Spatiotemporal Analysis of the Termokarst Lakes Evolution Using Remote Sensing and Randomized Machine Learning and Forecasting

The work is devoted to the problem of forecasting the evolution of the area of thermokarst lakes in the Arctic permafrost zone using the analysis of test areas from several geographical regions. The approach proposed in the work is based on the Randomized Machine Learning method for constructing mathematical models of lake area from climate indicators, learning the model on real data and further forecasting. The results of modeling the dynamics of lake areas using linear static and dynamic models are presented and compared. It is shown that proposed dynamic model can significantly reduce the average modeling error