Abstract: In economic, social and urban studies, areal units under analysis frequently differ from areal units over which data are compiled. Most area-based analyses, hence face an unavoidable problem of transferring data across different zonal systems, that is, statistical population data requiring a spatial transformation. This is so called areal interpolation. This paper studies population interpolation based on GIS. A areal interpolation with population real distribution is pointed out based on traditional areal interpolation by partitioning the zone into "occupied" and unoccupied'classes. The new formula is deducde. A population density algorithm is developed based on recursion as well. (1) The study region is partionde into two subregions-a sparsely populated region and a more densely populated one. (2) Through estimation the population density of sparsely populated region, the population density of a more densely populated region can be calculated. (3) The more densely populated region is divided into two subregions- a new sparsely populated region and a new more densely populated one if necessary. This process can take many iteration until the real population distribution model is obtained.

Key words: population density, areal interpolation, recursion algorithm