Researchers from the Aerospace Information Research Institute (AIR) of the Chinese Academy of Sciences utilized the Sustainable Development Goals Science Satellite 1 (SDGSAT-1) glimmer imagery and deep learning techniques to produce a high-resolution population map of the Guangdong-Hong Kong-Macao Greater Bay Area. This study, published in International Journal of Digital Earth, offers valuable insights for urban planning, disaster preparedness, and sustainable development. 

Accurately understanding population distribution is crucial for effective planning and resource allocation. Traditional mapping methods often lack the resolution required for detailed applications in high-density urban areas. This study fills this gap by utilizing SDGSAT-1's 40-meter spatial resolution data, combined with a new deep learning model called FinePop-net, to improve population mapping accuracy.

The researchers employed FinePop-net to work with SDGSAT-1 data, achieving higher spectral resolution and accuracy compared to traditional nighttime light (NTL) sources. This method also reduced the average error rates by over 30% compared to standard NTL-based population datasets. These results showed the potential of SDGSAT-1 data to achieve high-resolution population mapping.

The high-resolution population map generated by this study provided a detailed view of the Greater Bay Area at a 40-meter resolution. This granularity enabled targeted analysis for essential applications, such as optimizing the placement of healthcare facilities and planning efficient emergency responses.

"SDGSAT-1's glimmer imagery allows us to view population distribution patterns at an unprecedented resolution," said Dr. DUAN Haoxuan, first author of this study. 

This study, by integrating high-resolution imagery with deep learning techniques, underscores the potential of Earth observation for improving decision-making in urban planning and resource management globally.