Author(s): M. Braccia
Linked Author(s):
Keywords: Split-sample validation; Calibration; Fly Brook; AWBM; Hydrologic models; Uncertainty
Abstract: Uncertainty in hydrologic models is interpreted through split-sample validation. This can however, lower model performance by limiting the amount of data used for calibration. The Fly Brook catchment in south-west Western Australia has continuous streamflow records from 1962 to 1999 and from 2007 to 2009. Australian Water Balance Models were developed to model the missing period of record. Four calibration scenarios based on data prior to 1999 were studied. The scenarios were selected to determine whether an extensive calibration period provides the best model outputs or whether split-sample validation is essential when estimating streamflow. Observed data post 2007 was used as a reference to assess which calibration best simulates the recent flows. The study highlighted that validation is important to indicate the reliability of flow estimates not included in calibration as all the models performed poorly when compared to observed flow from 2007 to 2009.
Year: 2011