Fig. 8 and Fig. 9 show cross-sections through the synthetic and recovered models. For the sake of clarity, both the synthetic (input) and recovered (output) models are plotted as perturbations from the AuSREM starting model. In general, the pattern of anomalies is well retrieved, but there is both a loss in amplitude and a blurring of the input spikes. This is a well recognised phenomenon of regularised inversions based on the assumption of local linearisation (Rawlinson et al., 2014a).