Evaluating the Impact of Sampling Pattern on the Validity of Derived DEMs: Wadi Yiba Basin as an Experimental Environment

Abstract

The study aims to know the effect of the sample pattern on the validity of the DEM, relying on the analytical inductive approach through two stages. The first begins with deriving the DEMs via the Spline tool, and with a fixed sample size for the three samplings: (random, random with added 50m between the samples, and systematic), whereas the second depends on the evaluation both visually (hillshade and contour) and quantitatively (calculating the vertical accuracy based on the RMSE of the derived surface, and finding the differences between the surfaces of the three samples) of the derived models based on a reference of DEM generated from Spot 5 with clarity of 2.5m. The study concluded that the validity of the DEM is affected by the sample distribution pattern, as the model extracted by the random sampling with an added distance of 50m between samples achieved the highest proximity to the reference values compared to the DEMs derived by the other samples. Through the hillshade and contour, both balance and consistency appeared in the distribution of surface details, undulations, and texture. It also recorded the lowest value for the probability of error, achieving a vertical accuracy of 3.66m at a CI of 95%.

Keywords: Spatial Interpolation, Systematic Sampling, Random Sampling, Sample pattern, Vertical Accuracy.