Gridded datasets for Japan: total, male, and female populations from 2001-2020

Japan is one of the highly urbanized and severely aging society. In an aging society, chronic disease and disability are prevalent, and the population is sensitive to environmental issues and climate change. To detect the impacts of population changes, formulate the population and public health policies, and assist environmental applications, the high-resolution and accurate gridded population dataset is strongly desired. To provide basic data for these studies, we create an open access annual dataset containing the total, male, and female population counts in each grid at a 500-m resolution from 2001 to 2020. The yearly population dataset is based on the 4th-level mesh data from the Statistics Bureau of Japan to make it easy to use. The dataset is provided here alongside the descriptions of the data and methods used in the fitting, cross-validation, and prediction processes. The total population data are stored in 20 attributes, named the “X0000” style. The “0000” of “X0000” represents the four-digit year. For example, the attribute “X2001” reports the total 2001 population in the mesh. The dataset also preserves the direct output from the random forest and the logarithms of the total population. The attributes of the logarithms of the total population in each year are named in “X0000_log” style. The “0000” of “X0000_log” also stands for the four-digit year. The attributes of the female population, the logarithm of the female population, the male population, and the logarithm of the male population are written as “X0000_fema”, “X0000_fe_l”, “X0000_male”, and “X0000_ma_l”, respectively. The accuracy of the models constructed to predict the total, male, and female populations are 92.09%, 91.92%, and 92.16%, respectively.