| Abstract: | Let X(t) be the flow of a certain river at time t. A geometric Brownian motion process is used as a model for X(t) and is found to give very good forecasts of future flows. The forecasted values generated by this one‐dimensional model are compared with those provided by a deterministic model that requires the evaluation of 18 entries. Based on two important criteria, the stochastic model is superior, on average, to the deterministic model for forecasts up to 4 days ahead. Copyright © 2002 John Wiley & Sons, Ltd. |
