Parameter Uncertainty

In order to perform a stochastic simulation range and distribution of each model parameter must be defined. REBEKA2 allows choosing uniform, normal, lognormal and triangle distributions. Ranges can be defined as absolute or relative. The above screenshot shows a part of the model parameter table that allows defining the parameter range and distribution.

table of parameter uncertainties

It is currently not possible to consider correlations between parameters although they exist (e.g. between reservoir time constant and contributing area and reservoir time constant). It is very difficult to take parameter correlation into account because either the correlation is unknown or not well defined. The program RANDSAMP of the library UNCSIM which is used to generate the parameter samples allows accounting for parameter correlation (Reichert, 2001) in principle but it is not implemented in REBEKA2. It will be probably considered in a future version.


Reichert P. (2001). UNCSIM - A program package for uncertainty analysis and Bayesian inference: User Manual. Dübendorf, Switzerland, EAWAG: 36.