Implementation of the National Water Act in South Africa requires that an ecological Reserve be determined for all significant resources. The ecological Reserve determination is the estimation of the amount of water required to maintain the system in a particular ecological condition. Because aquatic habitats are defined in terms of local hydraulic variables rather than amounts of water, hydraulic analysis provides a crucial link in relating hydrological conditions and river ecosystem integrity. Over the last decade, considerable effort has been devoted to developing hydraulics for the Reserve determination. The hydraulics needs for Reserve determination are primarily for low flow analysis, and appropriate methods still need to be developed. This thesis deals with hydraulics under low flow conditions. Its emphasis is on developing appropriate methods for describing the hydraulic characteristics of South African rivers under conditions of low discharge, and the influence of vegetation and large bed roughness. The following methods have been developed: · A new equation for prediction of overall flow resistance under large-scale roughness, and a new approach for estimation of intermediate-scale roughness resistance that distinguishes the influences of large and intermediate scale roughness components. · Prediction methods for velocity distributions with large roughness elements. Under low flows, rocks and boulders may control the local velocity and depth distributions. Distributions of velocities and depth are related to rapidly spatially varied flow caused by the boundary geometry rather than flow resistance phenomena. With increasing discharge, the multiple local controls become submerged and the flow tends towards a resistance controlled condition. Available information addressing the distinction between resistance controlled and multiple local controls conditions is limited. This thesis contributes to understanding the transformation between multiple local controls and the resistance controlled conditions. · Practical conveyance prediction methods for three situations pertaining to the occurrence of vegetation in rivers and wetlands. In-channel and riparian vegetation makes an important contribution to the creation of physical habitats for aquatic animals, but also has significant effects on flow resistances that need to be predicted.

This study investigates whether the flood regime in a catchment is sensitive to the condition of riparian vegetation along the river network. The research is based on a comprehensive assessment and synthesis of field and laboratory measurements of vegetation flow resistance. A new numerical model is developed to estimate the roughness characteristics of multi-species riparian assemblages at a cross-section. Reach-scale and catchment-scale flood routing models are then applied to estimate the impact of vegetation on flood characteristics at successively larger scales. The investigation reveals that when riparian vegetation is removed at catchment-scale, peak stage declines as channel capacity increases but is also increased as the upstream catchment responds more rapidly to rain. In fact, the two competing impacts tend to cancel out leaving flood peak stage relatively insensitive to riparian condition. However, the overbank duration of a flood and flow speeds (including wave celerity) were both found to be sensitive to vegetation condition; respectively increasing and decreasing with density of vegetation. The first stage of this research examines the magnitude of the vegetation contribution to overall channel roughness, and established a means to predict it. The features of the flow resistance generated by six plant types (mature trees; grasses; aquatic plants; flexible saplings; and large woody debris) were distilled from a comprehensive review of over 160 existing publications (Chapter 2).