OSSE Evaluation of Rapid Airborne Ocean Observing Strategies in the Gulf of Mexico

During the Deepwater Horizon (DWH) oil spill, nine airborne surveys conducted by NOAA WP-3D hurricane research aircraft collected upper-ocean temperature, salinity, and velocity profiles over the interior eastern Gulf of Mexico. The goal was to improve the accuracy of data-assimilative ocean analyses and forecasts used to predict the transport and dispersion of the oil. There is also interest in conducting rapid airborne surveys ahead of approaching hurricanes to improve ocean model initialization in coupled hurricane forecast models. Because rapid airborne surveys have not been subjected to rigorous design studies in the past, Observing System Simulation Experiments (OSSEs) are performed using a new OSSE system recently validated over the interior Gulf of Mexico. These experiments demonstrate that airborne surveys substantially reduce errors in data-assimilative ocean analyses beyond the error reduction achieved by satellite altimetry assimilation. Several questions were addressed involving temporal and spatial sampling resolution, profile depths, and temporal delays in the availability of the profiles for assimilation. Because of rapid error growth in the Gulf of Mexico, typically half of the error reduction achieved by assimilating airborne profiles is lost within 5-7 days following each survey. Horizontal resolution is important, with substantial additional error reduction achieved when profiles are sampled at one-half versus one degree resolution. Instrument type and profile depth are also important. Surveys conducted with airborne XBTs sampling temperature to 400 m substantially reduce upper-ocean temperature and heat content errors, which is important for hurricane forecasting. However, surveys conducted with airborne XCTDs sampling temperature and salinity to 1000 m further reduce errors in ocean currents and upper-ocean salinity, demonstrating advantages of using this instrument in lieu of airborne XBTs. Finally, OSSEs illustrate the importance of making these quality-controlled observations available for assimilation within 1-2 days after they are taken because errors increase rapidly for longer lags.