UrbanTALES: The biggest urban climate simulation dataset ever released

Understanding why air moves the way it does between buildings has long been one of urban climate science’s toughest challenges. Airflow in streets and between towers is shaped by a maze of real-world factors: building heights, street layouts, wind direction, and even the gaps between structures. 

For decades, researchers could only study these effects using small collections of computer simulations, often just a handful at a time, because the models required massive computing power. A new project led by Dr Negin Nazarian and Dr Jiachen Lu at UNSW Sydney represents a stepchange in urban modelling capacity.

Negin and her collaborators, featuring contributions from researchers in Australia, Canada, the UK and Spain, have created UrbanTALES, the largest and most detailed dataset of its kind: at least 540 high-resolution simulations of wind and turbulence across neighbourhoods around the world, with more being added regularly.

To put that leap in context, researchers a decade ago would routinely work with only 10 to 15 simulations. UrbanTALES delivers more than fifty times that number, produced using over 35 terabytes of data and 3 million CPU hours, the equivalent of a personal laptop working 24 hours a day, seven days a week for 100 years.

Using a powerful computer model called PALM (Parallizable Large Eddy Simulation), the team recreated both idealized urban layouts (for controlled experiments) and real neighbourhoods from Australia and 21 other countries, mapped using OpenStreetMap. 

These neighbourhoods vary widely, from low-rise suburbs to dense districts resembling Tokyo or Barcelona, capturing global diversity in city form. Each simulation reveals how air flows between buildings, how turbulence forms, and how efficiently air mixes and escapes from the ‘urban canopy’.

Explainer: What is the urban canopy?

The urban canopy refers to the part of the city atmosphere that extends from ground level up to the average building height, where airflow is directly shaped by streets, building layouts, and other roughness elements. Within this layer, wind patterns become highly variable and turbulent due to the complex geometry of the built environment.

Why does this matter? UrbanTALES helps answer long-standing questions about how city design influences ventilation, heat, and air quality. For example, the dataset shows that simple metrics like building density don’t always predict airflow well; instead, a new measure called street connectivity (the uninterrupted paths air can travel) correlates more strongly with real-world flow behaviour. For instance, people may assume that denser areas always have slower wind, but the data shows that’s not necessarily true.

Negin said: “The field is still grappling with fundamental questions about how realistic urban neighborhoods influence airflow, and how these processes can be effectively represented in models.”

Crucially, UrbanTALES is open access, meaning scientists, planners, architects, and even AI developers can use it to improve weather and climate models, design cooler and safer streets, or train machine-learning tools for predicting wind at ground level.

This research was led by the School of Built Environment at UNSW and the ARC Centre of Excellence for 21st Century Weather, featuring Professor Melissa Hart and Senior Research Fellow Dr Mathew Lipson, with contributions from the University of Tasmania, the University of Guelph, the UK Met Office at the University of Reading, and the Centre for Energy, Environmental and Technological Research (CIEMAT) in Madrid, Spain. 

The research has been published in the Bulletin of the American Meteorological Society. You can access the full, freely available data set via this online platform: https://urbantales.vercel.app/

This article was written using a combination of human and artificial intelligence