Exploring the Feasibility of a Unified D-region Ionosphere Model

Direct measurements of the D-region ionospheres (6090 km in altitude) electrical properties are difficult with current technology. The D-region ionosphere lies above the altitude ranges that can be reached by balloons and below the altitudes that can be reached by satellites. Sounding rocket measurements are possible, and have been done in the past; however, they are costly and difficult to use on a large scale. Instead, very low frequency (VLF, 3-30 kHz) waves have been used to infer various properties of the D-region. Recently, machine learning techniques have been used to predict Wait and Spies waveguide parameters (h and ) for the D-region across a network of VLF transmitters and receivers with excellent temporal resolution (seconds). This model was limited to daytime conditions and was only used to model a couple of days. By adjusting a few assumptions in the original machine learning model, we are able to extend the technique into the night. With these changes, we modelled the ambient nighttime D-region for over 150 days. We explore both night to night variation as well as longer term seasonal variations and compare to expected trends. We also examine the differences in autocorrelation of daytime predictions and nighttime predictions. With improvements to machine learning based techniques, it may be possible to incorporate other sources of VLF into the model. Broadband emissions from lightning, known as sferics, have been used to produce ionospheric maps with spatial resolution on the order of 100 km. We explore the possibility of using machine learning to combine these two VLF sources into a single unified D-region model. By using both VLF transmitters and sferics, we hope to achieve a model with both excellent spatial and temporal resolution.