South African commercial plantations occupy an estimated 1.5 million hectares of the country and as the demands for timber products increase, this area is expected to increase. However, further expansion is limited, not only by the suitability of land, but also by the pressures from other water users. As a result the need has arisen for simulation models that can aid decisionmakers and planners in their evaluation of the water requirements of forestry versus competing land uses at different spatial scales. Different models exist to perform such tasks and range from simple empirical models to more complex physically-based models. The policies of the National Water Act (1998) relating to forestry serve to highlight the requirements of a model used for the assessment of afforestation impacts and these are discussed in this document. There is a perception that physically-based distributed models are best suited for estimation of afforestation impacts on a catchment's water yield since their physical basis allows for extrapolation to different catchments without calibration. Furthermore, it is often stated that the model parameters have physical meaning and can therefore be estimated from measurable variables. In this regard, a review of physically-based modelling approaches and a comparison of two such hydrological models forms the main focus of this dissertation. The models evaluated were the South African ACRU model and the Australian topography-based Macaque model. The primary objective of this research was to determine whether topography-based modelling (Macaque model) provides an improved simulation of water yield from forested catchments, particularly during the low flow period, compared to a physically-based model (ACRU model) that does not explicitly represent lateral sub-surface flow. A secondary objective was the evaluation of the suitability of these models for application in South Africa. Through a comparison of the two models' structures, the application of the models on two South African catchments and an analysis of the simulation results obtained, an assessment of the different physically-based modelling approaches was made. The strengths and shortcomings of the two models were determined and the following conclusions were drawn regarding the suitability of these modelling approaches for applications on forested catchments in South Africa:• The ACRU model structure was more suited to predictive modelling on operational catchments, whilst the more complex Macaque model's greatest limitation for application in South Africa was its high input requirements which could not be supported by the available data. • Despite data limitations and uncertainty, the Macaque model's topography-based representation of runoff processes resulted in improved low flow simulations compared to the results from the ACRU simulations, indicating that there are benefits associated with a topographically-based modelling approach. • The Macaque model's link to the Geographic Information System, Tarsier, provided an efficient means to configure the model, input spatial data and view output data. However, it was found that the ACRU model was more flexible in terms of being able to accurately represent the spatial and temporal variations of input parameters. Based on these findings, recommendations for future research include the. verification of internal processes of both the ACRU and Macaque models. This would require the combined measurement of both catchment streamflow and processes such as evapotranspiration. For the Macaque model to be verified more comprehensively and for its application in operational catchments it will be necessary to improve the representation of spatial and temporal changes in precipitation and vegetation parameters for South African conditions.

Thesis (M.Sc.)-University of Natal ,Pietermaritzburg, 2002.