The Naval Oceanographic Office (NAVOCEANO) deployed a Texas Automated Buoy System (TABS) during January 2003 to collect data expressly for the assessment and improvement of our operational Shallow-Water Analysis and Forecast System (SWAFS) model. TABS observed ocean temperatures, salinity, depth, and currents, along with air temperature, humidity, pressure and winds. These data were stored and relayed at half-hour intervals via Iridium, processed, archived, and made available to NAVOCEANO modelers in near-real-time. Using time series and inferred Lagrangian drifts, comparisons are made between these observations and ocean properties predicted daily by 48-hour SWAFS forecasts. The information is used to make model adjustments leading to improvements in forecast current speed, direction, and phase. The decay of model skill over the forecast period is examined, and we discuss how these comparisons can be used to provide users of SWAFS (and similar ocean models) with guidance on model uncertainties, strengths, and weaknesses.