Performance Evaluation of Three Event-Based Rainfall-Runoff Models for a Small Tropical Watershed

Comparison of the performance of different rainfall-runoff models helps to identify an appropriate model for use in a specific situation at a reasonable cost. An event-based hydrologic model gives an idea about the response of a drainage basin to a single rainfall event and is useful for locations which do not have data for a long time period. Rainfall and discharge data for long periods of time are required for continuous rainfall-runoff modelling and its availability is a major challenge as most rivers and streams in India are either ungauged or poorly gauged. In this work, the rainfall-runoff process in a small tropical watershed is simulated using three event-based models-the fully distributed TREX model, the semi-distributed HEC-HMS model, and the GIUH model, which is a lumped empirical model. The study area selected is an ungauged watershed located in Thiruvananthapuram District of Kerala State, India, namely, the Moozhy sub-basin of the Karamana River. Automatic rain gauges were installed at three locations in the watershed for collecting rainfall data. Runoff from the watershed was estimated using the stage measured with a staff gauge installed at the outlet of the watershed and the velocity measured using a flow prob. SRTM DEM analysed using various GIS-based tools is a major input to all the three models along with land use and soil data. Data pertaining to three rainfall events were used to calibrate these models. The models were further validated using another set of three observed rainfall and runoff events. Results of the statistical analyses performed indicated that the performance of the TREX model was reasonably good when compared to those of the HEC-HMS and GIUH models.