عنوان مقاله [English]
Considering the importance and status of monitoring the quality characteristics of water resources, it seems necessary to use technologies such as remote sensing that increase the speed and accuracy of water quality monitoring information. Turbidity is one of the key water quality characteristics which is one of the main sources of sediment in water bodies and directly affects issues such as sediment transport, aquatic life, and management of lakes and reservoirs. This makes water turbidity monitoring an important concern in the field of water resources management. Traditional and laboratory methods of monitoring are accurate but time-consuming and expensive. The use of scientific capabilities such as remote sensing can therefore be beneficial. In this research experimental methods and models developed based on neural network algorithms are used to estimate opacity parameter zoning maps. The main goal of this study is to investigate and compare different band combinations to estimate the turbidity parameter using the spectral data of Sentinel-2 and Landsat-8 satellite images. Considering the importance of the mentioned quality parameter and also the location of the study area, ground measurement of these parameters was a challenge; As a result, it is necessary and practical to use neural network and remote sensing capabilities for modeling. According to the investigations carried out in this study it can be concluded that due to the spectral data of Sentinel-2 and Landsat-8 satellite images, the zoning maps of water turbidity parameter can be estimated with good accuracy. Four different forms of band combinations (single bands, band ratio, spectral derivative ratio and normal difference index) have been used for which the detection coefficient value were 0.86, 0.83, 0.83 and 0.89 and the RMSE error were 3.26, 3.52, 3.43 and 2.79, respectively. Accordingly, each of the compounds is considered independently as the model input in the modeling process. The results showed that the band combination of the normalized difference index estimates much better results. However, considering the fact that the single band combinations also had good results with little difference from the normal difference index combinations’ derivative model and since it could reduce calculations, single band combinations can be used as best band combinations to estimate the water turbidity parameter in this study area.