Estimates of Groundwater Recharge beneath Banksia Woodland on the Swan Coastal Plain Using a Vertical Flux Model (WAVES): Sensitivity Analysis C. Xu a , R.P. Silberstein b and A.D. Barr b a Water Corporation of WA, Leederville, WA 6007. b CSIRO Land and Water, Floreat, WA 6913. Abstract: To aid management of the regional groundwater system around the city of Perth, a new coupled unsaturated vertical flux and saturated 3-D groundwater flow model, based on MODFLOW has been developed. The Vertical Flux Model (VFM) used depends on land-use, but for pastured areas, pine plantations and native bushland is based on WAVES, a detailed biophysical model linking transpiration, and soil-water uptake to climate and soil conditions. Application of WAVES requires a number of parameters that characterise the vegetation and soil hydraulic properties, some of which can only be determined approximately. Uncertainty in the model input parameters would affect the accuracy of recharge estimates. This paper investigates the sensitivity of groundwater recharge estimates to a wide range of vegetation parameters, climate conditions and soil hydraulic properties. Sensitivity analysis indicates that estimated groundwater recharge is relatively sensitive to leaf area index (LAI), light extinction coefficient and the maximum rooting depth of the vegetation, soil water holding capacity and, of course, the rainfall. Keywords: WAVES, Groundwater recharge modelling, Sensitivity analysis 1. INTRODUCTION The Water Corporation, the Water and Rivers Commission and CSIRO, have developed a groundwater model for the Perth Region known as the Perth Region Aquifer Modelling System (PRAMS). The PRAMS model consists of two coupled components: a Vertical Flux Model (VFM) package which calculates the net recharge/discharge into/from the watertable (Barr et al., 2003) and a saturated groundwater model based on MODFLOW for flows in the multi-layer aquifer system below (Yu et al., 2002). The VFM package consists of a number of modules including a Recharge Manager and several recharge models for different land-uses with the key module being the WAVES model developed by CSIRO (Zhang and Dawes, 1998), which is used to estimate the recharge beneath pine plantation, banksia woodland and pastures and crops. WAVES is a one-dimensional, daily time step model that simulates the fluxes of water and energy between the atmosphere, vegetation, and soil systems. It is a process-based model that couples these systems by modelling the interaction and feedback between them. WAVES uses an efficient numerical solution to solve Richards equation for unsaturated water flow. Daily transpiration is estimated by the Penman- Monteith equation and is extracted from the soil profile using weighting factors determined by the modelled root density and a normalised weighted sum of the matric and osmotic soil water potential of each layer. This model has been shown to simulate water dynamics and vegetation growth correctly for a wide variety and combinations of climate, soil and vegetation type (Zhang et al., 1996; Zhang et al., 1998). The WAVES model requires inputs of daily climatic and watertable data, parameters that characterise the vegetation type and soil parameters describing the water holding capacity and hydraulic properties of soil layers. The performance of the model application will, to some extent, be dependent upon reliable estimates of these parameters. However, most of these data have to be measured from plot scale field measurements. Due to the spatial and temporal variability of the measured attributes, extrapolation of the point measurements to a regional scale may introduce large uncertainty. The uncertainty in the model parameters will result in inaccuracy in the modelling results. It is therefore important to understand how sensitive the estimate of groundwater recharge/discharge is to these model parameters so limited resources can be best directed to reduce the uncertainty of parameters that have a significant effect on the model results. The varied land uses on the Swan Coastal Plan include native bush, pine plantation, urban, market garden horticulture and pasture grazing. The Gnangara groundwater mound, located north of Perth is a major groundwater resource for