Investigating the Effects of Surface Roughness on Tornadoes using High-Resolution Numerical Modeling

Recent studies have begun investigating the role that friction/surface roughness plays on tornado and supercell processes (e.g., Kellner and Niyogi 2014; Schankman et al. 2015; Roberts et al. 2016; 2018; Markert et al. 2019; Hua and Chavas 2019; Houser et al. 2020; Jiang and Dawson 2023; Wang et al. 2023). However, most of these studies treat surface roughness in a peripheral sense using turbulence parameterizations. This study employs a coupled fine-resolution CFD model with spatial resolution of 0.01 m at the center of the grid domain and storm-scale Cloud Model 1 (CM1). The CFD model explicitly resolves turbulence rather than parameterizing it. The model is initialized by downscaling 10-m resolution CM1 output from a violent tornado representative of the 24 May 2011 El Reno, OK tornado environment, which also serves as the boundary conditions.

High resolution tornado simulations are generated to investigate how variations in surface roughness change tornado characteristics including minimum central pressure, radial, tangential and vertical wind components and derived kinematic quantities, tornado structure, and path. To ensure a realistic treatment of surface roughness, representative roughness lengths from five types of land covers from the National Land Cover Dataset (NLCD) are considered in the coupled model, including smooth (roughness length of 0.01 m), grass, rough, medium density urban and mixed forest. The land cover is modeled on the lower boundary, which is treated like a wall function, by using surface roughness length. Simulations are performed with homogeneous surface roughness over the lower boundary domain, and simple partitions of the lower boundary domain into 2 halves or 4 quadrants, each having a different surface roughness length, to mimic a transition from one land cover type to another. These model runs are compared to an idealized, free-slip, smooth ground control run with zero roughness length, to provide a baseline tornado windcfield associated with a frictionless lower boundary.

Each type of land cover induces minor fluctuations on the tornado path compared to the control run, but the effects of land cover on the tornado’s translation are marginal to negligible. However, for tornado intensity, the effects of surface roughness appear to have a goldilocks-like outcome: Initial increases in the surface roughness length from free slip conditions result in a lower central pressure, stronger vertical velocities, and modestly stronger tangential winds. However, further increases past an intermediate roughness length weaken the tornado as indicated by smaller central pressure deficits and reduction in velocity components (at 10-m AGL). The relationship between land cover and tornado core radius size is complicated and not simply monotonic. However, the variation of core radius and the maximum horizontal velocity over the vortex’s vertical dimension suggests that the range of influence of surface roughness is mainly concentrated below 50 m.