Modelamiento de la serie temporal del valor de agua precipitable usando una red neuronal artificial
- Authors
- Romero, Ricardo; Pilapanta, Christian
- Format
- Article
- Status
- publishedVersion
- Description
Proyecto PIC-13-IGM-002: Determinaci?n del marco de referencia geod?sico
The Precipitable Water Vapor (PWV) data is fundamental physical parameters in climatology, meteorology, cIiriifite change, besides other fields. When such series are able to be modeled with high precision, a better prediction is given. Our study propose to train an Artificial Neural Network (ANN) of Radial Basis Functions (RBF) to predicting PWV. The time series data were obtained from the continuous operation station EPEe and MET. The EPEe belongs at SIRGAS ( SIstema de Referencia para las AmericaS) Network, and in the MET the temperature, pressure and humidity are obtained. The GAMIT/GLOBK software was used in the data processing, and the PWV was calculated each two hours from GPS week 1798 until 1825. The training neural network process, the time (in hours) and the PWV were used in the input layer and output layer respectively. Test points were used in the ANN trained to make the PWV prediction and to evaluate the learning of the net. The differences between known PWV and the predicted PWV with ANN--RBF were calculated. The results present a mean arithmetic of 0.1 mm, a standard deviation of 2.1 mm, and correlation coefficient of 0.92.
- Publication Year
- 2015
- Language
- spa
- Topic
- CLIMATOLOG?A
METEOROLOG?A
RED NEURONAL ARTIFICIAL
FUNCIONES DE BASE RADIAL
MET
PWV
- Repository
- Repositorio SENESCYT
- Rights
- openAccess
- License
- openAccess