Date of Award

Spring 1992

Document Type


Degree Name

Doctor of Philosophy in Civil Engineering - (Ph.D.)


Civil and Environmental Engineering

First Advisor

Eugene B. Golub

Second Advisor

Michael Bruno

Third Advisor

Paul C. Chan

Fourth Advisor

Robert Dresnack

Fifth Advisor

Murray Lieb


The purpose of this dissertation was to develop a better method for dealing with the problems of flood prediction in Urban Watersheds. It has long been realized that urbanization activity such as increased imperviousness, drainage improvements, etc. increases runoff volumes. Therefore, traditional flood prediction methods using the Log Pearson III distribution underestimate flood frequencies when applied to urban watersheds without modification.

In attempt to compensate for the effects of urbanization on streamflow, previous workers usually employed regional analysis techniques involving a number of different watersheds at various degrees of urbanization. Results obtained by this approach leave room for improvement primarily due to heterogeneities in hydrologic characteristics of watersheds. In contrast, the method developed in this thesis characterizes a watershed using a time based analysis in which the basin response patterns are studied through as long a period as data exists.

The method proposed is based on the hypothesis that basin response to small storms after dry periods derives mainly from impervious areas and hence provides a measure of the basin's state of development. By analyzing the peak flows resulting from drought period small storms over a long period of time, a trend equation may be established indicating the growth pattern of runoff contributed largely by impervious surfaces. This relationship in turn forms the basis for separating runoff components from pervious and impervious areas during major, wet period storms.

Next, the impervious surface runoff contribution is updated to present conditions equivalent flow by again applying the above trend equation, while the pervious surface contribution is updated by the ratio of the pervious surface in the present year to the pervious surface in the year in consideration.

Finally, the composite update ratios thus calculated are applied on the historic record of annual peak flows and the Log Pearson III technique applied to predict future floods.

The above ideas were illustrated using the Saddle River Basin in New Jersey. The maximum update ratio obtained was about 1.8 and the predicted floods increased in the range of 1.05 for the 100 year flood to 1.4 for the 2 year flood.