3.5 A Comparison of Multi- and Single-Core Convection-Allowing Ensembles

Monday, 7 January 2019: 3:30 PM
North 232C (Phoenix Convention Center - West and North Buildings)
Benjamin T. Blake, IMSG and NOAA/NCEP/EMC, College Park, MD; and J. R. Carley, M. E. Pyle, G. S. Romine, E. Rogers, E. Aligo, J. J. Levit, and J. Kain

As part of the unification effort to simplify the NCEP production suite, EMC plans to move toward a single-core, FV3-based convection-allowing ensemble with comparable performance to a multi-core ensemble. The current operational High Resolution Ensemble Forecast system (HREF) is underpinned by two separate dynamical cores, having diversity in physics as well as initialization time. Ongoing validation efforts, which have often focused on convective fields, continue to reveal that current experimental single-core ensembles have yet to outperform the multi-core HREF. In addition, it is well known that single-core, convection-allowing ensembles often tend to be underdispersive. Furthermore, while multicore ensembles tend to be more dispersive they can exhibit undesirable characteristics, such as clustering by dynamical core, and they are much more difficult to maintain and develop over time.

Comprehensive statistics and analyses are needed to identify where the most significant differences arise between multi- and single-core ensembles, such that development efforts may be appropriately concentrated toward improving the single-core ensemble to reduce the gap in skill. This presentation will share and discuss verification results comparing the performance of the HREF to the NCAR convection-allowing ensemble, a single-core ensemble that ran through 2017.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner