Population Estimation Using a 3D City Model: A Multi-Scale Country-Wide Study in the Netherlands

Area of Science:

• Geographic Information Science
• Demography
• Spatial Analysis

Background:

• Remote population estimation is crucial for demography, traditionally relying on 2D geographic data.
• Increasing availability of semantic 3D city models presents new opportunities for population estimation.
• Existing 2D methods often use maps and satellite imagery, avoiding extensive fieldwork.

Purpose of the Study:

• To investigate the utility of semantic 3D city models for remote population estimation.
• To compare the effectiveness of 3D city models against traditional 2D methods.
• To evaluate population estimation accuracy at various administrative levels using 3D data.

Main Methods:

• Utilized two methods: disaggregation (areal interpolation) and statistical modeling with 3D city models.
• Assumed housing space (volume) as a proxy for resident population.
• Employed a Dutch census dataset and a 3D model of 9.9 million buildings for analysis.

Main Results:

• 3D city models provide accurate population estimates for large geographical areas (e.g., countries).
• The 3D volume-based estimation method demonstrated clear advantages over 2D methods (footprints, floor areas).
• Estimation quality varied with administrative levels, with 3D models proving effective for broader scales.

Conclusions:

• Semantic 3D city models are valuable tools for large-scale remote population estimation.
• The 3D approach offers superior accuracy and advantages compared to 2D methods in population studies.
• Further research can explore the impact of varying semantic detail in 3D models on estimation accuracy.