Serviços Personalizados
Journal
Artigo
Indicadores
- Citado por SciELO
Links relacionados
- Similares em SciELO
Compartilhar
Geoacta
versão On-line ISSN 1852-7744
Resumo
CARRUPT MACHADO, Wagner; OLIVEIRA CANCORO DE MATOS, Ana Cristina; BLITZKOW, Denizar e DO NASCIMENTO GUIMARAES, Gabriel. The use of neural network for gravity anomalies interpolation - application to geoid model computation in Santa Catarina - Brasil. Geoacta [online]. 2016, vol.41, n.2, pp.32-43. ISSN 1852-7744.
To assess the impact of using gravity anomalies interpolated with Artificial Neural Networks (ANN) over geoid model compuation, this technique was used to interpolate Bouguer, free-air and Helmert gravity anomalies into a 5’ regular grid to compute the Santa Catarina State geoidal model. Three geoidal models were computed using the grids determined with ANN. The GEOIDE2014, computed by LTG (Geodesy and Topography Laboratory), was used for comparison. Therefore, the four geoidal models were assessed using information, provided by IBGE (Brazilian Institute of Geography and Statistics), of 53 leveling stations in which the geodetic altitude were determined with GNSS observations, called in this paper as GNSS/RN. The 53 geoidal heights were obtained from the difference between geodetic height and normal-orthometric height from the geometric (spirit) levelling. Such comparison has the aim of identify which models present better consistence with these points. It was obtained RMS of 0.17, 0.13, 0.17 and 0.17 m, where the geoidal model computed with the grid of Helmert anomaly interpolated with ANN presented the best result.
Palavras-chave : Neural Network; Geoid; GNSS; Spirit levelling; Height.