A Process Oriented Areal Interpolation Technique: A Coastal County Example
The Modifiable Areal Unit Problem (MAUP) is the classic term for describing different totals observed from spatially different aggregation units. In a typical analytical problem (e.g. estimating total population within a watershed from census unit totals) the spatial distribution of populations within the census units are modeled. To minimize MAUP errors, areal interpolation techniques are used to model such sub-unit population distributions. Areal interpolation techniques are highly dependent on ancillary data (e.g. land use/cover data) and typically do not include "intelligent" relations about where people choose to live, other than a weighted association between nominal land cover/use and population density The purpose of this research was to design and implement an "intelligent" areal interpolation method for housing data in coastal environments, validate the accuracy, and compare to other techniques. This study was conducted for Miami-Dade County in Florida at census scales from county to block. Parcel boundary data was used as a reference layer to validate each technique. Not surprisingly, all techniques perform best at finer spatial resolutions (e.g. block level) with error increasing at coarser resolutions. The accuracy of the dasymetric technique is directly related to the accuracy of ancillary data. The new intelligent technique, (referred to as the process-oriented technique from here onwards) models the relationship between housing unit density distribution and proximity to the coast. This process-oriented technique performed better than the areal weighting and the dasymetric mapping technique. Combining the 'process-oriented' technique with a dasymetric technique provided the least amount of error.