Abstract
In terms of economic damage, flood is the number one natural disaster in Australia. Extremes of floods and droughts are becoming more frequent now-a-days in Australia. Impact of climate change and anthropogenic activities on hydrologic catchments are assumed to be the driver of these changes in flood extremes. Design flood estimation is an established method used worldwide to minimise flood risk and in hydraulic design of different types of infrastructures. Annual maximum flood (AMF) data are widely used for designing flood estimation and flood risk assessment. In design flood estimation these AMF data are assumed to follow stationary assumptions i.e., all flood events result from same climate condition (homogenous) and all events are independent. In the context of climate change this assumption can be grossly violated. Investigation of trends in AMF data is the first step to understanding long term changes in flood data. This research study analysed the characteristics of trends in recorded AMF times series data. The selected study area is New South Wales (NSW) state of Australia. Initially 176 stream gauging stations are selected where the minimum record length of the AMF data in each station is 20 years. After visual and statistical data quality check, 36 stations are finally selected. The minimum and maximum record length of AMF data of these stations are 50 years (1971–2020) and 91 years (1930–2020), respectively. As stream flow have strong natural variability with large-scale periodic behaviour of climate, higher record length as considered in this study to captures multiple and long-term climate variability cycles whereas a shorter record length may provide misleading trends in AMF data. As catchment’s hydrologic behaviour may change significantly with the increase of catchment size, therefore all stream gauging stations are selected with maximum catchment area below 1000 km2 spreading over flood plains, mountainous and coastal regions of NSW. The widely used non-parametric Mann–Kendall (MK) tests are used in this study to detect statistical trends. Trend tests are conducted with 5 and 10% significance levels. The MK test result shows significant trends in 14 station’s AMF record at 10% significance level, and in 9 stations with 5% significance level. The study shows that at 10% significance level, 39% of the selected station’s AMF series have significant trends. The study also shows that the average AMF in these 14 stations has decreased by a minimum of 4% to a maximum of 64% with an average decrease of mean by 39%. To investigate the behaviour of statistical parameters and trends in AMF data with different time ranges, AMF data series of each station is divided into two sub-series with two equal periods of record. The linear trend of each sub-series is plotted, and the mean, standard deviation and skewness of these sub-series are calculated. It is found that the statistical properties between two periods have been changed with significant decrease of mean and standard deviation in the second period of the data which is a violation of the stationarity assumption of AMF data used in the flood frequency analysis (FFA). The study also shows a general downward trend (without considering significance level) in the AMF data exists in 32 stations of the selected 36 stations. This study results suggest that there exist decreasing trends in most of the station’s AMF data in NSW and stresses the importance of adopting non-stationary FFA for design flood estimation in NSW.
Original language | English |
---|---|
Title of host publication | Water Resources Management and Sustainability: Solutions for Arid Regions |
Editors | Mohsen Sherif, Vijay P. Singh, Ahmed Sefelnasr, M. Abrar |
Place of Publication | Switzerland |
Publisher | Springer |
Pages | 83-93 |
Number of pages | 11 |
ISBN (Electronic) | 9783031245060 |
ISBN (Print) | 9783031245053 |
DOIs | |
Publication status | Published - 2023 |