Total electron content prediction using singular spectrum analysis and autoregressive moving average approach.

Continuous monitoring of ionospheric behavior and subsequent development or improvement of models for the prediction of its parameters with consistent accuracy remains an ongoing challenge. In this sense, an integrated approach by combining the signal extraction technique Singular Spectrum Analysis (SSA) with Autoregressive Moving Average (ARMA) is presented in this work to predict the ionospheric Total Electron Content (TEC) values that are responsible for causing ionospheric delays in the trans-ionospheric signal propagation associated with satellite-based communication, navigation, and timing applications. In general, SSA is a nonparametric spectral estimation procedure that decomposes the signals into interpretable and physically significant components. The observed TEC from two Global Positioning System (GPS) stations across the low latitude Saudi Arabian region are considered during the year 2017 that falls in the descending phase of solar cycle-24. The performance of the proposed hybrid model is evaluated by comparing with the sole estimation from the ARMA model and the observed GPS–TEC dataset for two different geomagnetic conditions: a) the regular geomagnetically quiet period of 15 to 29 December, 2017 (Ap < 24 and Dst > − 30 nT) and b) the geomagnetic storm period from 7 to 9 September, 2017 (Dst min = − 142 nT). The corresponding average Precision, Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) of the proposed SSA–ARMA model predictions are 1.79 TECU, 1.23 TECU, and 13.02%. In contrast, the respective values in the exclusive ARMA model are 2.01 TECU, 1.37 TECU, 14.42% at Oman station. The corresponding values for Magna station are 0.92 TECU, 0.61 TECU, and 10.76% (SSA–ARMA) and 1.01 TECU, 0.75 TECU, and 11.33% (ARMA). The results show an improved computational efficiency with minor improvement in the TEC predictions with the proposed SSA–ARMA method compared to the sole employment of the ARMA model by disregarding the extraneous components. [ABSTRACT FROM AUTHOR]