As global temperatures climb and cities swelter through longer, more frequent heatwaves, the reflex response has been to crank up the air conditioning. But that fix comes with a cruel irony: the more machines hum to cool indoor air, the more waste heat they dump outside, intensifying the very urban heat islands they’re meant to help residents escape. Now, researchers at the University of Manchester think they’ve found a way to break that cycle — by turning rooftops into rain-fed cooling systems.
In a study published in the journal Earth’s Future, a team led by Dr. Zhonghua Zheng designed and tested a system that collects rainwater runoff from building roofs, stores it in a tank, and automatically sprays it back onto the roof surface whenever temperatures climb past a set threshold. Using Tokyo as a real-world test case, the researchers found that this rainwater harvesting and roof sprinkling system, which they call RWTSP, meaningfully cut the energy needed for air conditioning while also lowering outdoor temperatures, shrinking the number of heatwave days, and reducing the flooding-prone runoff that pours off city roofs during storms.
“Cities around the world are facing growing challenges from extreme heat,” said Zheng, who is Co-Lead for Environmental Data Science and AI at the Manchester Environmental Research Institute and a senior lecturer in data science and environmental analytics. “Air conditioning can help keep people safe and comfortable, but it also consumes large amounts of energy and releases additional heat into the urban environment. Our study shows that harvesting rainwater from roofs and using it strategically for cooling could provide a practical way to reduce both energy demand and urban temperatures.”
Why Roofs, and Why Now
The stakes are considerable. The World Health Organization has estimated that the number of people exposed to extreme heatwaves rose by roughly 125 million between 2000 and 2016, a trend tied to declining worker productivity and rising heat-related mortality. Meanwhile, the United Nations Environment Programme has projected that global electricity demand for air conditioning could triple by 2050 — an increase roughly equivalent to the combined 2019 electricity consumption of the United States, Europe, and Japan. Because AC units expel the heat they remove from indoor air into the streets and alleys outside, that surge in cooling demand threatens to make cities even hotter, deepening a feedback loop between climate and energy systems.
Roof watering isn’t an entirely new idea. Prior laboratory work, including recent experiments published in Energy and Buildings, has shown that spraying roofs can cool their surfaces by anywhere from about 5.5 to nearly 20 degrees Celsius, depending on the roofing material. What’s been missing, the Manchester team argues, is a way to pair that cooling effect with a sustainable water source, and a rigorous method for figuring out how big the tank should be, how much water to use, and when to turn the sprinklers on.
“The rainwater tank also provides an additional co-benefit on reducing the extreme urban runoff,” said Junjie Yu, a PhD researcher on the project. “This approach exemplifies a ‘natural solution to natural challenges,’ in which rainwater serves as a natural resource to mitigate both thermal stress and hydrological extremes.”
Building a Digital City to Test It
To evaluate the system without installing tanks and sprinklers on thousands of real rooftops, the researchers built a new module inside the Community Land Model Urban, a widely used urban climate model that simulates how heat, moisture, and energy move through buildings, streets, and rooftops. They grounded their simulations in real atmospheric data collected from a flux tower in the Yoyogi neighborhood of Tokyo, a compact, mid-rise residential area, and validated the model’s output against a decade of observed temperature and energy readings before trusting it to run experiments.
The team then faced a classic engineering trade-off: three variables — the size of the rainwater tank, the intensity of sprinkling, and the roof-surface temperature that triggers the system to switch on — interact in complicated, non-linear ways. Testing every possible combination directly in the climate model would have taken enormous computing time. So the researchers turned to machine learning, training a transformer-based “tabular foundation model” called TabPFN to approximate the climate model’s behavior, then using a genetic optimization algorithm to search for the combinations of tank size, intensity, and trigger temperature that delivered the best cooling for the least water storage. The AI-generated predictions were later checked against real climate-model runs and closely matched the ground truth, the team reported.
Timing Matters More Than Tank Size
The clearest finding to emerge from that analysis was that when the sprinklers switch on matters far more than how big the tank is or how much water gets sprayed. Using an interpretability technique called SHAP analysis, the researchers found that the temperature threshold for triggering sprinkling had a bigger influence on air conditioning energy use than either of the other two parameters.
The study also found diminishing returns on tank size. Growing the rainwater tank from nothing up to a modest size delivered a large chunk of the possible energy savings, but pushing the tank nearly five times bigger only squeezed out a small additional reduction. Excess water sprayed onto an already-wet roof, the researchers noted, often just sits there rather than evaporating and contributing to further cooling.
Applied across a roughly 0.6 square kilometer study area containing nearly 2,000 rooftops, the optimized systems reduced hours of extreme roof-surface and air temperatures substantially, and cut the number of heatwave days as tank size increased, though with a trade-off: bigger tanks reduced heatwave frequency and duration more, but delivered smaller marginal gains in energy savings. Notably, the benefits scaled with atmospheric warmth — the hotter the year, the more energy the system saved, a finding the authors say suggests the strategy could become more valuable, not less, as climate change intensifies.
Not a Fix for Everywhere
The researchers are careful to note that roof rainwater cooling won’t suit every city. Its usefulness will hinge on local rainfall, regulations governing rainwater harvesting, and the moisture risks — such as mold growth or roof deterioration — that come with regularly wetting a building’s exterior. Regions with high cooling demand and reliable rainfall stand to benefit most, the team suggests, while arid cities with scarce rainwater may find the approach less practical. The study also acknowledges its limits: it relies on model estimates rather than a physically built and monitored rooftop system, and the strength of the roof-to-street heat exchange it simulates may not translate identically to cities with different building layouts.
Still, the framework itself — pairing physics-based climate simulation with machine learning and optimization — is designed to be exportable. The authors say it could be adapted to evaluate other “blue infrastructure” cooling strategies, including road sprinkling and urban misting systems, and could eventually be scaled up within global Earth system models to assess rooftop cooling’s potential city by city, worldwide.
For now, the message for urban planners is a relatively simple one: before investing in bigger tanks, get the timing right.
Sources
Yu, J., Oleson, K. W., Qin, Y., Zhao, L., Topping, D. O., & Zheng, Z. (2026). Optimizing the rainwater harvesting and roof sprinkling system to adapt to urban extreme heat. Earth’s Future, 14, e2026EF008876. https://doi.org/10.1029/2026EF008876
University of Manchester. “Harvesting rainwater from rooftops could help cities stay cool and cut the number of heatwave days.” EurekAlert!, August 5, 2026. https://www.eurekalert.org/news-releases/1138805
World Health Organization. “Heatwaves.” who.int/health-topics/heatwaves
United Nations Environment Programme (2021). Beating the Heat: A Sustainable Cooling Handbook for Cities.
Chaumont, M., Filaine, F., Parison, S., Hendel, M., & Royon, L. (2025). Impact of roof watering on urban cooling during heat waves. Energy and Buildings, 337, 115693.
COPY II (2-3 PARAGRAPHS)
IMAGE CREDIT: NASA.

