Loading...

A novel analysis of critical water pollution in the transboundary Aras River using the Sentinel-2 satellite images and ANNs

Fouladi Osgouei, H ; Sharif University of Technology | 2022

137 Viewed
  1. Type of Document: Article
  2. DOI: 10.1007/s13762-022-04129-4
  3. Publisher: Springer Science and Business Media Deutschland GmbH , 2022
  4. Abstract:
  5. Recently, remote sensing considered as important tool in studies of water quality issues. The Aras River flows across a transboundary basin in northern Iran. In this study, the aim is to model the water quality parameters (WQPs) using remote sensing and an artificial neural network (ANN), which is a new method proposed to find WQPs based on multivariate regression approaches. The relationship between WQPs and digital data from the Sentinel-2 satellite was determined to estimate and map the WQPs in this river. Using the field data and digital image data, the obtained results show that multivariate regression approaches and high-resolution remote sensing could monitor and predict the distribution of WQPs in this basin. There was a meaningful correlation between bands calculated from image data and the river’s WQPs, including electrical conductivity (EC) (R2 = 0.91), Na+ (R2 = 0.81), Mg2+ (R2 = 0.90), SO42− (R2 = 0.89), Cl− (R2 = 0.81), and Ca2+ concentrations (R2 = 0.60). Likewise, the WQPs including EC, SO42−, Mg2+, and Cl− in the river are predicted using ANN. This powerful model is built for the rapid assessment and forecasting of experimentally mentioned WQPs at any location in the domain of interest. The results represent the good ability of ANN to simulate the mentioned WQPs. Simulation accuracy, measured by the determination coefficient (DC), varies between 0.982 and 0. 991 for training and overfitting EC test data. It is concluded that the management of transboundary rivers can be achieved by using remote sensing techniques and machine learning methods, which help the beneficiary countries in the sustainable operation. © 2022, The Author(s) under exclusive licence to Iranian Society of Environmentalists (IRSEN) and Science and Research Branch, Islamic Azad University
  6. Keywords:
  7. ANN ; Multivariate regression approaches ; Water pollution ; Learning systems ; Neural networks ; Regression analysis ; Remote sensing ; River pollution ; Rivers ; Water quality ; Electrical conductivity ; Multivariate regression ; Multivariate regression approach ; Remote-sensing ; River flow ; Satellite images ; Sentinel-2 ; Trans-boundary ; Water quality issues ; Water quality parameters ; Sensitivity analysis ; Artificial neural network ; Concentration (composition) ; Multivariate analysis ; River water ; Satellite imagery ; Sentinel ; Araks River
  8. Source: International Journal of Environmental Science and Technology ; Volume 19, Issue 9 , 2022 , Pages 9011-9026 ; 17351472 (ISSN)
  9. URL: https://link.springer.com/article/10.1007/s13762-022-04129-4