An Examination of SuperDARN Backscatter Modes Using Machine Learning Guided by Ray‐Tracing. Issue 9 (20th September 2022)

• Record Type: Journal Article
• Authors: Kunduri, B. S. R.; Baker, J. B. H.; Ruohoniemi, J. M.; Thomas, E. G.; Shepherd, S. G.
• Is Part Of: Space weather. Volume 20:Issue 9(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-09-20
• Subjects: Machine learning -- SuperDARN -- Ray‐tracing; Space environment -- Periodicals; 551.509992
• DOI: 10.1029/2022SW003130
• Languages: English
• ISSNs: 1542-7390

Abstract

Abstract: The Super Dual Auroral Radar Network (SuperDARN) is a network of High Frequency (HF) radars that are typically used for monitoring plasma convection in the Earth's ionosphere. A majority of SuperDARN backscatter can broadly be divided into three categories: (a) ionospheric scatter due to reflections from plasma irregularities in the E and F regions of the ionosphere, (b) ground scatter caused by reflections from the ground/sea surface following reflection in the ionosphere, and (c) backscatter from meteor trails left by meteoroids as they enter the Earth's atmosphere. Due to the complex nature of HF propagation and mid‐latitude electrodynamics, it is often not straightforward to distinguish between different modes of backscatter observed by SuperDARN. In this study, we present a new two‐stage machine learning algorithm for identifying different backscatter modes in SuperDARN data. In the first stage, a neural network that "mimics" ray‐tracing is used to predict the probability of ionospheric and ground scatter occurring at a given location along with parameters like the elevation angles, reflection heights etc. The inputs to the network include parameters that control HF propagation, such as signal frequency, season, UT time, and geomagnetic activity levels. In the second stage, the output probabilities from the neural network and actual SuperDARN data are clustered together to determine the category of the backscatter. Our model can distinguish between meteor scatter, 1/2 hop E‐/F‐region ionospheric as well as ground/sea scatter. We validate our model by comparing predicted elevation angles with those measured at a SuperDARN radar. Plain Language Summary: The Super Dual Auroral Radar Network (SuperDARN) is an international network of radars that operate in the High Frequency (HF) range (between 8 and 18 MHz). The radars are primarily used for monitoring the motion of plasma in the charged portion of Earth's upper atmosphere, called the ionosphere. A key feature of HF radio waves is that they are refracted or "bent" by the ionosphere. As a result, these frequencies are usually reflected back by decameter‐scale density structures in the ionosphere or by the ground, depending on the density of the ionosphere and operational characteristics of the radar. SuperDARN data is therefore mostly composed of measurements from the ionosphere, ground/sea, and meteor trails. At high latitudes, it is usually straightforward to distinguish between fast‐moving ionospheric and the slow‐moving ground observations. However, at mid‐latitudes, due to the similarities between ionospheric and ground measurements it is usually difficult to distinguish between these two types of measurements. In this study, we present a new methodology that uses a physics‐based approach to encode certain HF propagation characteristics into a machine learning model that can: (a) provide the ability to classify between different types of measurements, and (b) determine the uncertainties in the model ionosphere used to characterize HF propagation. Key Points: We developed a new machine learning model that is guided by ray‐tracing and the International Reference Ionosphere (IRI) to examine SuperDARN backscatter modes Our model can be used to classify SuperDARN backscatter into different categories such as meteor, E‐/F‐region ionosphere and ground/sea The model's error rate is lowest in winter and highest in summer, suggesting uncertainties in the IRI ionosphere are larger in summer