Research On The Inversion Of Three Water Quality Parameters In Caohai Lake, Guizhou Based On Zhuhai-1 Hyperspectral Satellit

Cachia is the largest karst plateau freshwater wetland in Guizhou and one of the three major plateau freshwater lakes in China.Affected by economic and social development and climate change,the aquatic ecosystem of Caohai wetland is facing major threats,and long-term monitoring of water quality is urgently needed.However,Caohai Wetland has a vast water area and complex hydrodynamic conditions.Traditional monitoring methods not only consume a large amount of manpower and material resources,but also cannot accurately reflect the water quality status of the entire wetland water area.Remote sensing has the advantages of large detection range,low cost,and high efficiency.It can quickly and continuously monitor large areas of water.In particular,hyperspectral remote sensing has the characteristics of high accuracy,multi band,and large amount of information.It can achieve quantitative analysis of water quality parameters,reflecting the spatial and temporal distribution and changes of water quality.In this paper,using the hyperspectral satellite image of Zhuhai-1,the total nitrogen（TN）,total phosphorus（TP）,and chlorophyll（Chl.a）parameters that reflect the state of water quality were measured in the field at the same time.The sensitive bands of the three water quality parameters are determined through analysis and inversion models are constructed respectively.Then,the water quality of Caohai since 2019 is retrieved and analyzed to provide data support for the water environment governance of Caohai.（1）Pearson correlation analysis was conducted using measured total nitrogen,total phosphorus,and chlorophyll-a concentration data and water surface spectral data extracted from the Zhuhai 1 satellite image to obtain sensitive wavebands for three water quality parameters.The analysis results show that the sensitive bands of total nitrogen include b2（466 nm）,b4（500 nm）,b9（580 nm）,b10（596 nm）,and b12（640nm）.The optimal band combination is b2+b9+b12,and the correlation with total nitrogen is-0.775;The sensitive wavebands of total phosphorus include b2（466 nm）,b10（596 nm）,b15（686 nm）,b22（780 nm）,and b31（926 nm）.The optimal waveband combination is b22+b31,with a correlation of-0.618 with total phosphorus;The sensitive wavebands of chlorophyll a include b21（776 nm）,b22（780 nm）,b24（820nm）,b29（896 nm）,and b32（940 nm）.The optimal waveband combination is（b21+b22）/b21,and the correlation with chlorophyll a is-0.701.（2）Using the optimal band combination of total nitrogen,total phosphorus,and chlorophyll-a as independent variables,and the measured water quality parameters as dependent variables,fitting analysis was conducted to establish water quality parameter inversion models,and the inversion models were verified by sampling data and monitoring vessel data.The results show that among the total nitrogen inversion models,the best fitting effect is:y=85.7960 x^0.4841,and the average absolute error and average relative error of the model are 0.39 mg/L and 19%,respectively;Among the total phosphorus inversion models,the model with the best fitting effect is:y=-2E-07x²+0.0002x+0.0621,and the average absolute error and average relative error of the model are 0.006 mg/L and 15%,respectively;Among the chlorophyll a inversion models,the best fitting model is:y=-20.843x²+57.085x-16.634,with an average absolute error and an average relative error of 1.67μg/L and 20%,respectively.The accuracy of the three models can be used to retrieve the water quality of Caohai wetland.（3）Using the model to retrieve water quality from 2019 to 2022,the results show that the concentrations of total nitrogen,total phosphorus,and chlorophyll-a have a similar trend of change since 2019,gradually decreasing over time.In terms of spatial distribution,the concentration in the east is higher,the concentration in the west is lower,the concentration near the water inlet is higher,and the concentration away from the water inlet is lower.