Validation of Pleiades Tri-Stereo DSM in Urban Areas

Emmanouil Panagiotakis,Dimitris Poursanidis,Vasiliki Charalampopoulou,Nektarios Chrysoulakis

doi:10.3390/ijgi7030118

Abstract

We present an accurate digital surface model (DSM) derived from high-resolution Pleiades-1B 0.5 m panchromatic tri-stereo images, covering an area of 400 km2 over the Athens Metropolitan Area. Remote sensing and photogrammetry tools were applied, resulting in a 1 m × 1 m posting DSM over the study area. The accuracy of the produced DSM was evaluated against measured elevations by a differential Global Positioning System (d-GPS) and a reference DSM provided by the National Cadaster and Mapping Agency S.A. Different combinations of stereo and tri-stereo images were used and tested on the quality of the produced DSM. Results revealed that the DSM produced by the tri-stereo analysis has a root mean square error (RMSE) of 1.17 m in elevation, which lies within the best reported in the literature. On the other hand, DSMs derived by standard analysis of stereo-pairs from the same sensor were found to perform worse. Line profile data showed similar patterns between the reference and produced DSM. Pleiades tri-stereo high-quality DSM products have the necessary accuracy to support applications in the domains of urban planning, including climate change mitigation and adaptation, hydrological modelling, and natural hazards, being an important input for simulation models and morphological analysis at local scales.

Highlights

Satellite remote sensing (optical and synthetic aperture radar (SAR)) and airborne Light Detection and Ranging (LiDAR) technologies have been exploited for their capabilities on the production of large area high resolution digital surface model (DSM) applied in urban studies [1,2,3,4,5], hydrological modelling [6], and natural hazards [7,8,9]
While point-based accuracy is the standard tool for the assessment of a DSM, line profiles can provide a different view on the structure of a produced DSM when compared to a reference dataset [18,19]
Ten evenly-distributed GCPs across the study area have been used to produce the DSM, which reflect a sufficient number for the total area of 400 km2

Summary

Introduction

Satellite remote sensing (optical and synthetic aperture radar (SAR)) and airborne Light Detection and Ranging (LiDAR) technologies have been exploited for their capabilities on the production of large area high resolution DSM applied in urban studies [1,2,3,4,5], hydrological modelling [6], and natural hazards [7,8,9]. The Pleiades-1 mission has attracted attention due to its unique tri-stereo image acquisition providing almost simultaneous images from three different views (fb = forward-backward, fn = forward-nadir, Figure 1) for the same area, with a stereo angle varying from ~6◦ to ~28◦ with the same spatial resolution This configuration, along with the use of the near-nadir image, allows a better retrieval of heights over terrains where the performance of classic photogrammetry with forward–backward looking only stereo pairs is limited [12]. By using 15 control points extracted by a LiDAR-based DSM, they achieved an RMSEz of 0.75, while Perko et al [14] following a similar approach over Innsbruck city, Austria

Objectives

Methods

Results

Conclusion