The implication of information theory for estimating and detangling hydrological modelling uncertainty

In the present article, the Theory of Information is applied to hydrological time series to quantify their informational content. Uncertainty is defined as the gap between available and required knowledge. This approach is more intuitive than the traditional treatment of uncertainty as confidence intervals. Moreover, a theoretical framework is developed, in which the components of total uncertainty are explicitly defined and a novel method of splitting epistemic uncertainty into its structural and parametric components is applied. The extensive computations to detangle these components are compacted into a single methodology, which is applied to two hydrological basins, to evaluate various methods of aerial integration of rainfall point data. Five lumped monthly models are utilized to simulate the runoff. For each model, a characteristic curve can be drawn which can be used to select the appropriate model depending on the available information of the data set.