Modeling and forecasting the peak flows of a river

A stochastic model is found for the value of the peak flows of the Mistassibi river in Quebec, Canada, when the river is in spate. Next, the objective is to forecast the value of the coming peak flow about four days in advance, when the flow begins to show a marked increase. Both the stochastic model proposed in the paper and a model based on linear regression are used to produce the forecasts. The quality of the forecasts is assessed by considering the standard errors and the peak criterion. The forecasts are much more accurate than those obtained by taking the mean value of the previous peak flows.

Mots clés

Stochastic modeling; Lognormal distribution; Linear regression; Correlation coefficient; Peak criterion; Hydrology; Science; Mathematics; General and Civil Engineering; Engineering (General). Civil engineering (General); TA1-2040; Technology and Engineering; Mathematics; Stochastic modeling; Lognormal distribution; Linear regression; Correlation coefficient; Peak criterion; Hydrology; QA1-939; Mathematics and Statistics; Technology