Hybrid Areal Interpolation of Census Counts from 2000 Blocks to 2010 Geographies

Area of Science:

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

Background:

• Census unit boundaries change over time, complicating population change analysis.
• Areal interpolation is a common method to estimate data across different geographic units.
• Existing methods may lack accuracy when census boundaries are not aligned.

Purpose of the Study:

• To assess various areal interpolation models for estimating 2000 census data in 2010 census units.
• To identify the most accurate model for measuring population changes across non-aligned census boundaries.
• To provide publicly accessible demographic estimates for research and planning.

Main Methods:

• Tested 8 binary dasymetric (BD) and 8 target-density weighting (TDW) models using ancillary data (densities, imperviousness, roads, water bodies).
• Developed 2 hybrid models combining the best-performing BD and TDW approaches.
• Evaluated model accuracy for estimating 2000 population characteristics within 2010 census units.

Main Results:

• A hybrid model, weighting TDW and BD methods based on the rate of change, demonstrated the highest accuracy.
• While most estimates showed minor differences from simpler models, significant variations occurred in certain areas.
• The final model provides estimates with lower and upper bounds for over 1,000 characteristics.

Conclusions:

• The developed hybrid areal interpolation model offers a more accurate approach to estimating population changes across shifting census boundaries.
• The model's flexibility in weighting different interpolation techniques enhances its applicability.
• Publicly available estimates via NHGIS facilitate historical demographic research and spatial analysis.