Heavy rainfall events are of great interest due to their significant impacts on many parts of the economy, community and natural systems. Projections of heavy rainfall at regional scales are crucial for adaptation planning for infrastructure, agriculture, and disaster risk management sectors, amongst others.

Whilst projections of rainfall from global climate models (GCMs) can provide useful information at large scales, it is well understood that GCMs are limited in their ability to represent the processes that produce rainfall extremes, and so this information is of limited utility at catchment and station scales. In order to better represent rainfall extremes it is desirable to resolve the processes that cause heavy rainfall through use of convection-resolving or convection-permitting models.

We demonstrate the ability of a convection-permitting model – the non-hydrostatic CSIRO Cubic Conformal Atmospheric Model (CCAM) run at 2km resolution – to produce a realistic climatology of heavy rainfall events over domains covering the Sydney area and south-east Queensland. Simulations were performed for the period 1980-2012, with the external forcing provided through spectral nudging towards the ERA Interim reanalysis. Output was then compared to the Australian Water Availability Project (AWAP) 0.05 degree gridded rainfall data set.

While these model results are promising for the present day, we will discuss ongoing work to benchmark modelled rainfall extremes against station data, and to identify the best configuration for downscaling from imperfect GCMs for the purposes of producing climate projections for use in impact assessments.