Oral Presentation AMOS Annual Meeting and International Conference on Tropical Meteorology and Oceanography

Constraining projections of extreme rainfall over Australia (#190)

Margot Bador 1 , Sugata Narsey 2 , Lisa Alexander 1
  1. CLEX and UNSW/CCRC, Sydney, NSW, Australia
  2. CLEX and Monash University , Melbourne, VIC, AUSTRALIA

Water resources are essential and many sectors depend on it. Over Australia, rainfall is strongly influenced by the different modes of climate variability and vary considerably from one season to the next. Physical considerations of the water cycle response to warming indicate an intensification of rainfall, at a faster rate for extreme than for mean rainfall (Collins et al. 2013, IPCC AR5). These thermodynamical considerations are assessed at global and annual scales while large uncertainties remain at regional and seasonal scales because of the dynamical and microphysical contributions to the change in extreme rainfall (Pfahl et al. 2017). Australia exhibits some of the largest uncertainties in future rainfall extremes compared to other regions of the globe because of large inter-model differences and large uncertainties due to internal variability (Bador et al. 2018).

Our work aims to better evaluate the seasonal changes in extreme rainfall over Australia. We first evaluate the CMIP5 models with a large ensemble of observations from different sources (in situ, satellite and reanalysis) to better estimate observational uncertainties. This allows us to constrain the models based on how they reproduce the annual cycle of precipitation and mean and extreme precipitation climatologies. This framework also gives us the opportunity to investigate whether observations and models agree on the synoptic conditions leading to extreme rainfall events. We first analyse the feature size of the precipitation system leading to extreme rainfall at each grid cell. We then investigate the most important large-scale conditions leading to extreme rainfall at selected locations across Australia. Each of these locations belong to different regions of co-varying extreme rainfall and are selected using a clustering algorithm. Our results allow a narrowing of the projections of rainfall extremes over Australia by using different constraints applied to the CMIP5 model ensemble.