Tuesday, 24 January 2017: 9:15 AM
607 (Washington State Convention Center )
Airborne ocean profile surveys, with instruments typically deployed in a “lawnmower” pattern, have been conducted prior to a number of tropical cyclones (TCs) with the goal of improving ocean model initialization in coupled TC prediction models. Ocean Observing System Simulation Experiments (OSSEs) are performed to quantitatively assess the impact of airborne expendable CTD (AXCTD) surveys on reducing RMS error and bias in ocean analyses prior to North Atlantic hurricanes Edouard and Gonzalo (2014). A particular focus of these experiments is to provide recommendations for horizontal profile separation distance. The synthetic ocean surveys conducted prior to both storms substantially reduce RMS error and bias in upper-ocean dynamical and thermodynamical fields important for the model to accurately predict ocean feedback and SST cooling in response to TC forcing. Surveys with characteristic profile separation of 0.5 degrees provide substantially greater RMS error reduction than surveys with profiles separated by 1.0 degrees, demonstrating that high horizontal resolution is necessary for such surveys to be effective. To clarify this result, idealized OSSEs are performed that take advantage of the high-resolution representation of the truth provided by the Nature Run (NR) to quantify RMS error correction in wavenumber space. Idealized synthetic CTD profile surveys separated by 2.5, 1.0, and 0.5 degrees covering a 15 x 15 degree box in the open North Atlantic demonstrate that RMS error correction is confined to wavelengths exceeding the Nyquist wavelength (about 600, 240, and 120 km, respectively). Over smaller wavelengths, RMS error tends to increase. This situation arises from the inability of individual ocean CTD profiles to accurately correct smaller-scale field structure in their vicinity. This correction structure (the increment) results from the spreading of innovation by the background error covariance matrix tapered by the applied localization radius. The smoothing properties of the background error covariance prevent smaller-scale structure in model fields from being accurately corrected.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner