Inter-comparison and coupling of ensemble and variational data assimilation approaches for mesoscale modeling

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 5 February 2014: 11:15 AM
Room C201 (The Georgia World Congress Center )
Jonathan Poterjoy, Penn State University, University Park, PA; and F. Zhang

A fully coupled EnKF-4DVar (E4DVar) data assimilation and prediction system has been developed for the WRF model. The E4DVar system requires that EnKF and 4DVar run independent of one another with two coupling steps: 4DVar uses the ensemble forecast mean and perturbations at the beginning of the observation time window, and the minimizing solution replaces the posterior EnKF analysis after assimilating the available data. The hybrid variational scheme reduces the number of inner-loop iterations by about 30% for the multi-incremental form of 4DVar that is used in this study, owing to the smooth first guess and ensemble estimation of background error covariance. Results are presented from a series of cloud-permitting analyses and forecasts to examine the performance of E4DVar in assimilating real observations for a case of tropical cyclogenesis (Hurricane Karl, 2010). The assimilated data include routinely collected observations as well as dropsondes from the Pre-Depression Investigation of Cloud Systems in the Tropics (PREDICT) field campaign. The coupled data assimilation method leads to a reduction in wind and moisture error in the vicinity of the developing tropical wave, which translates into more accurate forecasts for the timing and location of genesis. Furthermore, analysis increments from the E4DVar method reveal a complex relationship between the specified error covariance and four-dimensional minimization using the adjoint model. This study demonstrates the effectiveness of E4DVar at the mesoscale, while examining possible savings in terms of computation cost.