Trend analysis of Australian annual maximum flood data: exploring relationship with climate and catchment characteristics

Abstract

The potential impacts of natural climate variability or/and anthropogenic climate change have challenged the assumptions of stationarity and independence in flood frequency analysis. It is thus essential to investigate hydrological variables for trend behaviour and to assess the potential causes for the trends that are found. This study tests for changes in annual maximum (AM) streamflow time series using the most extensive streamflow database compiled in Australia to date, comprising records from 491 small to medium sized catchments with record lengths from 30 to 97 years. Three study periods, all ending in 2004, were examined using the non-parametric Mann-Kendall (MK) test and Sen’s slope estimator. The temporal and spatial dependency within the flood data are accounted for by utilizing the trend free pre-whitening approach and a bootstrap resampling method. The linkage of the identified trends in streamflow with precipitation and catchment characteristics is also investigated. Results indicate significant downward trends in the Australian AM flood data. This downward trend is dominant in the eastern, south-east and south-west regions of the continent. Further investigation shows that these downward trends are field significant at the 10% significance level. Changes in precipitation do not appear to be sufficient to explain the observed trends in streamflow, whereas a statistically significant positive relationship between streamflow trends and mean basin rainfall is found. Partial MK analysis show reduction in the total number of stations with statistically significant downward trends after accounting for three large-scale climatic modes (Southern Annular Mode, El Niño Southern Oscillation and Interdecadal Pacific Oscillation), suggesting that many of the observed trends may be associated with these climate modes. However, given the significant covariability between these three large-scale climatic modes and the global warming trend, the extent to which observed changes in AM flood are due to natural variability rather than anthropogenic climate change cannot yet be conclusively determined.