2023 Fall Lectures in Climate Data Science

Biweekly on Thursdays || Sept. 7, 2023 - Dec. 7, 2023

Click to see past Lectures in Climate Data Science from Spring 2023 and Fall 2022.

THURSDAY || SEPT. 7, 2023

JOSEPH KO (Columbia)

Hawkins: “Community Land Model Parameter Estimation”
Ko: “3-D Reconstruction and Unsupervised Clustering of Ice Crystal Particles to Inform Depositional Growth Parameterizations”

THURSDAY || SEPT. 21, 2023

Pacific Northwest National Laboratory

Improving Aerosol in Earth System Models with AI / ML
The role of aerosol in the Earth system remains a major source of uncertainty due to insufficient data, understanding, and computational power. In this talk, I will introduce the effort of developing faster and better representation of aerosol and aerosol-cloud interactions for a next-generation Earth system model by improving physics, increasing resolution, and utilizing machine learning techniques. In this effort, we have generated hundreds of terabytes of data that consists of process model, large-eddy, convection-permitting, and climate-scale perturbed physics simulations to train ML models. These ML models are used as new parameterizations in an Earth system model representing aerosol radiative properties, droplet nucleation, warm rain processes, and cloud adjustments, and as new diagnostic tools to reveal causal links between aerosol and the intricate Earth system. I will discuss our progress as well as challenges we encountered when integrating the ML models in the Earth system model. 

This Lecture is also a featured event of Climate Week NYC 2023 (September 17-24, 2023).

THURSDAY || OCT. 5, 2023

Duke University

Data Valuation in Federated Learning
To enable practical federated learning, we not only have to improve the efficiency but also address the incentive and fairness concerns. In this talk, I will delve into the related challenges and share our recent endeavors in valuation and personalization in federated learning. Particularly, valuation in federated learning seeks to allocate credits to participants in a just and equitable manner. Personalization in federated learning addresses the individual needs of participants. In the context of federated learning, those problems have some unique challenges. For example, the federated learning process is often run only once and many participants may not be consulted in a round. I will discuss the intuitions and ideas of our latest methods and also discuss some challenges and opportunities for future work.

THURSDAY || NOV. 2, 2023


ClimaX: A Foundation Model for Weather and Climate
Most state-of-the-art approaches for weather and climate modeling are based on physics-informed numerical models of the atmosphere. These approaches aim to model the non-linear dynamics and complex interactions between multiple variables, which are challenging to approximate. Additionally, many such numerical models are computationally intensive, especially when modeling the atmospheric phenomenon at a fine-grained spatial and temporal resolution. Recent data-driven approaches based on machine learning instead aim to directly solve a downstream forecasting or projection task by learning a data-driven functional mapping using deep neural networks. However, these networks are trained using curated and homogeneous climate datasets for specific spatiotemporal tasks, and thus lack the generality of numerical models. We develop and demonstrate ClimaX, a flexible and generalizable deep learning model for weather and climate science that can be trained using heterogeneous datasets spanning different variables, spatio-temporal coverage, and physical groundings. ClimaX extends the Transformer architecture with novel encoding and aggregation blocks that allow effective use of available compute while maintaining general utility. ClimaX is pre-trained with a self-supervised learning objective on climate datasets derived from CMIP6. The pre-trained ClimaX can then be fine-tuned to address a breadth of climate and weather tasks, including those that involve atmospheric variables and spatio-temporal scales unseen during pretraining. Compared to existing data-driven baselines, we show that this generality in ClimaX results in superior performance on benchmarks for weather forecasting and climate projections, even when pretrained at lower resolutions and compute budgets.

THURSDAY || NOV. 16, 2023

NOA URBACH (Teachers College)
  (Teachers College)

Agonafir: “Recent Advances in Urban Flooding Research”
Urbach/Torres: “New York City Educators’ Perceptions of Students’ Engagement with Climate Change”

THURSDAY || DEC. 7, 2023


The Rise of Machine Learning in Weather Forecasting
Over the last year, there has been a rise in machine learning methods being used to create highly accurate weather predictions. These methods have been applied by some of the world’s leading tech companies with impressive results being shown for many applications. Some works have even made claims that these machine learning methods are more accurate than the existing state-of-the-art numerical weather models. But is this claim justified? In particular, how do machine learning methods cope with extreme weather events, which are some of the most difficult and most important events to forecast. In this talk I will present an evaluation of the current state of machine learning in weather forecasting and reflect on the opportunities and challenges for the future.