On the Predictability of Multiday Episodes of Severe Convective Storms During the Mesoscale Predictability Experiment

Logan C Dawson, Purdue University

Abstract

Severe convective storms and associated hazards pose significant threats to life and property. Therefore, accurately predicting these storms, especially in multiday outbreaks of significant severe weather, serves an important benefit to society. With the development and proliferation of convection-allowing models, high-resolution operational numerical weather prediction models are capable of explicitly representing convective storms, which has proven beneficial in severe weather forecasting practices. The skill of convective-scale forecasts is heavily dependent on accurate representation of synoptic scale and mesoscale features present in the atmosphere when the forecasts are initialized. This is particularly challenging when convective-scale forecasts are initialized while convection is ongoing or when upscale feedbacks of prior convection have modified the mesoscale and synoptic scale environment. This research utilized the Weather Research and Forecasting model and the Data Assimilation Research Testbed toolkit to generate convective-scale forecasts of four severe weather outbreaks that occurred in the central Great Plains region of the United States in May 2013 during the Mesoscale Predictability Experiment. An objective of this field campaign was to quantify upscale feedbacks of deep convection and determine how these feedbacks affected the predictability of subsequent convective storms. In this work, a focus was placed on quantifying the predictability of supercell thunderstorms, which commonly produce severe convective hazards. It was shown that radar-derived rotation track observations were advantageous in verifying forecasts of updraft helicity and low-level vertical vorticity. In addition, further experimentation was conducted with forecasts of the 19 May and 20 May outbreaks to assess how antecedent convection on 19 May and resultant surface cold pools affected the predictability of subsequent severe convective storms on 20 May. This experimentation employed mesoscale data assimilation and modified microphysical heating tendencies to experiment with the model representation of the 19 May convection and resultant feedbacks. These experiments found that appropriate representation of processes responsible for cold pool formation and assimilation of conventional meteorological observations impacted the forecast skill in a positive manner. This work supports additional utilization of radar-derived rotation track data for verifying convective-scale forecasts and further incorporation of ensemble data assimilation in convective-scale forecasting practices.

Degree

Ph.D.

Advisors

Trapp, Purdue University.

Subject Area

Atmospheric sciences

COinS