Evaluating and enhancing general circulation model simulations for water resources climate change impact assessments

Human induced climate change will, in many parts of the world, create additional pressure on water supply systems. It is therefore vital for water resources managers to understand likely changes in future water availability. The only way of quantifying these changes is to use the outputs of General Circulation Models (GCMs). Due to time and resource limitations, it is often not possible to use the full suite of GCMs for which data is available. To decide which GCMs to use, it is necessary to evaluate the models with respect to their intended application. Following evaluation, corrections to GCM outputs may be required so that they match observations better. This thesis presents methods that can be used to evaluate and enhance GCM simulations for water resources impact assessments.For the purposes of evaluating GCM simulations, the thesis presents four techniques and case studies. The first, known as the Variable Convergence Score (VCS), compares the reliability of different climatic variables across time and space. It is developed for Australia to demonstrate that there is little confidence in precipitation projections for the future, compared to temperature and pressure projections. The VCS is also used globally to identify the reliability of GCMs for different regions. The third evaluation assesses the ability of GCMs to represent observed trends in pan evaporation. It is found that many of the GCMs do not capture the observed changes. Uncertainty in future projections is introduced by variability in vapour pressure deficit and wind speed. Finally, the modelling of variability on multiple time scales is investigated using a wavelet based skill score to assess the representation of large scale climate modes in GCMs.Methods to enhance GCM outputs are developed in the second part of the thesis. A Nested Bias Correction method is developed to post-process GCM simulations so that they are representative of observations across a range of time scales, achieved by matching the distributional and persistence attributes of the GCM and observed time series. The Nested Bias Correction method is applied to assess changes in future water availability across Australia. The bias correction approach is also compared to simple scaling approaches to demonstrate the importance of correcting for low-frequency persistence