Dissolved organic carbon and its potential predictors in eutrophic lakes

Abstract

Understanding of the true role of lakes in the global carbon cycle requires reliable estimates of dissolved organic carbon (DOC) and there is a strong need to develop remote sensing methods for mapping lake carbon content at larger regional and global scales. Part of DOC is optically inactive. Therefore, lake DOC content cannot be mapped directly. The objectives of the current study were to estimate the relationships of DOC and other water and environmental variables in order to find the best proxy for remote sensing mapping of lake DOC. The Boosted Regression Trees approach was used to clarify in which relative proportions different water and environmental variables determine DOC. In a studied large and shallow eutrophic lake the concentrations of DOC and coloured dissolved organic matter (CDOM) were rather high while the seasonal and interannual variability of DOC concentrations was small. The relationships between DOC and other water and environmental variables varied seasonally and interannually and it was challenging to find proxies for describing seasonal cycle of DOC. Chlorophyll a (Chl a), total suspended matter and Secchi depth were correlated with DOC and therefore are possible proxies for remote sensing of seasonal changes of DOC in ice free period, while for long term interannual changes transparency-related variables are relevant as DOC proxies. CDOM did not appear to be a good predictor of the seasonality of DOC concentration in Lake Võrtsjärv since the CDOM-DOC coupling varied seasonally. However, combining the data from Võrtsjärv with the published data from six other eutrophic lakes in the world showed that CDOM was the most powerful predictor of DOC and can be used in remote sensing of DOC concentrations in eutrophic lakes.

Keywords: Boosted Regression Trees; Coloured dissolved organic matter; Dissolved organic carbon; Eutrophic lakes; Remote sensing.