Premium
Applying artificial neural network to the short‐term prediction of electron density structure using GPS occultation data
Author(s) -
Zeng Zhen,
Hu Xiong,
Zhang Xunjie
Publication year - 2002
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1029/2001gl013656
Subject(s) - global positioning system , artificial neural network , occultation , feedforward neural network , longitude , ionosphere , total electron content , remote sensing , meteorology , radio occultation , space weather , backpropagation , tec , universal time , computer science , feed forward , geodesy , latitude , artificial intelligence , geology , geography , physics , engineering , geophysics , telecommunications , astronomy , control engineering
Artificial neural network (ANN) is used for assimilating of GPS ionospheric occulted data in order to take full advantage of the abundant GPS occulted data. A feedforward, full‐connected network is chosen based on the back‐propagation algorithm. Universal time, latitude, longitude, height, Kp index, and F 10.7 solar flux are chosen as the input vectors of the network while the electron density as the output vectors. The GPS occultation data on May 24th, 1996 were taken as training samples to train an ANN, and then the well‐trained ANN was used to predict the electron density on 25th. Comparison of the predicted results and observed data demonstrated that ANN is a promising method in assimilating the GPS occulted data to establish the ionospheric weather prediction model. Furthermore, the accurate and abundant observations are essential for ensuring the good performance of ANN.