On the Evaluation of the Neural Network Khartoum Geoid Model

(0 User reviews)   238   122
By Research Archive Posted on Jan 4, 2023
In Category - Research
Kamal Abdellatif Sami, Ammar Mohammed Maryod Aborida American Journal of Science, Engineering and Technology(Science Publishing Group) 2022
English Sudan
Abstract
This study was carried out to establish and evaluate an Artificial Neural Networks (ANN) geoid model for the Khartoum State. In the first stage the geometrical geoid heights were obtained from the differences between observed ellipsoidal heights and known orthometric heights of 48 geodetic Ground Control Points (GCP) in the study area. This followed by generating an ANN geoid model to extract the geoid heights from 42 ground control stations in the same study area in the Khartoum State. The main objective of this research study is to apply an ANN to model the Geoid surface using the back propagation algorithm in Khartoum state, through supervised training by geoidal undulations values. The WGS84 GPS/levelling geoid is computed then their results were used for comparison and evaluation of the determined ANN Geoid surface. In this study the geometrical geoid model was determined using the well-known geometrical geoid determination approach taking consideration of the distribution of the existing vertical control points in Khartoum area, with an intention of determining the orthometric heights of any point of unknown heights with uncertainties of less than 5cm. The ANN geoid uncertainties were evaluated and tested at 6 geodetic ground control points. The average difference between the derived geoid heights obtained from the geometrical geoid model, and their corresponding ANN geoid heights was found to be in the range of ±3 cm. Based on the test results of the statistical analysis and the study of a trained artificial neural networks model, the authors were able to estimate the geoid model with acceptable accuracy and can interactively be available for end users. This study showed that, the geoid heights in Khartoum State can be determined with the ANN method with typical accuracy of better than 5cm.
GPS, GNSS, Geoid, WGS84, UTM, Ellipsoidal Heights, Orthometric Heights

This study was carried out to establish and evaluate an Artificial Neural Networks (ANN) geoid model for the Khartoum State. In the first stage the geometrical geoid heights were obtained from the differences between observed ellipsoidal heights and known orthometric heights of 48 geodetic Ground Control Points (GCP) in the study area. This followed by generating an ANN geoid model to extract the geoid heights from 42 ground control stations in the same study area in the Khartoum State. The main objective of this research study is to apply an ANN to model the Geoid surface using the back propagation algorithm in Khartoum state, through supervised training by geoidal undulations values. The WGS84 GPS/levelling geoid is computed then their results were used for comparison and evaluation of the determined ANN Geoid surface. In this study the geometrical geoid model was determined using the well-known geometrical geoid determination approach taking consideration of the distribution of the existing vertical control points in Khartoum area, with an intention of determining the orthometric heights of any point of unknown heights with uncertainties of less than 5cm. The ANN geoid uncertainties were evaluated and tested at 6 geodetic ground control points. The average difference between the derived geoid heights obtained from the geometrical geoid model, and their corresponding ANN geoid heights was found to be in the range of ±3 cm. Based on the test results of the statistical analysis and the study of a trained artificial neural networks model, the authors were able to estimate the geoid model with acceptable accuracy and can interactively be available for end users. This study showed that, the geoid heights in Khartoum State can be determined with the ANN method with typical accuracy of better than 5cm.

There are no reviews for this PDF.

0
0 out of 5 (0 User reviews )

Add a Review

Your Rating *
There are no comments for this PDF.
You must log in to post a comment.
Log in

Related PDF