Can supervise: YES
Thi, PC, Ball, JE & Dao, NH 2018, 'Uncertainty estimation using the glue and Bayesian approaches in flood estimation: A case study-Ba River, Vietnam', Water (Switzerland), vol. 10, no. 11, pp. 1-15.View/Download from: UTS OPUS or Publisher's site
© 2018 by the authors. In the last few decades tremendous progress has been made in the use of catchment models for the analysis and understanding of hydrologic systems. A common application involves the use of these models to predict flows at catchment outputs. However, the outputs predicted by these models are often deterministic because they focused only on the most probable forecast without an explicit estimate of the associated uncertainty. This paper uses Bayesian and Generalized Likelihood Uncertainty Estimation (GLUE) approaches to estimate uncertainty in catchment modelling parameter values and uncertainty in design flow estimates. Testing of join probability of both these estimates has been conducted for a monsoon catchment in Vietnam. The paper focuses on computational efficiency and the differences in results, regardless of the philosophies and mathematical rigor of both methods. It was found that the application of GLUE and Bayesian techniques resulted in parameter values that were statistically different. The design flood quantiles estimated by the GLUE method were less scattered than those resulting from the Bayesian approach when using a closer threshold value (1 standard deviation departed from the mean). More studies are required to evaluate the impact of threshold in GLUE on design flood estimation.
© 2017 Engineers Australia Estimation of parameter values is an essential step in the application of catchment modelling systems. This step is time-consuming and requires considerable effort. While a variety of approaches have been developed to accelerate the process, this paper proposes a method to reduce significantly the number of parameters for a large catchment when a semi-distributed catchment modelling system is applied. Past studies have reported on the use of a scaling parameter to adjust parameter values from their initial values, introduced herein is the use of a scaling parameter together with a variation coefficient. This enables the spatial variation of changes in parameter values across the catchment to be considered. A case study was conducted for a 14,000 km 2 catchment to assess the validity of this approach where the focus of the catchment modelling was the prediction of a design flood statistic. This catchment was divided into 155 subcatchments with 5 sensitive parameters per subcatchment. Hence, a total of 775 parameters needed to be considered. Using the proposed approach, the number of parameters considered during the calibration was reduced to 8 coefficients which was reasonable for a calibration and validation process that also enabled an estimate of the parameter variability.
Cu, PT & Ball, JE 2017, 'The influence of the calibration metric on design flood estimation using continuous simulation', International Journal of River Basin Management, vol. 15, no. 1, pp. 9-20.View/Download from: UTS OPUS or Publisher's site
© 2016 International Association for Hydro-Environment Engineering and Research. Estimation of design ﬂood flow has been and remains a concern for both hydrologic research and hydrologic practice. Knowledge of design flood flows provides a basis for sustainable ﬂood management, which has the aim of reducing ﬂood risk, thereby protecting people's lives and property. Design ﬂoods for a given location can be estimated by a number of approaches including analysis of past flood statistics and the use of catchment modelling. When catchment modelling approaches are applied estimation of design flood flows, there is a need to calibrate the model parameters. As part of this calibration process, a calibration metric, or fitness measure, is needed to enable assessment of alternative sets of parameter values. Presented herein is an investigation into design flood quantiles derived from predictions obtained from a continuous catchment modelling system when alternative calibration metrics are used to assess the suitability of parameter values. Two alternative calibration metrics are considered with one calibration metric aimed at ensuring replication of recorded hydrographs and the second calibration metric aimed at ensuring replication of the statistical characteristics of the annual maxima series. It was found that use of the later calibration metric resulted in better reproduction of the flood probability model estimated from the historical data while reproduction of the recorded hydrographs (i.e. the first calibration metric) did not ensure reproduction of the flood probability model.
Ball, JE & Cu, PT 2015, 'Daily Rainfall Disaggregation For Flood Estimation', E-proceedings of the 36th IAHR World Congress, Congress of IAHR, the International Association of Hydro-Environment Engineering and Research, IAHR, The Hague, The Netherlands.View/Download from: UTS OPUS
Thi, PC & Ball, JE 2015, 'Estimating design flood magnitude for a Vietnamese catchment', The Art and Science of Water - 36th Hydrology and Water Resources Symposium, HWRS 2015, Hydrology and Water Resources Symposium and the International Conference on Water Resources and Environment Research, Engineers Australia, Hobart, Australia, pp. 1370-1379.
© 2015, Engineers Australia. All rights reserved.Flood is a common phenomenon in many tropical countries. Estimation of design flood flow has been a concern for many years in both hydrologic research and in hydrologic practice. Design flood magnitudes provide a basis for sustainable flood management which has the aims of reducing flood risk, and protecting people's lives and property. Design flood magnitudes for a given location can be estimated by a number of approaches including analysis of past flood statistics or the use of catchment modelling approaches like design storm methods or continuous simulation. The aim of this paper is to apply Annual Maximum Series fitting method for design flood estimation in continuous simulation with particular reference to a monsoon catchment. In this aspect, the annual maximum series was used as a performance measure rather than reproduction of individual hydrographs. This approach was used as the focus was on reproducing the observed frequency curve. For this purpose, a case study is performed for a large catchment, namely the Ba River, located in central Vietnam. This catchment is subject to a monsoonal climate and also to tropical cyclones.
Cu, P & Ball, JE 2014, 'Daily Rainfall Disaggregation for a Monsoon Catchment in Vietnam', Proceedings of the 35th Hydrology and Water Resources Symposium 2014, Hydrology and Water Resources Symposium and the International Conference on Water Resources and Environment Research, Engineers Australia, Perth, Western Australia, pp. 501-508.View/Download from: UTS OPUS
Rainfall is one of the more important factors influencing flows in a river and particularly so for flood flows in a river. Hence there is a need to adequately model rainfall if flows in rivers are to be predicted with reliability. There are three components to a rainfall model, namely the rainfall depth, the spatial distribution of rainfall and the temporal distribution of rainfall. This last aspect of a rainfall model is the focus of the study that will be presented in this paper. In many Asian countries, the data for assessing the sub-daily temporal distribution of rainfall is limited by the availability of continuous records. It is far more common for the available data to comprise daily rainfall records. There is a need, therefore to disaggregate these daily rainfall records. One method that has been applied to the problem of disaggregation of daily rainfall records is the Method of Fragments. Presented herein will be a discussion of the application of the Method of Fragments to the disaggregation of daily rainfall records in a monsoonal catchment in Vietnam. The case study catchment is the Ba River where 14 daily-read gauges and 12 hourly rainfall stations across the basin were available. The similarity of statistical characteristics of these gauges was evaluated by Welch Two Sample t-test and Bartlett test of homogeneity of variances. The model is validated at 3 stations, where recorded hourly data was available. Hourly rainfall of 14 stations is then disaggregated from daily data based on the historical observed hourly data from 12 hourly stations in the basin.