Influence of global climate model selection on runoff impact assessment

The future rainfall series used to drive hydrological models in many climate change impact on runoff studies are informed by rainfall simulated by global climate models (GCMs). This paper assesses how the choice of GCMs based on their abilities to reproduce the observed historical rainfall can affect runoff impact assessment. The 23 GCMs used in IPCC 4AR are considered together with 1961-2000 observed rainfall data over southeast Australia. The results indicate that most of the GCMs can reproduce the observed spatial mean annual rainfall pattern, but the errors in the mean seasonal and annual rainfall amounts can be significant. The future mean annual rainfall projections averaged across southeast Australia range from -10% to +3% change per degree global warming, which is amplified as -23% to +4% change in the future mean annual runoff. There is no clear difference in the future rainfall projections between the better and poorer GCMs based on their abilities to reproduce the observed historical rainfall, therefore using only the better GCMs or weights to favour the better GCMs give similar runoff impact assessment results as the use of all the 23 GCMs. The range of future runoff in impact assessment studies is probably best determined using future rainfall projections from the majority of available archived GCM simulations. (C) 2009 Elsevier B.V. All rights reserved.