Bayesian estimation of extreme flood quantiles using a rainfall-runoff model and a stochastic daily rainfall generator

摘要: Abstract Extreme flood estimation has been a key research topic in hydrological sciences. Reliable estimates of such events are necessary as structures for conveyance continuously evolving size and complexity and, result, their failure-associated hazards become more pronounced. Due to this fact, several techniques intended improve frequency analysis reducing uncertainty extreme quantile have addressed the literature last decades. In paper, we develop Bayesian framework indirect quantiles from rainfall-runoff models. proposed approach, an ensemble long daily rainfall series is simulated with stochastic generator, which models amounts upper-bounded distribution function, namely, 4-parameter lognormal model. The rationale behind generation model that physical limits amounts, consequently floods, exist by imposing appropriate upper bound probabilistic model, plausible can be obtained those very low exceedance probabilities. Daily time converted into streamflows routing each realization synthetic through conceptual hydrologic Rio Grande Calibration parameters performed nonlinear regression means specification statistical residuals able accommodate autocorrelation, heteroscedasticity nonnormality. By combining outlined steps structure analysis, one properly summarize resulting estimating accurate credible intervals set interest. method was applied American river catchment, at Folsom dam, state California, USA. Results show most including exceptionally large non-systematic events, were reasonably estimated approach. addition, accounting uncertainties modeling step, obtain better understanding influential factors formation dynamics.