INVESTIGATION AND DEVELOPMENT OF URBAN HYDROLOGIC MODELS

JI HAN, Purdue University

Abstract

The effort which has been expended in the past two decades on the development of urban hydrologic models is quite significant. However, a considerable amount of work still needs to be done on these models in several areas. The present study deals with three of these areas, namely: (1) the optimal estimation of parameters, (2) the investigation of objective functions in urban rainfall-runoff models, and (3) the least cost design of urban drainage systems by using urban rainfall-runoff models. Optimal parameter estimation methods are needed to eliminate the subjectivity inherent in the trial and error methods. The superiority of the optimal parameter estimation methods over the trial and error methods has not been clearly demonstrated to the practitioners. Such a demonstration would provide the basis for greater confidence in and acceptance of these models. In the present study, the regeneration and prediction performances of three urban runoff models, ILLUDAS, SWMM, and MINNOUR, in which optimal parameter estimates are used are compared with the performance of these models in which the parameter values are arbitrarily specified. The complexity of model structure does not guarantee better performance. Very few investigators have examined the problem of selection of objective functions which are used with the runoff models. Those who have investigated the problem have often used only simple runoff models. Hence, the validity of their conclusions is untested when more complex models such as ILLUDAS are used. ILLUDAS was used in this study to establish guidelines for selecting objective functions. Two sets of the objective functions were tested by using the data from the Upper Ross-Ade and Oakdale Avenue watersheds. The sum of the squared deviations between the observed and calculated hydrograph ordinates has been the most frequently used objective function in the past and the results of the present study show that this objective function gives best overall performance. The importance of hydrologic models in planning, design, and operation of urban drainage systems has been repeatedly mentioned and is generally accepted. However, application and use of such models to explore alternative designs and their associated costs has not been widely reported. Several of the least cost design models assume that the inflow rates are accurately known. Without a good hydrologic model, these inflow rates cannot be accurately estimated and hence the least cost design model may not be used with confidence. In the present study, a simple, readily usable, and theoretical sound model which includes ILLUDAS and a dynamic programming subroutine is developed. The model is tested by using data from Upper Ross-Ade and Bar Barry Heights watersheds. The model is used to investigate the variation of system costs with the variation in design parameters and to optimally size the detention storage volume for the least cost of the system.

Degree

Ph.D.

Subject Area

Civil engineering|Cartography

COinS