Georeferencing Multi-source Geospatial Data Using Multi-temporal TerraSAR-X Imagery: a Case Study in Qixing Farm, Northeast China

Geodata, including optical remote sensing (RS) images and topographic vector data, can be collected from multiple sources such as surveying and mapping agencies, commercial data acquisition companies, and local research institutes. These multi-source data have been widely used in past RS and geographic information system (GIS) studies in various applications. However, spatial inconsistencies inherent in the multi-source data require accurate georeferencing to be applied. This is challenging for study sites with limited accessibility and few reference maps. To address this challenge, this paper proposes an approach for generating ground control points (GCPs) using TerraSAR-X (TSX) data. In a case study, TSX images were used to georeference multi-source data covering the Qixing Farm in Northeast China. First, a stack of five multi-temporal TSX images were processed into one reference image to retrieve GCPs. These were then used to georeference the other datasets including Huanjing (HJ), Landsat 5 (LS 5), FORMOSAT-2 (FS-2), and RapidEye (RE) satellite images, as well as topographic vector datasets. Identifying tie points in the multi-source datasets and the corresponding GCPs in the TSX reference image enables georeferencing without field measurements. Finally the georeferencing accuracies for the optical RS images were assessed by using independent check points. Good results were obtained for the HJ, LS 5, FS-2 and RE images, with an absolute error of 7.15 m, 6.97 m, 8.94 m and 10.52 m, respectively. For the topographic vector datasets, ideal visual results were achieved, attributable to the rubber sheeting algorithm. These results demonstrate that the TSX reference image is suitable for georeferencing multi-source data accurately and cost-efficiently. The developed procedure can be applied in other study regions and is especially valuable for data-poor environments.