An Analysis of Catchment Attributes and Hydrological Response Characteristics in a Range of Small Catchments
1193/1/02
Nov 2002

EXECUTIVE SUMMARY

A fundamental but largely unresolved problem in hydrology has been to establish relationships between the physical attributes of a catchment and the flow characteristics of the stream or river leaving the catchment. These relationships are commonly sought to improve model-based predictions of stream flows, often by the procedure of extrapolating research results obtained from relatively few, intensively studied catchments to a broader region. Flow characteristics such as total stormflow and baseflow volumes, baseflow recession rates and peak stormflow volumes have major implications for downstream assurance of water supply, risk of flood damage, water quality or catchment erosion potential, and need to be predicted with useful accuracy if water resources are to be effectively managed. Predicting flow characteristics is especially important where changes in land use are anticipated and where alterations to the flow regime need to be assessed.

In the southern African context, accurate streamflow simulations are especially vital for the management and allocation of the scarce water resources. Allocation of water rights is already a contentious issue, and will become even more so as demand for water grows and as the country moves towards a more participatory system of water management, in which all major interest groups are represented in the decision-making process. Added complexity derives from the wide temporal fluctuations in river flows caused by highly variable rainfall regimes.

The complexities of predicting catchment flow regimes are great. Therefore, it is common to make use of computer models to simulate the major hydrological processes occurring in catchments. The ACRU hydrological model (Schulze, 1995) is widely used in southern Africa, and is particularly well suited to scenario analyses of the hydrological impacts of various land use options. Many past ACRU simulations on both large and small catchments have highlighted a general problem of correctly simulating flows from small catchments. Over larger catchments, spatial variations in geological and soil characteristics tend to average out, allowing easier simulations of the combined flows from the many smaller sub-catchments. At the smaller scale, however, individual catchment characteristics express themselves strongly through the hydrograph, and highlight our incomplete understanding in the link between rainfall and streamflow.

The goal of this project was to search for links between catchment physical descriptors, patterns of streamflow, and ACRU input parameter values required to simulate these flows. Such links are needed to offer guidelines to ACRU modellers on choice of parameter values, particularly those controlling streamflow generation. The research strategy adopted was to model runoff from a range of small catchments (<100 km² displaying a variety of physical features. A selection of 14 catchments (13 from South Africa, 1 from USA) was eventually chosen to represent a broad range of climate, topography, vegetation and geology. Daily rainfall and streamflow data were in every case considered to be of a high quality, spanning a continuous period of five years or longer. No significant impoundments, or major changes in land use over time, occurred in these catchments. ACRU was configured for each catchment, and simulations assessed for realism, with particular emphasis on the relationship between simulated and observed flows. Those parameter values most directly controlling the characteristics of flow were varied, both manually and with the parameter estimation software PEST, to obtain an acceptably good fit between simulated and observed flows.