Research On Establishment And Optimization Of Hyperspectral Water Quality Inversion Model

Water is the basic need to maintain life and health.Human production and life are inseparable from water.Excessive nitrogen and phosphorus in the water body lead to eutrophication of the water body,resulting in excess nutrients in the water,resulting in deterioration of water quality and a wide range of impacts.The traditional water quality monitoring method requires water quality testing personnel to collect water samples at the sampling point,and send the collected water samples to the relevant quality inspection agencies for chemical analysis as soon as possible.This method requires a lot of manpower and material resources,and cannot be achieved.Large-scale real-time water quality monitoring.Hyperspectral image data has high spectral resolution and rich and continuous data information,which can describe the features of ground objects more finely,and can be used to study water pollution and water quality parameter content.Hyperspectral remote sensing is more and more widely used in the field of inland water quality monitoring.Based on this research,in order to reduce the influence of specific factors of outdoor water bodies,a laboratory experimental system is built by simulating external conditions in the laboratory,and the laboratory hyperspectral water quality is constructed.The inversion model can improve the inversion accuracy and expand the scope of application.The details are as follows:（1）Using laboratory simulation experiments,40 standard solutions of sodium phosphate with different concentration gradients in the concentration range of0-2.5mg/L and 40 standard solutions of ammonium chloride with different concentration gradients in the concentration range of 0-20mg/L were prepared.The hyperspectral images of all standard solutions were obtained,and the spectral responses of water quality parameters total phosphorus and total nitrogen were analyzed,and the corresponding sensitive bands were found around 420 nm,720 nm and 410 nm,respectively.The data set of hyperspectral water quality inversion was established by principal component analysis,and the laboratory hyperspectral water quality inversion model was constructed by means of radiometric calibration of hyperspectral images,Savitzky-Golay（SG）filtering,and BP artificial neural network method.The coefficients of determination of the constructed laboratory hyperspectral water quality inversion model for total phosphorus and total nitrogen were 0.9802 and0.8602.（2）In order to verify the inversion accuracy and applicable range of the laboratory hyperspectral water quality inversion model,a certain indoor river in Yixing,Jiangsu Province was taken as the research object,and the model and data set were applied to the outdoor UAV-mounted hyperspectral imaging system.On the data of,five points were selected to calculate the results separately,and the inversion accuracies of the mean values of total phosphorus and total nitrogen concentrations were 95.00% and 93.52%.The outdoor hyperspectral water quality inversion model constructed by using the traditional method to directly draw water from the observation points of the river to be tested inverts the mean values of total phosphorus and total nitrogen concentrations obtained from the same five observation points,and the inversion accuracies are 86.87% and 86.48%.Comparing the inversion results of the two groups,it is found that the inversion accuracy of 90% of the spectral inversion results obtained by the laboratory hyperspectral water quality inversion model constructed in this study is slightly higher than the inversion accuracy of the outdoor water quality inversion model,which confirms that the research can be carried out.Effective prediction of the content of total phosphorus and total nitrogen in the river to be measured can also provide certain technical support for the hyperspectral remote sensing inversion of total phosphorus and total nitrogen in water.