Fine-grained gridded population products could provide essential support for scientific research as well as the designation of government policies. However, for developing countries like China, most of these products are validated using township-level data which are much larger than the grid cells and therefore hinters the credibility of their applications at fine-grained levels. Here, we develop a bottom-up mapping approach and use the micro population distribution data of mobile phones to map the 1km gridded population distribution in 36 major Chinese cities. Our product is compared with existing products using the township-level census data as well as anonymous user distribution data of mobile applications which enables product comparisons at finer resolutions from 0.01 arc-degree to 0.05 arc-degree. The comparisons reveal that our products are more accurate than the state of art grided population estimates across different resolutions. This study provides a more accurate gridded population dataset for large Chinese cities and also sheds new light on techniques for mapping and validating population distribution at the fine-grained level.
The code is used to replicate the work "Mapping fine-grained population distribution in 36 major Chinese cities with microdata".