Afternoon and nighttime differences in IRI-2026 electron-content partitioning over Hainan

Abstract. Does replacing the modeled bottomside with a locally observed profile improve total electron content (TEC) estimates? We examine this question at Hainan using 1,666 manually traced ionosonde profiles from 852 dates in 2015–2018 and the International Reference Ionosphere (IRI-2026). We replace the modeled profile below the F2 density peak with the traced bottomside, retain the modeled topside to 1000 km, and compare the original and combined columns with contemporaneous GNSS vertical TEC. Replacement increases overall mean absolute error from 5.064 to 5.898 TECU, but the two sampled local times respond differently. Error rises from 6.305 to 8.440 TECU in the afternoon and falls from 3.808 to 3.325 TECU after midnight. Under their respective integration limits, the modeled afternoon bottomside contains 4.338 TECU more than the traced bottomside on average. This extra lower-column content partly offsets the positive GNSS-minus-combined difference, making the original total closer to GNSS. The opposite responses persist when the two integrals meet at the observed peak height. At Hainan, closer agreement in total TEC can therefore conceal a substantial difference in the lower profile, and introducing that observed profile can expose the discrepancy. These Hainan results show that regional TEC updating needs to assess the upper contribution alongside the observed bottomside, particularly in the afternoon, while accounting for GNSS electron content above 1000 km.