NONLINEAR, DISCRETE FLOOD EVENT MODELS .2. ASSESSMENT OF STATISTICAL NONLINEARITY

The first paper (Part 1) of this series presented a Bayesian procedure for the estimation of parameters in nonlinear, discrete flood event models. Part 2 begins with a discussion of the concept of nonlinearity in parameter estimation, its consequences, and the need to assess its extent. Three measures of nonlinearity are considered. They are Beale's measure, a bias calculation, and maximum curvature measures. A case study is presented, using the model and data described in Part 1. The results show quite clearly that care is required in the application of all three measures to calibrated flood models, and in the interpretation of the measured values.