Taiwan Ionospheric Model (TWIM) prediction based on time series autoregressive analysis

As described in a previous paper, a three-dimensional ionospheric electron density (N_e) model has been constructed from vertical N_e profiles retrieved from the FormoSat3/Constellation Observing System for Meteorology, Ionosphere, and Climate GPS radio occultation measurements and worldwide ionosonde f_oF₂ and f_oE data and named the TaiWan Ionospheric Model (TWIM). The TWIM exhibits vertically fitted α-Chapman-type layers with distinct F₂, F₁, E, and D layers, and surface spherical harmonic approaches for the fitted layer parameters including peak density, peak density height, and scale height. To improve the TWIM into a real-time model, we have developed a time series autoregressive model to forecast short-term TWIM coefficients. The time series of TWIM coefficients are considered as realizations of stationary stochastic processes within a processing window of 30 days. These autocorrelation coefficients are used to derive the autoregressive parameters and then forecast the TWIM coefficients, based on the least squares method and Lagrange multiplier technique. The forecast root-mean-square relative TWIM coefficient errors are generally <30% for 1 day predictions. The forecast TWIM values of f_oE and f_oF₂ values are also compared and evaluated using worldwide ionosonde data.