| Dongting Lake and Poyang Lake,as the two main river-connected lakes in the middle reaches of the Yangtze River,have always had the problem of eutrophication,and total nitrogen(TN)and total phosphorus(TP)are important indicators reflecting the eutrophication status of lakes.Traditional water environment monitoring methods are time-consuming and laborious,and cannot monitor and assess the water environment conditions of large-scale water bodies.Satellite remote sensing monitoring has the characteristics of intuition,efficiency and large range,and has important advantages in TN and TP monitoring.Based on the machine learning method,this study established the remote sensing inversion model of TN and TP by using Landsat satellite images and measured data,evaluated the accuracy of the model,and selected the optimal model.Based on the optimal model,the spatial distribution of TN and TP in Dongting Lake and Poyang Lake from 1996 to 2021 was obtained,the temporal and spatial variation characteristics and trends were analyzed,and the influencing factors were analyzed and discussed in detail.The main research work and results are as follows:(1)The Landsat data is corrected by the dark pixel atmospheric correction method and the sensor difference correction method,which improves the accuracy of the model input data.Based on the original spectrum,28 kinds of spectral features were constructed.By calculating the correlation between the spectral features and TN and TP,the corresponding optimal spectral features are selected to improve the sensitivity of the input data.(2)The remote sensing inversion models of TN and TP are constructed based on the optimal spectral features by using machine learning method,and the optimal inversion model is selected by evaluating the accuracy of the model.The coefficients of determination(R~2)of Dongting Lake and Poyang Lake TN models were 0.73 and 0.67,the root mean square errors(RMSE)were 0.48 mg/L and 0.30 mg/L,and the Mean Absolute Percentage Errors(MAPE)were 0.20and 0.29.The accuracy of the Dongting Lake and Poyang Lake TP models were R~2=0.70 and0.69,RMSE=0.057 mg/L and 0.058mg/L,MAPE=0.23 and 0.47.(3)Using Landsat image data and based on the inversion model,the concentration distribution data of TN and TP in Dongting Lake and Poyang Lake were respectively obtained,and their temporal and spatial variation characteristics were analyzed.The TN of Dongting Lake showed a fluctuating upward trend as a whole,and the increase was significant in the wet season.The overall TP of Dongting Lake showed a decreasing trend,and the TP in spring was higher than that in other seasons,and the TP in the dry season showed a significant downward trend.The spatiotemporal variation trend of TN in Poyang Lake was the same as that in Dongting Lake,and the TP showed a fluctuating and decreasing trend as a whole.In the eastern half of the lake,TP decreased significantly,and the concentration in spring and summer was higher than that in autumn and winter,and the decline trend was significant in the dry season.(4)The influence of hydrometeorology and human activities on TN and TP of Dongting Lake and Poyang Lake was analyzed,and it was found that hydrological elements were positively correlated with TN and negatively correlated with TP.The construction of the Three Gorges Dam has changed the hydrological conditions and sandy characteristics of the lake,which has reduced the deposition of nitrogen and phosphorus elements in the lake and increased the concentration;agricultural wastewater has carried a large amount of nitrogen-containing substances into the lake,increasing the concentration of TN in the lake. |
PDF Full Text Request