Modeling stream dissolved organic carbon concentrations during spring flood in the boreal forest: A simple empirical approach for regional predictions

Changes in dissolved organic carbon (DOC) concentration are clearly seen for streams in which chemistry is measured on a high-frequency/episode basis, but these high-frequency data are not available in long-term monitoring programs. Here we develop statistical models to predict DOC concentrations during spring flood from easily available geographic information system data and base flow chemistry. Two response variables were studied, the extreme DOC concentration and the concentration during peak flood. Ninety-seven streams in boreal Scandinavia in two different ecoregions with substantially different mean water chemistry and landscape characteristics (covering a large climatic gradient) were used to construct models where 56% of the extreme DOC concentration and 63% of the concentration during peak flood were explained by altitude. This highlights important regional drivers (gradients in altitude, runoff, precipitation, temperature) of material flux. Spring flood extreme DOC concentration could be predicted from only base flow chemistry (r(2) = 0.71) or from landscape data (r(2) = 0 .74) but combining them increased the proportion of explained variance to 87%. The best model included base flow DOC (positive), mean annual runoff (negative), and wetland coverage (positive). The root mean square error was 1.18 mg L-1 for both response variables. The different ecoregions were successfully combined into the same regression models, yielding a single approach that works across much of boreal Scandinavia.