A predictive surge model based on Artificial Neural Networks (ANN) was recently developed for the Texas coast and its waterways. The model was implemented to make web accessible water level predictions at several stations of the Texas Coastal Ocean Observation Network (TCOON). Present operational predictions are based on past water levels and wind measurements at the target station as well as at an additional station. The models are being upgraded to include wind predictions from the National Centers for Environmental Prediction (NCEP) Nonhydrostatic Mesoscale Model (NMM) recently integrated within the Weather Research and Forecasting (WRF) framework. These predictions are uploaded four times per day through a collaboration with the local office of the National Weather Service. The talk will include a brief description of the training and optimization of the operational models as well as a description of their implementation on the World Wide Web. Challenges related to the addition of atmospheric predictions to the operational models will be discussed. The performance of the ANN models with and without wind predictions will then be discussed and compared with other simple water level models. While the models were initially not designed for tropical storms and hurricanes, their performance for up to 12 to 24 hour predictions in Texas bays and estuaries has proved quite accurate and will be discussed along with recommendations on how model predictions should be used and communicated during such events.