Presentation Information
[swOhB7iC-05]Unveiling Coastal Dynamics: Integrating Global Ice Extent and Climate Data to Improve Tide Predictions in the Bay of Bengal
*Kazi Rafid Raiyan (1. Independent Researcher (Japan))
Rising sea levels and extreme weather events are leading causes of increasing challenges in coastal regions, particularly in deltaic areas such as Bangladesh. These difficulties should be addressed effectively, and to do so, precise predictions of tidal height are crucial. This paper predicted daily high and low tides at coastal stations in Bangladesh utilizing various regression models, including Machine Learning (polynomial, support vector, XGBoost, random forest) and Deep learning (neural network).
The paper originated a novel dataset for Bangladesh that incorporates atmospheric data, tide height, and additionally, introduced a completely new feature, Ice extent, that has not been used in previous work to estimate tidal heights. The research indicated that including this new feature to predict tide height has improved the R2 score of regression models by 2\%-5\%. Eventually, after running all the models, the study found that the Neural regressor outperformed all the other models (with approximately 88\%-94\% R2 score) in predicting tide height (HIGH) for all of the stations. However, XGB Regressor outperformed in some stations, such as Chitalkhali and Kuakata (with around 83\%-95\% R2 score) for predicting tide height (LOW). In summary, This study provided detailed insights into the strenghts and weaknesses of various regression models, incorporated unique features and enhanced the prediction of coastal tide heights. Additionally, the method and findings of this study offer significant insight for developing strategies to manage coastal difficulties and adapt to climate change in an ever changing atmosphere.
The paper originated a novel dataset for Bangladesh that incorporates atmospheric data, tide height, and additionally, introduced a completely new feature, Ice extent, that has not been used in previous work to estimate tidal heights. The research indicated that including this new feature to predict tide height has improved the R2 score of regression models by 2\%-5\%. Eventually, after running all the models, the study found that the Neural regressor outperformed all the other models (with approximately 88\%-94\% R2 score) in predicting tide height (HIGH) for all of the stations. However, XGB Regressor outperformed in some stations, such as Chitalkhali and Kuakata (with around 83\%-95\% R2 score) for predicting tide height (LOW). In summary, This study provided detailed insights into the strenghts and weaknesses of various regression models, incorporated unique features and enhanced the prediction of coastal tide heights. Additionally, the method and findings of this study offer significant insight for developing strategies to manage coastal difficulties and adapt to climate change in an ever changing atmosphere.
Comment
To browse or post comments, you must log in.Log in