The genesis of Hurricane Karl (2010) examined through cycling ensemble data assimilation experiments using PREDICT observations

The Pre-Depression Investigation of Cloud Systems in the Tropics (PREDICT) field campaign operated 25 flight missions through eight Atlantic tropical disturbances in 2010 to test a set of hypotheses regarding how hurricanes form from easterly waves. In this study, an ensemble Kalman filter (EnKF) is applied to examine the genesis of Hurricane Karl (2010) using routinely collected observations as well as those taken during PREDICT. Flight missions through this disturbance span five days, thus providing an excellent case study for exploring the impact of observations in modeling the characteristics of the pre-genesis environment. Forecasts initialized days before the genesis event have little skill in predicting the formation of Karl when conventional observations are used to produce the analysis. Nevertheless, the genesis forecasts improve greatly following the assimilation of PREDICT observations. The dropsondes are found to increase the lower- and upper-level circulation and column moisture in the analyses, which result in a 24-h improvement in predicting the genesis event. The accuracy of the genesis simulations are improved further when a two-way coupled EnKF-4DVar data assimilation system is applied to assimilate all observations during the cycling period. In addition to creating detailed analyses of the pre-Karl disturbance, this study examines the effectiveness of ensemble data assimilation using two domain configurations. Analyses are performed on 13.5- and 4.5-km grid spacing domains to evaluate the practicality of high-resolution data assimilation for the event. In general, this study evaluates the dynamics of the evolving weather system in an attempt to explain the observed changes in forecast error from a set of ensemble data assimilation experiments with field observations.