Investigation of non-stationarity of extreme rainfalls and spatial variability of rainfall intensity-frequency-duration relationships: a case study of Victoria, Australia.

ABSTRACT Water infrastructure and flood mitigation projects are currently designed assuming a stationary climate. However, increased frequency and magnitude of extreme rainfall events questions the stationary climate assumption. The reality of non-stationary rainfall extremes should be properly considered in the design of water infrastructure and flood mitigation projects, because the extreme value distribution models with constant parameters may no longer be valid under non-stationary climate conditions. In this study, first extreme rainfall frequency analysis in Victoria (Australia) was conducted through generalized extreme value ( GEV) models under non-stationary and stationary climate conditions, and superiority of non-stationary GEV models over stationary models was investigated. High quality extreme rainfall data (i.e. annual maximums) from 23 stations in Victoria for storm durations ranging from 10 min to 48 h were used for frequency analysis. In developing the non-stationary extreme rainfall models, both the time dependency and the dependency to indices of climate oscillations affecting Australian rainfall variability were investigated. It was found that none of the non-stationary GEV models was superior to stationary GEV models. Therefore, the stationary GEV models were used to determine the spatial variability of rainfall intensity-frequency-duration relationships in Victoria. In general, low rainfall intensity estimates in the Western Region (except southern portion for long storm durations) were found in Victoria, whereas high rainfall intensity estimates were detected mostly in the Central and Northern Regions for short storm durations and in the Gippsland Region for long storm durations. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]