Ecological Systems and Devices Annotation << Back ARTIFICIAL NEURAL NETWORKS FOR PREDICTING CHANGES IN SURFACE CONCENTRATIONS OF THE MAJOR GREENHOUSE GASES A.V. Shichkin, A.G. Buevich, M.S. Remezova, A.P. Sergeev, E.M. Baglaeva, I.E. Subbotina, M.V. Sergeeva, A.Yu. Mamedalieva This work discusses several types of artificial neural networks (ANNs) for predicting the dynamics of changes in the concentration of the major greenhouse gases. The study is based on data on the surface concentration of the three main greenhouse gases (methane, carbon dioxide and water vapor), which were obtained during the monitoring of the dynamics of changes in the concentration of the main greenhouse gases on the Arctic island Belyy, Yamalo-Nenets Autonomous District, Russia. The forecast was obtained using the following types of ANNs: nonlinear autoregressive neural network with external input (NARX), Elman’s neural network (Elman) and Multilayer perceptron (MLP). For the study, a time interval of 168 hours was selected in the summer period of 2016, which was characterized by a pronounced daily cycle of changes in greenhouse gas concentrations. The forecast was carried out for a 20-hour time interval. The ANN model type NARX has built the most accurate prediction of concentration changes for all greenhouse gases. Keywords: greenhouse gases; time series; artificial neural networks; forecasting. DOI: 10.25791/esip.09.2021.1248 Pp. 10-18.