Evaluation of rainfall-runoff models application on Sodong Watershed, Central Java

Abstract

Rainfall-runoff model is a model that cannot be separated with hydrology analysis. This model converts precipitation data into discharge and can be found in daily and monthly basis. Referring to several researches, “monthly basis model can simulate annual discharge better than annual model” and “certain model has advantages and limitations for a basin”. By considering watershed’s data capability, watershed location, watershed area, and watershed conditions this thesis aims to evaluate 3 selected models. Sodong watershed was chosen as study location. Sodong watershed area is 44.87 Km2. It is located in Central Java and a tributary of Paku River. More than 60% of the watershed is porus and its rainy season is normal with approximately 10 days late. The watershed has 1 climate post and 3 rainfall stations. The analysis uses monthly data from year 2003 to year 2008. NRECA, HBV 96, and NAM model were chosen as the calculation model. The analysis consists of: sensitivity analysis, parameter calibration, verification, and evaluation. Sensitivity analysis was done by plotting the changes in parameters, calibrating was done by comparing simulated and observed discharge, verification was done by comparing parameter result with geological map, and evaluation was done by comparing each model duration curves. The research results are: the watershed porosity is high, models can simulate the watershed very well (NS>0.50), and duration curves comparison shows that overall HBV 96 model can simulate the watershed better. However in terms of simulating low flow, NAM model can simulate them better.