Functional classification and evaluation of hydrographs based on Multicomponent Mapping (Mx)

Abstract

The literature offers a wealth of different performance measures to evaluate model results. However, visual graphical evaluation based on the simple plotting of two curves is still the most intuitive and favoured approach by many modellers. This paper introduces a performance measure, which is based on this method (the Multicomponent Mapping (M^x )). The hydrograph is subdivided into box areas to which membership values according to the distance to the hydrograph are assigned. The box size is influenced by the expected effective observation error structure. It is demonstrated that this method can be used not only to calculate a quantitative performance measure but also to classify hydrograph outputs of Monte Carlo runs into functional classes in order to enable, for example, the cascading of uncertainties in large modelling systems. A comparison to traditional measures like the Nash‐Sutcliffe and the Cumulative Absolute Error is performed.

Keywords:

• Uncertainty
• model calibration
• hydrograph evaluation
• Nash‐Sutcliffe
• cluster analysis