Sea surface temperature evaluation of the Coastal Ocean Forecast System

The Coastal Ocean Forecast System (COFS) is an experimental forecast system for the U.S. East Coast that is based on operationally forcing the Princeton Ocean Model with daily forecast fluxes of momentum and heat derived from the meso-Eta atmospheric forecast model. The focus of this paper is on the skill assessment of COFS by comparing the 24-hour forecast sea surface temperature (SST) to available observations. For example, horizontal maps of predicted SST are qualitatively compared to remotely-sensed MCSST data. In addition, quantitative comparisons are made between the predicted SST and in situ observations from National Data Buoy Center (NDBC) fixed buoys. Time series from representative NDBC buoys and from the nearest model grid points are compared for a two year time period beginning October 1, 1993 and ending September 31, 1995. Root mean square differences and correlation coefficients assess the overall skill of the forecasts at the buoy locations, and standard cross-spectral techniques identify frequency bands of coherence between data and model. This preliminary evaluation suggests that there is some remarkable agreement between data and model; however, depending on the physical regime there is considerable room for improvement. Work presently underway is directed toward the assimilation of satellite-derived SST data, and also sea surface height (SSH) data derived from satellite altimeters, into the ocean model. It is anticipated that SST and SSH assimilation will result in improvements in forecast temperature and currents in the entire three-dimensional COFS domain.