2025 SUMMER LECTURES IN CLIMATE DATA SCIENCE
May 29, 2025 - July 31, 2025
- 12:00 – 1:30 pm (EDT)
- IN-PERSON at the Tang Family Hall (Rm 202) at the Columbia Engineering Innovation Hub (2276 12th Ave, New York, NY)
- VIRTUAL ATTENDANCE available
- Add to Calendar
Click to see past Lectures in Climate Data Science from Spring 2025, Fall 2024, Summer 2024, Spring 2024, Fall 2023, Spring 2023, and Fall 2022.
THURSDAY || MAY 29, 2025
ANDRE NOGUEIRA SOUZA
MIT
“Using Generative AI to Predict the Ocean Interior from the Ocean Surface”
Understanding subsurface ocean dynamics is essential for quantifying oceanic heat and mass transport, but direct observations at depth remain sparse due to logistical and technological constraints. In contrast, satellite missions provide rich surface datasets-such as sea surface height, temperature, and salinity-that offer indirect but potentially powerful constraints on the ocean interior. Here, we present a probabilistic framework based on score-based diffusion models to reconstruct three-dimensional subsurface velocity and buoyancy fields, including the energetic ocean eddy field, from surface observations. Using a 15-level primitive equation simulation of an idealized double-gyre system, we evaluate the skill of the model in inferring the mean circulation and the mesoscale variability at depth under varying levels of surface information. We find that the generative model successfully recovers key dynamical structures and provides physically meaningful uncertainty estimates, with predictive skill diminishing systematically as the surface resolution decreases or the inference depth increases. These results demonstrate the potential of generative approaches for ocean state estimation and uncertainty quantification, particularly in regimes where traditional deterministic methods are underconstrained or ill-posed.
THURSDAY || June 5, 2025
TOBIAS FINN
École des Ponts ParisTech / CEREA
“Going With the Flow: Towards Physically Consistent Data-Driven Earth System Models”
The AI revolution has transformed Earth system modelling. Deep neural networks trained on large datasets for short-term forecasts predict the temporal evolution of the atmosphere and other components of the Earth system at unprecedented speeds, while reaching the accuracy of the best geophysical models. However, due to their training protocol, these deterministic models have issues with long-term stability and physical consistency. Yet, there is hope to resolve these issues with generative models, which have recently revolutionised image and video generation.
In this talk, I take you on a journey that brings us from simple dynamics like Lorenz 1963 to the rough waters in the Arctic Ocean and around Antarctica. Starting from first principles, we will see how these generative models learn to iterate from noise to clean data. When we apply them to forecasting, we can produce large ensembles from a single set of initial conditions. Equipped with this capability, I will show that these models can learn emergent properties of the system, while outperforming deterministic forecasting models. Furthermore, we will exploit prior-known physical properties to efficiently train models for local-to-global scales, which, applied to forecasting, can stably project to long-term conditions. Additionally exhibiting a previously unseen physical consistency, these models are a leap forward in the efficiency and realism of data-driven Earth system models.
THURSDAY || June 26, 2025
LEAP ML JOURNAL CLUB
(Brenowitz, N. D., Ge, T., Subramaniam, A., Gupta, A., Hall, D. M., Mardani, M., … & Pritchard, M. S. (2025). Climate in a Bottle: Towards a Generative Foundation Model for the Kilometer-Scale Global Atmosphere. arXiv preprint arXiv:2505.06474.)
AI emulators offer a path to compressing, boosting limited ensembles, and improving the latency of interacting with petabyte-scale climate prediction data. However, prevailing auto-regressive paradigms offer limited flexibility, and are challenging to train on climate time horizons due to drifts, instabilities and component-coupling challenges. Conditionally generative models offer an appealing alternative. In this context we demonstrate a generative diffusion-based framework — Climate in a Bottle (cBottle) — for emulating global km-scale climate simulations and reanalysis on the equal-area HEALPix grid. cBottle consists of two model stages: a globally-trained coarse-resolution image generator that generates 100km (50k-pixel) fields given monthly average sea surface temperatures and solar conditioning, followed by a locally-trained 16x super-resolution stage that generates 5km (12.5M-pixel) fields; global super-resolution is made affordable using an overlapping patch-based multi-diffusion. Overall, cBottle shows promise as an emulator across a battery of climate model diagnostics, including diurnal-to-seasonal scale variability, large-scale modes of variability, tropical cyclone statistics, and trends of climate change and weather extremes. Moreover, cBottle is a step towards a foundation model, by bridging multiple data modalities (reanalysis and simulation) with corresponding utility beyond emulation to tasks such as zero-shot bias correction, climate downscaling, and channel in-filling.
TUESDAY || July 1, 2025
STEFANO CAMPANELLA
National Institute of Oceanography and Applied Geophysics – OGS
“Learning Global Ocean Forecasting on a Hierarchical Triangle Mesh”
The accuracy of medium-range weather forecasts has steadily improved over time. However, effective risk mitigation and decision making related to extreme weather events depend on reliable longer-range predictions.
Subseasonal-to-seasonal forecasts help bridge this gap and have traditionally relied on ensembles of physics-based atmosphere-ocean coupled simulations. Recently, data-driven models have emerged as a strong alternative but often lack an explicit representation of the ocean, which is crucial at these timescales.
To address this limitation, we introduce a data-driven model for global subseasonal-to-seasonal ocean forecasting using graph neural networks. We outline the rationale behind our architectural choices and present enhancements inspired by finite element methods to handle irregular domains more effectively.
THURSDAY || July 3, 2025
MARTÍN BASCOPÉ
Pontifical Catholic University of Chile
“Building Teacher Learning Pathways for Climate Change Education”
This presentation offers a reflective account of the research and development journey undertaken by the Center for Local Development (CEDEL-UC) at the Pontificia Universidad Católica de Chile, which has led to the design of structured learning trajectories for teacher professional development in climate change education (CCE). Rooted in over a decade of work in southern Chile and aligned with international collaboration through the Latin American STEM Network, this initiative integrates educational research, teacher engagement, and community participation to build locally relevant, pedagogically sound, and socially responsive educational practices.
A key pillar of this initiative is that teacher training is framed within a broader vision of education for citizenship, where science is not only taught as content but mobilized as an applied and ethical practice for social and environmental transformation. Grounded in scientific, civic, and territorial dimensions, and informed by both a comprehensive literature review and empirical analysis of teacher-generated practices, the proposed trajectory offers a replicable model for empowering educators as transformative agents. By bridging local knowledge with global challenges, and integrating interdisciplinary and emotional competencies, this work contributes to the reimagining of teacher professional development for a sustainable future.
THURSDAY || July 10, 2025
SARAH LEWIS
Teachers College-Columbia University
“Any Questions? Seeking Public Interest in Climate Change”
An effective response to climate change demands an engaged public—through behavior, buy-in, and expressions of concern and priorities. Yet this engagement can be hindered by gaps in trust and understanding between the public and scientists, policymakers, and industry leaders. The UNFCCC’s Action for Climate Empowerment framework emphasizes that public awareness—alongside formal education—is essential for closing these gaps.
To get there, researchers argue we need to engage the public in dialogue that reflects who they are and how they see the world. To date, most research efforts to understand the public have relied on top-down measurements of attitudes, ideology, and literacy using close-ended survey questions. This study builds on those efforts by analyzing over 900 responses to an open-ended survey question asking what Americans, in their own words, want to know about climate change.
Using a mixed-methods approach allowed us to both capture the texture of public curiosity—what people actually wonder about, and how they frame it—and to extract statistically meaningful and actionable patterns from a complex dataset. After inductively coding responses by theme, we used Latent Class Analysis to identify distinct profiles of questioners, followed by bivariate analysis to explore variation across demographic and attitudinal groups.
Our findings reveal three profiles of Americans—Response Seekers, Impact Probers, and Consensus Challengers. While each profile reflects known associations between climate change beliefs and broader group characteristics, each profile is internally diverse, complicating simple assumptions about who cares about what, and why.
THURSDAY || July 17, 2025
CHAD SMALL
University of Washington
“The Art (and Science) of Science Communication for Earth Scientists
Science communication – clearly and comprehensively delivering scientific information to non-technical audiences – is becoming increasingly important. Whether it’s for helping people understand the world around them, or building public trust in the scientific process, science communication is being featured more prevalently in researchers’ toolkits. For geoscientists, science communication is becoming essential as anthropogenic climate change tips the scales toward more environmental crises. But if we’re supposed to increase scientific literacy on both general and policymaking levels, there needs to be an understanding of which parts of our work are most salient to those outside our field. In other words, we need to know which parts of our research tell the stories that need to be heard the most. This interactive workshop will unpack how we should communicate research findings to different audiences. There will be a special focus on how geoscientists should respond when asked for comments by the media, and also how we can craft our own independent research-based stories.
THURSDAY || July 24, 2025
KATIE DAGON
NSF NCAR
“Machine Learning-Based Detection of Precipitation Extremes and Regional Climate Impacts”
Extreme precipitation events have wide-ranging impacts on humans and the environment. Machine learning-based detection algorithms can help with the automated classification of the synoptic-scale weather features that produce extreme precipitation events, such as fronts and atmospheric rivers (ARs). Here we use a suite of deep learning algorithms to identify weather fronts and ARs in high resolution Community Earth System Model (CESM) simulations and evaluate the results using observational and reanalysis products. To study how these features might change with climate change, we compare results between CESM simulations using present-day and future climate forcing. We further investigate regional climate impacts such as precipitation associated with fronts and ARs over the United States. We find that detected synoptic events in CESM have seasonally varying spatial patterns and responses to climate change and are found to be associated with modeled changes in large scale circulation.
THURSDAY || July 31, 2025
LEAP ML JOURNAL CLUB
(Myhre G, Hodnebrog Ø, Loeb N, Forster PM. Observed trend in Earth energy imbalance may provide a constraint for low climate sensitivity models. Science. 2025 Jun 12;388(6752):1210-1213. doi: 10.1126/science.adt0647. Epub 2025 Jun 12. PMID: 40504900.)
Climate forcings by greenhouse gases and aerosols cause an imbalance at the top of the atmosphere between the net incoming solar radiation and outgoing longwave radiation from Earth. This Earth energy imbalance has strengthened over the period 2001 to 2023 with satellite data. Here, we show that low climate sensitivity models fail to reproduce the trend in Earth energy imbalance, particularly in the individual longwave and shortwave contributions to the imbalance trend. The inability to produce a strong positive shortwave and strong negative longwave Earth energy imbalance trend is found to be a robust feature in the low climate sensitivity models, especially for models with a climate sensitivity below 2.5 kelvin. The negative longwave contribution to Earth energy imbalance is driven by surface temperature increases and is therefore most pronounced in high climate sensitivity models, whereas the shortwave contribution is generally positive and amplified by greater surface warming.