Modeling of surface dust concentrations using neural networks and kriging

International Conference of Numerical Analysis and Applied Mathematics Icnaam 2019

et al. 2020

The paper predicts the changes in the concentration of one of the main greenhouse gases -methane (CH4). The forecast was made for three different time periods, each of which had its own characteristics of the dynamics of changes in the concentration of CH4. Data for the study were collected while monitoring the content of the main greenhouse gases in the surface layer of atmospheric air in the Russian Arctic (Bely Island, Yamalo-Nenets Autonomous Okrug). We compared the results of the models prediction based on the two types of artificial neural networks: Elman and nonlinear autoregressive neural network with external input (NARX). NARX showed a high prediction accuracy for all studied time intervals.

show abstract

Section: Figure 1 Place Of the Measurements (Google Earth)mentioning

confidence: 99%

The forecast of the methane concentration changes for the different time periods on the Arctic island Bely

Медведев

International Conference of Numerical Analysis and Applied Mathematics Icnaam 2019

et al. 2020

show abstract

“…Therefore, all the developed scenarios of the temporal dynamics of the global content of greenhouse gases are prognostic and can be carried out with a certain probability. For the implementation of such predictions, models based on artificial neural networks (ANN) [1] - [16], and, in particular, networks like NARX [17]- [20], are well suited. The NARX network is a recurrent dynamic multi-level feedback network.…”

Section: Introductionmentioning

confidence: 99%

Prediction the dynamic of changes in the concentrations of main greenhouse gases by an artificial neural network type NARX

Buevich

International Conference of Numerical Analysis and Applied Mathematics Icnaam 2019

et al. 2020

The paper considered the use of one of the most accurate artificial neural networks for predicting time seriesa nonlinear autoregressive neural network with external input (NARX) for predicting the dynamics of changes in the concentrations of the main greenhouse gases. The data were obtained in the course of monitoring the dynamics of changes in the main greenhouse gases on the Arctic island Belyy, Russia. The data of the surface concentration of methane, carbon dioxide, carbon monoxide and water vapor were used. A time interval of 168 hours was chosen for the study during the summer period (July-August 2016). The NARX model accurately predicted concentration changes for all greenhouse gases.

show abstract

Analysis of time series of greenhouse gas concentrations in the Russian Arctic using the artificial neural networks

Buevich

AIP Conference Proceedings

et al. 2018