Design Flood Estimation in Ungauged Catchments: A Comparison Between the Probabilistic Rational Method and Quantile Regression Technique for NSW

2013

Hydrological Processes

Estimation of design flood in ungauged catchments is a common problem in hydrology. Methods commonly adopted for this task are limited to peak flow estimation, e.g. index flood, rational and regression-based methods. To estimate complete design hydrograph, rainfall-runoff modelling is preferred. The currently recommended method in Australia known as Design Event Approach (DEA) has some serious limitations since it ignores the probabilistic nature of principal model inputs (such as temporal patterns (TP) and initial loss) except for design rainfall depth. A more holistic approach such as Joint Probability Approach (JPA)/Monte Carlo Simulation Technique (MCST) can overcome some of the limitations associated with the DEA. Although JPA/MCST has been investigated by many researchers, it has been proved to be difficult to apply since its routine application needs readily available regional design data such as stochastic rainfall duration, TP and losses, which are largely unavailable for Australian states. This paper presents regionalization of the model inputs/parameters to the JPA/MCST for eastern New South Wales (NSW) in Australia. This uses data from 86 pluviograph stations and six catchments from NSW to regionalize the input distributions for application with the JPA/MCST. The independent testing to three test catchments shows that the regionalized JPA/MCST generally outperforms the at-site DEA. The developed regionalized JPA/MCST can be applied at any arbitrary location in eastern NSW. The method and design data developed here although primarily applicable to eastern NSW can be adapted to other Australian states and countries.

Section: Introductionmentioning

confidence: 99%

Development of regionalized joint probability approach to flood estimation: a case study for Eastern New South Wales, Australia

Caballero

2013

Hydrological Processes

“…The annual maximum flood database in this paper has been taken from the current revision of the regional flood estimation method in Australian Rainfall and Runoff Project 5 (Rahman et al, 2012). The selected stations contain annual maximum flood data of minimum and maximum record lengths of 26 years and 58 years, respectively, with an average of 33 years.…”

Section: Study Area and Datamentioning

confidence: 99%

Trend analysis of flood data in Australia: a case study for Victoria

Hossain

Piantadosi, J., Anderssen, R.S. And Boland J. (Eds) MODSIM2013, 20th International Congress on Modelling and Simulation

Haddad

et al. 2013

The possible impact of climate change on water resources management has received a great deal of attention from scientists all over the world. It has been noted that climate change will have a significant impact on various aspects of hydrologic cycle such as rainfall and floods. Due to climate change, the flood data can no longer be assumed stationary. Previous studies have shown that about 15% of Australian stream gauging stations show trends in annual maximum flood data, mostly downward in the south & south-east regions and upward in the northern region. These studies were based on the flood data till 2005 as part of Australian Rainfall and Runoff Revision Project 5. In this study, an updated database has been prepared for the annual maximum floods covering till year 2011 for the state of Victoria. This updated database is then used in this paper to identify trends in annual maximum flood series data using Mann-Kendall test. It has been found that 26% of the selected stations exhibit a downward trend at 5% level of significance, and none show an upward trend.The trend analysis was also repeated for 10% and 1% significance levels as well. Here, 60 stations show trends (i.e. 39% of the stations) at 10% level of significance and 14 stations show trends (i.e. 9% of the stations) at 1% level of significance. In all cases, the demonstrated trends are much higher than that would have been shown by mere sampling variability. Furthermore, no noticeable relationship between trends and catchment areas is identified. The reason behind these trends is being examined, and will be reported in future publications.

“…mean, standard deviation and skewness) of the log-transformed annual maximum flood series to estimate flood quantiles from a regional LP3 distribution. Rahman et al (2011) and Palmen and Weeks (2011) compared the QRT against the PRM and found that QRT outperformed the PRM. Haddad et al (2011a, b, c), Rahman (2011, 2012) and Hackelbusch et al (2009) found that the QRT-GLS method outperformed the QRT-OLS method while a region-of-influence (ROI) approach (Burn, 1990) outperformed the fixed region approach.…”

Section: Adopted Rffa Modelmentioning

confidence: 99%

Regional flood modelling in the new Australian rainfall and runoff

Haddad

Piantadosi, J., Anderssen, R.S. And Boland J. (Eds) MODSIM2013, 20th International Congress on Modelling and Simulation

2013

This paper presents the regional flood estimation model being developed for the 4 th edition of Australian Rainfall and Runoff (ARR). This covers the basic steps in the adopted modelling framework i.e. data collation, formation of regions, model form, and estimation of the parameters of the model. The adopted modeling framework revolves around the principles of reduction of uncertainty and the best possible use of the regional data by the reduction of the regional heterogeneity to enhance the accuracy of at-site design flood estimation. The ARR regional flood estimation model uses a Bayesian generalised least squares regression-based approach in the data-rich regions of Australia that consists of 619 stations. In the formation of the regions, it divides Australia into a number of regions. A region-of-influence approach is adopted to form regions where there are over 50 stations in the region. It also adopts a parameter regression technique where three parameters of the Log Pearson Type 3 distribution are regionalized. A Monte Carlo simulation technique is adopted to estimate the model uncertainty. The RFFA model is being re-calibrated with the new IFD data and the most up-to-date streamflow records and is expected to be released by Dec 2013.