Accuracy Issues of a Public-Domain Area-Wide Streamflow Model Semi-Automatically Built From Data Readily Downloaded From The Internet

Hanlie Pretorius, William James and Julian Smit


The aim of the major study was to determine if sufficient data at no charge is available on the Internet to use as input to a free and open source hydrological model for use in a streamflow monitoring system. The study area is the C83A quaternary catchment (746 km2) in the Northern Free State, part of the Vaal primary catchment in South Africa. The catchment has a response time of approximately 13 hours and so sub daily calculation steps were required in the hydrological model. The data that were available on the Internet to use as input to a hydrological model were: catchment boundaries, river lines, a digital elevation model, soil texture, land use, evaporation rates, flow rates and rainfall estimates from the Tropical Rainfall Measurement Mission (TRMM) 3B42 data product. After determining the requirements for the model and evaluating a number of models, the USEPA Storm Water Management Model (SWMM) was chosen. Only two years of observed rain and runoff were suitable for analysis. The coarseness of the available input data necessitated a coarser model than is the norm for SWMM. Nevertheless, the calibrated model performed well overall, having Nash-Sutcliffe (N-S) correlations of 0.80 and 0.73 for the calibration and evaluation periods respectively. The derived model-and-data system, denoted UCTPCSWMMC83A.1, is assessed in terms of the accuracy achievable by a public-domain area-wide streamflow model semi-automatically built from data readily downloaded from the Internet.

Permanent link: