Exploring the Sensitivity of Convective Storm Characteristics and Dynamics to Grid Spacing in Convection-Allowing North American Mesoscale Forecast System Simulations

Open Access
Colbert, Michael Robert
Graduate Program:
Master of Science
Document Type:
Master Thesis
Date of Defense:
June 13, 2017
Committee Members:
  • Paul Markowski, Thesis Advisor
  • Yvette Richardson, Committee Member
  • David Stensrud, Committee Member
  • NAM
  • numerical weather prediction
  • severe weather
  • convection
  • tornadoes
  • resolution
  • grid spacing
  • supercells
  • CAM
  • convection-allowing models
In support of the Next Generation Global Prediction System (NGGPS) project, two severe weather cases (28 April 2014 and 6 May 2015) are simulated using the North American Mesoscale Forecast System (NAM) 1.33-km Fire Weather nest and 4-km CONUS nest, each with 5-min output. On 28 April 2014, convection along and ahead of a cold front traversing the southeast U.S. was responsible for 359 storm reports (121 of which were tornado reports). On 6 May 2015, convection that initiated ahead of a dryline resulted in 144 storm reports (59 of which were tornado reports) in the central Great Plains region. Comparisons are made between the simulations and observations to explore how convection evolves differently on the two model grids. The analysis focuses on the models’ representations of the environment (e.g., instability and vertical wind shear), storm attributes (e.g., cold pools, storm motion, and mesocyclones), and dynamical processes within the storms (e.g., the baroclinic generation of vorticity.) The results suggest that the increased resolution of the Fire Weather nest promotes the simulation of convective storms with more realistic sizes and structures. The Fire Weather simulations also tend to perform better with the timing and location of convection initiation. However, as time progresses in the model simulations, the Fire Weather nest seems to lose some of its advantages over the CONUS nest. In the 28 April 2014 case, cold pools in the Fire Weather simulation ultimately became too cold and large and the convection moved faster than in the observations. Among other factors, the relative decrease in benefits of the Fire Weather nest late in the simulations may be related to its limited domain, its tendency to produce spurious convection patterns, and parameterizations which are not optimally calibrated for simulations at 1.33-km grid spacing.