The CMIP5-ensemble of GCM simulations, are potentially a valuable source of information on future climate conditions. The ensemble contains simulations for numerous different state-of-the-art GCMs for a number of different future greenhouse-gas concentration pathways. It offers adaptation researchers the ability to consider uncertainties in future climate associated with modelling uncertainties and future greenhouse gas concentrations. However, like previous generations of GCMs, the spatial resolution of the model output is too coarse to be used by agricultural production models that take climate data at a farm scale as input. Furthermore, though significant advances in global climate modelling have been made since the CMIP3 GCMs were built, the CMIP5 GCMs are not perfect representations of the climate system. This means that simulation outputs have biases relative to corresponding climate observations. Using unprocessed CMIP5 output as input to agricultural production simulations can therefore lead to unrealistic results, both for historical climate conditions and projected future conditions.

The presentation describes a new dataset derived from the CMIP5 ensemble that is suitable as input to farm-scale agricultural simulations and, potentially, other types of simulations relevant to climate change impacts. The dataset is generated by spatially interpolating monthly CMIP5 output to observing sites, bias-correcting these and then disaggregating the monthly data into daily data using a weather generator. A summary of rainfall and temperature data from the dataset will be presented for the NSW wheat belt. Biases in CMIP5 GCMs for this region and future changes in climate conditions, and their uncertainty, will be discussed