Illinois experts chart future of urban climate research: solving the ‘trilemma’

Urban climate experts at the University of Illinois Urbana-Champaign’s Discovery Partners Institute are defining a new vision for urban climate modeling and forecasting.

In a perspective piece for the Nature portfolio journal Urban Sustainability, researchers Peiyuan Li and Ashish Sharma of the Discovery Partner Institute (DPI) Climate Hub provide with international collaborators a roadmap for integrating three powerful but traditionally distinct research thrusts. They outline how physics-based modeling, artificial intelligence and hyperlocal measurements can synergize to create accurate and informative urban climate intelligence.

Photo Credit: Courtesy of Ashish Sharma and Peiyuan Li.

“By 2050, two-thirds of the world's population will live in urbanized settings,” said Li, the article’s lead author. “Understanding how weather and climate impact cities — from metropolitan areas to neighborhoods and even down to individual blocks — directly informs policy decisions tied to public health and economic development that impact people’s safety, wellbeing and livelihoods. It is a ‘grand challenge’ that needs to be met with complementary tools and capabilities.”

“We call it an ‘urban climate trilemma’ because there are three distinct problems that any framework needs to address simultaneously: granularity, coverage and interpretability,” said Sharma, the DPI Climate Hub Lead. “That is to say, the field must be able to resolve very fine features, cover very large geographic features and yield predictions that can be understood and acted on. Traditional tools try to address two of these problems at once, but nothing has attempted to address all three together.”

Urban climate systems are such a difficult problem because they operate across multiple interacting scales. Techniques used to model weather and climate across an entire region cannot capture events on the scale of individual city blocks, but block-scale processes are driven by large-scale trends. Machine learning models based on hyperlocal data help, but these models often do not incorporate an understanding of the physical processes at play, limiting their applicability.

“Unlike rural areas, cities are incredibly complex and heterogeneous ecosystems,” Sharma said. “Our work in Chicago has shown that there can be temperature differences as high as 4 degrees Fahrenheit in a span of two blocks. This sort of information is vital for policies related to heat management for residents. But the challenge is to make future predictions when all these hyperlocal areas are interacting with each other and with regional effects.”

“It’s especially urgent since the effects of many policy decisions won’t be felt for decades,” Li added. “If a city decides to plant 1 million trees to help manage heat, it will likely take upwards of 20 years. Models and predictions need to be reliable over long timescales, a goal that microclimate modeling has not yet achieved.”

Researchers at the DPI Climate Hub are spearheading a multi-disciplinary approach to this problem, combining the strengths of physics-based models, machine learning models and remote sensing of hyperlocal environments. Regional-scale physics-based models are interpretable and provide coverage over large areas. Urban-scale physics-based models provide interpretability over very small areas. Machine learning models are used to bridge granular resolution and large-area coverage. Sharma and Li believe that advancing the capabilities of urban climate modeling will require drawing on the unique strengths of each approach.

“Regional modeling, local modeling and data-driven modeling are traditionally distinct research communities,” Li said. “However, in our own work and watching the field evolve, we believe that there is a trend towards convergence.”

“Looking forward, researchers will need to recognize this trend and the capabilities it offers,” Sharma added. “We’re confident that research focuses and funding decisions in the future will increasingly recognize that the ‘trilemma’ requires a three-pronged approach.”

The article includes input from international experts, with contributors from Argonne National Laboratory, Spain’s Department of Environment, the Indian Institute of Technology Bombay, the University of Birmingham, The Hong Kong University of Science and Technology, the University of Chicago and Singapore Management University.

The perspective, “Unraveling the intractable trilemma in urban weather and climate modeling,” is available online. DOI: 10.1038/s42949-026-00388-z

Illinois Grainger Engineering affiliations
Ashish Sharma is the Climate and Urban Sustainability Lead at the Discovery Partners Institute. He is also affiliated with the Department of Climate, Meteorology and Atmospheric Sciences and the National Center for Supercomputing Applications.