Towards improved parameter estimation in streamflow predictions using the ACRU model.

Abstract

An unresolved problem in hydrology has been to establish relationships between catchment attributes and the flow characteristics of the stream. Such information is commonly sought to improve streamflow predictions, often in a process of extrapolating research results obtained from relatively few, but intensively studied catchments, to a broader region. This study has attempted to clarify terminology related to streamflow generation processes and mechanisms, and to investigate relevant physiographic and climatic characteristics which critically influence the hydrological responses of catchments. Fourteen catchments were selected for this study. They comprised both operational and research catchments. These catchments were selected to be representative of variations in climate, topography, vegetation and geology occurring throughout the Republic of South Africa (RSA). The selection of catchments was also restricted to areas less than 100 krrr', and to the higher rainfall regions of the country, where runoff is significant and any land use changes may lead to marked changes in evapotranspiration and streamflow. A catchment was also selected from an arid zone in the USA, to capture the flow characteristics that are typical of such areas. A frequently applied simulation model on RSA catchments is the ACRU model. While physical-conceptual in structure it contains some parameters which, while not determining total streamflow magnitudes, governs the time distribution of the streamflows generated. Two such parameters from the ACRU model selected were the coefficient of baseflow response (COFRU) and the quickflow response fraction of the catchment (QFRESPj. These parameters are not explicitly physically based, and therefore improved guidelines of initial parameter values are required. Relationships between catchment characteristics and these two parameters were sought to provide guidelines for effective parameterisation of these parameters in future studies. Trends between QFRESP and COFRU, and catchment physical and climatic attributes such as catchment area, average depth of the soil profile, maximum basin relief, MAP and profile plant available water were identified, and could prove useful to future users of the ACRU model and guide experimentation in estimating initial parameter values. However, only a single significant multiple regression model was obtained for the baseflow release fraction COFRU from a catchment using MAP, catchment area and profile plant available water.