Evaluating two numerical methods for developing a local geoid model and a local digital elevation model for the Red Sea Coast, Egypt

Abstract

Global Navigation Satellite System (GNSS) provides ellipsoidal heights, whereas engineering projects need orthometric heights of points. Thus, to take advantage of GNSS in engineering work, the relation between orthometric and ellipsoid height must be determined as a function of position by geoid modeling. The Red Sea coastal cities have various natural potentials that render them economically promising cities that are attractive to numerous investors and visitors alike. Over the past years, there has been an increased governmental awareness of setting up large projects in these cities. With the increasing utilization of GNSS in a number of these projects, there is an imperative need to develop a local geoid model in this region. This study is an attempt to use a geometric method for the study area in order to develop a local geoid model and a local digital elevation model. In addition, the performance of EGM2008, SRTM, and ASTER is evaluated. In order to develop the local geoid model, the polynomial regression method and Artificial Neural network (ANN) technique are employed. Performance evaluation of the studied methods is based on a compilation of various statistical parameters and goodness-of-fit measures, followed by a comparison of methods outputs. The study indicates significant improvements in the local geoid model when using both methods. Additionally, the model resulting from the ANN appears more reliable compared to the calculated model using polynomial regression. In this study, the accuracy of EGM2008 is about ±0.466 m. Moreover, regarding the global digital elevation model, ASTER outperforms SRTM by about ±0.905 m.