Research On CDOM Semi-analytical Remote Sensing Inversion Model Of Inland Turbid Water

Chromophoric Dissolved Organic Matter(CDOM)plays an important role in the biogeochemical processes such as the distribution,migration and transformation of organic pollutants,nutrients,and heavy metals,and its concentration content can indicate the degree of contamination of water bodies and eutrophication status to some extent.Remote sensing technology has played an important role in the monitoring of water color elements such as CDOM and water ecosystems.However,for inland turbid water bodies,in order to obtain high CDOM inversion accuracy,it is necessary to strengthen the research on the optical properties of CDOM in regional water bodies and to futher establish CDOM analysis/semi-analysis remote sensing inversion models based on relevant theories.In this paper,Taihu Lake was taken as the research area,the temporal and spatial distribution characteristics of CDOM absorption spectrum characteristics were analyzed,combined with the measurement data in the field and in the room.The correlation between the CDOM concentration(characterized by the CDOM absorption coefficient ag(440)at 440 nm)and the values of the parameters(suspension,chlorophyll-a and dissolved organic carbon concentration and total particulate matter absorption coefficient,non-pigmented particulates and phytoplankton pigments absorption coefficient)were investigated in detail.Based on the above,the parameters of the three semi-analytical inversion models established by Lee et al.(2002),Li et al.(2014)and Dong et al.(2012)were adjusted,and thus The absorption coefficient at(λ)and its decomposition coefficient(absorption coefficient of phytoplankton pigment particles aph(λ),sum of non-pigments particulate matter and CDOM absorption coefficient adg(λ)etc.)were estimated combined with the bio-optical model and the intrinsic optical data.Combining the three CDOM separation methods:based on the total particle absorption coefficient,the non-pigment particle absorption coefficient,and the sumof the pigment particle and the CDOM absorption coefficient,a corresponding three-category CDOM semi-analysis remote sensing inversion model was established.The best model was selected and improved,the inversion accuracy of the improved model was greatly improved,providing a scientific basis for remote sensing monitoring of CDOM in Taihu Lake.The main conclusions are as follows:(1)Overall,the ag(440)ranged from 0.28-1.85 m-1 with an average of 0.84±0.4 m-1.In terms of time,the average value of ag(440)showed a decreasing trend from 2004 to 2010 and increased in 2017.Spatially,the average values of ag(440)in the northern and western lakes affected by the import of land resources were characterized as follows:Northern Lake>Western Lake>Eastern Lake>Central Lake>Southern Lake.In addition,there was a negative correlation between the Sg value of the CDOM absorption spectrum and ag(440),and the average spatial-temporal distribution feature of the CDOM was roughly opposite to ag(440).The average contribution of CDOM absorption at 440 nm(0.27±0.58%)was much lower than the average contribution of non-pigmented particulates(0.54±0.12%).Differences in temporal and spatial correlations between ag(440)and suspended matter,chlorophyll-a and dissolved organic carbon concentration and total particulate matter absorption coefficient,non-pigmentary particles and phytoplankton pigment particles absorption coefficient were significant,suggesting that the source of CDOM was complex and diverse under different time and space conditions.(2)Through the evaluation of the inversion accuracy of the established three-category CDOM semi-analysis remote sensing inversion model,the inversion accuracy of the CDOM inversion model with the separation method based on the sum of absorption coefficients of pigment particle and CDOM was highest in all CDOM inversion models.In addition,the inversion accuracy of the CDOM inversion model with separation method based on the separation method of the total particles a bsorption coefficient was higher than that of the CDOM inversion model with the separation method based on non-pigmented particles absorption coefficient.The inversion accuracy of the CDOM inversion model based on the reference band 726 nm was the highest.(3)Based on the CDOM inversion model established above,the model with higher inversion accuracy was selected for improvement.Although the inversion accuracy of the CDOM inversion model established with the separation method based on the sum of pigment particles and CDOM absorption coefficient was highest,since the absorption value of the pigment particles in this paper was too low and the estimation error was large,the model was not further developed.Instead,the three inversion models with relatively high inversion accuracy in the CDOM inversion model established with separation method based on the total particle absorption coefficient were further improved.In general,the inversion accuracy of the three improved CDOM inversion models was greatly improved compared with that before improvement.The same point with the inversion accuracy evaluation of the pre-improvement model was that the inversion accuracy of the CDOM inversion model based on the reference band 726 nm was also the highest.Combined with the improved Sg inversion model,the CDOM absorption spectrum was parameterized,and the result showed that the inversion accuracy gradually decreases with the increase of the band.