Document Type
Presentation
Publication Date
8-1-2014
Abstract
Time of concentration (ToC) is the most frequently utilized time-scale parameter in hydrology which must be estimated accurately to ensure correct simulation of many different hydrological processes. Hydrologists have developed many empirical and semi-empirical methods for estimating ToC which are regional, watershed, and site-specific. Modellers are often confused by the number of ToC estimation methods and formulas and often select an equation without evaluating its correctness which leads to inaccurate simulation results. The importance of deriving and using regional ToC equations has been highlighted in many studies. In this paper, a methodology is proposed for deriving ToC equation(s) for watersheds located in a specific geographic region using GIS and Genetic Programming (GP). The use of GIS data allows for easy extraction of multiple characteristics of a large number of watersheds and sub-watersheds. Also, integration of GIS maps into the TR-55 model enables the determination of “true” TOC values for the watersheds under study. The obtained physical and hydrological characteristics of the watersheds are combined with rainfall characteristics and computed ToC values to form a large database. GP is then used as a data mining tool for conducting symbolic regression and deriving the most accurate set of equations for the watersheds of the region. In a case study, the proposed methodology is applied to 72 watersheds and sub-watersheds in Khorasan Razavi province, Iran. The method provides a set of different ToC equations to be used for watersheds with different sizes in the region. The equations proposed by GP are evaluated and compared to other conventional ToC estimation methods. The set of equations found by GP provides insight on the relationship between ToC and other watershed and rainfall characteristics and highlights the potential role of GP as an attractive and effective Knowledge Discovery tool.
Comments
Session S6-01, Special Session: Evolutionary Computing in Water Resources Planning and Management I