| Carbon is the largest component of biogenic elements.Organic carbon is the main component of organic matter and the main form of ecosystem material and energy storage.Dissolved organic carbon(DOC)is the largest organic carbon reservoir in the water,which is equivalent to the atmosphere carbon storage,and plays an important role in the water carbon cycle.The study of the concentration and time and space distribution of DOC is of great significance to study the decay of UV-B radiation,carbon in inland watert and the geochemical cycle of carbon.Based on the field observation and indoor analysis,this paper analyzed the variation law of DOC source and CDOM fluorescence characteristics of inland lake water,and discussed the relationship between CDOM fluorescence characteristics and absorption characteristics in water body.On this basis,a method of water classification based on CDOM optical characteristics was proposed,and the optical estimation model of DOC concentration was established for each type of water body.The conclusions drawn mainly include the following aspects:(1)Using the parallel factor analysis method,the CDOM three-dimensional fluorescence matrix of Taihu Lake and Chaohu Lake was decomposed into four components,which include three humus components and one kind of protein component.The humus components mainly include peak A ultraviolet light humus,peak C visible humus,and peak M marine humus;the class of protein component includes peak T-type tryptophan.CDOM fluorescence intensity and the proportion of humus in the lake area,which is influenced by the rivers,are larger than that of the lake.There was a good correlation between the fluorescence intensity and the absorption coefficient of CDOM,and the fluorescence intensity was positively correlated with acdom(355)(R2=0.62).The shape of the CDOM absorption coefficient curve can indicate the composition of CDOM,and the absorption coefficient curve of CDOM is segmented by spectral differentiation technique.The ratio of CDOM absorption curve slope ratio S260-290/S240-260 and the proportion of humus in CDOM have good correlation(R2=0.66).(2)Inland lakes can be divided into two types by S260-290/S240-260:when S260-290/S240-260 is less than 1.1,the source of DOC is mainly endogenous biological effect and DOC concentration and Chla have a good correlation;when S260-290/S240-260 is more than 1.1,the DOC contribution of rivers is significant,and humus content is higher.The DOC concentration of two kinds of water bodies can be expressed by CDOM absorption coefficient acdom(355)or acdom(412):the model determining coefficient R2 values of the typel water body are above 0.85,MAPE values are lower than 15.15%and the root mean square error RMSE values are within 0.56mg/L.The accuracy of the DOC concentration estimation of the model was improved obviously after classification.For type 2 water body,The RMSE values are less than 0.66,MAPE values are less than 11.31%,RMSE values are within 0.54 mg/L,and the accuracy of the DOC concentration estimation is higher than that before classification,which shows that the method of classification modeling can effectively improve the estimation accuracy of DOC concentration in inland lake water.(3)The method of water classification based on remote sensing reflectance was established by using random forest regression algorithm.The remote sensing reflectances were used to estimate S260-290/S240-260 and the water body was divided into two categories,DOC source mainly affected by endogenous biological function of the lake or mainly affected by the external rivers.For each type of the water body,a DOC concentration estimation model based on remote sensing reflectance Rrs(710)/Rrs(560)was constructed.The independent validation data shows that the RMSE of the model is 0.89 mg/L and the MAPE is 20.4%.The results show that the method of estimating the DOC concentration in the water body based on different DOC sources can better estimate the DOC concentration in the inland lake water. |
PDF Full Text Request