Satellite remote sensing and deep learning for aerosols prediction
Satelitsko osmatranje i duboko učenje za predviđanje aerosola
Abstract
Introduction/purpose: The paper presents a new state-of-the-art method that involves NASA satellite imagery with the latest deep learning model for a spatiotemporal sequence forecasting problem. Satellite-retrieved aerosol information is very useful in many fields such as PM prediction or COVID-19 transmission. The input data set was MODAL2_E_AER_OD which presents global AOT for every 8 days from Terra/MODIS. The implemented machine learning algorithm was built with ConvLSTM2D layers in Keras. The obtained results were compared with the new CNN LSTM model. Methods: Computational methods of Machine Learning, Artificial Neural Networks, Deep Learning. Results: The results show global AOT prediction obtained using satellite digital imagery as an input. Conclusion: The results show that the ConvLSTM developed model could be used for global AOT prediction, as well as for PM and COVID-19 transmission.
Uvod: Izložena je unapređena metoda koja uključuje Nasine satelitske snimke sa najnovijim modelom dubokog učenja koji se odnosi na problem predviđanja prostornovremenskih signala. Informacija o aerosolima sa satelitskih snimaka je vrlo značajna za predviđanje disperzije čestica u atmosferi i prenosa virusa COVID-19. Ulazni podaci MODAL2_E_AER_OD predstavljaju globalni AOT za osam dana sa Terra/MODIS. Algoritam mašinskog učenja je sačinjen od kompozitnih neuronskih slojeva ConvLSTM2D u biblioteci Keras. Dobijeni rezultati su upoređeni sa novim modelom CNN LSTM. Metode: Proračunske metode mašinskog učenja, veštačke neuronske mreže, duboko učenje. Rezultati: Rezultati prikazuju globalno predviđanje optičke debljine aerosola sa digitalnim satelitskim snimcima koji su korišćeni kao ulazni podaci. Zaključak: Pokazano je da je razvijeni model ConvLSTM pogodan za globalno predviđanje atmo-sferske debljine aerosola, kao i za prenos atmosferskih čestica i virusa COVID-19.
Keywords:
aerosol optical thickness / NASA Earth observations / ConvLSTM2D / COVID-19 / particulate matter dispersion / optička debljina aerosola / NASA Earth Observations / ConvLSTM2D / COVID-19 / disperzija česticaSource:
Military Technical Courier : Vojnotehnički glasnik, 2023, 71, 1, 66-83Funding / projects:
- Ministry of Education and Science and Technological Development of the Republic of Serbia
Collections
Institution/Community
VinčaTY - JOUR AU - Mirkov, Nikola AU - Radivojević, Dušan AU - Lazović, Ivan AU - Ramadani, Uzahir AU - Nikezić, Dušan PY - 2023 UR - https://vinar.vin.bg.ac.rs/handle/123456789/13687 AB - Introduction/purpose: The paper presents a new state-of-the-art method that involves NASA satellite imagery with the latest deep learning model for a spatiotemporal sequence forecasting problem. Satellite-retrieved aerosol information is very useful in many fields such as PM prediction or COVID-19 transmission. The input data set was MODAL2_E_AER_OD which presents global AOT for every 8 days from Terra/MODIS. The implemented machine learning algorithm was built with ConvLSTM2D layers in Keras. The obtained results were compared with the new CNN LSTM model. Methods: Computational methods of Machine Learning, Artificial Neural Networks, Deep Learning. Results: The results show global AOT prediction obtained using satellite digital imagery as an input. Conclusion: The results show that the ConvLSTM developed model could be used for global AOT prediction, as well as for PM and COVID-19 transmission. AB - Uvod: Izložena je unapređena metoda koja uključuje Nasine satelitske snimke sa najnovijim modelom dubokog učenja koji se odnosi na problem predviđanja prostornovremenskih signala. Informacija o aerosolima sa satelitskih snimaka je vrlo značajna za predviđanje disperzije čestica u atmosferi i prenosa virusa COVID-19. Ulazni podaci MODAL2_E_AER_OD predstavljaju globalni AOT za osam dana sa Terra/MODIS. Algoritam mašinskog učenja je sačinjen od kompozitnih neuronskih slojeva ConvLSTM2D u biblioteci Keras. Dobijeni rezultati su upoređeni sa novim modelom CNN LSTM. Metode: Proračunske metode mašinskog učenja, veštačke neuronske mreže, duboko učenje. Rezultati: Rezultati prikazuju globalno predviđanje optičke debljine aerosola sa digitalnim satelitskim snimcima koji su korišćeni kao ulazni podaci. Zaključak: Pokazano je da je razvijeni model ConvLSTM pogodan za globalno predviđanje atmo-sferske debljine aerosola, kao i za prenos atmosferskih čestica i virusa COVID-19. T2 - Military Technical Courier : Vojnotehnički glasnik T1 - Satellite remote sensing and deep learning for aerosols prediction T1 - Satelitsko osmatranje i duboko učenje za predviđanje aerosola VL - 71 IS - 1 SP - 66 EP - 83 DO - 10.5937/vojtehg71-40391 ER -
@article{ author = "Mirkov, Nikola and Radivojević, Dušan and Lazović, Ivan and Ramadani, Uzahir and Nikezić, Dušan", year = "2023", abstract = "Introduction/purpose: The paper presents a new state-of-the-art method that involves NASA satellite imagery with the latest deep learning model for a spatiotemporal sequence forecasting problem. Satellite-retrieved aerosol information is very useful in many fields such as PM prediction or COVID-19 transmission. The input data set was MODAL2_E_AER_OD which presents global AOT for every 8 days from Terra/MODIS. The implemented machine learning algorithm was built with ConvLSTM2D layers in Keras. The obtained results were compared with the new CNN LSTM model. Methods: Computational methods of Machine Learning, Artificial Neural Networks, Deep Learning. Results: The results show global AOT prediction obtained using satellite digital imagery as an input. Conclusion: The results show that the ConvLSTM developed model could be used for global AOT prediction, as well as for PM and COVID-19 transmission., Uvod: Izložena je unapređena metoda koja uključuje Nasine satelitske snimke sa najnovijim modelom dubokog učenja koji se odnosi na problem predviđanja prostornovremenskih signala. Informacija o aerosolima sa satelitskih snimaka je vrlo značajna za predviđanje disperzije čestica u atmosferi i prenosa virusa COVID-19. Ulazni podaci MODAL2_E_AER_OD predstavljaju globalni AOT za osam dana sa Terra/MODIS. Algoritam mašinskog učenja je sačinjen od kompozitnih neuronskih slojeva ConvLSTM2D u biblioteci Keras. Dobijeni rezultati su upoređeni sa novim modelom CNN LSTM. Metode: Proračunske metode mašinskog učenja, veštačke neuronske mreže, duboko učenje. Rezultati: Rezultati prikazuju globalno predviđanje optičke debljine aerosola sa digitalnim satelitskim snimcima koji su korišćeni kao ulazni podaci. Zaključak: Pokazano je da je razvijeni model ConvLSTM pogodan za globalno predviđanje atmo-sferske debljine aerosola, kao i za prenos atmosferskih čestica i virusa COVID-19.", journal = "Military Technical Courier : Vojnotehnički glasnik", title = "Satellite remote sensing and deep learning for aerosols prediction, Satelitsko osmatranje i duboko učenje za predviđanje aerosola", volume = "71", number = "1", pages = "66-83", doi = "10.5937/vojtehg71-40391" }
Mirkov, N., Radivojević, D., Lazović, I., Ramadani, U.,& Nikezić, D.. (2023). Satellite remote sensing and deep learning for aerosols prediction. in Military Technical Courier : Vojnotehnički glasnik, 71(1), 66-83. https://doi.org/10.5937/vojtehg71-40391
Mirkov N, Radivojević D, Lazović I, Ramadani U, Nikezić D. Satellite remote sensing and deep learning for aerosols prediction. in Military Technical Courier : Vojnotehnički glasnik. 2023;71(1):66-83. doi:10.5937/vojtehg71-40391 .
Mirkov, Nikola, Radivojević, Dušan, Lazović, Ivan, Ramadani, Uzahir, Nikezić, Dušan, "Satellite remote sensing and deep learning for aerosols prediction" in Military Technical Courier : Vojnotehnički glasnik, 71, no. 1 (2023):66-83, https://doi.org/10.5937/vojtehg71-40391 . .