Seasonal Variation of the D‐Region Ionosphere: Very Low Frequency (VLF) and Machine Learning Models

Abstract

AbstractThe D‐region ionosphere (6090 km) plays an important role in long‐range communication and response to solar and space weather; however, it is difficult to directly measure with currently available technology. Very low frequency (VLF) radio remote sensing is one of the more promising approaches, using the efficient reflection of VLF waves from the D‐region. A number of VLF beacons can therefore be turned into diagnostic tools. VLF remote sensing techniques are useful and can provide global coverage, but in practice have been applied to a limited area and often on only a small number of days. In this work, we expand the use of a recently introduced machine learning based approach (Gross & Cohen, 2020, https://doi.org/10.1029/2019JA027135) to observe and model the D‐region electron density using VLF transmitting beacons and receivers. We have extended the model to cover nighttime in addition to daytime, and have applied it to track D‐region waveguide parameters, h’ and , over 400 daytimes and 150 nighttimes on up to 21 transmitter‐receiver paths across the continental US. Using an exponential fit, h’ represents the height of the ionosphere and represents the slope of the electron density. Using this data set, we quantify diurnal, daily and seasonal variations of the D‐region ionosphere for both daytime and nighttime D‐region ionosphere. We show that our model identifies expected variations, as well as producing results in line with other previous studies. Additionally, we show that our daytime predictions exhibit a larger autocorrelation at higher time lags than our nighttime predictions, indicating a model with persistence may perform better.