Abstract
Wildfires are a recurrent disturbance across northern Queensland, with significant ecological and atmospheric consequences. This study examines wildfire occurrence and associated variations in atmospheric carbon monoxide (CO) column density across five Local Government Areas (LGAs), namely, Charters Towers, Croydon, Etheridge, Mareeba, and Tablelands over the period 2020–2025 using a multi-sensor remote sensing framework implemented in Google Earth Engine (GEE). Sentinel-2-derived vegetation and burn indices, including the Normalized Difference Vegetation Index (NDVI) and Normalized Burn Ratio (NBR), were integrated with Sentinel-5P TROPOMI CO column density and ERA5-Land meteorological variables (temperature and wind speed), complemented by ESA WorldCover land classification and historical fire-extent polygons. Annual summaries were used to examine interannual patterns and associations among vegetation condition, fire-related landscape characteristics, atmospheric CO column density, and meteorological variables. Because Sentinel-5P measures total atmospheric CO column density and does not distinguish wildfire emissions from anthropogenic emissions, transported pollution, or background atmospheric CO, the CO observations are interpreted as fire-associated atmospheric patterns rather than source-specific wildfire emission estimates. Observed CO column density was strongly associated with meteorological conditions, with temperature showing a positive correlation (r = 0.94) and wind speed a negative correlation (r = −0.58). NDVI and NBR were negatively correlated with CO (r = −0.62 and r = −0.45, respectively), indicating that lower vegetation greenness and lower burn-ratio values tended to coincide with higher CO columns, although these relationships are influenced by fire intensity, fuel condition, atmospheric transport, and spatial-scale mismatch. The study demonstrates the value of integrating multi-source Earth observation data for regional fire-smoke and air-quality assessment while highlighting the source attribution and spatial-resolution limitations of TROPOMI for local administrative-scale analysis.