Assessment of IRI-2016 hmF2 model options with digisonde, COSMIC and ISR observations for low and high solar flux conditions

The attempt of this work is to evaluate the three options (BSE-1979, AMTB-2013, SHU-2015) embedded in IRI-2016 for the F2 layer peak height (hmF2) estimation under low and high solar flux levels. We compare the model hmF2 with digisonde observations from Qaanaaq, Milllstone Hill, Ramey, Jicamarca and Port Stanley as well as Constellation Observing System for Meteorology, Ionosphere and Climate (COSMIC) ionospheric radio occultation data over the five stations spanning from northern hemisphere to southern hemisphere. In the case of Millstone Hill, we also use incoherent scatter radar (ISR) measurements. For the comparison with digisonde data, SHU-2015 is the most accurate option for reproducing hmF2 at low and middle latitude stations (Milllstone Hill, Ramey, Jicamarca) with the smallest Root-Mean-Square-Error and the highest correlation with observations. At high latitude (Qaanaaq and Port Stanley), BSE-1979 becomes the best option. This is true for both low and high solar flux conditions. Similar results are obtained for the comparison between ISR and IRI-2016 hmF2 model options over Millstone Hill, where SHU-2015 option has the smallest deviations from the observation data in both low and high solar activity conditions. As for the comparison with COSMIC data, SHU-2015 edges out the other two options under low solar activity level with better performance at three stations. There is no optimal option under high solar activity level, which is to our surprise, cause SHU-2015 was developed based on a large amount of radio occultation data. At Jicamarca, all options show dramatic deviations from COSMIC observations, showing even negative correlations. We propose that the performance of the three options do not vary much with solar condition, while they differ a lot with location, for example, latitude.