Development of a kriging-based regional flood frequency analysis technique for South-East Australia

Abstract

Flood is one of the most common natural disasters, which also triggers other natural disasters such as erosion and landslides. Flood damage can be minimised by ensuring optimum design of drainage infrastructure and other flood management tasks, which depends largely on reliable estimation of flood quantiles. This study investigates flood quantile estimation in ungauged catchments using a kriging-based regional flood frequency analysis (RFFA) technique. Three main research objectives are addressed in this study. Firstly, kriging-based RFFA models are developed using 558 catchments from eastern Australia in the range of frequent to rare flood quantiles (2, 5, 10, 20, 50 and 100 years of average recurrence intervals (ARIs)). Secondly, a validation of the models by adopting a leave-one-out (LOO) validation technique is undertaken to identify the best and the worst performing catchments across eastern Australia. Finally, a detailed comparison is made for the kriging-based RFFA technique with a generalised least-squares-based quantile regression technique, known as ‘RFFE model 2016’ using the same dataset to evaluate whether there are general patterns of the performance in different catchments. The study shows that for eastern Australia (a) the developed kriging-based RFFA model is a viable alternative for flood quantile estimation in ungauged catchments, (b) the 10-year ARI model Q10 performs best among the six quantiles, which is followed by the models Q5 and Q20, and (c) the kriging-based RFFA model is found to outperform the ‘RFFE model 2016’.

Peaks-Over-Threshold-Based Regional Flood Frequency Analysis Using Regularised Linear Models

“…The GDR divides the coastal reg of south-eastern Australia into coastal and inland plains. The rationale for including b areas is based on the previous studies of Ali and Rahman [20]…”

“…The GDR divides the coastal region of south-eastern Australia into coastal and inland plains. The rationale for including both areas is based on the previous studies of Ali and Rahman [20] and Zalnezhad et al [21], which suggest considering both inland and coastal as a single region for RFFA. A total of seven catchment characteristics are adopted as independent variables in this study.…”