Application of Markov Chain simulation for model calibration

Using a 3-parameter model that predicts the alongshore current flowing on a beach we apply a stochastic technique known as Markov Chain simulation to find best-fit parameter values and their uncertainty. Because this technique is computationally demanding, we are particularly interested to see how the best-fit values and their uncertainty are affected by the amount and characteristics of the data added to the calibration procedure. The findings include that the amount of data greatly affects the best-fit values and that the common assumption that parameters are time-independent is violated (at least, for the present model). The parameter uncertainty is additionally used to obtain inference about the model