Application of Bayesian approach to hydrological frequency analysis
ABSTRACT An existing Bayesian flood frequency analysis method is applied to quantile estimation for Pearson type three (P-III) probability
distribution. The method couples prior and sample information under the framework of Bayesian formula, and the Markov Chain
Monte Carlo (MCMC) sampling approach is used to estimate posterior distributions of parameters. Different from the original
sampling algorithm (i.e. the important sampling) used in the existing approach, we use the adaptive metropolis (AM) sampling
technique to generate a large number of parameter sets from Bayesian parameter posterior distributions in this paper. Consequently,
the sampling distributions for quantiles or the hydrological design values are constructed. The sampling distributions of
quantiles are estimated as the Bayesian method can provide not only various kinds of point estimators for quantiles, e.g.
the expectation estimator, but also quantitative evaluation on uncertainties of these point estimators. Therefore, the Bayesian
method brings more useful information to hydrological frequency analysis. As an example, the flood extreme sample series at
a gauge are used to demonstrate the procedure of application.
KeywordsBayesian theory–hydrological frequency analysis–Markov Chain Monte Carlo–prior distribution–posterior distribution
- SourceAvailable from: link.springer.com[Show abstract] [Hide abstract]
ABSTRACT: To develop national economy and use the water resources and hydropower resources sufficiently, a lot of high arch dams, with the height of more than 200 m, have been and will be built in China. Although arch dams have good mechanical behavior, there is still failure possibility due to the huge water pressure and high stress level in dam, complex topographic and geologic conditions, formidable environment and high intensity earthquake. As one of the three main aspects concerning the safety of high arch dam, the study on global destruction, has been elaborated in the literatures, and research advance in the other two aspects, namely the failure risk and local damage of high arch dams, will be reviewed in this paper. In recent years, the failure risk of high arch dams has been investigated, and the model for identifying dam failure risk factors has been established. It is shown that the foundation deterioration and strong earthquake are the major risk sources for high dam failure. With the fault tree method, the failure mode and failure probability of high arch dams are studied, and the principle for determining failure mode and the method of calculating failure probability are proposed. Meanwhile, the determination principle of acceptable risk standard for high arch dam was proposed, and the acceptable risk rate and the acceptable standard value of various risk losses were given. For the local damage of arch dam, it is pointed out that the local damage belongs to the strength failure at material level. The study on local failure mechanism of arch dam is reviewed, based on the theories that from traditional strength theory to damage mechanics and meso-mechanics theory. Aiming at the cracking, the main pattern of local failure of high concrete dam, the research advances in the analysis methods and cracking criteria for smeared crack model and discrete crack model are summarized, and the research findings of preventive measures for local failure are shown.Chinese Science Bulletin 57(36). · 1.37 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Parameter optimization of a hydrological model is an indispensable process within model development and application. The lack of knowledge regarding the efficient optimization of model parameters often results in a bottle-neck within the modeling process, resulting in the effective calibration and validation of distributed hydrological models being more difficult to achieve. The classical approaches to global parameter optimization are usually characterized by being time consuming, and having a high computation cost. For this reason, an integrated approach coupling a meta-modeling approach with the SCE-UA method was proposed, and applied within this study to optimize hydrological model parameter estimation. Meta-modeling was used to determine the optimization range for all parameters, following which the SCE-UA method was applied to achieve global parameter optimization. The multivariate regression adaptive splines method was used to construct the response surface as a surrogate model to a complex hydrological model. In this study, the daily distributed time-variant gain model (DTVGM) applied to the Huaihe River Basin, China, was chosen as a case study. The integrated objective function based on the water balance coefficient and the Nash-Sutcliffe coefficient was used to evaluate the model performance. The case study shows that the integrated method can efficiently complete the multi-parameter optimization process, and also demonstrates that the method is a powerful tool for efficient parameter optimization.Chinese Science Bulletin 57(26). · 1.37 Impact Factor