In ionospheric research, and radio wave propagation prediction, knowledge of the state of the ionosphere is critical. One way to obtain this information is by examining ionospheric traces, called ionograms, which are graphical plots of echo delay versus frequency. Ionograms are typically referred to as the virtual height profile of the F layer and can be used to identify its shape, and to infer other ionospheric parameters such as critical frequencies.

This article analyses ionogramme a novel use of an advanced artificial intelligence model for automated ionogram processing. The model, called FC-DenseNet24, is trained to recognize the signals in ionograms as one of three classes: E region, Es layer, or F layer. The resulting accuracy was found to be significantly higher than the traditional manual processing using the Peru auto-encoder, which is widely used for ionogram scaling.

Demystifying the Ionogram Standard Blood Test: A Comprehensive Guide to Electrolyte Balance

In addition, the model was able to detect features of the shape of the F layer peak, and thus may be a more accurate approach to inferring critical frequencies from ionograms than traditional methods. However, tests also demonstrated that the model does not always find a good fit, and so it is important to perform additional physical checks on ionospheric peaks when the quality of the data is uncertain.

Nevertheless, the ability to automatically scale and invert ionograms will make the process of ionospheric parameter extraction much faster and less labor intensive. This is especially beneficial when working with large volumes of ionospheric data, as has recently become the case in the development of digital ionospheric sounder systems.