Geophysical processes and biosphere: article

Аnnotation: Here we provide short description of an expert system for prediction of the ice jamming power in the area of the Severnaya Dvina river and verification procedure of the system. Expert system based on hydrological and meteorological data at time interval 1911-2016. To process data we use machine learning technique and adjacent mathematics, because there is no mathematical model of ice jamming process and time series of observations are too short to apply classical statistics approach. The expert system was developed early at 2012; to tune it data written at 1991-2010 seasons at hydrological pickets was used. In this work, we used additional data sets at 2011-1016 seasons to repeat learning and estimate quality of the system. It was shown that system has reliable efficiency: forecast results coincide with observations for overall added data (seasons 2011-2016). One should keep in mind additional data do not change the forecast accuracy, which stay approximately 85 % like in previous study. All developed software is cross-platform, written on C++ language and it is implemented as command line application. The software can be easily adopted to operate as part of Severnaya Dvina river area realtime monitoring service.