Jornal de Sensoriamento Remoto e GIS

Abstrato

Accuracy Assessment of Spatial Interpolation Methods to Derives DEMs of Small Islands with Relative Topographic Variations

Hameid NA*, Bannari A, Kadhem G, Abdelhadi A

Responding to Climate Change in Small Island Developing States (SIDS), accurate Digital Elevation Model (DEM) can support the Sea level Rise (SLR) scenarios and sequenceit is impacts on coastal zone for proper adaptation. The DEM accuracy may vary to a certain degree following different interpolation algorithms and the data acquisition method. Indeed, numerous mathematical interpolation methods have been developed on spatial interpolation for topographic information densification and DEM restitution. The aim of this study focuses on the accuracy assessment of high spatial resolution DEM (at 2.5 m pixel size) regenerated from high topographic contour lines map at scale of 1:5,000 applying four different interpolation algorithms. Three deterministic methods were considered including the IDW with variable and fixed parameters, the Spline with regular and tension conditions, and the Natural Neighbor. While, for stochastic methods, the ordinary and simple Kriging were analyzed according to the semi-variogram adjustment considering five mathematical functions: Stable, Circular, Spherical, Exponential and Gaussian. For validation purposes, a datasets of 400 ground control points (GCPs) uniformly distributed over the study site, to cover all the existing altitude classes, were used. These were measured using Differential Global Position System (DGPS) with ± 1 cm and ± 2 cm for planimetric and altimetric accuracies, respectively. The results obtained show that ordinary and simple kriging methods, based on the exponential function, achieved a similar DEMs restitution with the best RMSE (± 0.65 m), which proved to be less than the tolerance or the total deviation (± 0.78 m). Consequently, these two Kriging methods are more accurate for DEM production for small island applications such as the evaluation of coastal zones vulnerability to SLR, flooding, detection of topographic features and hydrological modeling.