Detailed Differences Between HRRRv3 and HRRRv4 Forecasts for the 20 May 2019 Severe Weather Event

A severe weather outbreak occurred across the southern Plains on 20 May 2019, which had been forecast as a “High Risk” by the Storm Prediction Center. The operational HRRR (hereafter HRRRv3) provided a particularly ominous forecast of the event the day before by predicting open warm sector supercells in addition to those forming along larger-scale airmass boundaries. Despite particularly skillful objective forecasts of composite reflectivity in HRRRv3 compared to other deterministic CAM forecasts comprising the High Resolution Ensemble Forecast (HREFv2.1) system, which will be discussed initially, indeed such storms did not materialize. The real-time parallel experimental version of the HRRR (hereafter HRRRv4), which is projected to replace HRRRv3 in mid-2020, had a slightly different forecast evolution, backing off on the coverage of the warm sector storms, and performing objectively somewhat better.

A detailed analysis of the near-storm environment in both HRRRv3 and HRRRv4 forecasts (re-forecasting where appropriate) will be presented to investigate two particularly pertinent questions: 1) why did open warm sector storms develop in the model? And 2) why did these storms exhibit unexpectedly low-UH tracks compared to the environment in which they were birthed? The discussion will reference trajectory analyses and a UH track climatology of the HRRRv3 to help answer these questions.

Note: This talk is very similar in scope to the submitted abstract "A Review of NCEP’s Convection-Allowing Model Guidance for the 20 May 2019 Southern Plains High Risk Day" by Dawson et al. (EMC) and we request to have the two talks adjacent to each other.