New approach for predicting of air pollution near highway caused by burning peat bog

Introduction. Peat deposits are found in many places around the world, but the world’s largest peatlands are the West Siberian Lowland, the Hudson Bay Lowland, and the Mackenzie River Valley. Peat fires are significant sources of carbon dioxide (a greenhouse gas) and carbon oxide (a toxic gas). In addition, peat fires release mercury into the atmosphere at a rate 15 times greater than upland forests, which may be a serious human health concern. If a peat fire develops near a highway, the smoke from the burning peat-bog reduces the visibility, makes the breathing difficult, affect the human nervous and cardiovascular systems and may finally result in traffic accidents or in an emergency. Modelling methodology. K-theory approach . According to Berlyand, such parameters as instant concentrations of CO pulsed deviations from these values and the velocity of the CO diffusion should be taken into consideration while developing an emission model of the peat deposits burning near the highway. The problem is simplified by the application of the turbulent diffusion model. Using this approach, also known as K-theory, together with reasonable approximations and assumptions, there was established that the concentration of the pollutant emitted from the unregulated square source, such as a burning peat bog, is as follow in the Russian normative document OND-86. At the same time, this approach is time-consuming and doesn’t specify inaccurate problem parameters derived from the measurements. To solve these problems, we offer to apply a neural network approach. On the base of the measurements, there was developed a neural network model with parameters (weights) tuned via optimization methods. The RProp method and the combination of “cloud” and RProp methods were in use. The neural network model of the complex system can gather pieces of heterogeneous information - differential equations, conservation laws, equations of state, symmetry conditions, etc. The information exchange via neural network parameters between different levels of hierarchy makes computing less resource consuming. Results and discussion. Case study 1. Visualizes the joint results of experimental and simulated measurements of the peat fire-related CO concentrations near the Federal Highway R-255 “Siberia”. The concentrations of CO are expressed in terms of Limit Value Units: 20 minutes CO limit value is 5 mg/m³. The calculations were realized using the software program Ecolog 4 (Integral Co. Ltd., Saint Petersburg, Russia). The results of the measured and simulated CO concentrations reaching values of 0,8-1,2 mg/m³ were later used as input heterogeneous data for the calculations by the neural network technique described above. Case study 2. Turbulent diffusion loses importance when modeling the transfer of the smog clouds from the peat fire over long distances. In addition, there is possible not only a smouldering peat fire but a burning peat fire followed by the emission of hot gases. We have developed an original neural network model, based on the Gaussian dispersion, to estimate these physical phenomena. Assume that the average cross-section of a peat fire smog cloud, migrating in the vicinity of a highway, is similar to the Gaussian distribution having a plume profile. Show’s the dynamic development of the pollution in this area at the wind in the direction of the highway (4 neurons). Parametric model allows predicting the level of peat fire-related air pollution at different wind directions (Project No. 14-01-00733-А supported by the grant of the Russian Foundation for Basic Research).

автомагистраль, торфяной пожар, угарный газ, опасное загрязнение воздуха, транспортный коллапс, информационный процесс, нейросетевая модель, motorway, peat fire, carbon monoxide, dangerous air pollution, traffic collapse, information process, neural network model

1. Fraser L. H., Keddy P. A. (eds.). The World’s largest wetlands: Ecology and conservation.-Cambridge, UK : Cambridge University Press, 2005. -488 p. DOI: 10.1017/cbo9780511542091.

2. Biester H., Bindler R. Modelling past mercury deposition from peat bogs - the influence of peat structure and 210Pb mobility // Working Papers of the Finnish Forest Research Institute. - 2009. - No. 128. -P. 483.

3. De Groot W. J. Peatland fires and carbon emissions/Natural Resources Canada, Canadian Forest Service. -Great Lakes Forestry Centre, Saul Ste. Marie, Ontario Frontline Express, 2012. -No. 50. -2 p.

4. Fokeeva E. V., Safronov A. N., Rakitin V. S., Yurganov L. N., Grechko E. I., Shumskii R. A. Investigation of the 2010 July-August fires impact on carbon monoxide atmospheric pollution in Moscow and its outskirts, estimating of emissions // Izvestiya, Atmospheric and Oceanic Physics. - 2011. - Vol. 47, Issue 6. -P. 682-698. DOI: 10.1134/s0001433811060041.

5. Konecny K., Ballhorn U., Navratil P., Jubanski J., Page S. E., Tansey K., Hooijer A., Vernimmen R., Siegert F. Variable carbon losses from recurrent fires in drained tropical peatlands // Global Change Biology. - 2016.-Vol. 22, Issue 4. -P. 1469-1480. DOI: 10.1111/gcb.13186.

6. Gaveau D. L. A., Salim M. A., Hergoualc’h K., Locatelli B., Sloan S., Wooster M., Marlier M. E., Molidena E., Yaen H., DeFries R., Verchot L., Murdiyarso D., Nasi R., Holmgren P., Sheil D. Major atmospheric emissions from peat fires in Southeast Asia during non-drought years: evidence from the 2013 Sumatran fires // Scientific Reports. - 2014. - Vol. 4, Issue 1. - Article No. 6112. DOI: 10.1038/srep06112.

7. Pouliot G., Pierce T., Benjey W., Ferguson S. A. Wildfire Emission Modeling: Integrating BlueSky and SMOKE // Proceedings of 14th International Emission Inventory Conference “Transforming Emission Inventories - Meeting Future Challenges Today”, 11-14 April 2005. - 9 p. URL: https://www.researchgate.net/publication/228674671_Wildfire_emission_modeling_integrating_BlueSky_and_SMOKE (дата обращения: 20.05.2017).

8. Benson P. E. A review of the development and application of the CALINE3 and 4 models // Atmospheric Environment. Part B. Urban Atmosphere. - Vol. 26, Issue 3. - P. 379-390. DOI: 10.1016/0957-1272(92)90013-i.

9. Berkowicz R. OSPM - a parameterized street pollution model // Urban Air Quality: Measurement, Modelling and Management, 2000.-P. 323-331. DOI: 10.1007/978-94-010-0932-4_35.

10. Berlyand M. E. Prediction and regulation of air pollution // Atmospheric and Oceanographic Sciences Library. -1991.-Vol. 14. -320 p. DOI: 10.1007/978-94-011-3768-3.

11. ОНД-86. Методика расчета концентраций в атмосферном воздухе вредных веществ, содержащихся в выбросах предприятий. -М. : Гидрометеоиздат, 1987. -93 с.

12. Genikhovich E. L., Gracheva I. G., Onikul R. I., Filatova E. N. Air pollution modelling at an urban scale - Russian experience and problems // Water, Air & Soil Pollution: Focus. - 2002. - Vol. 2, Issue 5-6. -P. 501-512. DOI: 10.1023/A:1021336829300.

13. Genikhovich E. L. Comparison of United States and Russian complex terrain diffusion models developed for regulatory applications // Atmospheric Environment. - 1995. - Vol. 29, Issue 17. - P. 2375-2385. DOI: 10.1016/1352-2310(95)00053-2.

14. Lo_kin V., Lo_kina O., Uљakov A. Using K-theory in geographic information investigations of criticallevel pollution of atmosphere in the vicinity of motor roads // World Applied Sciences Journal. - Vol. 23, Issue 13. -P. 96-100. DOI: 10.5829/idosi.wasj.2013.23.pac.90020.

15. Lozhkina O., Nevmerzhitsky N., Lozhkin V. Evaluation of air pollution by PM10 and PM2.5 on Saint Petersburg ring road: mobile measurements and source apportionment modelling // Proceedings of 10th International Conference on Air Quality: Science and Application, Milano, 14-18 March 2016. -Hertfordshire : University of Hertfordshire, 2016. -P. 176.

16. Lozhkina O. V., Lozhkin V. N. Estimation of road transport related air pollution in Saint Petersburg using European and Russian calculation models // Transportation Research. Part D: Transport and Environment. -2015.-Vol. 36. -P. 178-189. DOI: 10.1016/j.trd.2015.02.013.

17. Tarkhov D. A., Vasilyev A. N.Newneural network technique to the numerical solution of mathematical physics problems. II: Complicated and nonstandard problems // Optical Memory and Neural Networks (Information Optics). -2005.-Vol. 14. -P. 97-122.

18. Васильев А. Н., Тархов Д. А. Нейросетевое моделирование. Принципы. Алгоритмы. Приложения. -СПб. : Изд-во Политехнического университета, 2009. -527 с.

19. Vasilyev A. N., Tarkhov D. A. Mathematical models of complex systems on the basis of artificial neural networks // Nonlinear Phenomena in Complex Systems. -2014. -Vol. 17, No. 3. -P. 327-335.

20. Haykin S. Neural Networks and Learning Machines. 3rd ed.-NewYork : Prentice Hall, 2009.-936 p.

21. Сухоиванов А. Ю. Моделирование процессов переноса в атмосфере и воздействия на окружающую среду вредных продуктов горения, образующихся при пожаре : дис.…канд. техн. наук. -СПб., 2001.-202 с.