Global climate is unequivocally changing with warming trends detected in several regions throughout the world evident from both observations and global climate model analyses (IPCC, 2007). Warming of the climate system will intensify the global hydrological cycle that lead to changes in the pattern of occurrences of extreme hydrologic events (i.e. floods and droughts) and increases in their magnitudes and frequencies. These changes in extreme hydrologic events pose significant and serious challenge to hydrologists and water resource managers, since a failure to consider such variability in flood risk assessment can lead to an underestimation of the likelihood and sequences of floods, which in turn has vital implications for the planning and management of water resources and relevant infrastructure. The presence of changes in flood behaviour has also challenged the core assumption of the current flood risk assessment procedures, which assume flood frequency models are stationary. Hence, detecting and accounting for this variability in extreme floods due to climate variability and change are important in any future water resources management. This research focuses on quantifying non-stationarity evident in flood data for undisturbed catchments throughout the continent of Australia, identifying meteorological and climatic causes of this non-stationarity and then using this information to develop a non-stationary generalised extreme value model that best represents the changes in flood data. The research uses data from 491 catchments, which represent the most comprehensive and extensive flood database from the continent of Australia, to (i) detect non-stationarity in annual maximum (AM) flood data at local scale with the consideration of the impact of serial correlation; (ii) identify changes in AM flood data at regional scale by accounting for the effects of spatial dependency; (iii) analyse the natural driving sources of this non-stationarity in AM flood data, and (iv) develop a non-stationary generalised extreme value model that yields estimation of flood risk reflecting non-stationarity in flood data. The research is divided into four main parts. The first part advocates the use of the non-parametric tests including the Mann-Kendall, Pettitt and Sen's slope estimator to quantify non-stationarity in the form of gradual trend and abrupt shift in AM flood data for each catchment. Here, the Mann-Kendall method is also adapted to account for the impact of serial correlation on trend results using three different approaches, the pre-whitening, trend free pre-whitening and variance correction. Regional Mann-Kendall method and the global bootstrap resampling approaches are also used in this section to assess the potential impact of spatial correlation on trend results. The direction and the timing of the change-point along with the potential impact of the abrupt shift on the MK test results are also evaluated. The analyses are undertaken over three study periods along with the 'All Records' case. Resemblance in the results at local scale has been detected where the downward trends are identified as statistically significant in the majority of the catchments located along the south-east and south-west region of the continent, whereas upward trends are more visible in the north-west regions. Results have also revealed that many more of the trends would have been considered statistically regionally and field significant if the cross correlations have been ignored, particularly for shorter study period. Furthermore, similarity between the Pettitt test and MK results is noted for those catchments at which gradual trend are identified. For instance, negative shift in the mean is observed at catchments that exhibited a downward trend, and a positive shift in the mean is observed in the case of upward trend. The evaluation of the potential impact of abrupt shift on MK test results in the subseries, before and after the shift, indicated that in the most of the cases, none of these subseries exhibited significant trends once the change-points were accounted for. To appropriately estimate flood risk, it is necessary to consider the driving causes of non-stationarity in AM flood, which has been the emphasis of the second part of this research. Herein, large-scale climatic patterns, including Southern Annular Mode (SAM), El Nino Southern Oscillation (ENSO), Indian Ocean Dipole (IOD) and Interdecadal Pacific Ocean (IPO) are examined using the Partial Mann-Kendall approach. The analysis indicates that the variability described by these indices can explained a significant part of the observed temporal trends in the Australian AM flood data. However, given the significant co-variation of some of these indices (and particularly the SAM, filtered IPO) with the longer-term global warming trend, it is not possible to conclusively state whether the observed changes are due to natural or anthropogenic forcings. The third part of this research emphasises on assessing whether the trend outcomes are comparable when using monthly maximum and peak-over-theshold (POT) flood data as opposed to using AM floods. Using the Van Bell Hughes trend homogeneity test the trends in monthly maximum flood data over Victoria have found to have the same direction (i.e. homogeneous) for the majority of the selected catchments. A resemblance in the trend results between the AM, monthly maximum and POT flood magnitude index has been identified. Finally the fourth part of the research synthesises the previous three parts by developing non-stationary generalised extreme value model using the time and the climate index SAM to represent the source of non-stationarity in AM flood data. Based on the model selection criteria, the results revealed that the trend in location and in the log-transformed scale parameter along with a constant shape parameter is the optimal model to represent the variability in AM flood data.
Date of Award | 2014 |
---|
Original language | English |
---|
- climatic changes
- Australia
- floods
- droughts
- hydrologic models
- global warming
Effect of climate variability and change on flood magnitude and frequency in Australia
Ishak, E. H. (Author). 2014
Western Sydney University thesis: Doctoral thesis