Global gridded population datasets systematically underrepresent rural population.

Area of Science:

• Geospatial analysis
• Demography
• Sustainable development studies

Background:

• Global gridded population data are crucial for sustainable development initiatives.
• Existing datasets are primarily calibrated for urban environments, leaving rural accuracy largely unexamined.

Purpose of the Study:

• To systematically validate the accuracy of global gridded population datasets specifically within rural areas.
• To assess the extent of underestimation in rural population figures provided by major datasets.

Main Methods:

• Utilized reported human resettlement data from 307 large dam construction projects across 35 countries as a ground truth.
• Compared resettlement figures against estimates from prominent global gridded population datasets (WorldPop, GWP, GRUMP, LandScan, GHS-POP).

Main Results:

• Identified substantial discrepancies among datasets, with consistent and significant negative biases.
• Observed underestimation ranging from -53% (WorldPop) to -84% (GHS-POP), indicating rural populations are often halved in estimates.
• Even the most accurate dataset underestimated rural populations by 50% compared to reported resettlement figures.

Conclusions:

• Current global gridded population datasets exhibit critical underestimation of rural populations.
• This bias necessitates a re-evaluation of past and future applications of these datasets to ensure equitable resource distribution.
• Improving rural population data accuracy requires enhanced census methods, alternative data sources, and refined population modeling calibration.