Performance of artificial neural networks on kriging method in modeling local geoid

Abstract

Transformation of ellipsoidal heights determined by satellite techniques into local leveling heights requires geoid heights at points of interest. However, the geoid heights at each point are not available. In order to determine them, the local geoid in the transformation area must be modeled or computed by an appropriate method, one way of doing it, is to use control points both of whose ellipsoidal and local leveling heights are available. In this study, performance of geoid by ANN compared to Kriging method in modeling local geoid was presented. Moreover, the transformation ability of the methods was investigated through a geodetic test network in Bursa Metropolitan Area of Turkey. The results suggest that the model by ANN exhibit better results than the one by Kriging Method. Keywords: Geoid; Local Geoid Modeling; ANN; Kriging; Transformations. RESUMOA transformação das altitudes elipsoidais determinadas por técnicas satelitais em locais de altitudes niveladas requer o conhecimento da altitude geoidal no ponto de interesse. Entretanto, as altitudes geoidais de cada ponto nem sempre estão disponíveis. A fim de determiná-las, um geóide local de transformação deve ser determinado ou calculado através de métodos apropriados. Uma maneira de fazer isto, é usar pontos de controle de ambos cujas altitudes elipsoidais e niveladas estejam disponíveis. Neste estudo, apresenta-se os resultados da comparação entre o geóide obtido por ANN e o método da Krigagem na modelagem do geóide local. Akcin,H. ; Celik, C. T. Bol. Ciênc. Geod., sec. Artigos, Curitiba, v. 19, n o 1, p.84-97, jan-mar, 2013. 8 5 Além disso, a habilidade da transformação dos métodos foi investigada através de uma rede geodésica teste na área metropolitana de Bursa na Turquia. Os resultados sugerem que o modelo por ANN exibe melhores resultados que o método da Krigagem. Palavras-chave: Geóide; Modelamento do geóide local; ANN; Krigagem; Transformações.

Application of neural networks for the reliability design of a tunnel in karst rock mass

“…Much research was conducted on the comparison of Kriging and NN, to determine which surrogate model performs better in establishing complex correlation between parameters. Many of these suggest that the NN models are superior to the geostatistical Kriging model and exhibit higher accuracy, e.g., in geodesy (Akcin and Celik 2013); groundwater contamination (Chowdhury et al 2010); ionosphere mapping, especially when data set is spare (Jiang et al 2015); geotechnical site characterization (Samui and Sitharam 2010) or mapping of rock depth below soft deposits (Sitharam et al 2008 conditions prior to excavation when compared with the methods based on soft computing methods, while Santos et al (2015) concluded that model errors obtained with the different estimation methods (linear regression, geostatistical Kriging, and NN algorithms) are very similar. As the utilization of NN generated using evolutionary algorithms can be considered as an advanced surrogate model, compared to the traditionally used statistical and experimental methods, many researchers utilized its benefits in rock tunneling and underground rock engineering (Lee and Sterling 1992;Moon et al 1995;Benardos and Kaliampakos 2004;Yoo and Kim 2007;Mahdevari and Torabi 2012;Zhang and Goh 2015;Hasegawa et al 2019).…”

Geoid undulation prediction using Gaussian Processes Regression: A case study in a local region in Turkey

“…When the network is trained with a sufficient number of training data, predictions can be made based on previous learning. A few studies have used the ANN with a different network topology in the modeling of a local GNSS/leveling geoid surface (Lin, 2007;Akcin and Celik, 2013;Erol and Erol, 2013;Albayrak et al, 2020). For example, Kavzoglu and Saka (2003) predicted geoid undulation for Istanbul, Turkey using a feed-forward ANN learning algorithm for the application.…”

Using Particle Swarm Optimization to Establish a Local Geometric Geoid Model

“…Numerous studies have investigated the development of suitable local geoid undulation models by applying various fitting methods to measurements of ellipsoidal height h and leveling orthometric height H obtained using global positioning systems (GPSs) (Odera et al, 2014;Abdalla et al, 2011;Abdalla and Fairhead, 2011;Akcin and Celik, 2013;Abdalla, 2009;Featherstone et al, 2004;Kao et al, 2014 ;Kao, 2006;Kotsakis and Sideris, 1999;Kao and Bethel, 1992(a); Kao and Bethel, 1992(b) ;Ning, 2015;Lin, 2007;Tranes et al, 2007;Kiamehr, 2006;Ustun and Demirel, 2006;You, 2006;Yang and Chen, 1999). The geoid of each position in a location can be calculated to determine H for each position.…”

“…The geoid of each position in a location can be calculated to determine H for each position. According to a review of research related to the fitting of geoid undulation models, surface fitting has been performed with least squares collocation, neural networks, second-order curve-surface fitting, back-propagation neural networks (Akcin and Celik, 2013), multisurface function methods, genetic algorithms, least-squares support vector machines (Shen, 2011), and other artificial intelligence methods (Abdalla and Green, 2016;Abdalla and Elmahal, 2016;Daras, 2008;Ellmann, 2001;Ulotu, 2009). The fitting processes used in these geoid undulation modeling methods necessitate extensive computation and complex program development.…”