Estimating building height in China from ALOS AW3D30

Abstract

Building height plays an essential role in urban remote sensing and is of great significance for understanding the functioning of urban systems. However, there is little information about building height in China at the country-scale. Previous studies have employed a digital surface model (DSM) to map building height, but the slope correction for buildings in sloping areas remains understudied. In this context, we developed a method to estimate building height for all of China based on the Advanced Land Observing Satellite (ALOS) World 3D-30 m (AW3D30) DSM and other ancillary data including the Global Artificial Impervious Area (GAIA) dataset, the NASADEM dataset and the Global Roads Inventory Project (GRIP) dataset. The proposed method enabled us to accurately estimate building height with a special slope correction algorithm, improving the accuracy of building height estimation. The outcome of our procedure is a map of building height for China at a spatial resolution of 30 m. Compared to field-measured building height data and reference building height data from Baidu map, results indicate that the proposed method performed well (root mean square error (RMSE) of 4.26 m and 4.98 m, respectively). Moreover, the new building height product further demonstrates that our method alleviates the overestimation effect of buildings with small footprint area in the suburbs when compared with the existing building height product. Overall, the new building height map of China contributes to the improved management of urban areas and further studies of urban environments.