Australian fires and perinatal health risks

This study is investigating risk of birth outcomes and perinatal mortality from fire-related PM_2.5. The interdisciplinary team is developing fire modeling methods to estimate PM_2.5 specifically from fires for Australia with improved fire emissions inventories and Lagrangian modeling.

Investigation of smoke exposure during  Au stralia fire seasons: Phase 1. Importance of quantifying plume injection heights

Michelle L. Bell,¹ Josh Warren,¹ Xu Feng,² Loretta J. Mickley,² Gavin Pereira,³ Yuming Guo,⁴ Jenny A. Fisher⁵

¹Yale University, New Haven, USA; ²Harvard University, Cambridge, USA; ³Curtin University, Perth, Australia; ⁴Monash University, Melbourne, Australia; ⁵University of Wollongong, Wollongong, Australia

Backg round and objectives: In recent years Australia has experienced unprecedented wildfires (often referred to as “bushfires”), resulting in severe air pollution across much of the continent. To date, studies on fire smoke and health are limited, but the growing literature indicates increased risk such as for hospital admissions and mortality. Our literature reviews on fire smoke found that birth outcomes and perinatal mortality are understudied, despite the importance of adverse pregnancy outcomes, which are linked to mortality and other health outcomes. A first step in quantifying smoke exposure in Australia involves modeling the emissions and transport of smoke plumes, but incomplete knowledge of the injection heights of these plumes poses a major challenge.

Methods and approach: Here we use two methods to quantify the percentage of fire emissions injected above the planetary boundary layer (PBL), and we further investigate the impacts of plume injection heights on daily mean surface concentrations of PM_2.5 from smoke in key cities in southeastern and northern Australia from 2009 to 2020. For the first method, we rely on PBL heights from an assimilated meteorological dataset, together with long-term averaged vertical profiles of smoke emissions based on satellite data. For the second method, we develop a novel approach based also on satellite observations but coupled with a random forest, machine-learning model that allows us to predict the percentage of smoke injected above the PBL in each grid cell on each day. We apply the resulting injection percentages to the smoke PM_2.5 concentrations simulated by the Stochastic Time-Inverted Lagrangian Transport (STILT) model for each target city.  

Preliminary results: We find that characterization of the plume injection heights greatly affects estimates of surface daily smoke PM_2.5, especially during severe wildfire seasons, when intense heat from fires can loft smoke high in the troposphere. When compared to daily surface observations, the base model overestimates PM_2.5, but both plume injection methods diminish this overestimate. The use of machine learning to predict the smoke behavior leads to the best agreement, especially in northern and southeastern Australia, where many fires occur. We find that smoke PM_2.5 contributed 3% to 52% of total PM_2.5 in key cities across Australia during the wildfire seasons of 2009 to 2020. For example, in Sydney, smoke PM_2.5 accounted for 3-40% of total PM_2.5 over these years, with 34-40% during the intense 2015-2016 and 2019-2020 wildfire seasons.

Interpretation: Results show that use of machine learning to characterize  the injection heights of smoke plumes improves our estimates of PM_2.5 exposure. Smoke from wildfires significantly enhances PM_2.5 exposure in many Australian cities, especially during intense fire seasons. These results have implications for human health, which we will are currently investigating.