抄録
This paper analyzes the accuracy of count data transferred through the areal weighting interpolation with respect to the geometrical properties of zonal systems used for aggregating spatial data. A stochastic model is employed to measure the estimation error caused in data transfer between incompatible zonal systems. The relationship between estimation error and the geometrical properties of zones is approximately represented in analytical forms. The major results are as follows: 1) the perimeter of the target zone and the area and perimeter of the source zones are crucial to the accuracy of the areal weighting interpolation; 2) estimation error increases in proportion to the square root of the perimeter of the target zone; 3) concerning the lattice system, estimation error is proportional to the square root of the perimeter and the biquadratic root of the area of the cell, and inversely proportional to the biquadratic root of the number of cells; 4) the hexagonal lattice yields the most accurate estimates among all lattices.
