TIDAL SIMULATION USING REGIONAL OCEAN MODELING SYSTEM (ROMS) Xiaochun Wang 1,2 , Yi Chao 1 , Zhijin Li 1,2 , John Farrara 1,2 , Changming Dong 3 , James C. McWilliams 3 , C. K. Shum 4 , Yu Wang 4 , Koji Matsumoto 5 , Leslie K. Rosenfeld 6 , and Jeffrey D. Paduan 6 1 MS 300-323, JPL/Caltech, 4800 Oak Grove Dr., Pasadena, CA 91109, USA 2 Raytheon ITSS, 299 North Euclid Ave. Suite 500, Pasadena, CA 91101, USA 3 IGPP, UCLA, 405 Hilgard Ave., Los Angeles, California, 90095, USA 4 Ohio State University, 125 S Oval Mall, Columbus, Ohio 43210, USA 5 National Astronomical Observatory of Japan, 2-12, Hoshigaoka, Mizusawa, Iwate 023-0861, Japan 6 Department of Oceanography, Naval Postgraduate School, Monterey, California, 93943, USA ABSTRACT A three dimensional general circulation model is used to simulate tides along the central western coast of U.S. The model, which is configured from the Regional Ocean Modeling System (ROMS), is three-level nested with the finest resolution of 1.6 km in the Monterey Bay, Cali- fornia. Forced by tidal signal along the open boundaries in west, north and south directions, ROMS can simulate barotropic tides reasonably well in the region. The total discrepancy of the amplitudes of eight major tide con- stituents, as measured by root of summed squares, is 3.5 cm in the open ocean compared with tide amplitudes es- timated by Topex/POSEIDON along-track altimetry ob- servation. Along the coastal region, the discrepancy of amplitudes is 5.4 cm which is about 10% of the ampli- tude of the most energetic M 2 constituent. For these ma- jor tide constituents, the phase error is generally much less than half hour. The simulated sea surface tidal cur- rent, which is heavily influenced by internal tide activity, shows sensitivity to the stratification of the model and has large room to improve. 1. INTRODUCTION The presence of tidal signal poses a major challenge for the development of coastal operational forecasting sys- tems. The barotropic tide signal associated with sea sur- face height is relatively straightforward to predict and simulate. However, the interaction of barotropic tides and topography would generate internal tides and make the flow pattern very complex with the presence of time- dependent stratification and mean current. Another mo- tivation to develop a tidal-permitting circulation model is that the temperature, salinity and current collected by moving platforms (e.g., gliders or AUVs) contain both the circulation and tidal signals. To assimilate these data into a non-tidal-permitting model will introduce large er- rors in the model. For many applications of a coastal op- erational forecasting system, it is certainly desirable to simulate tide directly instead of providing a detided solu- tion. A three level nested model is configured for the Monterey Bay, California to simulate tide. The oceanic general circulation model used is the Regional Ocean Modeling System (ROMS), which is a community model designed for coastal applications [1]. The purpose of our research is to test the capability of ROMS in simulating tides. The research also serves as a necessary exercise to implement tides in an operational ocean forecasting system. In this paper, we emphasize the validation of the model tide simulation. The characteris- tics and energetics of tides of the region will be reported in separate publications. The paper is organized as following. After the brief Intro- duction, Section 2 discusses the model we use. Section 3 validates tidal simulation using tidal parameters from satellite altimetry observation and tide gauges. Section 4 presents the sea surface tidal current simulation and its comparison with high frequency coast radar observation. Section 5 summarizes our research. 2. MODEL CONFIGURATION AND TIDAL FORCING The model used is three-level nested (Fig. 1). The out- most model domain which has the coarsest-resolution (L0 model) covers the U.S. western coastal region from southern California to Oregon, and level 1 model (L1 model) covers the central and northern California coast and level 2 model (L2 model) zooms in for the Monterey Bay region. The nesting of the model is realized through the Adaptive Grid Refinement in Fortran (AGRIF) pack- age which is based on the use of pointers [2]. The pack- age is systematically tested in [3]. The testing indicates that the package can provide a continuous solution at the interface of coarse and fine model grid. Table 1 lists the domains and details of our nested model. The horizontal