Accuracy of hmF2 estimations, including IRI-2020 options and ionograms validated parameters, compared to ISR measurements at Millstone Hill

Abstract

Several empirical formulations used over time to estimate the fundamental ionospheric parameter hmF2 have been compared in this study. These are the first formulation proposed by Shimazaki (1955) (SHI-1955) as a function of the propagation parameter M(3000)F2, the more accurate BSE-1979 formula proposed by Bilitza et al. (1979) and firstly adopted by the International Reference Ionosphere (IRI) model, and the newest Altadill-Magdaleno-Torta-Blanch (AMTB-2013) (Altadill et al., 2013) and SHU-2015 (Shubin, 2015) models, obtained with a different approach with no explicit dependence on any ionospheric parameter and added as alternative options in the IRI-2016. The evaluation of the accuracy of the available formulation is performed by comparing the modeled values of hmF2 with those simultaneously obtained with independent measurements from the Incoherent Scatter Radar (ISR) installed at the Millstone Hill ionospheric station. The database considered consists of 3626 measurements, thus allowing the evaluation of the results for different heliogeophysical conditions. SHI-1955 and BSE-1979 formulations are evaluated also using input data manually scaled from ionograms recorded at the same location, with the aim of evaluating their accuracy when updated with validated data rather than modeled ones. The SHU-2015 is confirmed the best option in any condition, while AMTB-2013 turns out to perform poorly during night, when SHI-1955 and BSE-1979 fed by validated data can be used for trend analyses due to the high correlation with ISR data. Despite this, BSE-1979 performs better with modeled parameters as input, in terms of RMSE and mean deviation from ISR data. The use of SHI-1955 with CCIR-modeled M(3000)F2 is discouraged under daytime conditions even for long trend analyses.