期刊名称：Journal of Environmental Hydrology (ältere Jahrgänge)
印刷版ISSN：1058-3912
电子版ISSN：1996-7918
出版年度：2009
卷号：17
出版社：IAEH
摘要：Extended deficit analysis (EDA) is a useful procedure for estimating streamflow deficits and, hence,reservoir capacity for a given reliability of supply. In this study, EDA has been used to assess the 1 in 100 year deficit/reservoir capacity for a draft (demand) of 75% of the mean annual flow (MAF) for 30 unregulated streams in New Jersey, USA. The study demonstrates that for the same level of draft, streams draining northern catchments have relatively larger deficits and hence, higher storage capacities in comparison to those in the southern Coastal Plains. Three metrics, vulnerability, relative severity (also a measure of maximum number of years of water held in storage) and resilience have been used to assess stream reservoir performance (drought severity) for each of the 30 unregulated streams. An examination of the metrics revealed that drought in southern catchments could be less severe than those in the north for the chosen demand. It is proposed that higher hydraulic connectivity between groundwater and surface water (enhanced baseflow) for southern catchment streams may have been responsible for the observed trends in drought severity. Further, the study revealed that streams within the state could exhibit prolonged failure (drought) durations beyond a water supply level of 80% of MAF. Whilst annual vulnerability is predominantly related to stream interannual variability (measured by the coefficient of variation), the maximum number of years of water that can be held within a catchment at the given demand level (i.e. 0.75 MAF) is somewhat related to the lag-one serial correlation coefficient. This demonstrates that accurate replication of the coefficient of variation and the lag-one serial correlation coefficient in a streamflow time series could be paramount for appropriate prediction of future drought severity within the state. It is suggested that a water 'residence time' of at least a year could be combined with the lag-one serial correlation coefficient to characterize systems as predominantly carryover or within year systems for a given level of draft.
