Hydrologic Investigation of Low Gradient Watersheds

Abstract

Abstract : The aim of this project is to improve our understanding of the hydrologic characteristics of tidal low-gradient watersheds. Hydrologic and meteorologic measurements as well as distributed and lumped hydrologic modeling were used to gain such understanding. The analysis performed indicated that the tidal cycle, wind setup, and backwater effects dominate the hydrologic response of coastal watersheds to rainstorms. Therefore, predictive tools must incorporate these factors. Hence, and for the purpose of improving our understanding of physical characteristics of these watersheds, Artificial Neural Network and Non-Parametric Regression models have been developed. They were able to reproduce the complex and non-unique stage-discharge relationship. Furthermore, comparisons between distributed and lumped models showed that their predictive ability for midsize catchments was similar. Specifically, the sensitivity of both approaches to variations in the temporal sampling of rainfall was not significantly different. Their response to variations in the spatial density of rainfall information was also investigated. Efforts done to establish guidelines on the temporal and spatial rainfall sampling requirements for hydrologic predictions indicated that a resolution frequency of up to one hour is adequate for predicting runoff discharges.