Uncertainty Analysis for the Calculation of Marine Environmental Design Parameters in the South China Sea

Guilin Liu,Xinsheng Zhou,Yi Kou,Fang Wu,Daniel Zhao,Yu Xu

DOI: https://doi.org/10.1007/s00343-022-2052-y

2023-01-01

Journal of Oceanology and Limnology

Abstract:The calculation results of marine environmental design parameters obtained from different data sampling methods, model distributions, and parameter estimation methods often vary greatly. To better analyze the uncertainties in the calculation of marine environmental design parameters, a general model uncertainty assessment method is necessary. We proposed a new multivariate model uncertainty assessment method for the calculation of marine environmental design parameters. The method divides the overall model uncertainty into two categories: aleatory uncertainty and epistemic uncertainty. The aleatory uncertainty of the model is obtained by analyzing the influence of the number and the dispersion degree of samples on the information entropy of the model. The epistemic uncertainty of the model is calculated using the information entropy of the model itself and the prediction error. The advantages of this method are that it does not require many-year-observation data for the marine environmental elements, and the method can be used to analyze any specific factors that cause model uncertainty. Results show that by applying the method to the South China Sea, the aleatory uncertainty of the model increases with the number of samples and then stabilizes. A positive correlation was revealed between the dispersion of the samples and the aleatory uncertainty of the model. Both the distribution of the model and the parameter estimation results of the model have significant effects on the epistemic uncertainty of the model. When the goodness-of-fit of the model is relatively close, the best model can be selected according to the criterion of the lowest overall uncertainty of the models, which can both ensure a better model fit and avoid too much uncertainty in the model calculation results. The presented multivariate model uncertainty assessment method provides a criterion to measure the advantages and disadvantages of the marine environmental design parameter calculation model from the aspect of uncertainty, which is of great significance to analyze the uncertainties in the calculation of marine environmental design parameters and improve the accuracy of the calculation results.