Uncertainty Analysis of Areal Weighting Interpolation Using Simulated Census Boundaries
How to transfer data across incompatible zonal systems has been a big challenge as part of Modifiable Areal Unit Problem (MAUP) addressed by many researchers for years. There are plenty of areal interpolation methods discussed and compared by previous papers in estimation of variables across incompatible zonal systems. However, not many papers gave a close look on uncertainty analysis of the interpolation method. This study aims to evaluating the uncertainty in a simple but commonly used areal interpolation method—areal weighting method using Monte Carlo simulation. Source and target zone boundaries are randomly generated by aggregating 2010 Buffalo and Cheektowaga block group geography. Interpolation method is implemented using the data obtained from the 50 sets of randomly aggregated source zones to three sets of target zones with various sizes 24, 53 and 91 respectively. The uncertainty analysis shows that as the number of target zone increases, the accuracy of areal weighting method decreases. Furthermore, the areal weighting errors for each target zones are related with the percentage of populated area as well as some geometric parameters such as perimeter and area of the target zone.