Comparative Assessment of Satellite and Field Surveying Data Accuracy for Construction Projects in Iraq

Abstract

This study aims to evaluate whether freely available Sentinel-2 and Landsat-8 satellite imagery—when orthorectified using a local ground control point (GCP) network—can meet the positioning accuracy requirements of civil surveying for construction projects in Baghdad, Iraq, where economic sanctions and budgetary constraints preclude routine deployment of high-cost LiDAR systems. While satellite data offer broad coverage and low acquisition costs, their meter-scale geolocation errors often fall short of the centimeter-level precision required for legal boundary demarcation and structural design. To address this gap, we established a 25-point GCP network at the Corporation of Research and Industrial Development (CRID) site (59,600 m²) using TOPCON GR-5/GRS-1 GNSS and ES-105 total station, achieving a planimetric RMSE of 2.1 cm and vertical RMSE of 3.4 cm. These points were used to orthorectify Sentinel-2 and Landsat-8 imagery in ENVI 5.6 (FLAASH atmospheric correction, 2nd-order polynomial). Results show that GCP-constrained Sentinel-2 attained a horizontal RMSE of 6.8 m—an 85% improvement over raw data—while Landsat-8 stabilized at 12.4 m. Critically, this hybrid workflow reduced surveying costs by ~70% (from ~520 to 45 IQD/m²), without compromising boundary fidelity: the field-derived area (59,599.59 m²) matched official records (59,600 m²) within 0.4 m², whereas uncorrected satellite data overestimated the area by 499 m²,a discrepancy consistent with prior studies of urban cadastral applications. We conclude that while centimeter-grade field data remain essential for final design and legal documentation, GCP-enhanced Sentinel-2 offers a pragmatic, cost-efficient solution for preliminary planning, corridor mapping, and rapid cadastral updates in Iraqi urban settings.