Performance analysis of IRI-2016 and IRI-2020 models, and GPS and GLONASS-TEC variations, and their predictions using Artificial Neural Networks (ANNs) at low latitude station Agra, India

Abstract

The total electron content (TEC) data was collected using the newly installed receiver namely GPStation6 at the Agra, India station in 2016. The TEC data for Global Positioning System (GPS) and GLONASS navigation systems and the IRI-2020 and IRI-2016 models were statistically processed diurnally, monthly, seasonally, and yearly during the descending phase of solar activity in 2018. The entropy of the TEC variations was calculated to assess the chaotic behaviour of the data. Subsequently, the observed results were predicted using artificial neural networks (ANNs). The highest TEC values for GPS, GLONASS, and both IRI models were recorded in April (≈50 TECU and 35 TECU), whereas the lowest values were recorded in November and December (≈20–25 and 15–20 TECU). In terms of seasonal values, GPS and GLONASS-TEC were underestimated by the IRI models. The maximum seasonal TEC values (≈45 TECU) were recorded for GLONASS in the equinox, ≈35 TECU for GPS in the summer, and ≈30 TECU for both IRI models in the equinox, while the minimum TEC values were recorded in winter for GPS (≈30 TECU), GLONASS (≈35 TECU), and IRI models (≈25 TECU). A strong correlation was observed between TEC variations and IRI models. A weaker and negative correlation between magnetic storm activity (∑Kp) and observed TEC variations was observed, whereas a weaker and positive correlation was found between TEC and solar activity (F10.7). The entropy values were higher for GPS-TEC than for GLONASS-TEC. These variations were interpreted based on literature published by previous researchers.