Modelling in ungauged catchments using PyTOPKAPI : a case study of Mhlanga catchment.

Abstract

Hydrological modeling of rainfall-runoff processes is a powerful tool used in various water resources applications, including the simulation of water yield from ungauged catchments. Many rivers in developing countries are poorly gauged or fully ungauged. This gives rise to a challenge in the calibration and validation of hydrological models. This study investigated the applicability of PyTOPKAPI, a physically based distributed hydrological model, in simulating runoff in ungauged catchments, using the Mhlanga River as a case study. This study is the first application of the PyTOPKAPI model to simulate daily runoff on an ungauged catchment in South Africa. The PyTOPKAPI model was parameterised using globally available digital elevation data (DEM), satellite-derived land cover, soil type data and processed hydro-meteorological data collected from various sources. Historical 30-year (1980-2009) quaternary monthly streamflow (from a well-tested and calibrated model) and daily meteorological variables (rainfall, temperature, humidity and so on) were obtained. The rainfall data were subjected to double mass curve test to check for consistency. The monthly streamflow was transposed to the catchment and disaggregated to daily streamflow time step. The PyTOPKAPI model was calibrated using an average runoff ratio as an alternative to matching streamflow data that is usually used for model calibrations. The simulated results were thereafter compared with the disaggregated monthly quaternary data. The model results show good overall performance when compared with the average runoff ratio, monthly disaggregated streamflow and the expected mean annual runoff in the catchment. In general, PyTOPKAPI can be used to predict runoff response in ungauged catchments, and thus may be adopted for water resources management applications.