Modified solar flux index for upper atmospheric applications

Abstract

[1] The F10.7 solar flux index was modified in order to better describe short-term variations in solar extreme ultraviolet (EUV) irradiance for application in ionosphere and thermosphere studies. Several parameters were computed from the F10.7 time series with the assistance of an artificial neural network (ANN) technique, and the daily F10.7 index value was converted to a new solar flux index, MEI10.7. The ANN consists of an input layer that includes an experimental solar input and the day of the year to take seasonal factors into account, one hidden layer, and a target layer of ionospheric total electron content (TEC). The ANN training and validation data set covered one solar cycle from 1997 to 2008. The parameter set that yielded the smallest root-mean-square errors (RMSEs) between the observed and ANN-predicted TECs was adopted for modifying the solar flux index. The MEI10.7 index was evaluated via the same ANN technique. MEI10.7 yielded a smaller RMSE than the magnesium index (Mg II core-to-wing ratio) and a similar RMSE to the EUV index based on the integrated 26–34 nm emission measured by the Solar and Heliospheric Observatory. An advantage of MEI10.7 is long-term availability since the 1940s, unlike satellite measurements. A long-term trend analysis of the ionospheric critical frequency (foF2) at Kokubunji, Japan, conducted for the period from 1957 to 2010 examined the difference between the ANN-modeled and measured foF2 values. The linear regression error when foF2 was modeled by MEI10.7 was appreciably smaller than when it was modeled by F10.7.