Data-enriched Interpolation for Temporally Consistent Population Compositions

Area of Science:

• Geographic Information Systems (GIS)
• Demography
• Spatial Analysis

Background:

• Temporal incompatibility of census geographies hinders micro-scale population studies.
• Accurate multi-temporal population estimates are crucial for understanding demographic shifts.
• Existing methods struggle with consistent small-area population data over time.

Purpose of the Study:

• To evaluate areal interpolation combined with dasymetric refinement for demographic estimation.
• To assess the effectiveness of ancillary variables (GHSL, NLCD, buildings, ZTRAX) in improving population estimates.
• To create consistent, multi-temporal small-area population estimates for Massachusetts (1990-2010).

Main Methods:

• Areal interpolation techniques (Areal Weighting, Target Density Weighting, Expectation Maximization) were employed.
• Dasymetric refinement utilized ancillary datasets like Global Human Settlement Layer (GHSL) and building footprints.
• Comparison of methods across different time spans (1990-2010 vs. 2000-2010) and population groups.

Main Results:

• Dasymetrically refined areal interpolation significantly improves accuracy, especially over longer periods (1990-2010).
• Target Density Weighting (TDW) refined with building footprints or ZTRAX data yielded superior results.
• Current census urban areas overestimate urban population; refined methods provide more accurate spatial distribution.

Conclusions:

• Areal interpolation with dasymetric refinement offers a robust strategy for reliable multi-temporal population estimates.
• This methodology enhances micro-scale modeling of subpopulations, particularly urban populations, aiding urbanization studies.
• The approach provides a foundational method for advancing demographic composition analysis and future population projections.