Identifying linear and non-linear behaviour in reduced complexity modelling output using surrogate data methods

Validation of model output is an important issue in the environmental sciences. This is particularly the case for reduced complexity modelling approaches where verification of the underlying equations is often problematic. Hence, methods that go beyond validation based on average values or the histogram of model output, are clearly advantageous. In this paper we show that a validation method based on the serial properties of time–space data, and developed in geomorphology, is a useful alternative but is nevertheless still insensitive to certain asymmetric characteristics of the signal. Hence, use of this method for validation may be enhanced by the adoption of an accompanying test for asymmetry in the model output.