Search

This paper examines models for the errors in forecasts of river and (or) watershed flows produced by the PREVIS forecasting system, which is used by Alcan, among other companies. We analyzed the following statistical models: generalized Pareto, Laplace, and Gaussian distributions, depending on the flow value forecasted by PREVIS. These models enable us to quantify the precision of the forecasts produced by PREVIS, as well as the risk of seeing the flow exceed a certain critical threshold, given the forecasted flow. [ABSTRACT FROM AUTHOR]