Artificial intelligence will clarify the boundaries of the northern lights

Auroras arise after the emission of particles on the Sun and are a kind of indicator, which tells about the state of the Earth’s magnetic shield. The auroral oval defines the area where these auroras occur – a ring around the magnetic poles. Its width and position depend on space weather. In calm conditions this area is narrow, but with increasing emission of solar wind particles the ring expands, and then auroras may occur even in regions where they are not seen. It is necessary to know the boundaries of the auroral oval as precisely as possible in order to determine where observations, communication and navigation may deteriorate as a result of the interaction of the solar wind with the Earth’s magnetic field.

That is why a discovery made at the Institute of Solar-Terrestrial Physics, Siberian Branch of the Russian Academy of Sciences (Irkutsk) is important. Scientists have learned to determine the boundaries of the auroral oval using computer vision. Initially, the SIMuRG system was created, which is able to collect and process large amounts of data from global navigation satellite systems (signals from GPS satellites, as well as GLONASS and other global network systems), to then use them to explore the near-Earth space. Then the analysis of big data and computer vision methods were applied.

The complexity of the task faced by scientists turned out to be in the “sparse” data. Satellite receivers are not at a specific distance from each other or on a given line – there are oceans, sparsely populated and hard-to-reach areas. This can be thought of as image processing, where there are less than 30% of pixels and the remaining 70% need to be filled in. Artificial intelligence just adds the missing data, which allows you to determine the territory where communication and navigation can become worse. In addition, scientists had to take into account the noise of incoming data and the fact that the satellites transmitting the signals are constantly moving, as a result, the picture is changing all the time. This is difficult, because everything that does not actually relate to physical phenomena, everything that is associated with measurement errors, the algorithm should not take into account when determining the boundaries of the oval.

However, the scientists intend to cope with this problem and improve the model. Moreover, the use of both optical and magnetic data can not only serve the specific purpose of determining the boundaries of the auroral oval, but will also allow a deeper understanding of the essence of physical phenomena and processes occurring in the atmosphere, reports the publication “Science in Siberia”.