Adaptation of SMOGK deep convection detection system algorithms to high-orbit meteorological satellites "Arktika-M"

Modern climate research indicates that recent decades have been the warmest in the history of meteorological observations across Northern Eurasia. This, in turn, has led to an increased frequency and intensity of atmospheric convection processes, resulting in the formation of single- and multi-cell clusters of hazardous cumulonimbus clouds and mesoscale convective systems (MCS). Powerful convective clusters and systems trigger extreme weather events—intense thunderstorms, heavy rainfall, large hail, destructive squalls, and tornadoes. The projected rise in convective extreme events underscores the urgent need to develop effective methods for monitoring deep convective clouds (DCC) using satellite data across the entire Russian Federation, including sparsely populated regions of Siberia and the Far East, where limited radar coverage and synoptic stations further reduce the quality and detail of atmospheric process monitoring.

This study demonstrates the application of an interactive and automated Deep Convective Cloud Monitoring System (SMOGK) based on the analysis of individual DCC development cases in southern Russia in 2025. Automated detection of DCC contours and displacement trajectories was performed using thresholding methods and machine learning algorithms at the first stage, leveraging long-term data from the Meteosat-10 satellite in Rapid Scan mode (providing 5- and 15-minute imaging intervals for European Russia). Measurements in the infrared (IR) and visible ranges of the electromagnetic spectrum from the Russian Arktika-M №1-2 satellites were then used to assess their comparability and consistency with Meteosat-10 data, as well as their potential for automated DCC detection. It is important to note that Arktika-M №1-2 satellite data differ from the European satellite data initially used to develop the automated DCC recognition algorithm within SMOGK. Therefore, an analysis was conducted to evaluate the specific characteristics and typical values of brightness temperature distributions in various wavelength bands for DCC cloud tops, as observed by the respective satellites. Preliminary analysis revealed strong correlations between recognized DCC contours (based on qualitative expert assessment). This justified the subsequent adaptation of the DCC recognition algorithm to the spectral channels and data features of Arktika-M №1-2.

The results showed high consistency (~85%) in DCC detection and evolution tracking between different satellite systems and C-band Doppler radars (DMRL-C). Integrated visualization and comparison of multi-source data confirm the viability of using Arktika-M №1-2 for automated DCC monitoring—a critical capability for hazardous weather nowcasting and monitoring at latitudes north of 50°N over Russia, where geostationary satellite data are either scarce or distorted.

These findings highlight the promise of integrating multi-satellite data to improve forecasting accuracy for severe convective events in a changing climate.