AI-Enhanced Remote Sensing Detection and Forecasting of Harmful Algal Blooms in Lake Pontchartrain

Harmful Algal Blooms (HABs) and particularly toxic cyanobacterial Harmful Algal Blooms (CyanoHABs) have become a growing threat to the environment, economy, communities, and human health. This is particularly true for Lake Pontchartrain that is located in southeastern Louisiana and has been facing water quality challenges particularly associated with CyanoHABs, causing frequent health advisories and fish kills. This paper presents an Artificial Intelligence(AI)-enhanced satellite remote sensing approach for detection and forecasting of CyanoHABs in Lake Pontchartrain and beyond. Specifically, an AI-based forecasting model has been developed by combining the decision-tree-based ensemble machine learning algorithm XGBoost (Extreme Gradient Boosting) and the remote sensing data from MODIS (Moderate Resolution Imaging Spectroradiometer) Aqua and Sentinel-3 satellites. The model input variables are selected by using the random forest (RF) method. It was found that the occurrence of CyanoHAB is strongly affected by the antecedent environmental conditions 10 – 20 days before. The AI-based forecasting model was validated using the observed CyanoHAB data from 2018. Model testing and validation results indicate that the AI-based forecasting model is capable of forecasting CyanoHABs with the lead-time of 10 days, providing an efficient and effective tool for implementing management interventions and thereby reducing the health risk to the general public.