The accuracy assessment in areal interpolation: An empirical investigation

Abstract

Areal interpolation is the process of transferring data from source zones to target zones. While method development remains a top research priority in areal interpolation, the accuracy assessment aspect also begs for attention. This paper reports an empirical experience on probing an areal interpolation method to highlight the power and potential pitfalls in accuracy assessment. A kriging-based interpolation algorithm is evaluated by several approaches. It is found that accuracy assessment is a powerful tool to understand an interpolation method, e.g. the utility of ancillary data and semi-variogram modeling in kriging in our case study. However, different assessment methods and spatial units on which assessment is conducted can lead to rather different results. The typical practice to assess accuracy at the source zone level may overestimate interpolation accuracy. Assessment at the target zone level is suggested as a supplement.