Parameter optimization and uncertainty assessment for rainfall frequency modeling using an adaptive...

Area of Science:

• Hydrology
• Statistical Modeling
• Environmental Science

Background:

• Extreme rainfall frequency analysis is crucial for infrastructure design.
• Traditional methods like Maximum Likelihood Estimation (MLE) may provide overly optimistic results and insufficient uncertainty quantification.
• Bayesian inference offers a robust framework for parameter estimation and uncertainty assessment.

Purpose of the Study:

• To introduce a novel parameter optimization and uncertainty assessment procedure for extreme rainfall frequency modeling.
• To compare the efficiency of the adaptive Metropolis-Hastings (AM-H) algorithm with conventional Maximum Likelihood Estimation (MLE).
• To assess the practical application of the proposed method in Chaohu city, China.

Main Methods:

• Utilizing Bayesian inference with an adaptive Metropolis-Hastings (AM-H) algorithm.
• Employing a Markov chain Monte Carlo sampler to explore posterior distributions.
• Comparing AM-H with MLE for parameter estimation and uncertainty quantification.

Main Results:

• The adaptive Bayesian method (AM-H) demonstrated superior estimation for extreme rainfall return levels compared to MLE.
• MLE tended to produce overly optimistic estimates in the Chaohu city case study.
• AM-H proved more reliable for uncertainty quantification, providing narrower credible intervals for quantile estimates.

Conclusions:

• The AM-H algorithm is a more dependable approach for extreme rainfall frequency modeling and uncertainty quantification.
• The developed procedure is instrumental for accurate risk assessment in urban storm drainage planning.
• Bayesian inference with AM-H offers significant advantages over MLE for hydrological frequency analysis.