The 20 May 2013 Oklahoma tornado outbreak under pseudo-global warming

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 3 November 2014: 4:45 PM
Madison Ballroom (Madison Concourse Hotel)
Robert J. Trapp, Purdue University, West Lafayette, IN; and K. Hoogewind, M. E. Baldwin, and S. Lasher-Trapp

This research seeks to answer the question of how a current-day tornado outbreak might be realized under future anthropogenic climate change. We adapt the “pseudo-global warming” (PGW) methodology employed by Lackmann (2013) and others, who have investigated flooding events under climate change. Here, we exploit coupled atmosphere-ocean GCM data contributed to the CMIP5 archive, and take the mean 3D atmospheric state simulated during May 1990-1999 and subtract it from that simulated during May 2090-2099. Such 3D changes in temperature, humidity, geopotential height, and winds are added to reanalysis data (CDAS1), and this modified atmospheric state is then used for initial and boundary conditions for a real-data WRF model simulation at high resolution. Comparison of this simulation with a reanalysis-only (or control; CTRL) simulation facilitates assessment of PGW effects.

Consistent with our previous findings on connections between climate change and convective storms, the regional pre-convective environment on 20 May 2013 modified by PGW consisted of relatively higher CAPE and lower vertical wind shear. Nonetheless, the WRF simulation with the PGW conditions still yielded supercellular convection in central Oklahoma, as was observed on 20 May 2013, and was also simulated with the CTRL conditions. Thus, PGW did not lead to a change in convective mode, as has been speculated as a possible effect. Convective cells in the PGW simulation were, however, relatively more isolated, and initiated approximately two hours later than in the CTRL simulation. This can be attributed to the higher CIN. Quantitative differences in the characteristics of the convection will be discussed, as will be any sensitivities to the components of the modeling procedure.