Prediction of Harmful Algal Blooms Severity Using Machine Learning and Deep Learning Techniques

The prediction of harmful algal blooms is most important because harmful algal blooms are having a high impact on marine ecosystems and human health. In this research, proposed a method for predicting harmful algal blooms (HABs) using the light gradient boost machine (LightGBM) algorithm. The proposed method utilizes metadata and satellite imagery from NASA to forecast when HABs will arise, with the root mean square error (RMSE) used to evaluate the accuracy of the prediction algorithm. The method achieves high accuracy and low false alarm rates compared to existing boosting algorithms, namely XGBoost, CatBoost, and CNN. Experimental results are conducted in order to identify harmful algal species. These findings demonstrate the effectiveness of using boosting algorithms and metadata for HAB prediction, which enable more effective management of the harmful impacts of HAB.