A Fully Unsupervised Machine Learning Framework for Algal Bloom Forecasting in Inland Waters Using MODIS Time Series and Climatic Products

Progressively monitoring water quality is crucial, as aquatic contaminants can pose risks to human health and other organisms. Machine learning support the development of new effective tools for monitoring, including detection algal blooms from remotely sensed image series. Therefore, in this paper, we introduce Algal Bloom Forecast (ABF) framework, a fully automated framework bloom prediction inland bodies. Our approach combines machine learning, time series products (i.e., Moderate-Resolution Imaging Spectroradiometer (MODIS) images), environmental data spectral indices build anomaly models that predict occurrence events posterior period. assessments focused on application ABF equipped with vector (SVM), random forest (RF), long short-term memory (LSTM) methods, outcomes which were compared through different evaluation metrics such global accuracy, kappa coefficient, F1-Score R2-Score. Case studies covering Erie (USA), Chilika (India) Taihu (China) lakes are presented demonstrate effectiveness flexibility our approach. Based comprehensive experimental tests, found best predictions achieved by bringing together design RF model.