Abstract
Observing System Simulated Experiments (OSSEs) performed during the 2014 North Atlantic hurricane season quantify ocean observing system impacts with respect to improving ocean model initialisation in coupled tropical cyclone (TC) prediction systems. The suitability of the OSSE system forecast model (FM) with respect to the previously validated Nature Run is demonstrated first. Analyses are then performed to determine the calibration required to obtain credible OSSE impact assessments. Impacts on errors and biases in fields important to TC prediction are first quantified for three major components of the existing operational ocean observing system. Satellite altimetry provides the greatest positive impact, followed by Argo floats and sea surface temperature measurements from both satellite and in-situ systems. The OSSE system is then used to investigate observing system enhancements, specifically regional underwater glider deployments during the 2014 hurricane season. These deployments resulted in modest positive impacts on ocean analyses that were limited by (1) errors in the horizontal structure of the increment field imposed by individual gliders and (2) memory loss in the spreading of these corrections by nonlinear model dynamics. The high-resolution, three-dimensional representation of the truth available in OSSE systems allows these issues to be studied without high-density ocean observations.
