Replacing unstructured WW3 in the Great Lakes with a Recurrent neural network and a boosted ensemble decision tree

Investigated replacing unstructured WW3 in the Great Lakes with (i) a Recurrent Neural Network (RNN, especially an LSTM) developed by EMC and (ii) a boosted ensemble decision tree (XGBoost) developed by GLERL. These two AI models were trained on two decades of wave observations in Lake Erie and compared to the operational Great Lakes unstructured WW3.