||||||

Home » Archive of journals » Volume 16, No. 1, 2026 » Comparative analysis of neural network models for forecasting the dynamics of greenhouse gas concentrations in the surface air layer on Bely Island

COMPARATIVE ANALYSIS OF NEURAL NETWORK MODELS FOR FORECASTING THE DYNAMICS OF GREENHOUSE GAS CONCENTRATIONS IN THE SURFACE AIR LAYER ON BELY ISLAND

JOURNAL: Volume 16, No. 1, 2026, p. 53-63

HEADING: New technologies for the Arctic

AUTHORS: Bobakov, V.S., Buevich, A.G., Butorova, A.S., Sergeev, A.P.

ORGANIZATIONS: Institute of Industrial Ecology, Ural Branch of the RAS

DOI: 10.25283/2223-4594-2026-1-53-63

UDC: 504.064.2.001.18

The article was received on: 23.07.2025

Keywords: greenhouse gases, forecast, time series, neural networks, graph neural networks, Long Short-Term Memory

Bibliographic description: Bobakov, V.S., Buevich, A.G., Butorova, A.S., Sergeev, A.P. Comparative analysis of neural network models for forecasting the dynamics of greenhouse gas concentrations in the surface air layer on Bely Island. Arktika: ekologiya i ekonomika. [Arctic: Ecology and Economy], 2026, vol. 16, no. 1, pp. 53-63. DOI: 10.25283/2223-4594-2026-1-53-63. (In Russian).

Abstract:

In recent years, the application of machine learning technologies has become increasingly relevant for monitoring greenhouse gas emissions in the Arctic, where intense permafrost degradation and increasing methane emissions into the atmosphere are observed. In the study, we have developed and compared recurrent and graph neural network-based models to predict the dynamics of concentrations of major greenhouse gases in the surface air on Bely Island, the Yamalo-Nenets Autonomous Area, Russia. The original dataset consisted of surface concentration measurements of four gases for the summer months of 2016—2017: carbon dioxide (CO2), methane (CH4), carbon monoxide (CO), and water vapor (H2O). For graph forecasting, we used the Multivariate Time Series Graph Neural Network (MTGNN) model; for recurrent forecasting, we used the Long Short-Term Memory (LSTM) and Long Short-Term Memory Network (LSTNet) models. The forecast accuracy of the models was assessed using the following metrics: mean absolute error (MAE), root mean square error (RMSE), normalized root mean square error (NRMSE), and correlation coefficient (Corr). In general, the MTGNN model demonstrated lower error values compared to recurrent models (LSTM, LSTNet). Compared to LSTM and LSTNet, MTGNN had 25—55% lower errors. The more accurate graph model required significantly more time to train and had a significantly larger number of parameters for optimization.

Finance info: The study was supported by Ministry of Science and Higher Education of the Russian Federation, project FUMN-2024-0003. The authors are grateful to the Center for Collective Use of Arctic Environmental Research at the Institute of Industrial Ecology of the Ural Branch of the Russian Academy of Sciences for providing equipment for measuring greenhouse gas concentrations on Bely Island.

References:

1. Liu J., Ma Z. Forecasting Housing Price Using GRU, LSTM and Bi-LSTM for California. 2024 IEEE 2nd International Conference on Control, Electronics and Computer Technology (ICCECT). [S. l.], 2024, pp. 1033—1037.

2. He Z., Zhao C., Huang Y. Multivariate Time Series Deep Spatiotemporal Forecasting with Graph Neural Network. Applied Sciences, 2022, vol. 12 (11), p. 5731.

3. Widiasari I. R., Efendi R. Utilizing LSTM-GRU for IOT-Based Water Level Prediction Using Multi-Variable Rainfall Time Series Data. Informatics, 2024, vol. 11 (4), p. 73.

4. Sangeetha S. K. B., Mathivanan K. M., Rajadurai H., Jaehyuk C., Sathishkumar V. E. A multi-modal geospatial–temporal LSTM based deep learning framework for predictive modeling of urban mobility patterns. Scientific Reports, 2024, vol. 14, p. 31579.

5. Zhang X., Tian Z., Shi Y., Guan Q., Lu Y., Pan Y. STFGCN: Spatio-Temporal Fusion Graph Convolutional Networks for Subway Traffic Prediction. IEEE Access, 2024, vol. 12, pp. 194449—194461.

6. Dai Z., Wang C., Hu D., Chen J., Fu S. Intercity round-trip multi-region demand prediction based on multi-task fusion recurrent graph attention network. Neural Computing & Applications, 2025, vol. 37, pp. 14829—14848.

7. Jiang W., Luo J. Graph neural network for traffic forecasting: A survey. Expert Systems with Applications, 2022, vol. 207, p. 117921.

8. Cheng K., Ye J., Lu X., Sun L., Du B. Temporal Graph Network for continuous-time dynamic event sequence. Knowledge-Based Systems, 2024, vol. 304, p. 112452.

9. Liang Y., Zhang W., Sheng Z., Yang L., Xu Q., Jiang J., Tong Y., Cu B. Towards Scalable and Deep Graph Neural Networks via Noise Masking. arXiv preprint, 2025, 2412.14602.

10. Li C., Li S., Gao Y., Zhou L., Li W. Static graph convolution with learned temporal and channel-wise graph topology generation for skeleton-based action recognition. Computer Vision and Image Understanding, 2024, vol. 244, p. 104012.

11. Bui K. N., Cho J., Yi H. Spatial-temporal graph neural network for traffic forecasting: An overview and open research issues. Applied Intelligence, 2022, vol. 52, pp. 2763—2774.

12. Wu Z., Pan S., Chen F., Long G., Zhang C., Yu P. S. A Comprehensive Survey on Graph Neural Networks. IEEE Transactions on Neural Networks and Learning Systems, 2021, vol. 32 (1), pp. 4—24.

13. Wu F., Souza A., Zhang T., Fifty C., Yu T., Weinberger K. Simplifying Graph Convolutional Networks. Proceedings of the 36th International Conference on Machine Learning, PMLR, 2019, vol. 97, pp. 6861—6871.

14. Buevich A. G., Sergeev A. P., Shichkin A. V., Baglaeva E. M., Subbotina I. E., Butorova A. S. Model for predicting surface methane concentration in the Arctic region based on an artificial neural network with a long chain of short-term memory elements and wavelet transform of the initial data. Arctic: Ecology and Economy, 2023, vol. 13, no. 3, pp. 428—436. (In Russian).

15. Buevich A., Sergeev A., Shichkin A., Baglaeva E. A two-step combined algorithm based on NARX neural network and the subsequent prediction of the residues improves prediction accuracy of the greenhouse gases concentrations. Neural Computing & Applications, 2021, vol. 33, pp. 1547—1557.

16. Baglaeva E., Buevich A., Sergeev A., Rakhmatova A., Shichkin A. Forecasting of some greenhouse gases content trend in the air of the Russian Arctic region. Atmospheric Pollution Research, 2021, vol. 12 (3), pp. 68—75.

17. Bogoyavlensky V. I., Sizov O. S., Nikonov R. A., Bogoyavlensky I. V., Kargina T. N. Land degassing in the Arctic: genesis of natural and anthropogenic methane emissions. Arctic: Ecology and Economy, 2020, no. 3 (39), pp. 6—22. (In Russian).

18. Bogoyavlensky V. I., Sizov O. S., Nikonov R. A., Bogoyavlensky I. V. Monitoring changes in methane concentration in the Arctic atmosphere in 2019—2021 according to TROPOMI spectrometer data. Arctic: Ecology and Economy, 2023, vol. 12, no. 3, pp. 304—319. (In Russian).

19. Serykh I. V., Tolstikov A. V. Climatic changes in air temperature in the western part of the Russian Arctic in 1940—2099 according to ERA5 data and CMIP6 models. Arctic: Ecology and Economy, 2024, vol. 14, no. 3, pp. 334—349. (In Russian).

20. Subbotina I. E., Baglaeva E. M., Buevich A. G., Sergeev A. P., Shichkin A. V. Dynamics of greenhouse gas content in the surface layer of atmospheric air of the Arctic Bely Island in the summer period 2015—2017. Arctic: Ecology and Economy, 2022, vol. 12, no. 1, pp. 68—76. (In Russian).

21. Baglaeva E. M., Sergeev A. P., Buevich A G., Shichkin A. V., Subbotina I. E. Nonparametric assessment of the predictive accuracy of models using the example of time series of methane concentrations in the atmospheric air of the Arctic Island Bely. Arctic: Ecology and Economy, 2024, vol. 14, no. 4, pp. 500—510. (In Russian).

Download »

DOI 10.25283/2223-4594