Skip to main content
1 × 1 km grid cells with population data

Dasymetric Population Mapping Using Building Data

Citation: Pirowski, Tomasz, and Bartłomiej Szypuła. Dasymetric Population Mapping Using Building Data. Annals of the American Association of Geographers. 2024; 1–19.

Abstract: The goal of this research was a quantitative-spatial high-resolution analysis of population distribution based on residential building data extracted from topographic objects database. Attribute information on residential buildings (location, volume, function) provides opportunities to estimate the number of residents. The recalculation of the population from the urban units of Cracow into new spatial units was based on the area-weighted aggregation method. The location of residential buildings constituted a limiting variable, and the total square meterage (calculated as the area of the buildings and the number of their floors) constituted the binding variable. The introduction of additional binding variables related to the type of building and its location, as well as various methods of determining the square meterage per building type, resulted in the creation of a total of nineteen maps of population. As a result, the best methods for the correct geographic scale and segmentation of residential building type—single family or multifamily—were identified. For the input data, based solely on the amount of population in urban units, the calculated value of the mean absolute percentage error (MAPE) in the 1 × 1 km grid was 310.8 percent, and for the root mean square error (RMSE) was 1,476 people. In the dasymetric method, directly associating the population with the volume of residential buildings, the errors fell to 21.9 percent and 632 people, respectively. The best result was obtained for the variant based on minimizing the RMSE, associating the number of residents to single-family buildings (2.88 people/building) and associating the number of residents to the square footage in multifamily buildings (37.1 m2/person; MAPE = 19.2 percent, RMSE = 556 people).