Development, implementation, and validation of a California coastal ocean modeling, data assimilation, and forecasting system. (May 2018)

• Record Type: Journal Article
• Authors: Chao, Yi; Farrara, John D.; Zhang, Hongchun; Armenta, Kevin J.; Centurioni, Luca; Chavez, Francisco; Girton, James B.; Rudnick, Dan; Walter, Ryan K.
• Is Part Of: Deep sea research. Volume 151(2018)
• Page Start: 49
• Page End: 63
• Publication Date: 2018-05
• Subjects: Coastal ocean modeling -- Model validation -- Californa Current system; Oceanography -- Periodicals; Ocean bottom -- Periodicals; Marine biology -- Periodicals; 551.46
• DOI: 10.1016/j.dsr2.2017.04.013
• Languages: English
• ISSNs: 0967-0645

Abstract

Abstract: A three-dimensional, near real-time data-assimilative modeling system for the California coastal ocean is presented. The system consists of a Regional Ocean Modeling System (ROMS) forced by the North American Mesoscale Forecast System (NAM). The ocean model has a horizontal resolution of approximately three kilometers and utilizes a multi-scale three-dimensional variational (3DVAR) data assimilation methodology. The system is run in near real-time to produce a nowcast every six hours and a 72-hour forecast every day. The performance of this nowcast system is presented using results from a six-year period of 2009–2015. The ROMS results are first compared with the assimilated data as a consistency check. RMS differences in observed satellite infrared sea surface temperatures (SST) and vertical profiles of temperature between observations and ROMS nowcasts were found to be mostly less than 0.5 °C, while the RMS differences in vertical profiles of salinity between observations and ROMS nowcasts were found to be 0.09 or less. The RMS differences in SST show a distinct seasonal cycle that mirrors the number of observations available: the nowcast is less skillful with larger RMS differences during the summer months when there are less infrared SST observations due to the presence of low-level clouds. The larger differences during summer were found primarily along the northern and central coasts in upwelling regions where strong gradients exist between colder upwelled waters nearshore and warmer offshore waters. RMS differences between HF radar surface current observations and ROMS nowcasts were approximately 7–8 cm s −1, which is about 30% of the time mean current speeds in this region. The RMS differences in sea surface height (SSH) between the AVISO (Archiving, Validation and Interpretation of Satellite Oceanographic) altimetric satellite observations and ROMS nowcasts were about 2 cm. In addition, the system realistically reproduces the interannual variability in temperatures at the M1 mooring (122.03°W, 36.75°N) in Monterey Bay, including the strong warming of the California coastal ocean during 2014. The ROMS nowcasts were then validated against independent observations. A comparison of the ROMS nowcast with independent profile observations of temperature and salinity shows RMS differences of 0.7 to 0.92 °C and 0.13 to 0.17, which are larger (by up to a factor of 2) than the differences found in the comparisons with assimilated data. Validation of the depth-averaged currents derived from Spray gliders shows that the flow patterns associated with California Current and California Undercurrent/Davidson current systems and their seasonal variations are qualitatively reproduced by the ROMS modeling system. Lastly, the impact of two recent upgrades to the system is quantified. Switching the lateral boundary conditions from a U.S. west coast regional model to the global HYCOM (HYbrid Coordinate Ocean Model) model results in an improvement in the simulation of the seasonal and interannual variations in the SSH, especially south of Pt. Conception (120.47°W, 34.45°N). The assimilation of altimetric satellite SSH data also results in an improvement in the model surface currents when compared to independent surface drifter observations.