Incoherent scatter radar measurements and modeling of high-latitude solar photoionization

A 9-year database of sunlit E region electron density altitude profiles (N_e(z)) measured by the Sondrestrom incoherent scatter radar (ISR) has been partitioned over a parameter space of 10.7 cm solar radio flux (F _10.7) and solar zenith angle (χ) to investigate long-term solar and thermospheric variability and to validate a contemporary EUV photoionization model. A two-stage filter, which rejects N_e(z) profiles with large Hall-to-Pedersen conductance ratio and incorporates an MLT-dependent correction factor for low-energy precipitation, is used to mitigate auroral contamination. Resultant filtered mean sunlit N_e(z) is compared with subauroral N_e measured for the same F_10.7 and χ conditions at the Millstone Hill ISR in order to confirm adequate high-energy auroral rejection. Mean N_e, as expected, increases with solar activity and decreases with large χ. Radar model comparison indicates that across all parameter space and for altitudes from 105 to 180 km, the GLOW model estimates are within 5% of the ISR mean with the contribution from photoelectrons accounting for 30 to 50% of equilibrium ion density. Above 180 km, the GLOW model slightly overestimates the height of the F1 layer. Radar model comparison also reveals a low-altitude N_e enhancement for high solar activity at altitudes commensurate with 3 to 7 nm XUV and H Lyman-β radiances. The variance of the ISR mean N_e is shown to be greatest at low F_10.7 (solar minimum). Simulated N_e variance envelopes, given by perturbing the GLOW model neutral atmosphere input by the measured A_p, F _10.7, and T_e extrema, are narrower than ISR derived geophysical variance envelopes at solar minimum. We find no evidence for solar cycle control of low-energy precipitation and thus we attribute the observed N_e variance at solar minimum to variability in solar EUV flux. In order to address estimation of N_e at altitudes where GLOW model photochemical equilibrium assumptions are invalid, we provide an empirical model for Sondrestrom quiet time photoionization N_e(z) as a function of F_10.7 and χ.