The process of formation and rotting of ice on lakes is an integral part of the hydrological cycle of many lakes. The conditions of the ice regime significantly influence the ecological system of lakes. The article includes calculation and analysis of errors in the determination of the spatial ice distribution (spatial resolution of 4–6 km) on Lake Onego, Lake Ladoga, Lake Segozero and Lake Vigozero within the period of 2006−2017 according to National Snow and Ice Data Center (NSIDC), National Oceanic and Atmospheric Administration National Environmental Satellite, Data, and Information Service (NOAA NESDIS) data with regard to reliable Moderate Resolution Imaging Spectroradiometer (MODIS) data (spatial resolution of 500 m). It was established that within the monitoring period, NSIDC data have the minimum mean values of errors in determining the spatial distribution of ice on lakes (3−10%) compared to NOAA NESDIS data (11−19%) and are also of more practical interest in estimating the ice coverage of lakes. The dependence of the mean value of errors that occur in the determination of the spatial distribution of ice (according to NSIDC, NOAA and NESDIS data) on the actual value of ice coverage (according to MODIS) was revealed. The results show that the NSIDC data allow estimating adequately the phases of the ice regime; however, the formation of a daily time series of ice coverage during freeze-up and break-up phases is possible only with a significant error (mean value of absolute deviations according to MODIS data is up to 35%).
The ice cover on lakes is one of the most influential factors in the lakes’ winter aquatic ecosystem. The paper presents a method for predicting ice coverage of lakes by means of multilayer perceptrons. This approach is based on historical data on the ice coverage of lakes taking Lake Onega as an example. The daily time series of ice coverage of Lake Onega for 2004–2017 was collected by means of satellite data analysis of snow and ice cover of the Northern Hemisphere. Input signals parameters for the multilayer perceptrons aimed at predicting ice coverage of lakes are based on the correlation analysis of this time series. The results of training of multilayer perceptrons showed that perceptrons with architectures of 3-2-1 within the Freeze-up phase (arithmetic mean of the mean square errors for training epoch
) and 3-6-1 within the Break-up phase (
) have the least mean-squared error for the last training epoch. Tests within the holdout samples prove that multilayer perceptrons give more adequate and reliable prediction of the ice coverage of Lake Onega (mean-squared prediction error MSPE = 0.0076) comparing with statistical methods such as linear regression, moving average and autoregressive analyses of the first and second order.