Measuring the contribution of built-settlement data to global population mapping

Area of Science:

• Geospatial analysis
• Population dynamics
• Environmental modeling

Background:

• Top-down population modeling is crucial for global public health, planning, and sustainability.
• Remote-sensing (RS) derived built-environment data are key covariates in these models.
• Gaps in spatial and temporal coverage of RS data necessitate interpolation, but its utility is understudied.

Purpose of the Study:

• To determine the utility of modeled built-settlement extents in top-down population modeling.
• To assess the value of interpolated settlement data for population density prediction.
• To evaluate the effectiveness of modeled settlement data compared to existing RS data.

Main Methods:

• Utilized modeled global built-settlement extents (2000-2012) from spatio-temporal disaggregation of settlement growth.
• Applied random forest-informed dasymetric disaggregations for annual population modeling.
• Analyzed data across 172 countries over a 13-year period.

Main Results:

• Modeled built-settlement data consistently ranked as the second most important covariate for predicting population density.
• Annual lights at night were the primary covariate, followed by modeled settlement data.
• Modeled settlement data often provided more predictive information than annually available RS data and last observed settlement extents.

Conclusions:

• Modeled built-settlement extents are a valuable and informative covariate in global population modeling.
• Interpolated settlement data can effectively fill spatial and temporal gaps in RS data for population studies.
• This approach enhances the accuracy and utility of population models for various applications.