| This objective of this research was to design, develop, implement and assess a new methodology for producing accurate coastal predictions of water levels and currents. The culmination of this research is an operational Coastal Prediction System (CPS) that reliably produces predictions that are generally more accurate and spatially resolved than the existing harmonic analysis method.; The architecture of the Coastal Prediction System consists of automated, integrated, subsystems for acquisition, processing and quality control of data; three-dimensional numerical circulation modeling; and dissemination of results to the Internet. Tampa Bay, Florida was selected as the test bed for this research.; Predictions made by the system can be for past, present, future or specified time frames (hindcast, nowcast, forecast or on-demand).; The hindcast protocol is operated in a manual mode, and is used for the purposes of providing a detailed calibration of the numerical model, and a subsequent framework for conducting detailed scientific and management studies.; Both the nowcast and forecast protocols are fully automated. The nowcast model performs model integration updates on the order of every 6 to 12 minutes. The forecast protocol performs 24 hour forecasts every 4 hours. The on-demand protocol is manually invoked, but can automatically perform simulations spanning hindcast to forecast time frames.; Skill assessment of the hindcast model using National Ocean Service methods obtained water level skill scores (SD, SA, and SL), of 96.4%, 95.9%, and 94.1%, and current skill scores of 90.5%, 88.6%, and 89.9%. The mean error (ME) and mean absolute error (MAE) of the modeled salinity and observed salinity was less than 0.23 and 0.85 psu, respectively.; Skill assessment of the automated nowcast protocol, during the year 2000, obtained water level skill scores of 95.6%, 95.5%, and 94.6%, and current skill scores of 91.7%, 91.5%, and 90.5%.; A method for constructing the open boundary forecast water level was developed that showed during year 2000, improvements relative to tidal harmonics were 39% and 77% for MAE and frequency of absolute errors greater than 0.15 meters (outliers).; Skill assessment of the automated forecast protocol, during June through December 2001, obtained National Ocean Service water level skill scores of 90.2%, 91.9%, and 93.9%. With respect to the MAE, the forecast model water levels were 50.0% more accurate than the harmonics. The forecast model obtained skill scores for currents of (SD, SA, and SL) of 89.3%, 86.0% and 87.2%. The harmonic current skill scores were slightly higher, with values of 91.1%, 88.4% and 88.3%.; Analysis indicates that during year 2000, the automated nowcast protocol routinely provides disseminated products within a lag of 0.27 hours of the present time, while the forecast protocol reliably produces predictions with 24 hour leads.; The results of this research soundly demonstrate that the CPS is a feasible method for reliably producing accurate predictions of water levels and currents. Although developed in a research mode, this technology can be readily implemented in a fully operational mode. In this manner, this technology can be expected to radically enhance the present procedures for maritime, research, military and public sectors to obtain coastal predictions. |
