A coastal estuarine ensemble forecasting system
This research has been conducted in an effort to reduce the computational costs associated with high-resolution coastal estuarine modeling in ADCIRC. It was proposed that model scalability will be significantly enhanced if the conventional single-domain technique is replaced by a nesting method. Preliminary testing indicated the potential for 50% to more than 80% simulation runtime reduction (depending on the study site) by implementing a nesting technique. Once the concept was proved, the development of an operational, ensemble-based forecasting system, exclusively for the east coast of Central Florida and the IRL system, was implemented. The system, referred to as Ensemble Multistage, was built and placed in operation during winter 2019. Ensemble Multistage is released as opensource software and features the combined qualities of simplicity, utility, scalability, high-performance computer-independency, and probabilistic presentation of predictions. This work presents the multi-wind forcing ensemble modeling consisting of seven forecast simulations forced by three weather forecast models. For the duration of operation, January-June 2019, the system shows the capability of reflecting forecast wind uncertainties in forecast waves and circulation. In terms of forecast skill, great improvements in wave forecasting are achieved in near-shore regions where waves are monitored and real-time data are available. Improvements in water level predictions are event-dependent in that the model response might not well be captured by any forecast members during low-impact wind events in which the circulation is not wind driven. With respect to wind models, wave predictions made by NAM and SREF were in better agreements with observations.