Comparative Accuracy Evaluation of Fine-Scale Global and Local Digital Surface Models: The Tshwane Case Study I

Abstract

ABSTRACT:Conducted in the City of Tshwane, South Africa, this study set about to test the accuracy of DSMs derived from different remotely sensed data locally. VHR digital mapping camera stereo-pairs, tri-stereo imagery collected by a Pléiades satellite and data detected from the Tandem-X InSAR satellite configuration were fundamental in the construction of seamless DSM products at different postings, namely 2 m, 4 m and 12 m. The three DSMs were sampled against independent control points originating from validated airborne LiDAR data. The reference surfaces were derived from the same dense point cloud at grid resolutions corresponding to those of the samples. The absolute and relative positional accuracies were computed using well-known DEM error metrics and accuracy statistics. Overall vertical accuracies were also assessed and compared across seven slope classes and nine primary land cover classes. Although all three DSMs displayed significantly more vertical errors where solid waterbodies, dense natural and/or alien woody vegetation and, in a lesser degree, urban residential areas with significant canopy cover were encountered, all three surpassed their expected positional accuracies overall.

Validación De Modelos De Elevación Digital en Los Andes Centrales De Chile: Análisis De La Exactitud Y Precision Vertical a Escala Regional Y Local Con Datos GNSS Y Laser

“…Esta ventaja se explicaría por los ángulos oclusos que alcanza el modo de adquisición tri-estereo en comparación con el modo StripMap de TanDEMx y el bi-estatico de SRTM-SAR C/X, y por la resolución espacial siempre menor de los DEM de este subgrupo. Esta constatación es coherente con las estimaciones de la precisión de DEM construidos con datos de sensores ópticos bi/tri-estereo de alta resolución espacial como SPOT5, CartoSat-1, Pléiades1-2-3, y WorldView1-2 en zonas montañosas (Alganci, Besol y Sertel, 2018;Reinartz et al, 2006;Breytenbach, 2016;Mukherjee et al, 2015). El DSM AW3D30 v21 es particularmente interesante pues su fuente son datos SAR y datos ópticos de alta resolución, lo que lo hace único pues en zonas de baja penetración por nubes o alta reprodispersión, capturó donde otros no lo hicieron.…”