During the analysis of Weather Research and Forecasting (WRF) predictions of 100 severe weather outbreaks using NCEP Reanalysis as input, it became clear that a small number of cases were not forecast adequately. It seems prudent to investigate the sources of these poorly forecast cases since these will help to describe the degree to which the synoptic scale influences prolific severe weather outbreaks.

Much work has been done in the past to study the influence of the initialized dataset and model physics and numerics on the resultant forecasts. With this in mind, several tests will be conducted to determine possible sources of large forecast errors. Some of these tests include the use of various model physics, changed both individually and in conjunction; the use of higher resolution data as input to determine if scales more resolved than the synoptic scale influence particular outbreaks; the shift of inner domains in the model runs to determine the sensitivity of forecast convection and atmospheric conditions to computational parameters; and the use of a different model to determine the influence of model numerics on forecast quality.

Preliminary results indicate that forecast quality appears to be similarly poor no matter what physical schemes are used (e.g., choices in convective, microphysics, or boundary layer schemes) and that a higher resolution dataset does not greatly improve the forecast in any of the relevant cases. This suggests model choice and a lack of physical understanding of the processes involved in these outbreaks are the primary sources of error.