Overview
We are using machine learning to map residential property-parcel scale (re)development across Australian cities and measure its impact on built, impervious, and tree cover.
Using a combination of thermal remote sensing and microclimate models, we're assessing how (re)development has warmed Australian residential areas and identifying types of development that increase or reduce heat exposure.
A detailed analysis of existing planning and policy processes will be undertaken to assess their effectiveness for controlling heat-causing (re)development.
The Hot Cities project is funded by an ARC Discovery Project and is a collaboration between The University of Western Australia and RMIT.

Background
Australian cities are facing many challenges including housing growing populations while mitigating the harms of urban heat exposure. Due to economic and demographic drivers, cities are dynamic and under continual (re)development which manifests in expansion, densification, and upgrading of existing housing stocks. This (re)development changes built and natural environments and the thermal properties of residential areas in ways that can increase exposure to harmful heat.
Extreme heat kills more Australians than any other natural hazard, it affects mental and physical health, and limits economic productivity. The Sustainable Development Goals, the Australian Smart Cities Plan, and local government initiatives recognise the threat of extreme heat in urban environments. Urban (re)development occurs in piecemeal fashion, with varying degrees of planning oversight, and without consistent monitoring. Without detailed and consistent maps of where and how parcels are (re)developing, the magnitude of warming caused by (re)development remains unclear.
Using city-scale high-resolution land cover and built form datasets and geospatial machine learning and temperature modelling this project will address these data and knowledge gaps. It will generate a fine-scale picture of how cities are being developed, quantify the impacts of this development on built and nature environments and heat exposure, assess the effectiveness of existing planning controls, and generate new insights to guide planning and management of urban areas that are climate resilient. A detailed description of the project's work is provided below.

Work Packages
WP1: Mapping (re)development
Developing geospatial machine learning models to classify the development type on Australian property parcels.
WP2: Built and natural environment change
Combining city-scale maps of (re)development with high-resolution surface cover data to map change in built and natural cover.
WP3: Heat modelling
Using microclimate models and thermal imaging to assess how (re)development has warmed Australian residential areas.
WP4: Planning and policy analysis
Assessing how existing planning and policy processes control (re)development that exacerbates heat exposure.

News & Updates
August, 2025
GIScience, Canterbury, New Zealand
Presentation: A deep learning urban development type classification system: an application to millions of Australian property parcels.
May, 2025
Planning Institute of Australia, Planning Congress
Presentation: Exposing the true nature of residential development: An automated property parcel classification approach using multi-modal deep learning.
April, 2025
Western Australian Local Government Association, Urban Forest Working Group
Presentation: ARC Discovery Project – Impact of development typologies on urban heat and canopy.
October, 2024
New South Wales Department of Planning, Housing and Infrastructure, Green and Sustainable Cities Branch
Presentation: Why is (Re)development hot? Measuring Cumulative Heat in Australian Cities.
October, 2024
Western Australian Local Government Association
Presentation: Why is (Re)development hot? Measuring Cumulative Heat in Australian Cities.
October, 2024
Victorian Government, Department of Transport and Planning
Presentation: Why is (Re)development hot? Measuring Cumulative Heat in Australian Cities.
October, 2024
ISPRS Technical Commission IV Symposium, Fremantle
Presentation: Residential redevelopment through the lens of AI. Automating property parcel classification using multi-modal deep learning.
May, 2024
Hydropolis 2024: Creating a green-blue Perth and Peel
Presentation: City scale monitoring of vegetation and heat.