Comparison of TEC prediction methods in mid-latitudes with GIM maps

Abstract

There are many long-term and short-term prediction methods of Total Electron Content (TEC) that need to be tested for each specific region. Recently, much attention has been paid to testing TEC models in high-, low-latitude and equatorial regions. This paper compares the TEC prediction methods in the mid-latitude zone according to the data of the Juliusruh, Rostov, Manzhouli stations in 2008 and 2015. For a long-term prediction, the IRI-Plas and NeQuick models are compared with the Global Ionospheric Maps (GIM) presented by the Jet Propulsion Laboratory (JPL) and the Technical University of Catalonia (UPC). For a short-term prediction, the Standard Persistence Model (SPM) method, a 27 day median model, and the proposed short-term prediction method are compared for one day ahead. It is shown that for all stations the IRI-Plas model provides better compliance with GIM maps than the NeQuick model irrespective of a solar activity level. An average absolute error lays in the range of 3–3.5 TECU, relative root square mean (RMS) error in the range of 22–27% in 2015 and 1.7–2 TECU, 20–25% in 2008. For the NeQuick model, these estimates were 6.7–8.2 TECU and 42–45% in 2015 and 2.2–3.6 TECU, 30–37% in 2008. For the short-term forecast, the best results were obtained by the SPM method with an average absolute error in the range of 1.95–2.15 TECU in 2015 and 0.59–0.98 TECU in 2008, a relative RMS error in the range of 17–21% in 2015, 11.5–15% in 2008. For the proposed short-term prediction method, these errors were 2.04–2.2 TECU and 12–14% in 2015 and 0.7–1.0 TECU, 7–11% in 2008. Using medians, the errors were 3.1–3.4 TECU and 17–21% in 2015 and 1.0–1.3 TECU, 10–15% in 2008. The dependence of results on the Dst-index was obtained.