Estimation method to analyze parameter and prediction uncertainty of River Storage Mode using the Bayesian inference and segment mixture likelihood

摘要

An embodiment of the present invention is obtained by assuming a prior probability distribution of parameters of a river storage model based on a time concentration curve collected from a measurement target river, and sampling the first parameter from the prior probability distribution. a prior probability distribution assumption step; a model simulation step of simulating a river reservoir model based on the sampled parameters; The first likelihood is calculated through a preset formula, obtained by sampling a second parameter different from the first parameter, a new river reservoir model is simulated based on the obtained parameter, and the second likelihood is calculated a likelihood calculation step of calculating; a third parameter obtaining step of calculating a ratio for the first likelihood and the second likelihood, and sampling and obtaining a third parameter according to the magnitude of the calculated ratio; a posterior probability distribution deriving step of obtaining parameters after the third parameter until a preset convergence condition is satisfied, and deriving a posterior distribution of parameters of the river storage model with the obtained parameters; and a step of deriving a confidence interval for deriving a 95% prediction confidence interval based on the derived posterior probability distribution. Evaluating the parameter uncertainty and prediction uncertainty of the river reservoir model by applying the Bayesian inference technique and the curve division mixed likelihood. discloses a method to