-
Idealized simulations of supercell thunderstorms are performed using the compressible, nonhydrostatic WRF model, version 3.7.1 (Skamarock et al., 2008). The model configuration is detailed in Table 1, and is similar to the idealized supercell simulations in Johnson et al. (2016, 2019). Storms are simulated for 2 h on a 200 km × 200 km grid with a 1 km horizontal spacing. The vertical grid extends to 20 km in height with an approximate 500 m grid spacing. The Weisman and Klemp (1982) thermodynamic sounding with a veering quarter-circle wind profile (clockwise shear of 5.23 × 10−3 s−1 up to 2.3 km, unidirectional shear of 5.69 × 10−3 s−1 above to 7 km) is employed for the atmospheric environment, resulting in a convective available potential energy (CAPE) of approximately 2163 J kg−1 and storm-relative helicity in the 0–3 km layer of approximately 180 m2 s−2. The storm is initiated using an ellipsoidal thermal bubble with a maximum potential temperature perturbation of 3 K. Radiation, land surface, cumulus, and planetary boundary layer parameterizations are turned off.
WRF model configuration Run time 120 min Δt 6 s Sound wave Δt 1 s Model output interval 10 min Horizontal domain 200 km × 200 km Model lid 20 km Δx 1 km Δy 1 km Δz ~500 m Time integration scheme Third order Runge–Kutta Horizontal momentum advection Fifth order Vertical momentum advection Third order Horizontal scalar advection Fifth order Vertical scalar advection Third order Upper level damping 5000 m below model top Rayleigh damping coefficient 0.003 Turbulence 3D 1.5-order turbulent kinetic energy closure Horizontal boundary conditions Open Table 1. WRF model input.
-
As mentioned earlier, three MP schemes with varying degrees of complexity in representing hydrometeors, as available in WRF v3.7.1, are used in the supercell simulations. They are, respectively, the HUCM “full” SBM (Khain and Sednev, 1996; Khain et al., 2004), the fully two-moment NSSL (Mansell et al., 2010), and the partially two-moment Thompson (Thompson et al., 2008) BMP schemes.
The HU-SBM “full” scheme prognoses the PSDs of liquid (one category spanning all drop sizes), three ice crystals (plates, columns, dendrites), (snow) aggregates, graupel, and hail, which are discretized into 33 mass-doubling bins ranging from 3.35 × 10−11 to 1.44 × 10−1 g. There are no processes in the HU-SBM “full” scheme that convert ice crystal habit to other habits after nucleation (in which the ice crystal destination is determined by ambient temperature). An alternative “fast” (in contrast to “full”) version of HU-SBM available in WRF prognoses the PSDs of one liquid and fewer ice categories, including ice crystals/aggregates, and graupel/hail, that are discretized into 33 or 43 bins. Studies have shown that the HU-SBM “fast” version has skill simulating deep convection (e.g., Khain et al., 2016; Shpund et al., 2019). In this study, we choose to evaluate the HU-SBM “full” version because of its inclusion of more ice categories (six versus two), which we believe are important for supercell storms. The use of 33 bins with smaller maximum diameters than the available 43 bins in the “fast” version does impose some limitation; therefore, results related to the maximum bin sizes do not necessarily carry over to the “fast” version. Hereafter, HU-SBM with no qualifier refers to the “full” version.
The NSSL BMP prognoses the mass mixing ratio qx and total number concentration NTx (x refers to species) of cloud water, rain, cloud ice, snow, graupel, and hail, and additionally the particle volume of graupel and hail, which can be used to predict the bulk density. The Thompson scheme prognoses the qx of cloud water, rain, cloud ice, snow, and graupel, and the NTx of cloud ice and rain. Among many available BMPs, the NSSL scheme is one of the most sophisticated two-moment BMPs and has been shown to outperform other BMPs in supercell simulations (Johnson et al., 2016, 2019), while the Thompson scheme is employed in the U.S. operational High-Resolution Rapid Refresh forecasting system (Benjamin et al., 2016) and has generally good performance for precipitation forecasting.
It is important to note that the hydrometeors in each MP scheme may contain different assumptions of PSDs and particle properties. Liquid, plates, graupel, and hail in the HU-SBM contain constant bulk densities of 1000, 900, 400, and 900 kg m−3, respectively, across their discretized mass bins, while column, dendrite, and snow bulk densities decrease at larger mass. Rain, graupel and hail in the NSSL scheme assume gamma PSDs (e.g., Ulbrich, 1983) with shape parameters of α = 0, 0, and 1, respectively. Cloud water, cloud ice, and snow have mass-dependent (rather than the commonly utilized diameter-dependent) gamma distributions with shape parameters of α = 0, 0, and −0.8 (Zrnic et al., 1993) respectively. Cloud water, cloud ice, rain, and snow have bulk densities of 1000, 900, 1000, and 100 kg m−3 respectively. Graupel and hail bulk densities are predicted via their bulk prognosed volumes. The Thompson scheme assumes an exponential PSD (gamma PSD with a shape parameter of α = 0) for cloud ice, rain, and graupel, and a gamma PSD (α = 12) for cloud water. Snow in the scheme follows a linear combination of exponential and gamma PSDs (Field et al., 2005; Thompson et al., 2008). Cloud water, cloud ice, rain and graupel have bulk densities of 1000, 890, 1000 and 500 kg m−3, respectively, while snow density decreases with increasing diameter (similar to HU-SBM snow).
WRF model configuration | |
Run time | 120 min |
Δt | 6 s |
Sound wave Δt | 1 s |
Model output interval | 10 min |
Horizontal domain | 200 km × 200 km |
Model lid | 20 km |
Δx | 1 km |
Δy | 1 km |
Δz | ~500 m |
Time integration scheme | Third order Runge–Kutta |
Horizontal momentum advection | Fifth order |
Vertical momentum advection | Third order |
Horizontal scalar advection | Fifth order |
Vertical scalar advection | Third order |
Upper level damping | 5000 m below model top |
Rayleigh damping coefficient | 0.003 |
Turbulence | 3D 1.5-order turbulent kinetic energy closure |
Horizontal boundary conditions | Open |