Reconstructing continuous population density surface from polygon-based data

socioeconomic data such as population are collected and represented as attributes of spatial polygons (census tract). And this representation ignores real internal variance and leads a limitation of the usage of such data. This paper proposes an approach to reconstructing continuous density surface which describes internal variation of each polygon. The approach keeps three rules: 1) The surface is smooth without steps; 2) The volume of each polygon is preserved; 3) The distribution of derived population density coincides with the real population distribution approximately. And an assessment of the approach is given by comparing data got by reaggregation with the census data of some polygons on a smaller scale. The results show the applicability of the approach and an improvement of the precision of the surface. And the surface of the population density can be applied in further researches.