Scholarlink

<p>Machine-learning based methods are increasingly employed for the prediction of storm surges and development of early warning systems for coastal flooding. The evaluation of the quality of such methods needs to explicitly consider the uncertainty of the prediction, which may stem from the inaccuracy in the forecasted inputs to the model as well as from the uncertainty inherent to the model itself. Defining the range of validity of the prediction is essential for the correct application of such models. A methodology is proposed for building a robust model for forecasting storm surges accounting for the relevant sources of uncertainty. The model uses as inputs the mean sea level pressure and wind velocity components at 10&#xa0;m above sea level. A set of Artificial Neural Networks are used in conjunction with an adaptive Bayesian model selection process to make robust storm surge forecast predictions with associated prediction intervals. The input uncertainty, characterised by comparing hindcast data and one day forecasted data, is propagated through the model via a Monte Carlo based approach. The application of the proposed methodology is illustrated by considering 24&#xa0;h target forecast predictions of storm surges for Millport, in the Firth of Clyde, Scotland, UK. It is shown that the proposed approach significantly improves the predictive performance of existing machine learning based models and provides a meaningful prediction interval that characterises feature, model and forecast uncertainty. The forecast system has negligible computational time requirements and showed very good agreement with observations acccording different metrics and achieving e.g., a correlation coefficient of 0.942 for 24&#xa0;h forecasted surge from 2021 to 2023. The mean absolute error was 0.06&#xa0;m for all observations and only 0.10&#xa0;m for observations above 0.75&#xa0;m showing its accuracy for predicting extreme events.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Robust storm surge forecasts for early warning system: a machine learning approach using Monte Carlo Bayesian model selection algorithm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月7日

查看全文
  • E. Macdonald,
  • E. Tubaldi,
  • E. Patelli

摘要

Machine-learning based methods are increasingly employed for the prediction of storm surges and development of early warning systems for coastal flooding. The evaluation of the quality of such methods needs to explicitly consider the uncertainty of the prediction, which may stem from the inaccuracy in the forecasted inputs to the model as well as from the uncertainty inherent to the model itself. Defining the range of validity of the prediction is essential for the correct application of such models. A methodology is proposed for building a robust model for forecasting storm surges accounting for the relevant sources of uncertainty. The model uses as inputs the mean sea level pressure and wind velocity components at 10 m above sea level. A set of Artificial Neural Networks are used in conjunction with an adaptive Bayesian model selection process to make robust storm surge forecast predictions with associated prediction intervals. The input uncertainty, characterised by comparing hindcast data and one day forecasted data, is propagated through the model via a Monte Carlo based approach. The application of the proposed methodology is illustrated by considering 24 h target forecast predictions of storm surges for Millport, in the Firth of Clyde, Scotland, UK. It is shown that the proposed approach significantly improves the predictive performance of existing machine learning based models and provides a meaningful prediction interval that characterises feature, model and forecast uncertainty. The forecast system has negligible computational time requirements and showed very good agreement with observations acccording different metrics and achieving e.g., a correlation coefficient of 0.942 for 24 h forecasted surge from 2021 to 2023. The mean absolute error was 0.06 m for all observations and only 0.10 m for observations above 0.75 m showing its accuracy for predicting extreme events.