The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLVIII-4/W8-2023
https://doi.org/10.5194/isprs-archives-XLVIII-4-W8-2023-453-2024
https://doi.org/10.5194/isprs-archives-XLVIII-4-W8-2023-453-2024
25 Apr 2024
 | 25 Apr 2024

BUILDING-LEVEL POPULATION ESTIMATION USING LIDAR-DERIVED BUILDING VOLUME DATA

K. A. Vergara

Keywords: population estimation, population, building volume, census, geographic information systems

Abstract. Population censuses serve as pivotal repositories for demographic and socioeconomic information. Conducted quinquennially in the Philippines, these censuses aggregate population data into administrative units, with the barangay as the smallest unit. However, this aggregation, when used in analyses, often presumes homogeneity within these units, potentially leading to overgeneralized results. With the advent of micro-level data, such as building-specific population counts, more nuanced spatial analyses become feasible. This study leveraged existing mathematical model to estimate residential building populations in Quezon City, utilizing 3D building information derived from elevation models, building footprints, local regulations, and land use types. The study yielded promising results, achieving a normalized absolute error (NAE) of 0.133 and an R2 value of approximately 0.976, indicating a high degree of model accuracy. However, the model also revealed systematic biases, notably underestimating populations in high-density areas and overestimating in low-density barangays. These findings underscore the complexity of factors influencing the model's performance, ranging from building digitization errors to assumptions about floor height and living area per person. The study thereby elucidates the valuable role that 3D building information can play in disaggregating population data for more granular analyses, while also highlighting areas for further model refinement to mitigate issues of overestimation and underestimation.