Overview of the performance of satellite fire products in China: Uncertainties and challenges

Satellite fire observations provide an essential constraint for the estimation of global biomass burning emissions. In this study, we present a comprehensive insight into the performance of the common satellite fire products in eastern China. Despite consistent spatial patterns, both polar-orbiting and geostationary satellite observations have large omission errors for the agricultural burning fires. Owing to a coarse resolution of 2 km, approximately 90% of the concurrent 375 m Visible infrared Imaging Radiometer (VIIRS) fires are not detected in Himawari-8 products. Nevertheless, the total amount of daily Himawari-8 fires is much higher than those of VIIRS and 1 km Moderate resolution imaging spectroradiometer (MODIS). The peak time of diurnal fire counts in eastern China has obvious seasonal variations, some of which are missed by polar-orbiting satellite detection. Validation by 3 m PlanetScope images shows that VIIRS and MODIS have a very high accuracy in detecting crop straw burning fires. However, Himawari-8 fires have obvious false alarms due largely to their algorithm defects. Also, the coarse resolution of Himawari-8 tends to make fire detection more likely to be obscured by dense smoke. The unmanned aerial vehicle (UAV) supervisions reveal prevalent small agricultural fires (<0.5 × 10⁴ m²) in the extensive croplands that are not detected in the common satellite fires. As control measures get more stringent, spatial-temporal patterns as well as the scales of biomass burning activities in China have undergone dramatic changes. Considering the crucial role of satellite fires in estimating biomass burning emissions, it is necessary to improve satellite fire detection with more advanced observations and retrieval methods.