1/27/25 | 11:00am-4:45pm | 32-155
Operations Research/Machine Learning for Climate and Sustainability
Description: This seminar highlights innovative solutions to environmental challenges through optimization and machine learning. There will be five talks throughout the day that explores a diverse range of topics, including the analysis of extreme weather impacts, optimization and decarbonization of energy infrastructure, and strategies for sustainable forest management.
Date: Monday, January 27th
Place: 32-155
Schedule
Coffee and Refreshments – 10:30am-11:00am
Talk 1 – 11:00am-11:45am
Ana Trisovic
Research Scientist
MIT, FutureTech Lab
| Title |
Analysis of Co-Exposure to Extreme Heat, Wildfire Burn Zones, and Wildfire Smoke in the Western US (2006–2020)
| Abstract |
| Climate change drives three key heat-related hazards: extreme heat (EH), wildfire burn zones (WFBZ), and wildfire smoke (WFS). This study utilized daily census tract-level data from 2006–2020 to analyze the spatiotemporal patterns and demographic characteristics associated with these hazards across eleven Western US states. Among the 18,106 census tracts analyzed, at least one hazard was observed on an average of 32 days annually, accounting for 581,867 tract-days per year. Co-exposures, particularly EH-WFS, showed an upward trend over the study period and represented the most common pairing, with an annual average of 38,218 tract-days. EH-WFS exposure patterns exhibited significant interannual variability, while co-exposures involving WFBZ were spatially restricted and showed no temporal increase. On average, the highest tract-days of triple exposure (EH-WFBZ-WFS) per year occurred in California (n=35), Arizona (n=25), and Oregon (n=24). Census tracts with the greatest exposure to EH-WFBZ-WFS disproportionately contained populations characterized by higher proportions of older adults, individuals with disabilities, people living in poverty, and non-White racial/ethnic groups. American Indian and Alaska Native populations faced particularly high risks across all co-exposure scenarios. As the impacts of climate change intensify, systematic tracking of multi-hazard co-exposures offers critical insights for resource allocation and public health intervention strategies, enabling more equitable protection for vulnerable communities. |
|
| Bio |
| Dr. Ana Trisovic is a Research Scientist at the FutureTech Lab at MIT, focused on studying the impacts of emerging technologies on scientific innovation and societal systems. Dr. Trisovic’s interdisciplinary research portfolio encompasses the analytics of scientific processes (science of science), scalable frameworks for open data and open-source software, and the operationalization of AI models, hardware systems, and research reproducibility strategies. Previously, Dr. Trisovic held a position as a Research Associate at the Harvard T.H. Chan School of Public Health and served as a Postdoctoral Scholar at Harvard’s Institute for Quantitative Social Science. There, she contributed to the development and deployment of platforms such as the Harvard Data Commons and Dataverse. As a CLIR Postdoctoral Fellow at the University of Chicago, she specialized in advancing methodologies for data preservation and accessibility. During her Ph.D. at the University of Cambridge, in collaboration with CERN, her work centered on data systems for the LHCb experiment, CERN Open Data, and CERN Analysis Preservation. Dr. Trisovic currently serves as an editor for the Journal of Open-Source Software (JOSS) and as a review board member for the Journal of Systems Research (JSys). As a member of the U.S. National Committee for CODATA under the National Academies of Sciences, Engineering, and Medicine, Dr. Trisovic is committed to advancing open science paradigms and democratizing access to AI-driven solutions. |
Talk 2 – 12:00pm-12:45pm
/
Ruaridh Macdonald
Research Scientist
MIT Energy Initiative
| Title |
Optimization Methods in Energy Infrastructure Planning
| Abstract |
| This talk explores how optimization methods are used in energy infrastructure planning to evaluate the potential impacts of emerging technologies and policies and guide their future R&D and deployment. After highlighting examples of recent applications, I will outline challenges in the field and then discuss ongoing work at MIT to address them. This includes new approaches to incorporating non-linear features in models, model reduction, and sectoral and spatial Benders decompositions. Finally, I will discuss future directions for this research, including why simulation-based approaches may become more prominent. |
|
| Bio |
| Dr. Ruaridh Macdonald is a research scientist at the MIT Energy Initiative. His research uses multi-sector infrastructure planning models to investigate how best to decarbonize the electricity grid and other sectors to reduce the cost of the energy transition while also ensuring grid resilience and security. He is developing novel methods which allow for higher resolution multi-sector models to be optimized over long time periods. He received his PhD in Nuclear Science and Engineering from MIT. |
Lunch Break – 12:45pm-2:00pm
Talk 3 – 2:00pm-2:45pm
Rohit Parasnis
Postdoctoral Researcher
MIT LIDS
| Title |
Aligning Sustainability and Welfare in Forest Management: A Network Game Approach
| Abstract |
| We address the challenge of promoting sustainable practices in production forests managed by strategic entities (agents) that harvest agricultural commodities under concession agreements. These entities engage in activities that either follow sustainable production practices or expand into protected forests for agricultural growth, which leads to unsustainable production. Our study uses a network game model to design optimal pricing policies that incentivize sustainability and discourage environmentally harmful practices. Specifically, we model interactions between agents, capturing both intra-activity (within a single activity) and cross-activity (between sustainable and unsustainable practices) influences on agent behavior. We solve the problem of maximizing welfare while adhering to budgetary and environmental constraints, particularly, limiting the total unsustainable effort across all agents. Although this problem is NP-hard in general, we derive closed-form solutions for various realistic scenarios, including cases with regionally uniform pricing and the use of sustainability premiums or penalties. Remarkably, we find that we can achieve both welfare improvement and reduction in unsustainable practices without reducing any agent’s utility, even when there is no external budget for increasing premiums. We also introduce a novel node centrality measure to identify key agents whose decisions most influence the aggregate level of unsustainable effort. Empirical validation confirms our theoretical findings, offering actionable insights for policymakers and businesses aiming to promote sustainable resource management in agricultural commodity markets. Our work has broader implications for addressing sustainability challenges in the presence of network effects, offering a framework for designing incentive structures that align economic objectives with environmental stewardship. |
|
| Bio |
| Rohit Parasnis is a postdoctoral researcher in the Laboratory for Information and Decision Systems at the Massachusetts Institute of Technology. From 2022 to 2023, he was a postdoctoral researcher in the Department of Electrical and Computer Engineering at Purdue University. He received the Ph.D. degree in Electrical Engineering from the University of California San Diego in 2022 and dual degrees (B.Tech in Electrical Engineering and M.Tech in Applied Mechanics) from IIT Madras in 2016. His research interests include network analysis and control, game theory, optimization, and matrix analysis, with applications to the sustainability of natural and engineered systems. He was awarded the Henry Booker Award for socially responsible research in 2021 and the Charles Lee Powell Fellowship in 2016. |
Talk 4 – 3:00pm-3:45pm
Yuki Miura
Assistant Professor
New York University
| Title |
Identifying, Measuring, and Managing Socioeconomic Impacts of Floods Amid Climate Change
| Abstract |
| Climate change worsens the impacts of hurricane-induced flooding, making it crucial to develop effective methods for identifying, measuring, and managing these escalating risks. Addressing such risks triggered by climate change and natural hazards is challenging due to inconsistent historical data, unpredictable weather patterns, conflicting stakeholder objectives, and a multitude of potential adaptation solutions. My research offers practical tools for assessing the socioeconomic impacts of floods and other climate-related risks. These tools are designed for real-world application, incorporating feedback from a range of stakeholders—including municipal offices, emergency management teams, and water resource experts. This talk will cover analyses at multiple scales, including local (New York City), national, and global levels. |
|
| Bio |
| Dr. Yuki Miura is an assistant professor at the Department of Mechanical and Aerospace Engineering and the Center for Urban Science and Progress (CUSP) at Tandon School of Engineering at New York University. She is also a faculty advisory board member at the Volatility and Risk Institute at NYU Stern School of Business. Previously, she was at Morgan Stanley in climate risk management and quantitative strategy. She serves as a NYC Panel on Climate Change. Her research is at the intersection of engineering, climate science, finance, and social sciences. She has collaborated with the National Center of Atmospheric Research and the governments of New York State/City. Her work has been recognized through publications in leading journals and featured in The New Yorker and The New York Times. She holds PhD, MPhil, and MS from Columbia University and BE from Keio University. |
Talk 5 – 4:00pm-4:45pm
Michael Howland
Assistant Professor
MIT, CEE
| Title |
Powering the Renewable Energy Transition with Optimization of Joint Physics- and Data-Driven Models
| Abstract |
| To meet net-zero carbon emissions targets by mid-century, an order of magnitude increase in renewable power capacity is required. Yet it is not clear where wind and solar generation should be sited to maximally support resilient power systems with low cost. To guide energy system planning and operation, wind and solar resources are estimated using uncertain numerical models of the multi-scale Earth system. Planning for decarbonization, therefore, depends on the quality of meteorological data that is used, including resolution, accuracy, and uncertainty, and how it is used to represent renewable power in energy system modeling. We determine the required spatial resolution of meteorological data and how to integrate it with energy system modeling to plan cost-effective decarbonized energy systems. We design minimum-cost decarbonized energy systems in three diverse geographic regions (ISO-NE, CAISO, ERCOT) and we interpret these kilometer-scale energy system optimizations based on the geophysical drivers and constraints for generalizable insights. The supply of wind and solar power generation relies on spatiotemporal variations and correlations within and across the wind and solar resources. Using downscaled meteorological data at km-scale yields lower cost compared with typical meteorological data at resolutions over 30 km (i.e. standard reanalysis) by revealing opportunities for complementarity between spatiotemporal variations in wind and solar supply to align with demand. Further, current renewable energy tax incentives in the United States reward total energy production – this study suggests that wind and solar complementarity can lead to a more cost-effective energy system design. The wind and solar siting locations that minimize the energy system cost differ significantly from the locations with the highest wind/solar resource potential on average. Moving forward, critical uncertainties in future climate conditions and their impacts on renewable supply and energy demand will pose risks to the resilience of future energy systems if they are not addressed in planning. We leverage gradient-free optimization and machine learning for a thousand-fold speed-up in the quantification of uncertainty in meteorological models. This acceleration enables energy system optimization under uncertainty to uncover practical strategies for renewable siting to minimize risk to climate change and extreme events. |
|
| Bio |
| Michael F. Howland is the Esther and Harold E. Edgerton Assistant Professor of Civil and Environmental Engineering at MIT. He was a Postdoctoral Scholar at Caltech in the Department of Aerospace Engineering. He received his B.S. from Johns Hopkins University and his M.S. from Stanford University. He received his Ph.D. from Stanford University in the Department of Mechanical Engineering. His work is focused at the intersection of fluid mechanics, weather and climate modeling, uncertainty quantification, and optimization and control with an emphasis on renewable energy systems. He uses synergistic approaches including simulations, laboratory and field experiments, and modeling to understand the operation of renewable energy systems, with the goal of improving the efficiency, predictability, and reliability of low-carbon energy generation. He was the recipient of the Robert George Gerstmyer Award, the Creel Family Teaching Award, and the James F. Bell Award from Johns Hopkins University. At Stanford, he received the Tau Beta Pi scholarship, NSF Graduate Research Fellowship, a Stanford Graduate Fellowship, and was awarded as a Precourt Energy Institute Distinguished Student Lecturer. At MIT, he has received the Maseeh Excellence in Teaching Award and the Office of Naval Research (ONR) Young Investigator Program (YIP) award. |
If you have any questions, please contact us via email: orc_iapcoordinators@mit.edu.
