Degree Type


Date of Award


Degree Name

Master of Science


Geological and Atmospheric Sciences

First Advisor

William A. Gallus, Jr.


Convective morphology was analyzed with 3-km WRF-ARW simulations for 37 events during the warm season from 2006 to 2010. Ten classifications were used to identify convective modes displayed in each event. An objective scoring method, based on normalized time and the type of mode exhibited, was developed to measure the accuracy of the modeled morphologies when compared to radar observations. Trends in the simulated evolution were discussed, as well as common discrepancies between the model and observed events. Environmental conditions before convective initiation were obtained from RUC analyses, and statistically significant associations between the parameters and the model accuracy scores were found. Finally, four cases highlighting the main morphological issues in the simulations were investigated further.

Overall, the simulations entailed more cellular modes and fewer linear modes than the observed systems. Bow echoes and linear systems with trailing stratiform rain regions were especially difficult for the model to forecast. Of the 21 cases with an observed bow echo, only 8 featured a simulated bow echo. Twelve cases included a missed squall line with trailing stratiform rain. Cellular modes were simpler to forecast, as 75% of the forecast comparisons with an observed cellular mode also featured a modeled cellular mode. The simulations usually portrayed convective evolution more accurately when the initial synoptic environment included strong deep-layer shear and cool potential temperatures at the level of maximum theta-E. Major timing errors with convective initiation or dissipation in the simulations usually occurred when initial 0-6 km shear was very low, surface potential temperatures were cool, and when potential temperatures quickly warmed with height.

The case studies showed that differences in wind shear and cold pool strength and development between the simulations and observations were responsible for many convective mode discrepancies. Strong boundary-normal shear, a warm cold pool, and very dry air within the rear inflow jet inhibited the formation of a trailing stratiform rain region in the simulation for the first case study. Weak deep-layer shear and the lack of a fully-developed cold pool in the model did not allow the simulated line to bow out in the second case study. Weaker forcing near the surface in the simulation compared to the observations allowed for clustered cells instead of a broken line in the third case study. Mid-layer shear not oriented along the boundary did not provide for proper cold pool merging for linear development in the WRF output for the fourth case study. Sensitivity tests were also conducted for microphysical schemes and initial and lateral boundary conditions for the first and second case studies. None of the microphysics schemes produced the observed convective mode, but GFS initial conditions in the first case study produced a trailing stratiform region.

Copyright Owner

Darren Victor Snively



File Format


File Size

75 pages

Included in

Meteorology Commons