IRI Total Electron Content in the Canadian Sector: A comparison to GPS observations and recommendations

Total Electron Content (TEC) measurements from ten dual-frequency GPS receivers in the Canadian High Arctic Ionospheric Network (CHAIN) are used to evaluate the performance of IRI-2007 within the Canadian Arctic. In all regions, we see systematic underestimation of daytime TEC, particularly at solar maximum, where summer RMS TEC errors range from up to 10 TECU at sub-auroral stations to up to 14 TECU in the polar cap region. Performance during winter periods was exceptional, with RMS errors well constrained below 2 TECU at all stations. Diurnal variability in TEC is found to be underestimated during equinox periods by up to 40% at sub-auroral latitudes and up to 70% in the polar cap region. During the winter, the IRI performs much better, where diurnal variations are matched to within an overestimation of 40% or an underestimation of 20% across all regions. Using a CHAIN ionosonde, collocated with the Resolute GPS receiver, we are able to diagnose the main contributing layer of the ionosphere to the observed TEC errors. At the Resolute station, the IRI is found to produce bottomside TEC within 1 TECU of observation; however, topside TEC demonstrates significant errors of up to 7 TECU during equinox and summer periods at high solar activity. Winter topside TEC is found to be within 1 TECU, even at solar maximum. Diurnal variability in topside TEC is found to be almost entirely driven by NmF2 variability. It is also shown that the IRI's use of the highly smoothed IG12 index to drive solar activity can result in significant errors during periods of large amplitude, short-term changes in solar activity. Using a monthly IG index to drive the model, rather than IG12, we find that IRI TEC errors reduce from 5 TECU to less than 1 TECU during a three month, ~100% increase in solar activity between October and December, 2011. Overall, all stations showed an average improvement of 0.33 TECU while using monthly IG rather than the IRI's default IG12 index.