Monitoring Water Quality Based On 3D-EEM And RS On Arid Region In The Ebinur Lake Watershed

The water resources management problem in the arid region of Central Asia is particularly prominent.It is imminent to study the water quality in the arid region of Central Asia under the background of global warming.The water quality of the Ebinur Lake Basin is rapidly deteriorating,which seriously threatens the smooth implementation of the major strategy in the West and the sustained economic development of the minority areas.Therefore,study on the selection of the Water management problem has become prominent in arid central Asia in recent years,and studying the water quality under the background of global warming in arid central Asia is extremely urgent.As Ebinur Lake Watershed is located in the northern foot of Tian Mountain,global warming and human activities in this region have caused serious environmental problems and have affected the economic growth and the implementation of the national strategy called One Belt One Road.Thus,this study selected Ebinur Lake Watershed as a case study,took Watershed Water Quality Monitoring as a main line,used low altitude remote sensing and three dimensional fluorescence technologies as method,took Neural network technology as the medium,revealed the experimental model of joint technology on the basis of stereoscopic experiment,studied internal responding regulation of water quality parameters,natural reflection characteristics of water constituents,and water reflectance,analyzed the relationship between water quality parameters and natural optical characteristics of water constituents to achieve using quantitative spectral method to reveal water quality,to set up a water quality monitoring model integrating low altitude remote sensing and three dimensional fluorescence technologies.Ebinur Lake Basin as the research target,with "water monitoring" as the main line,remote sensing(Air ground joint)and three-dimensional fluorescence technique,three-dimensional experimental study combined with technology integration mode based on the intrinsic response laws between the groups of water-water quality indicators reveal the inherent reflection the apparent characteristics of water at a rate of three,quantitative characterization of water quality indexes and water components of inherent optical characteristics,the apparent relationship between the reflectance of water.On this basis,the water quality monitoring model of the fusion of low altitude remote sensing and three dimensional fluorescence technology is established and the water quality is evaluated.The conclusions are as follows:(1)This study developed evaluation water quality index(WQI)for Central Asia arid areas of water quality evaluation based on combined with the surface water quality monitoring standards of Chinese and WHO and water reflection characteristics of the spectrum for water material content.WQI is used in Ebinur Lake watershed,the results as water quality of upstream is good in Jinghe rivers,Bortala rivers,it's meet the people drinking,Ebinur lake water is the worst,the WQI value is as high as 4000.(2)Spectral characteristics is the basis of the implementation of water quality monitoring based on remote sensing and 3D fluorescence technology.Fractional derivative,amplification combining techniques are used,in order to better mining the information of spectral data,to achieve the extraction of water spectral information,realizes the matching of Sentinel-2,Sentinel-3 data with ground measured hyperspectral,and three-dimensional fluorescence spectral data bands,and lays the foundation for fusion of "satellite-ground" combined spectral data and feature extraction of spectral data.Spectral data fusion is the basis for the implementation of multi-technology integration.In this study,multi-source spectral data fusion has been achieved through machine learning amplification and combination algorithms,spectral indices(DI,RI and NDI),and fluorescence baselines,and water quality suitable for arid areas has been extracted.The relationship between the sensitive spectral data monitored,the spectral sensitivity index and the water quality WQI in 0.01 significant levels of R2 were greater than 0.6.(3)Combined with the Sentinel 3OLCI 0-order differential value and fluorescence baseline data(FLH)implementation of water quality monitoring based on remote sensing and fluorescence spectroscopy technology,the results found that the optimal and best model for the PSO algorithm optimized SVR model.The model has an R2 of 0.81 and an RPD of 1.72.The stability of the model is strong.However,the accuracy of the model is lower than that of the Sentinel 2-3 derivative data.The inversion mapping of Ebinur Lake was implemented based on this model,.The study provided data foundation and theoretical guidance for water quality management and monitoring of Ebinur Lake.(4)Combining remote sensing hyperspectral sensitivity band,remote sensing hyperspectral exponent,three-dimensional fluorescence component data and three-dimensional fluorescence spectrum index as input data to build an estimation model based on 1.6-order differential,and using the GS,GA,POS optimization algorithm to support the Ebinur Lake watershed River Water quality estimation model finds that the optimal and optimal model is the PSO algorithm optimized SVR model.The model's R2 is 0.92,and the RPD is 2.89.The model has strong stability and reliable prediction capability.In terms of application,this study can improve the monitoring accuracy of water quality indicators by means of joint technologies,and break through the bottleneck of single technology in the low accuracy of water quality monitoring and incomplete monitoring parameters.To promote the monitoring and research process of water quality in inland arid regions and provide guarantees for water safety in the west and the smooth implementation of the national“One Belt and One Road" major strategy.In terms of theory,this study introduced a spectral amplification combination algorithm to achieve the rearrangement of spectral data features,enriched the spectral research methods,the application of fractional differentials in data processing,and broke through the limitations of integer order differential in spectral data processing for a long time,to lay the foundation for the processing of spectral data.