Estimation of Base-Year Floor Space Data

Abstract

Base-year floor space data are essential for integrated land use transport models. This information has traditionally been estimated on the basis of limited population or employment data provided by the census, with unknown accuracy. Advances in geographic information systems (GIS) and remote sensing technology enable the accuracy of this conventional method to be evaluated with geographic data of high precision. This study assessed the accuracy of the census-based approach in estimating base-year floor space data; the paper describes the use of a hybrid method that combines lidar and geographic vector data (referred to as the “lidar-GIS method”) to develop a ground truth building floor space database for the City of Fredericton, New Brunswick, Canada. A ground survey proved that the developed ground truth building data set provided highly accurate estimates of building heights, footprint areas, and therefore floor space. Through a case study, the floor space at the dissemination area level estimated with the traditional census-based method was compared with the floor space estimated with the lidar-GIS method for various land use categories. The results showed that for the residential floor space estimation, the average absolute percentage error of the census-based approach was 15%. The accuracy of the conventional method was much lower for nonresidential land use categories, with average errors of 50% or higher. These statistics indicate that the traditional census-based approach is unreliable and inaccurate for use in the estimation of base-year floor space and suggests that a method that incorporates the above-mentioned modern information technologies be used.