Assessment of water quality in natural river based on HJ1A/1B CCD multi-spectral remote sensing data

This study selects the typical middle and lower reaches of Han River as the study area and focuses on water quality evaluation methods and water quality evaluation of the surface water of the river basin. On the basis of the field survey, the author conducted a water quality sampling survey in the study area in spring and summer in 2012. The main excessive factors in the study area are determined as TN and TP. Using HJ1A/1B CCD multi-spectral data, the multiple linear regression inversion model and neural network inversion model are established for content of TN and TP. In accordance with these inversion results, the single factor water quality identification indexes in the study area are obtained. The results show that, BP neural network model boasts the highest inversion accuracy and that the single factor water quality identification indexes resulting from its inversion results are highly accurate, reliable and applicable, which can really reflect the changes in water quality and better realize the evaluation of water quality in the study area. Water quality evaluation results show that the water pollution in the study area is organic pollution; the water quality of Han River experiences large differences in different regions and seasons; downstream indexes are superior to upstream indexes, and the indexes in summer are superior to those in spring; the TN index seriously exceeds the standard in spring and the TP index seriously exceeds the standard in some regions.