Prediction of TEC at low latitudes station using Neural Network model
- Resource Type
- Conference
- Authors
- Akir, Rohaida Mat; Chellappan, Kalaivani; Md Yusop, Noraide; Abdullah, Mardina; Homam, Mariyam Jamilah
- Source
- 2021 7th International Conference on Space Science and Communication (IconSpace) Space Science and Communication (IconSpace), 2021 7th International Conference on. :94-98 Nov, 2021
- Subject
- Aerospace
Communication, Networking and Broadcast Technologies
Fields, Waves and Electromagnetics
Geoscience
Atmospheric modeling
Artificial neural networks
Predictive models
Prediction algorithms
Data models
Reliability
Root mean square
total electron content
forecasting
neuralnetwork
RMSE
IRI01-corr
- Language
- ISSN
- 2165-431X
The total electron content (TEC) of the ionospheric atmosphere is frequently used in studies on ionospheric physics and the effects of space weather. This paper focuses on a feasibility study of the Neural Network (NN) model for the TEC prediction in Peninsular Malaysia which is included in low latitude region. The selection data is based on available data from Universiti Kebangsaan Malaysia (UKM) station from the year 2011 until 2015 which was the ascending solar cycle period in cycle 24. A feedforward neural network with a back propagation algorithm was proposed as a prediction model. The NN parameters were selected based on the TEC influence’s factor which were the seasonal variation, diurnal variation, and solar activity. The TEC value predicted by the NN model was evaluated using the GPS-TEC and compared to IRI01-corr TEC models result. The result shows the proposed NN model give less root mean square error (RMSE) against the IRI01-corr TEC models with the average RMSE approximately 3.58 TECU and 14.72 TECU respectively. In addition, the best prediction falls during the solstice compared to the equinox.