Adaptive high-resolution mapping of air pollution with a novel implicit 3D representation approach

Abstract

Mapping air pollution at high spatial resolution is essential for understanding, managing, and mitigating the adverse impacts of air pollution. Current air pollution monitoring approaches suffer from limited spatial coverage and resolution. Artificial intelligence holds great promise for tackling these challenges, yet its application in air pollution monitoring remains nascent, facing limited transferability regarding low-quality labeled and non-uniform spread data. Here, we introduce Height-Field Signed Distance Function (HF-SDF), an innovative 3D implicit representation, to reconstruct air pollution concentration maps from coarse, incomplete data, which achieves both extensive spatial coverage and fine-scale results with powerful transferability. HF-SDF learns a continuous and transferable mapping model that integrates an auto-decoder network with a geometric constraint, offering flexible resolution. The evaluation uses reanalysis data and satellite observations, reaching accuracy rates of 96% and 91%, respectively. HF-SDF reveals immense promise in advancing air pollution monitoring by offering insights into the spatial heterogeneity of pollution distributions.