*3.4. Transfer Functions*

The TF models representing a one-layer or three-layer model are described by [22,23]. The one-layer model represents a vadose zone where the clay layer is absent and therefore perching and a requirement for drainage does not occur. The input is the time-varying *RZD* and the output is the time-varying irrigation recharge rate. The three-layer model was defined to represent a vadose zone where perching may occur due to the presence of a clay layer. The input is the time-varying *RZD* and the output is the time-varying irrigation recharge rate along with a time varying drainage rate. The drainage occurs when a perched water table reaches the surface of the model.

Both models are parameterized using the modified Maulem–van Genuchten function [32]. The parameters used were based on measured soil water retention data of Mallee soils [33].

A scripting process was developed to run the AWB, distribute *RZD* across the various recharge zones, and to call and run the TF models for each zone.

#### *3.5. Partial Calibration of Irrigation Recharge*

A partial calibration of the AWB-TF models can be performed using drainage data. This calibration can occur externally to any groundwater modelling, and several aspects of the drainage data can be used to constrain the model outputs, as follows:


In calibrating the models at Loxton and Bookpurnong, it was considered that two main parameters could be varied as part of the calibration:


#### *3.6. Groundwater Model and Integration of Transfer Function*

Various options can be employed to integrate the AWB and TF models with a groundwater model. In this study, a staged integration was undertaken to assess the merits of the approach. We considered a base case and two potential methods that integrate the AWB and TF models with the groundwater model focusing on di fferent calibration approaches (Table 1).



The base case (LB2011) is the existing Loxton–Bookpurnong groundwater model in MODFLOW from which recharge rates are derived inversely via manual calibration.

TF-A/B uses drainage data to calibrate the AWB and TF models and derived recharge rates that are passed to the groundwater model. The MODFLOW model is recalibrated in accordance with the revised irrigation recharge and the observed groundwater level data. In this instance, the AWB-TF model was calibrated manually and the MODFLOW model was calibrated using an automated calibration procedure incorporating the Parameter ESTimation code, PEST [34].

TF-C calibrates the integrated (AWB-TF-MODFLOW) model simultaneously using drainage and groundwater data. In this instance, the integrated model was calibrated using PEST, but a manual calibration process could also be applied.
