The distribution of daily rainfall in Australia and simulated future changes
Ian G. Watterson and Tony Rafter
Journal of Southern Hemisphere Earth Systems Science
67(3) 160 - 180
Published: 2017
Abstract
This study extends recent projections of monthly and daily precipitation over Australia by analysing the full frequency distribution of daily rain amounts and making projections of the new statistics wet-day fraction and top percentile of rain. Simulations from an ensemble of 33 CMIP5 models are used, together with six simulations from the downscaling model CCAM, with the data analysed on the model grids. Consistent with its higher resolution (0.5°), CCAM provides a more skilful simulation for the extreme grid point rainfall than most CMIP5 models. CCAM compares well with AWAP gridded data for wet-day fraction, while there is a wide range of CMIP5 results. In the future climate of 2080–2099 under the RCP8.5 scenario, changes in mean rainfall of both signs occur within the CMIP5 ensemble for most regions and seasons, although mean winter rainfall in southern Australia declines 5 to 30 per cent in most models and in CCAM. CCAM simulates increases in summer, and also more wet days, in contrast to CMIP5. Aside from the north in winter, the changes from CMIP5 become increasingly positive, on stepping from mean to top percentile to twenty-year extreme rainfall, a contrast of typically 25 per cent. There is much less contrast between these statistics from CCAM. The distributions of rain amounts shed light on these different projections. Averaged over Australia and four seasons, CCAM produces a broader distribution than the CMIP5 ensemble mean. However much of the future increase is in the 2 to 8 mm daily range, whereas CMIP5 distributions tend to shift towards amounts in the range 30 mm to 200 mm. Further assessment of such distributions in both these and newer versions of CCAM, ACCESS and other GCMs is recommended.https://doi.org/10.1071/ES17010
© Commonwealth of Australia represented by the Bureau of Meterology 2017. This is an open access article distributed under the Creative Commons Attribution-NonCommerical-NoDerivatives 4.0 International License (CC BY-NC-ND).