摘要
For many decades, synthetic streamflow series have been utilized in hydrology to analyze numerous stochastic problems whose solutions depend on the values of the streamflows and their temporal pattern. The stochastic generation of synthetic streamflows at a given time level can adopt two general approaches: the generation at the required time level by applying an appropriate model; or the generation of annual flows using a suitable annual model, followed by their disaggregation into flows at the required time level. The first approach is feasible for a seasonal or monthly level, but not for a daily level, while the latter can be applied to any level. It also has the advantage of allowing the preservation of the historical statistical properties at both the upper (year) and the lower (season, month or day) time levels. One of the simplest disaggregation models is the method of fragments. Based on the extensive application of that method to the generation of monthly flow series in more than 50 Portuguese river gauges (Silva and Portela, 2011, Hydrol Sci J 57(5): 942–955. doi: 10.1080/02626667.2012.686695, 2012), it was possible to establish a deterministic criterion to define the classes of fragments and to select the fragments that proved to be very robust. That criterion was revisited and modified and applied to the generation of synthetic daily flow series, with even better results. This paper describes the revisited method, presents the results from its application to a few case studies and discusses its relevance to analyze the uncertainty due to the temporal variability of the flow regime.