Regional flood frequency analysis based on characterization and classification of drainage networks in arid and semi-arid regions

Abstract

Regional frequency analysis procedures are widely used in the field of hydrology when little or no data are available at the sites of interest for design of hydraulic structures. This analysis is normally comprised of two steps: delineation of groups of homogeneous regions and information transfer within the groups. The quality of the final estimate depends largely on homogeneity of regions and the density of measurement gauges. Characterization of drainage networks is important for defining homogenous regions, since river networks can appear significantly distinct in different regions depending on climatic, physiographic, and topographic constraints. The main objective in the dissertation is to provide enhanced flood estimates based on regional frequency analysis focused on arid and semi-arid regions, by characterizing drainage networks. To characterize river network systems, we classify network types into different classes based on three measures i.e., drainage area increments, stream course irregularities, and tributary junction angles, and determine specific conditions under which river networks develop in these regions using topographic curvatures. In this regard, we developed a novel method for robust and accurate classification of drainage networks by using only the tributary junction angle and lengths of secondary tributaries. The method is based on deriving three indices inspired by methods in fractal analysis. These indices are used to directly discriminate between different channel network types and can also be utilized in regional frequency analysis as new physiographical variables. We also identified that the drainage network type has an effect on the hydrological response of the basin. Finally, we developed improved regional flood frequency analysis methods by integrating information of the climate represented through aridity index, and information of drainage network using variables such as the proposed indices, tributary junction angle and time of concentration (response time). The results of the improved models are demonstrated using data from several basins in different arid and semi-arid regions. It has been observed that the accuracy of flood estimates is improved when the proposed variables are used for regional frequency analysis.

Date of Award	Dec 2016
Original language	American English
Supervisor	Taha B. M. J. Ouarda (Supervisor)