Abstract |
Except in special cases, multivariate streamflow models developed in the past have been restricted to the preservation of first and second moments of historical streamflow frequency distributions, relying on the use of arbitrary nomalizing transformations to apply the models to those frequently encountered instances where special cases are not applicable. This work extends those models to preserve third moments in generated records of monthly flows, thus avoiding the use of arbitrary transformations. Methods are reported to either assign probabilities to the occurrence of extreme events of sub-month duration or to generate observations on maximum or minimum sub-month flows. Methods are then presented to superimpose daily and weekly sequences on monthly flows only as they may be required for system simulation. Empirical work is based on streams in North Carolina and other selected gages in the eastern and southeastern sections of the United States. (Author) |