Simulating Supercell Thunderstorms in a Horizontally-Heterogeneous

Convective Boundary Layer

Christopher Nowotarski, Paul Markowski, Yvette Richardson

Pennsylvania State University

George BryanNational Center for Atmospheric Research

25th Severe Local Storms ConferenceDenver, CO

September 14, 2010

Motivation

• Previous 3D numerical simulations of supercells and tornadoes generally use horizontally homogeneous environments and exclude surface fluxes.

• Observations and simulations suggest that supercells are favored in areas with significant vertical wind shear and instability (CAPE).

• Convective Boundary Layers (CBLs) are characterized by local variations in these quantities.

• Consequently, supercells simulated in such environments may behave differently than in a homogeneous environment.

Background- Mesoscale variations in low-level vertical

wind shear and moisture “profoundly influence the morphology of deep convective storms” (Richardson et al. 2007, Richardson 1999).

- Storms became more organized and stronger when moving to areas of increased shear.

- Storms propagated towards areas of increased low-level moisture in weak shear regimes

- Isolated supercells in areas of increased moisture showed both higher updraft speeds and stronger low-level rotation.

- Tornadoes tend to be more likely in environments with higher 0-1 km shear and lower lifting condensation levels (Markowski and Richardson 2009)

Background - HCRs- Boundary layer convection is a source of heterogeneity

in the atmosphere.

- Because of low-level shear requirements, supercell environments are likely characterized by rolls or disorganized convection.

- Variations in thermodynamic quantities result from increased convergence in updraft branches. (Weckwerth 1996)

-Potential temperature can be 0.5 K higher in updraft

-Mixing ratio is 1.5 – 2.5 g kg-1 higher in updraft

- Increased instability, lower LCLs, and favored regions for cloud formation in updraft branches.

- Roll axes tends to be aligned with mean CBL wind (Weckwerth 1999)

- Organized boundary layer convection results in local, periodic variations in low-level vertical wind shear.

- Maxima tend to be in areas without strong magnitudes of vertical velocity.

(from: Weckwerth 1996)

(from: Markowski and Richardson 2007)

Background

24 May 2008(from: NOAA comprehensive Large Array-Data Stewardship System)

Experiment Design

• Two high resolution simulations of supercells

• One simulation allows a convective boundary layer to develop before initializing deep convection.

• The other is a “typical” horizontally-homogeneous simulation.

• Compare simulations, focusing on behavior and structure of mature storms (rather than initiation).

Model Configuration• Model

– CM1, version 1, release 14 (with modifications)– dt = 0.75 s, 0.125 s for acoustic calculations– periodic lateral boundary conditions

• Grid – dimensions: 200 km x 150 km x 18 km– dx, dy = 200 m– dz = stretched from 50 m (below 3 km) to 500 m (above 9.5 km)

• Parameterizations– Ice microphysics (Lin et al. 1983)– land surface scheme using two-layer soil model (Noilhan and Planton 1989) CBL run only– Radiation (REFERENCE!!!!) CBL run only

What about mixing out shear?• No large-scale horizontal temperature

gradient or Coriolis force, so vertical wind shear is mixed out in CBL.

• To maintain requisite shear for HCRs and supercells, we need to artificially relax the horizontal winds at low-levels towards the initial state.

• Add constant velocity tendency to each gridpoint on a vertical level that nudges the average wind at that height towards the initial average.

• Some reduction in shear is still allowed!

• Modification of technique applied by Robe and Emmanuel (2001)

+

Base States

• CBL simulation– 35 m s-1 0-6 km shear– SBCAPE: 2819 J kg-1

– Small capping inversion to prevent widespread convection

• Homogeneous simulation– Average of CBL simulation after one hour

of simulation time– Low level moisture and temperature

have increased (CAPE too)– Some reduction in low-level shear from

mixing

Results• Simulated CBL (before storm initiation)

Results

Results

Results

Results

Summary

Future Work