Digital surface models from ZiYuan-3 triplet: performance evaluation and accuracy assessment

Abstract

At the beginning of 2012, a new optical satellite, called ZiYuan-3 (ZY-3), was launched from the Taiyuan Satellite Launch Centre. This article is aimed at the analysis and the assessment of the digital surface models (DSMs) and orthophotos from a ZY-3 triplet acquired over the city of Bolzano and the surrounding areas. The imagery processing chain was executed with four different software packages. In all cases the Rational Polynomial Functions (RPFs) orientation model was used, with rational polynomial coefficients (RPCs) supplied by the vendors. The accuracies of the extracted DSMs were compared with the public light detection and ranging (lidar) DSM. The DSM extracted with each software package was used to generate the corresponding orthophoto. The orthophoto accuracy was assessed against 1:5000/1:10000 maps. The results obtained show that ZY-3 imagery allow us to generate DSMs with an accuracy, in terms of root mean square error (RMSE), which can reach 1.5–4.2 m in the flat area, 3.9–6.9 in the urban area, and 5.7–8.2 m in the mountain area. These results are better than Shuttle Radar Topography Mission (SRTM). Regarding orthophoto generation, all software packages supply similar results with a significant bias of about 20 m, mainly caused by vendors’ RPC inaccuracy.