Design flood estimation based on GRADEX

Thomson, Philip (2017) Design flood estimation based on GRADEX. [USQ Project]

Text (Project)

Download (7MB) | Preview


The estimation of peak discharges for common Annual Exceedance Probabilities (AEP) is a task frequently undertaken by hydrologists. Australia’s Rainfall and Runoff recommends standard procedures for these estimations which are widely used throughout Australia. Several of these methods are based on rainfall instead of historical flood events. Simple design events are one of these methods. These are ideal methods for estimating design floods when there is limited local data available or when only a quick and simple estimation is necessary. One of these simple rainfall analysis methods is the GRADient of EXtreme values (GRADEX) concepts, which historically has received little attention in Australia. The GRADEX method was developed in France in 1967 and has been widely used there for the last 50 years. This project will apply GRADEX concepts in Australian conditions to consider their suitability and whether it is could be incorporated for everyday use.

A literature review into the history and methods of the GRADEX concept revealed the key assumption of the GRADEX method, that a catchment can only retain a set depth of rainfall either by infiltration into the subsoil or storage/detention in depressions in the land. After this rainfall has been retained any additional rainfall that falls on the catchment is converted to runoff. This assumption allows for simple estimation of discharge values utilising rainfall data. The literature review showed that whilst the GRADEX method has been widely considered in France where it was developed, America and South America, it has had little review in Australia, this suggested that an analysis into the method in Australian conditions was worthwhile.

21 sites in North Queensland between Townsville and Cooktown were selected from the Hydrological Reference Station Network for the comparison. The catchment area of these sites varies from 16.2 to 8638km2 and had a 24hr design rainfall of between 204 and 624mm. A 24hr event duration was adopted through undertaking a discharge to runoff plot for each site. A variety of discharge estimates were calculated for varying values for R (the rainfall retained within a catchment), then the site R value was determined through determination of best fit using a Nash-Sutcliffe E estimate, where the R value with an E values closest to 1 was adopted for the site. The resulting discharges are the GRADEX design discharges for the 21 sites, design discharges are all calculated for each site using ARRs Log Pearson 3 method. After the design discharges for the minor and major events were calculated an ANOVA analysis of variance was undertaken to determine the methods reliability.

Two of the sites, H7 the Herbert river at Abergowrie and MC3 Walsh river at Trimbles crossing produced clearly inaccurate results and were then disregarded from further analysis. This left 19 sites for the ANOVA comparison. The 1 in 10-year comparison produced very comparable results, in this comparison only two catchments apart of the Flinders-Norman basin produced results that were not significantly similar. The Flinders-Norman catchments are a part of the Semi-Arid region of the Areal rainfall reduction chart. The findings from the 1 in 10-year analysis suggest that drier catchments are not well suited to analysis by the GRADEX method.

The 1 in 100-year analysis produces results which are less significantly similar than those from the 1 in 10-year analysis. The catchments in the Semi-Arid region once again produced not statistically similar results. For further analysis, the major design discharges were broken into smaller data samples. One such data set was splitting the sites based on their respective 24hr design rainfall depth. These were broken into sites with greater than 375mm of rainfall and sites with less than 375mm of rainfall. The sites with rainfall greater than 375mm produced significantly similar results while the sites with less than 375mm produced results that were not significantly similar. The results from the 1 in 100-year analysis gives further evidence to the methods poor results in drier conditions.

The GRADEX method certainly is a suitable method for flood estimation in certain conditions. However, its dependency on rainfall depth and its tendency to produce poor results in arid conditions makes the method less appealing in Australian conditions. The method produces reliable results and is ideal in climates that receive plenty of rainfall and in these situations, it could be used to profound effect. The methods time-consuming nature and the uncertainty of the results reliability until the analysis is complete does makes the method less attractive. In Australian conditions, the GRADEX method could be well utilised as a comparison to standard ARR methods or in catchments that have limited historical data.

Statistics for USQ ePrint 40893
Statistics for this ePrint Item
Item Type: USQ Project
Item Status: Live Archive
Additional Information: Bachelor of Civil Engineering (Honours) (Civil)
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Civil Engineering and Surveying (1 Jul 2013 - 31 Dec 2021)
Supervisors: Brodie, Ian; Gillies, Malcolm
Date Deposited: 08 Sep 2022 03:34
Last Modified: 08 Sep 2022 03:34
Uncontrolled Keywords: GRADient of EXtreme values (GRADEX); flood; rainfall; catchment

Actions (login required)

View Item Archive Repository Staff Only