Prediction of algal blooms via data-driven machine learning models: an evaluation using data from a well-monitored mesotrophic lake

Abstract. With increasing lake monitoring data, data-driven machine learning (ML) models might be able to capture the complex algal bloom dynamics that cannot completely described in process-based (PB) models. We applied two ML models, gradient boost regressor (GBR) and long short-term memory (LSTM) network, predict blooms seasonal changes chlorophyll concentrations (Chl) a mesotrophic lake. Three predictive workflows were tested, one based solely on available measurements others applying two-step approach, first estimating nutrients have limited observations then predicting Chl using observed pre-generated environmental factors. The third workflow was developed hydrodynamic data derived from PB model as additional training features approach. performance of superior Chl. hybrid further improved prediction timing magnitude blooms. A sparsity test shuffling order testing years showed accuracy decreased with sample interval, varied training–testing year combinations.