and emphasizes p-values and statistical significance. This is the normative standard in the journals where most climate-change science is published. Yet a sampling distribution is not always meaningful (there is only one planet Earth). Moreover, scientific statements about climate change are hypotheses, and the frequentist tradition has no way of expressing the uncertainty of a hypothesis. As a result, in climate-change science, there is generally a disconnect between physical reasoning and statistical practice. In this talk, I explain how the frequentist statistical methods used in climate-change science can be embedded within the more general framework of probability theory, which is based on very simple logical principles. In this way, the physical reasoning represented in scientific hypotheses, which underpins climate-change science, can be brought into statistical practice in a transparent and logically rigorous way. The principles are illustrated through three examples of controversial scientific topics: the alleged global warming hiatus, Arctic-midlatitude linkages, and extreme event attribution. These examples show how the principles can be applied, in order to develop better scientific practice.

January 11th (9am Los Angeles)

Woosok Moon

NORDITA / Department of Mathematics, Stockholm University

Baroclinic instability and large-scale wave propagation on planetary-scale atmosphere

Midlatitude atmospheric variability is dominated by the dynamics of the baroclinically unstable jet stream that meanders and sheds eddies at the scale of the Rossby deformation radius. The eddies interact with each other and with the jet, affecting the variability at a wide range of scales, but the mechanisms of planetary-scale fluctuations of the jet are not well understood. Here, we develop a theoretical framework to explore the stability of planetary-scale motions in an idealized two-layer model of the atmosphere. The model is based on a combination of the vertical shear and the Sverdrup relation, providing the dynamic link between the two layers, with meridional eddy heat fluxes parameterized as a diffusive process with the memory of past baroclinicity of the jet. We find that a planetary-scale instability exists if the vertical shear of the jet does not exceed a particular threshold. The inclusion of the eddy memory effect enables westward or eastward propagation of planetary waves relative to the barotropic mean flow. Importantly, we find growing planetary waves that propagate slowly westward or are stationary, which could have important implications for the formation of atmospheric blocking events. Our theoretical results suggest that with the ongoing polar amplification due to global warming and the corresponding reduction of the vertical shear of the mean wind, the background conditions for the growth of planetary-scale waves are becoming more favorable. 

Links to recent relevant papers:

https://rmets.onlinelibrary.wiley.com/doi/abs/10.1002/qj.4232

https://rmets.onlinelibrary.wiley.com/doi/pdf/10.1002/qj.4030

https://www.tandfonline.com/doi/pdf/10.1080/16000870.2019.1697164

December 7th (9am Los Angeles)

Mary Silber

University of Chicago

Vegetation Pattern Formation in Drylands : It’s about the Water

A beautiful example of spontaneous pattern formation occurs in certain dryland environments around the globe. Stripes of vegetation alternate with stripes of bare soil, with striking regularity and on a scale readily monitored via satellites. Though the vegetation is a showstopping spectacle, water, which is the limiting resource for these ecosystems, is the unseen player behind the scenes. Water concentrates into the vegetated zones, essentially reinforcing vegetation patterning, via positive feedbacks, and its dynamics play out on the short timescales of the rare storms. In contrast, the vegetation may change very little over decades. Mathematicians, physicists and theoretical ecologists have had a blast creating and analyzing models of the slow vegetation dynamics. They have suggested, based on model simulations, monitoring changes to pattern characteristics as early warning signs of ecosystem collapse, e.g., under climate change. In this talk I will tell you a little bit about the empirical data,  the models, the analysis and simulations. I will also advocate for a shift in focus to the water resource that is entering the system, and the need to observe and model it better on its timescale. Our work suggests some alternative perspectives on these fascinating patterns, and new questions about how they might respond to changes in precipitation characteristics, such as seasonality, storm strength and storm frequency, that will likely occur as a consequence of climate change.

November 2nd (9am Los Angeles)

Sarah-Jane Lock

European Centre for Medium-Range Weather Forecasts

A representation of model uncertainty in the IFS due to the transport scheme

ECMWF ensemble forecasts include stochastic perturbations designed to represent model uncertainties. Their inclusion significantly increases the skill of the ensemble forecasts. Currently, the stochastic perturbations used operationally in the IFS represent uncertainties that are attributed to the parametrisations of atmospheric physics processes. Recent efforts have turned to exploring how to introduce stochastic perturbations to represent model uncertainties that arise within the dynamical core. 

Following work to improve the convergence of the IFS semi-Lagrangian transport scheme, a new stochastic perturbation scheme has been developed. The “STOCHDP” scheme introduces stochastic perturbations to the calculation of the departure point (DP) in the semi-Lagrangian method, motivated by Diamantakis & Magnusson (2016, MWR), who analysed the impact that the complexity of the flow field has on the rate of convergence of the iterative DP calculation. 

October 5th (9am Los Angeles)

Morgan O'Neill

Stanford University

Convective forcing of stratospheric water vapor and implications for climate

The presence of water vapor in the lower stratosphere is enormously consequential for the climate. Thermodynamically, it helps set the `cold reservoir’ temperature effectively experienced by the tropospheric heat engine, and an increase in water vapor in the lower stratosphere cools the stratosphere and warms the Earth’s surface. An increase in stratospheric water vapor speeds ozone destruction. The stratospheric water budget remains poorly constrained, both observationally and theoretically. Strong thunderstorms are known to be an important secondary source of water vapor to the lower stratosphere, and how they may feedback to large scales in a warming climate is almost completely unknown.

I will contextualize the role of water vapor in the upper atmosphere, and then discuss our recent work on the physics of storm tops responsible for water vapor injection. The most severe midlatitude supercell thunderstorms typically feature an Above-Anvil Cirrus Plume (AACP), which is a wake of ice and water vapor downstream of overshooting deep convection, several kilometers above the main anvil shield. The AACP is uniquely capable of lofting water high above the tropopause. Using high-resolution large eddy simulations, we show that the AACP is formed by the development of a new type of hydraulic jump at the tropopause. Immediately upon the jump onset, the simulated water vapor injection rate into the stratospheric overworld increases from less than 1 tonnes per second to greater than 7 tonnes per second, accompanied by horizontal windspeeds that top 110 m/s at the tropopause. The presence of a small-scale, dynamical threshold past which only some storms become very effective local hydrators of the lower stratosphere suggests a blind spot in large-scale climate models unable to resolve this behavior. The feedback on the environmental conditions for future storms due to a humidified lower stratosphere is an exciting area of future research.

July 6th (9am Los Angeles)

Laure Zanna

NYU

Climate Modeling in the Age of Machine Learning 

Numerical simulations used for weather and climate predictions solve approximations of the governing laws of fluid motions on a grid. Ultimately, uncertainties in climate predictions originate from the poor or lacking representation of processes, such as ocean turbulence and clouds that are not resolved on the grid of global climate models. The representation of these unresolved processes has been a bottleneck in improving climate simulations and projections. 

The explosion of climate data and the power of machine learning algorithms are suddenly offering new opportunities: can we deepen our understanding of these unresolved processes and simultaneously improve their representation in climate models to reduce climate projections uncertainty? 

In this talk, I will discuss the current state of climate modeling and its future, focusing on the advantages and challenges of using machine learning for climate projections. I will present some of our recent work in which we leverage tools from machine learning and deep learning to learn representations of unresolved ocean processes and improve climate simulations. Our work suggests that machine learning could open the door to discovering new physics from data and enhance climate predictions.

June 1st (9am London)

Kristian Strømmen

University of Oxford

Atmospheric Circulation Regimes: From Topological Foundations to Climate Model Projections

The idea of atmospheric circulation regimes has been considered in one way or another since the 1950's, motivated by the desire to truncate the complex and chaotic dynamics of the atmosphere into a finite number of simple flow states and their interactions. However, their application to climate science is still held back by some key issues. Firstly, there is no clear definition of what constitutes a regime, and the various techniques to diagnose them in the literature require ad-hoc choices which produce a wide variety of regimes that are not easily comparable. Secondly, the extent to which regimes can actually add information to concrete questions about weather forecasts or climate projections beyond what can be obtained with cruder techniques (such as linear regression) has been unclear. Thirdly, attempts to use regimes to analyse climate model projections has been seriously confounded by the often considerable discrepancies between the regime structure of climate models compared to the real atmosphere. 

In this talk I'll outline some recent ideas aiming to address these challenges. I'll show that 1) persistent homology, a tool from topological data analysis, offers one potential way to define regimes in terms of topological structure; 2) that following the logic of basic Markov chain maths to its natural conclusion can offer concrete insight into questions concerning mid-latitude predictability; and 3) that by filtering out a linear mode of atmospheric variability associated to the speed of the North Atlantic jet, one can obtain extremely stable and consistent regimes for the entire CMIP6 ensemble. I'll show how this allows, for the first time, a robust assessment of what components of the climate system determine the North Atlantic regime behaviour.

May 4th (9am Beijing)

Raffaele Ferrari

MIT 

A scaling theory for meridional heat transport in planetary atmospheres

The meridional temperature profile of the upper layers of planetary atmospheres is set through a balance between differential radiative heating by a nearby star, or by intrinsic heat fluxes emanating from the deep interior, and the redistribution of that heat across latitudes by turbulent flows. These flows spontaneously arise through baroclinic instability of the meridional temperature gradients maintained by the forcing. This turbulence takes the form of coherent vortices that mix the meridional temperature profiles and zonal jets that suppress that mixing. Using the framework of the two-layer quasigeostrophic beta-plane model, we derive a quantitative parameterization that predicts the meridional temperature profile in terms of the externally imposed heat flux.

April 6th (9am Beijing)

Tapio Schneider and Andrew Stuart

Caltech

Data-Informed Climate Models With Quantified Uncertainties

While climate change is certain, precisely how climate will change is less clear. But breakthroughs in the accuracy of climate projections, and in the quantification of their uncertainties, are now within reach, thanks to advances in the computational and data sciences and in the availability of Earth observations from space and from the ground. The talk will survey the design of a new Earth system model (ESM), under development by the Climate Modeling Alliance (CliMA). Key new concepts in the ESM will be covered, including turbulence, convection, and cloud parameterizations. Furthermore a novel approach to fast and efficient algorithms for parameter estimation from data, including the quantification of uncertainties, will be described. 

March 2nd (9am London)

Antoine Venaille

ENS de Lyon

Wave topology in the climate system

Over the last decades, the concept of topological protection has shone a new light on the physics of waves. It explains the robust emergence of unidirectional trapped waves at the boundary separating  materials characterized by different topological invariants. Born in condensed matter to explain the robustness of electronic transport at the edge of a sample, those ideas have now spread over all fields of physics, including geophysical fluid dynamics. We will explain how to compute such topological invariants and how they are related to discrete symmetries in the hierarchy of climate models, with applications to equatorial and internal wave dynamics.

February 2nd (9am London)

Stephan Rasp

Climate AI

The optimization dichotomy: Why is it so hard to improve climate models with machine learning? 

Despite decades of improvements in climate modeling, large uncertainties remain about how much our planet will warm in response to rising greenhouse gas concentrations. Convincing evidence points towards sub-grid parameterizations, approximations of unresolved physical processes, as the main culprit. Clouds, in particular, remain challenging because of their chaotic behavior at the intersection of turbulence, microphysics and radiation. 

Two recent developments have gotten climate modelers excited that new way of solving this conundrum might be on the horizon: first, increases in computing power have made it possible to explicitly resolve some of the most challenging processes in global simulations spanning several months. Second, machine learning has made significant advances in the last decade with astounding successes in the fields of computer vision, natural language processing and reinforcement learning.

The idea is simple: instead of heuristically designing parameterizations, is it possible to train a machine learning parameterization based on short-term, high-resolution data? First studies show that this is indeed plausible. However, as I will point out in this talk, this does not automatically lead to better climate simulations. I argue that we are currently optimizing our machine learning models for a target that is far away from the outcome we actually want to achieve. Solving, or circumventing, this dichotomy will be key to make machine learning work for climate modeling.

December 1st (9am London)

Nikki Vercauteren

FU Berlin

Towards stochastic modelling of turbulence in the stably stratified atmospheric boundary layer

Atmospheric boundary layers with thermally stable stratification are the least understood type of boundary layers due to suppressed turbulence and the presence of myriads of processes on multiple spatiotemporal scales that modulate the turbulence. Stable boundary layers (SBLs) are however the norm in Polar and winter alpine environments, and more generally at nighttime. In such environments, turbulence is typically unsteady and intermittent and can be triggered by motions on submeso scales such as density currents, wave-like motions or two-dimensional modes that represent a non-stationary forcing of turbulence. These motions are poorly understood and mainly not represented in models. Classical approaches to turbulence parameterisation fail to reproduce turbulent dissipation in SBL context and this is a known source of errors in larger scale atmospheric models, including climate models.

In this presentation I will approach the question of intermittency of turbulence and its partial modulation by wave-like non-turbulent motions based on multiscale data analysis and statistical clustering methods. Analyzing turbulence data based on statistically classified flow regimes helps unravel organizing principles in complex dynamics of near-surface SBL turbulent flows. Following data-driven studies, we suggest replacing a typically used stability correction function with a stochastic differential equation to include the impact of the variability in the submesoscale frequency band of near-surface flows. The nonstationary parameters of the proposed stochastic equation and their scaling are examplarily estimated from data of the FLOSS2 field experiment. Such new types of stochastic parameterisations may ultimately be used in weather and climate models to complement existing turbulence parameterisations.

November 3rd (9am Los Angeles)

Darryl Holm

Imperial College London 

Applying Kelvin’s circulation theorem in climate science.

The prediction of climate change and its impact on extreme weather events is one of the great societal and intellectual challenges of our time. It has three parts:

1. Make the distinction between weather and climate. 

2. Understand the dynamics of the fluctuations of the physical variables. 

3. Predict how the variances of the fluctuations are affected by statistical correlations in their fluctuating dynamics. 

This 1World CliMaths seminar investigates a stochastic geometric mechanics framework called LA SALT which can at least formally meet all three parts of the challenge for the problem of climate change, given a deterministic fluid theory derived from the variational principles of geometric mechanics.

References:

Lorenz, E.N.: Climate is what you expect. (unpublished) (1995).  https://eapsweb.mit.edu/sites/default/files/Climate_expect.pdf 

Holm, D.D.: Variational principles for stochastic fluid dynamics. Proc. R. Soc. A 471(2176), 20140963 (2015) http://dx.doi.org/10.1098/rspa.2014.0963

Drivas, T.D., Holm, D.D., Leahy, J.-M.: Lagrangian-averaged stochastic advection by Lie transport for fluids. J. Stat. Phys. (2020) https://doi.org/10.1007/s10955-020-02493-4

Alonso-Or ́an, D., Bethencourt de Le ́on, A., Holm, D.D., Takao, S.: Modelling the climate and weather of a 2D Lagrangian-averaged Euler–Boussinesq equation with transport noise. J. Stat. Phys. (2020) https://doi.org/10.1007/s10955-019-02443-9