Modelling streamflow is vital for helping improve management of water resources. This includes short term forecasting (e.g. estimating flood peaks) as well as longer term simulation of streamflow (e.g. looking at the impact of land use and/or climate change on water resources). Evaluation of the performance of models is often done using a point by point comparison of the observed and modelled outputs (e.g. least squares). Recent work has explored using correlation analysis to gain understanding of where a model may be under-performing, enabling exploration of alternate model structures. This project will explore how the correlation functions can vary through the time series using a wavelet approach, and test the potential for this to further increase our ability to test model performance.