Storm-scale Data Assimilation and Ensemble Forecasts of the 27 April 2011 Severe Weather Outbreak in Alabama, Part 2

In part 1 of the presentation, we describe a multi-scale ensemble analysis and forecasting system employing the WRF-ARW model and the Data Assimilation Research Testbed (DART) software. The system is tested for the outbreak of tornadic supercells in northern Alabama on 27 April 2011. In part 2, we take a closer look at the results, attempting to address, through sensitivity experiments and additional analysis, some of the questions raised in part 1 regarding biases in storm motion, insufficient ensemble spread, and limited predictability of some storms. Preliminary results indicate that reducing horizontal grid spacing of the storm-scale system from 3 to 1 km reduces bias errors in predicted storm motion but does not contribute significant additional ensemble spread. Therefore, other options for increasing spread in 0-1 h forecasts are being investigated. The storm-scale system generally predicts the significant isolated tornadic supercells with high probabilities, long lead times, and consistency from one initialization time to the next. In contrast, cases with multi-storm interactions tend to have high variability in mesocyclone forecasts from one initialization time to the next. Factors affecting predictability of the outcome of these multi-storm interactions will be examined through ensemble sensitivity analysis.

Note: This abstract describes the second part of a two-part presentation. We ask the program committee to consider scheduling the two presentations back-to-back.