Using Machine Learning to Classify Stratigraphic Layers of Snow According to the Snow Micro Pen Device

The observation of snow cover by the staff of the Geographical Faculty of Moscow State University of the meteorological observatory has long been researched. This article describes the snow accumulation features and the snow cover's stratigraphy. The third cyclone arrived in Moscow on the night of December 14. There had been a large number of snowdrifts since the beginning of the snow accumulation, and the 49 cm mark was recorded at the MSU weather station. The difficulties of classifying layers in the snow column have been investigated by many glaciologists, something that is also considered in this paper. Machine learning methods were used to classify stratigraphic layers in the snow column according to measurements from the snow micro pen device. The ice crystal shapes within the snow column, resulting from metamorphism (rounded, faceted, thawed), exhibit variations in both density and parameters derived from the snow micro pen device data processing. Specifically, MPF(N) represents the average resistance force, SD(N) denotes its standard deviation, and cv signifies its covariance. This diversity allows for the categorization of processed device data and the incorporation of new measurement data without relying on direct manual drilling results. The obtained device data underwent thorough processing. Through comparison with data from direct snow stratigraphy surveys, the stratigraphic layers of the snow column were classified. Subsequently, utilizing the classified data of the device's stratigraphic layers, K-nearest neighbors clustering enabled the classification of new data obtained from the device without the need for additional manual surveys in the future.

snow cover, spatial temporal heterogeneities, MSU, snow thickness, meteostation, winter season, snow layer, winter period, research, heterogeneity of the snow layers