Date of Award

Spring 1988

Document Type


Degree Name

Doctor of Engineering Science in Environmental Engineering


Civil and Environmental Engineering

First Advisor

Eugene B. Golub

Second Advisor

Robert Dresnack

Third Advisor

Paul C. Chan

Fourth Advisor

Murray Lieb

Fifth Advisor

Michael Bruno


Previous studies on multiyear droughts are often limited to the analysis of historic annual flow series. A major problem in these studies is the unavailability of long historic flow records, on which to perform the analysis. To overcome this difficulty, the present study has used synthetically generated mean annual flow series to suppliment the historic flows. For the purpose of generating flows, a general methodology was developed to propose a mathematical model based on the harmonic and stochastic analyses of the historic flow series. The main objective was to derive a large population of multiyear drought events from the generated flow series, and to utilize this population for the simualtion of statistical & stochastic behaviour of the drought parameters.

The methodology was applied to a study area which includes six watersheds in the northern part of New Jersey and one watershed in the central New Jersey area. Analyses reveal that the mean annual flow series recorded at each selected streamflow gaging station, represents a periodic-stochastic process. The best model was determined for each stream, and used to generate a long annual flow series. Multiyear point droughts were identified by analyzing both the historic and generated flow series at a fixed truncation level. Four important drought parameters, namely, the duration, severity, magnitude and the time of occurrence were determined for each stream. The statistical properties of each of these parameters were then evaluated. It was found that the generated drought events closely follow the same statistical behaviour as the historic drought events. Based on the statistical properties, classical probability distributions such as gamma and log normal were fitted to the generated drought parameters. The applicability of these distribution functions to predict the extreme drought events have been illustrated with examples. In addition, the cross-correlation structure of the time of occurrence parameter of droughts has been identified with regard to spatial distribution of the point droughts over the study area.