Modelling the health impacts of prescribed burns

Spatio-temporal analysis using earth observation data to identify adverse health effect and location correlations of landscape fire in the Perth metropolitan area of Western Australia.

The Challenge

Landscape fires (LFs) include wildfires (WFs) and prescribed/planned burns (PBs) and are defined as fires that occur in forest, scrub, or grassland (bushfires). These fires are a significant source of short-term increases in particulate air pollution. However, there is limited information on the health effects of LF smoke on the general population in Western Australia. Studies of health effects of LFs have used a variety of population exposure methods, most of which are limited. Methods to more precisely estimate exposure to LF effects, as well as gain an understanding of the spatio-temporal variations associated with LF effects, are needed in order to accurately correlate such events with health outcomes. This was the first WA study that employed spatio-temporal analysis and earth observation data to explore population smoke exposure and examine the effects of LFs on a large population.

Project Aims

This project had to main objectives which were to:

  1. Explore and improve methods to measure population landscape fire smoke exposure using satellite imagery
  2. Measure the impact of landscape fires on the general population, including the identification of vulnerable groups.

The project partners were the Western Australian Department of Health (DOHWA), Curtin University (Curtin), NGIS, Bureau of Meteorology (BOM WA Office), the Western Australian Department of Biodiversity, Conservation and Attractions, Parks and Wildfire Service (DBCA).

The Approach

The study area covered the whole Perth metropolitan area and the data used for the study were from July 2015 to December 2017. The statistical, epidemiological and spatial analyses were conducted in four steps:

  1. Identify smoke exposure – Analyse satellite images to identify smoke plume masks and affected areas
  2. Establish empirical smoke exposure models
  3. Conduct air quality relationship model fitting/validation assessment by linking earth observation data to air quality and climate data
  4. Conduct a relationship assessment between landscape fire smoke related particulate matter and health utilisation data (i.e. Hospital admissions, emergency department attendance and ambulance callouts).
Key Findings
  1. Smoke plume identification and spatial analysis was a useful and effective tool in more precisely identifying the movement of smoke and affected geographical areas, thus, in theory, providing better population exposure estimates.
  2. This study developed a systematic way of identifying, digitalising, and rasterising smoke plumes from satellite images into spatial grid cells was developed which could potentially be used in other similar studies.
  3. Landscape fire smoke-related particulate matter (PM2.5) was significantly associated with previous day PM2.5 levels, venting index, fire radiative power, aerosol optical depth, fire danger rating and smoke plume masks.
  4. A significant association was found between landscape fire smoke-related PM2.5 and emergency department attendances and hospital admissions for respiratory and cardiovascular conditions.
  5. Health service utilisation peaked on the same day and 1, 2, or 3 days after exposure to landscape fire smoke.

Both reports and the bulletin can be directly accessed via the following link:

To learn more contact us at