Repeatable and reproducible workflow to examine the human-environmental interaction: a spatially explicit GeoAI method

Abstract

This study roots in the Social Ecological Theory and employs a tri-environmental conceptual framework (i.e., social, built and natural environment) to examine the human-environmental interaction, in terms of how the combined exposure to heat and air pollution affect physical and mental health. It contributes, for the first time, the fine-grained nationwide investigation in Australia and highlights how such human-environmental interaction varies across inter- and intra-urban areas. We conducted an ecological study to explore the importance of heat and air quality to physical and mental health by considering 48 tri-environmental confounders through the global and local random forest regression models, as advanced machine learning methods with the advantage of revealing the spatial heterogeneity of variables. Our key findings are threefold. First, the social and built environmental factors are important to physical and mental health in both urban and rural areas, and even more important than exposure to heat and air pollution. Second, the relationship between temperature and air quality and health follows a V-shape, reflecting people's different adaptation and tolerance to temperature and air quality. Third, the important roles that heat and air pollution play in physical and mental health are most obvious in the inner-city and near inner-city areas of the major capital cities, as well as in the industrial zones in peri-urban regions and in Darwin city with a low-latitude. Our conceptual framework and analytical workflow can also be applied to examine the interaction between human and other environmental problems in the era of a warming climate.