内陆淡水湖博斯腾湖水质遥感反演及时空演变特征OA北大核心CSTPCD

Lakes serve as important water sources in arid regions,and changes in water quality of lakes in these areas play a crucial role in maintaining regional ecological balance and water cycle.Therefore,focusing on Chi-na's inland freshwater lake,Bosten Lake,Xinjiang,this study conducted pre-processing on Landsat series remote sensing images from 2001 to 2020,including atmospheric correction and radiometric calibration.An improved normalized index method was used to extract water bodies and analyze their area change trend.Combined with measured mineralization and chlorophyll-a concentration data,an empirical method was employed to compare and analyze the correlation between band and band combination reflectance values with mineralization and chlo-rophyll-a,selecting the bands with the best correlation to establish models.Mineralization inversion models were constructed using bands B5,B3/B4,and B3+B4,while chlorophyll-a concentration inversion models were devel-oped using bands B2,B3,and B2×B3.Subsequently,the accuracy of the selected two inversion models was vali-dated using determination coefficients and root mean square errors.The results indicate:(1)The lake area gradu-ally decreased from 2001 to 2013 and then began to increase in two phases with 2013 as the turning point.(2)Bosten Lake exhibits a spatial distribution of mineralization concentration higher in the lake center and lower in the lake periphery,while chlorophyll-a concentration distribution is opposite to mineralization distribution.(3)Over a long time series,mineralization concentration and chlorophyll-a concentration in Bosten Lake show an in-creasing followed by decreasing trend at an annual scale.The maximum mineralization concentration was 1023.8 mg·L-1 in 2013 and the maximum chlorophyll-a concentration was 5.04µg·L-1 in 2015.In the past few decades,significant progress has been made in China's water quality monitoring and management in Bosten Lake.Howev-er,further improvements are needed in terms of spatiotemporal coverage,index lists,the integration of human-na-ture interactions,inversion accuracy,and model generalization.