Mapping Three-dimensional Urban Structure by Fusing Landsat and Global Elevation Data

TL;DR

This paper presents a novel method to map 3D urban structures at a country scale by fusing Landsat imagery and global elevation data, achieving accurate estimates of building height and volume.

Contribution

It introduces a new fusion-based approach using freely available data to characterize 3D urban structures at large scales, which was previously limited by data availability.

Findings

Achieved RMSE of 1.61 m for building height

Achieved RMSE of 1,142 m³ for building volume

First open data-based 3D urban structure mapping at country scale

Abstract

To meet the challenges of global urbanization, earth observation information is greatly needed. The lack of global three-dimensional (3D) urban structure data has been a major limiting factor in important urban applications such as population mapping, disaster vulnerability assessment, and climate change adaptation. Due to limited data availability, remote sensing data haven been mainly used to characterize 3D urban structure at the city scale. In this study, we propose a method to map 3D urban structure using freely available Landsat and global elevation data. Building on an object-based machine learning approach, the synergy of Landsat and elevation data were used to estimate building height and volume at 30 m resolution. This method has been tested for the entire country of England and yielded a root-mean-square error (RMSE) of 1.61 m for building height and an RMSE of 1,142 m\textsuperscript{3} for building volume. The results in this study represent the first attempt to use open data for urban structure characterization at the country scale. Our results demonstrated the utility of these data for large-scale urban studies and the potential of generating global 3D urban structure data products using a fusion-based approach.