Resolving Risk in a Nonstationary World: Large Ensembles, Extremes, and Sustainability

Recording YouTube video

Framing Slides Framing Slides (PDF)

Abstract Today’s climate risks lie not in the averages but in the extremes. In a changing climate, the tails of probability distributions may spread far more rapidly than the drifting mean, fundamentally challenging how hazards are modeled, interpreted, and acted upon.

Dr. Sai Ravela (MIT) develops methods to resolve these risks by producing detailed, large-ensemble process representations that explicitly connect hazards to decisions within what he calls the “sustainability stack”—a framework that links hazard, vulnerability, and exposure to impact, risk, and ultimately, action.

This work shows how even modest physics-based simplifications, combined with statistical methods and generative AI, can improve our ability to capture risks at the tails despite limited observations. In particular, it extends classical parameterized approaches for tropical cyclones to the more complex dynamics of extratropical systems. These systems remain traditionally underserved in risk modeling, yet they play a critical role in shaping climate impacts. Leveraging remote sensing for exposure mapping and large language models for vulnerability mapping, Dr. Ravela demonstrates how the sustainability stack can be operationalized for local decision-making through serious games, translating modeling into practical action.

By bridging innovations in modeling with the realities of decision cycles, this talk offers a new perspective on how risk science can directly inform sustainable planning and adaptation under deep uncertainty.

Biography Dr. Sai Ravela is principal research scientist and director of the Earth Signals and Systems Group at MIT’s Department of Earth, Atmospheric and Planetary Sciences (EAPS) and has for more than 20 years been a research scientist at MIT, pioneering work toward Earth sustainability and computer vision. His enduring research interest is to design and use methods that can answer queries about the behavior of stochastic spatio-temporal processes. Combining climate statistics, atmospheric modeling and physics to assess the risk of future extreme events, Ravela is a pioneer of understanding risk of extreme weather in an interconnected world. He broadly studies estimation, control, decision and information theories. He has over 125 publications and patents to his name. He received the MIT Infinite Kilometer Award for exceptional research and mentorship in 2016 and is co-founder of Windrisktech LLC, which quantifies cyclone-induced risk.

His recent focus has included earth systems estimation, specifically to design algorithms that overcome the problems of nonlinearity, dimensionality and uncertainty that is characteristic of earth problems and a real barrier to effective predictability. What makes Dr. Ravela’s work particularly compelling for our discussion on complex risk science is his groundbreaking research on how climate change is amplifying coastal hazards in vulnerable regions like Bangladesh. His recent work reveals how rising seas and intensifying cyclones create cascading risks that compress what were once generational threats into decade-scale realities. This intersection of physical modeling, machine learning, and real-world risk assessment exemplifies the kind of integrated approach we need to understand and manage complex, interconnected risks in our changing world.

You can read about specific research topics on this subject and others here.