Estimating residential development activity from Earth observation to assist with population forecasting.
Accurate population forecasting helps councils make strategic decisions for the future. The team at Informed Decisions generate these forecasts across Australia, taking key information such as current population, number of dwellings, and residential development activity. Residential development activity can be manually assessed from high-resolution aerial photography, but this is time-consuming and cannot be easily scaled out across Australia due to the availability of aerial photography. This project explored whether freely available satellite imagery could be used to automatically estimate residential development progress, allowing .id to streamline their collection of this important data.
FrontierSI worked directly with Informed Decisions.
Over a period of three months, the FrontierSI team developed a prototype workflow to estimate the status of known developments from freely available Sentinel-2 imagery from Digital Earth Australia. Residential developments could be classified as undergoing land-clearing, undergoing development, or completed. The FrontierSI team used machine learning to identify areas of bare soil, built up vegetation and water, and then crafted a customised ruleset to classify the status of each development site. The solution was built to run on large areas (e.g., Western Sydney) and produced classification maps across multiple years, as well as estimates on the amount of area converted from one type to another year-on-year.
The project established that Earth observation data can be used to automatically estimate residential development status across large areas at low cost. Having a solution that can be run automatically will save Informed Decisions significant effort when it comes to collecting data on residential development status across Australia.
Please contact us at email@example.com or Project Manager Caitlin Adams at firstname.lastname@example.org