Water Balance Uncertainty of a Hydrologic Model to Lengthy Drought and Storm Events in Managed Forest Catchments, Eastern Australia

Abstract

Interest has grown in applying hydrologic models in managed catchments despite uncertainties around model inputs and empirical relationships to simulate complex geo-hydrological processes of streamflow and sediment variations. Unquantified interactions between geophysical, climate and management indices can also increase simulation uncertainties. Calibration of model outputs against observed values allows identification of the most influential variables and their optimised ranges by which model performance can be enhanced. A rainfall-runoff Soil and Water Assessment Tool (SWAT) model was utilised for four catchments in northern New South Wales, Australia to simulate time series of streamflow across varying rainfall regimes, from dry seasons from 2002 towards rainy 2009. Parameters causing a substantial change on model streamflow outputs were first identified using a sensitivity analysis which indicated that hydrologic factors governing the sources of water supply were critical parameters. These sensitive variables were substantially derived from groundwater modules, basic flow in the main channel, and management practices. Statistical tests of between-catchment differences showed that model simulations performed better in a catchment where the sole rain gauge was installed, while also having the narrowest variations in simulated values (r-index = 0.02). In contrast, the highest uncertainty of model simulations was found in the furthest catchment from the rain gauge where there was not a satisfactory agreement with observed data. Yearly differences between 2002 and 2009 indicated an overestimation of streamflow during low flow periods. However, the calibration process performed well in most peak flows where estimations followed the respective observed values. Long-term dry periods between 2002 and 2007 resulted in an overestimated baseflow by predicting an unrealistic recharge infiltrating aquifers.