1. Introduction
Aerosols, both natural and anthropogenic, may impact clouds and precipitation by acting as cloud condensation nuclei (CCN). Given the same liquid water content, increasing CCN concentrations tends to increase resulting cloud droplet concentrations but decrease droplet size. This creates more narrow cloud droplet spectra and reduced collision efficiencies, thus inhibiting the warm-rain process (Hobbs et al. 1970; Eagan et al. 1974; Braham et al. 1981; Kaufman and Nakajima 1993; Rosenfeld 1999, 2000). This effect has been numerically simulated in deep convection (Wang 2005; van den Heever et al. 2006; Fan et al. 2007; Li et al. 2008; Storer et al. 2010) and supercells (Lerach et al. 2008; Lim and Hong 2010). In addition, higher CCN concentrations can lead to more supercooled water aloft in deep convection, which creates stronger updrafts aloft via enhanced latent heating (Wang 2005; Seifert and Beheng 2006; van den Heever et al. 2006; Carrió et al. 2010). Since hailstone growth principally occurs in the updraft where there is a close match between updraft velocities and the fall velocities of the stones (Foote 1984), enhanced CCN concentrations can aid in the production of larger hail within convection. van den Heever and Cotton (2004) and Gilmore et al. (2004) performed idealized numerical simulations of supercell thunderstorms using single-moment microphysics and 1-km horizontal grid spacing, finding that increasing the mean diameter of rain and hail distributions (all else being equal) reduced the net surface area of the hydrometeors, thereby reducing evaporative cooling and melting rates. This produced weaker low-level downdrafts and weaker, shallower cold pools.
While the precise mechanisms of supercell tornadogenesis are still up for debate, studies have suggested that tornadoes are often linked to the rear flank downdraft (RFD), which can transport vertical vorticity to the surface, baroclinically generate horizontal vorticity, and enhance convergence along gust fronts beneath the updraft (Burgess et al. 1977; Davies-Jones 1982a,b; Davies-Jones and Brooks 1993; Walko 1993; Brooks et al. 1994; Trapp and Fiedler 1995; Markowski 2002). Specifically, many have argued that tornadogenesis is more likely to occur when the temperature deficit within the RFD is small relative to the environment (Ludlam 1963; Lemon 1974; Eskridge and Das 1976; Nelson 1977; Brandes 1978; Leslie and Smith 1978; Davies-Jones 2000). Markowski et al. (2002) showed with Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX) observations that air parcels within RFDs tend to be less negatively buoyant and thus warmer in tornadic versus nontornadic supercells. This concept was further supported by idealized numerical simulations at 40-m horizontal and vertical grid spacing (Markowski et al. 2003), where tornadic vortices increased in intensity and longevity as downdraft parcel buoyancy increased, because colder parcels were more resistant to lifting. Snook and Xue (2008) extended the work of van den Heever and Cotton (2004) and Gilmore et al. (2004) to tornadogenesis using 100-m horizontal grid spacing, verifying that rain and hail size distributions favoring larger hydrometeors yielded warmer cold pools via reduced evaporative cooling. In addition, the larger hydrometeors, possessing greater terminal fall speeds, were not advected as far from the updraft before falling to the ground, reducing the areal coverage of precipitation. This positioned the gust front closer to the storm center, permitting vertically oriented updrafts and low- and midlevel vertical vorticity alignment. This, in turn, increased the dynamic suction effect by the mesocyclone (Rotunno and Klemp 1982) and associated low-level vertical stretching, thereby increasing the potential for tornadogenesis.
Lerach et al. (2008) performed idealized three-dimensional supercell simulations at 111-m horizontal grid spacing and found that increasing CCN and giant CCN (GCCN) concentrations reduced warm- and cold-rain (“cold rain” refers to rain formed from the melting of graupel and hail) processes within the RFD and forward flank downdraft (FFD). This resulted in lower precipitation rates, less evaporative cooling, and a weaker cold pool that provided a more favorable setup for tornadogenesis, where the low-level mesocyclone and near-surface vorticity provided by the RFD-based gust front remained vertically stacked. In contrast, reducing the aerosol concentrations resulted in stronger cold pools and eventual storm undercutting, which hindered tornadogenesis.
While the results of Lerach et al. (2008) provide insight into the possible role of aerosols in influencing supercell storms and tornadogenesis, the relative impact must be put into context with other environmental parameters. VORTEX observations and modeling studies suggest that tornado likelihood, intensity, and longevity increase as RFD-based CAPE increases, and higher relative humidity at low levels is more conducive to RFDs associated with relatively high buoyancy and higher tornadogenesis potential (Markowski et al. 2002, 2003). The goal of this study is to extend the work of Lerach et al. (2008) to comparing the role of aerosol indirect microphysics with those of low-level moisture and CAPE on supercell tornadogenesis. We perform an ensemble of numerical simulations of an idealized supercell thunderstorm, differing only in initial background CCN concentrations and environmental low-level moisture. The simulations are compared to assess which scenarios best promote vortex development under idealized environmental conditions. The results are put into context with previous numerical modeling studies, particularly Markowski et al. (2003).
2. Model setup
This study utilized the Regional Atmospheric Modeling System (RAMS; Pielke et al. 1992) version 4.3.0 (Cotton et al. 2003) in a Cartesian coordinate domain. RAMS makes use of the nonhydrostatic/compressible forms of the basic model equations (Tripoli and Cotton 1986) on a staggered Arakawa-C grid (Arakawa and Lamb 1981) with terrain-following sigma coordinates in the vertical (Tripoli and Cotton 1980). Time differencing is performed via a hybrid combination, with the calculation of the Exner function 
A bin-emulating, two-moment bulk microphysics scheme (Meyers et al. 1997; Feingold et al. 1998) was utilized in these simulations, in which the cloud droplet size spectrum was decomposed into two modes (cloud 1 and cloud 2) to represent the frequently bimodal distribution of cloud droplet spectra (Saleeby and Cotton 2004). The cloud-1 mode accounted for cloud droplets spanning 1–40 μm in diameter. The cloud-2 mode was for droplets 40–80 μm in diameter. The scheme explicitly predicted mixing ratios and number concentrations of pristine ice, snow, aggregates, graupel, hail, cloud-1 and cloud-2 droplets, and rain. Nucleation by CCN, GCCN, and ice nuclei (IN) were explicitly considered. CCN directly nucleated to form cloud-1 droplets, while GCCN directly nucleated to form cloud-2 droplets. We excluded the effects of terrain, surface fluxes, surface drag, radiation, and friction because of the time scales involved and the desire to simplify the experiment. Convection was explicitly resolved on all grids.
The initial soundings and vertical wind profile utilized were adapted from a previously employed setup found to generate storm-splitting and supercells (Grasso 2000; van den Heever and Cotton 2004; Gaudet and Cotton 2006; Lerach et al. 2008). This study focused on the right-moving storms. Convection was initiated by introducing a “warm, moist bubble” (10 × 10 × 1.5 km3, 3-K thermal perturbation, 20% moisture perturbation) at the surface. The model aerosol species were set initially horizontally homogeneously with prescribed vertical profiles of CCN, GCCN, and IN. Four simulations were performed. In two of them, the initial background CCN concentrations were set to represent a relatively “clean continental” environment. In the other two simulations, CCN concentrations were increased to represent an aerosol-rich environment due to the presence of dust (Koehler et al. 2009) or other pollutants. Chemically, the CCN were assumed to be ammonium sulfate. The background water vapor mixing ratios below 800 mb differed by 20% for each pair of CCN simulations, in order to maintain consistency with the simulations of Markowski et al. (2003). Figure 1 displays the initial profiles of temperature and dewpoint temperature (Figs. 1a,b), horizontal wind (Fig. 1c), CCN (Fig. 1d), GCCN (Fig. 1e), and IN (Fig. 1f). Because of the tendency of convection to propagate off of grid 1 (via computational restrictions on grid size), a constant mean storm motion vector of u = 14.1 m s−1, υ = 14.1 m s−1 was subtracted from the hodograph (Fig. 1c) at the time of initialization as in previous studies (Gaudet and Cotton 2006; Gaudet et al. 2006; Snook and Xue 2008). Table 1 summarizes the differences between the four simulations conducted. The simulation with clean continental background CCN and relatively low (high) relative humidity below 800 mb will be referred to as the CLN-DRY (CLN-WET) simulation. The simulation with dusty, or polluted, background CCN and relatively low (high) relative humidity below 800 mb will be referred to as the DST-DRY (DST-WET) simulation. CCN concentrations near the surface were set to 200 and 2000 cm−3 for the CLN and DST simulations, respectively (Fig. 1d), based on Cirrus Regional Study of Tropical Anvils and Cirrus Layers–Florida-Area Cirrus Experiment (CRYSTAL-FACE) measurements (van den Heever et al. 2006). The background initial profiles of GCCN (Fig. 1e) and IN (Fig. 1f) were held fixed for all four simulations. While GCCN and IN concentrations are thought to be important to convective processes, their effects are left for future work. Simulations lasted 110 min. Grid 2 was initialized at 40 min. Grid 3 was initialized at 60 min.
Initial background profiles of (a) temperature and dewpoint temperature for the low-moisture simulations, (b) temperature and dewpoint temperature for the high-moisture simulations, (c) horizontal wind represented by a hodograph, (d) CCN, (e) GCCN, and (f) IN. Note that in (c), the letter S represents the surface wind vector and vectors are labeled every 2 km. In (d), the solid and dashed curves represent the “clean continental” and dusty/polluted CCN profiles, respectively.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Initial background profiles of (a) temperature and dewpoint temperature for the low-moisture simulations, (b) temperature and dewpoint temperature for the high-moisture simulations, (c) horizontal wind represented by a hodograph, (d) CCN, (e) GCCN, and (f) IN. Note that in (c), the letter S represents the surface wind vector and vectors are labeled every 2 km. In (d), the solid and dashed curves represent the “clean continental” and dusty/polluted CCN profiles, respectively.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Initial background profiles of (a) temperature and dewpoint temperature for the low-moisture simulations, (b) temperature and dewpoint temperature for the high-moisture simulations, (c) horizontal wind represented by a hodograph, (d) CCN, (e) GCCN, and (f) IN. Note that in (c), the letter S represents the surface wind vector and vectors are labeled every 2 km. In (d), the solid and dashed curves represent the “clean continental” and dusty/polluted CCN profiles, respectively.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Experiment names and parameters. CCN concentrations and mixing ratios represent values at the surface.
3. Results
a. Storm evolution
Each simulation produces storm splitting, and both a right-moving, cyclonically rotating supercell and a left-moving, anticyclonically rotating supercell are evident at 55 min (not shown). The left mover propagates out of the grid domain and is not considered further. Figure 2 displays total condensate at 1 km above ground level (hereafter, all heights AGL) on grid 2 for all four simulations in 15-min increments from 65 to 95 min of simulation time. Updrafts stronger than 10 m s−1 and downdrafts stronger than −5 m s−1 at 3.5 km are overlaid (in this study positive vertical velocities denote upward motion and negative vertical velocities denote downward motion). For simplicity the positive y, negative y, positive x, and negative x directions will be referred to as north, south, east, and west, respectively. By 65 min, the simulated right movers possess noticeable hooks in the total condensate fields, associated with precipitation from the RFD wrapping cyclonically around the main updraft. The storms initialized with higher moisture values below 800 mb (hereafter HM) are greater in horizontal extent compared to the storms initialized with lower moisture values (hereafter LM), shown in Fig. 2 as larger regions of total condensate greater than 0.001 g kg−1. In particular, regions with total condensate greater than or equal to 0.5 g kg−1 that are associated with the RFD and FFD precipitation are noticeably larger in areal coverage in the HM cases throughout the simulated time span, and the cyclonically curved hook associated with the RFD is more pronounced in the HM cases. This is all to be expected, as the HM simulations were initialized with 20% more available moisture below 800 mb and therefore initialized with 59% higher CAPE (3517 vs 2207 J kg−1), suggesting more intense convection and precipitation in the HM simulations. This coincides with the fact that the HM supercells possess spatially larger updraft and downdraft cores at 3.5 km compared to the LM cases. The clean storms contain larger regions of downdrafts stronger than −5 m s−1 compared to the dusty storms, most noticeable after 65 min. Profiles of mean and peak vertical motion within both updraft and downdraft regions on grid 2 at 55 min are shown in Fig. 3. The HM simulations generally produced stronger peak updrafts than the LM storms (80 vs 70 m s−1) due to higher CAPE, with the DST-WET simulation producing peak updraft velocities roughly 5 m s−1 greater than those of the CLN-WET storm at 12.5 km (Fig. 3a). The CLN-DRY and DST-DRY simulations produced comparable updraft profiles (Fig. 3a). However, the CLN-DRY simulation created slightly larger peak updrafts at 10 km, while the DST-DRY simulation created greater peak updrafts at 17 km. The DST-WET storm generally produced stronger mean updrafts than the CLN-WET simulation between 3 and 10 km (Fig. 3c). On average, the HM simulations produced stronger peak downdrafts below 5 km compared to the LM simulations because of greater precipitation loading and associated evaporative cooling. The CLN simulations produced stronger downdrafts than the DST simulations (Fig. 3d) for similar reasons. Therefore, the CLN-WET simulation produced the strongest low-level downdrafts, while the DST-DRY simulation produced the weakest. The strongest peak downdrafts occurred below 0.5 km in the HM simulations and near 1.3 km in the LM simulations (Fig. 3b). These results coincide with the general differences seen in cold-pool evolution between simulations.
Total condensate at 1 km on grid 2 at (top to bottom) 65, 80, and 95 min for the (left to right) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations. Vertical velocities of −5, 10, and 20 m s−1 at 3.5 km are overlaid with thick (thin) contours for updrafts (downdrafts).
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Total condensate at 1 km on grid 2 at (top to bottom) 65, 80, and 95 min for the (left to right) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations. Vertical velocities of −5, 10, and 20 m s−1 at 3.5 km are overlaid with thick (thin) contours for updrafts (downdrafts).
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Total condensate at 1 km on grid 2 at (top to bottom) 65, 80, and 95 min for the (left to right) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations. Vertical velocities of −5, 10, and 20 m s−1 at 3.5 km are overlaid with thick (thin) contours for updrafts (downdrafts).
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Profiles of (a) peak updrafts, (b) peak downdrafts, (c) mean updrafts for all updrafts > 1 m s−1, and (d) mean downdrafts for all downdrafts < −0.5 m s−1 at 55 min on grid 2. Updraft profiles are shown up to 20 km; low-level downdrafts are shown up to 5 km.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Profiles of (a) peak updrafts, (b) peak downdrafts, (c) mean updrafts for all updrafts > 1 m s−1, and (d) mean downdrafts for all downdrafts < −0.5 m s−1 at 55 min on grid 2. Updraft profiles are shown up to 20 km; low-level downdrafts are shown up to 5 km.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Profiles of (a) peak updrafts, (b) peak downdrafts, (c) mean updrafts for all updrafts > 1 m s−1, and (d) mean downdrafts for all downdrafts < −0.5 m s−1 at 55 min on grid 2. Updraft profiles are shown up to 20 km; low-level downdrafts are shown up to 5 km.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Figure 4 displays near-surface (24 m) temperature on grid 2 for all simulations at 95 min. Both low-level moisture and CCN concentrations played contributing roles to the size and strength of the resulting cold pools. The CLN-WET supercell produced the largest and strongest cold pool, with minimum temperatures reaching 18°C and a significant region reaching 21°C. The DST-DRY supercell produced the smallest and weakest cold pool, with minimum temperature values only reaching 21°C over an approximate 2 × 2 km2 region. The CLN-DRY and DST-WET supercells produced comparable cold pools, with minimum temperatures around 20°C. However, the CLN-DRY region of 20°C air covered an area of roughly 6 × 6 km2, while that of the DST-WET simulation spanned an area of approximately 4 × 4 km2. Additionally, the minimum cold-pool temperatures were located back in the FFD region in the CLN-DRY simulation, while minimum cold-pool temperatures in the DST-WET simulation were located closer to the RFD and leading storm outflow. As a result, the DST-WET supercell had the second strongest horizontal temperature gradient across the RFD-based gust front. The strongest gradient was produced in the CLN-WET simulation.
Near-surface (24-m) temperature on grid 2 at 95 min for the (a) CLN-DRY, (b) CLN-WET, (c) DST-DRY, and (d) DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Near-surface (24-m) temperature on grid 2 at 95 min for the (a) CLN-DRY, (b) CLN-WET, (c) DST-DRY, and (d) DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Near-surface (24-m) temperature on grid 2 at 95 min for the (a) CLN-DRY, (b) CLN-WET, (c) DST-DRY, and (d) DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
The evolution of total condensate at 1 km was overall similar between CLN and DST simulations of the same initial low-level moisture (Fig. 2), suggesting that changing background CCN concentrations had little effect on the convection. However, assessment of the cold-pool strength between simulations suggested otherwise. The time evolution of precipitation rates on grid 2 is shown in Fig. 5. The precipitation rates in the HM cases reached maximum values near an hour into the simulations, with peak values greater than 150 mm h−1. At 65 min, the precipitation rates in the LM cases peaked at 50 mm h−1. Maximum values were slightly greater in the clean simulations, and peak precipitation rates were located within or near the RFD in all simulations at this time. After 65 min, however, peak precipitation rates in the clean simulations were located rearward within the FFD, while peak values in the dusty simulations generally coincided with the RFD. Therefore, the DST-WET cold pool produced a stronger temperature gradient across the RFD-based gust front than in the CLN-DRY case, even though minimum temperature values in the CLN-DRY cold pool spanned a greater area (Fig. 4).
Precipitation rates on grid 2 at (top) 65 and (bottom) 95 min for the (left to right) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Precipitation rates on grid 2 at (top) 65 and (bottom) 95 min for the (left to right) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Precipitation rates on grid 2 at (top) 65 and (bottom) 95 min for the (left to right) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
b. Microphysical effects on grid 2
The enhanced CCN concentrations in the dusty simulations resulted in cloud droplet concentrations that were on average 10 times greater than those produced in the clean supercells (1000 vs 100 cm−3 within updraft regions; not shown), and the droplets were of smaller sizes in the dusty simulations. This resulted in reduced collision efficiencies and more supercooled water aloft in the updraft regions. Figure 6 displays time series for various grid-cumulative microphysical parameters on grid 2. With more supercooled water aloft available for ice formation, the dusty simulations produced noticeably higher snow and pristine ice particle concentrations (and of smaller sizes) than the clean simulations with the same low-level moisture. Therefore, more ice mass was lifted to the upper levels of the storm in the dusty simulations (Fig. 6a), meaning that more ice reached the storm’s anvil rather than contributing to precipitation processes. This result is consistent with previous modeling studies of convective storms (Tao et al. 2007; Fan et al. 2007; Carrió et al. 2010). Available low-level moisture played a contributing role as well. The HM simulations produced more ice aloft and larger anvils than the LM simulations (Fig. 6a). As such, the DST-WET simulation yielded the most ice mass while the CLN-DRY simulation yielded the least.
Time series of grid-integrated cumulative (a) snow + pristine ice + aggregates mass above 10 km, (b) graupel + hail mass, and (c) rain mass below 5 km; (d) time series of maximum precipitation rates; and (e) grid-integrated accumulated precipitation on grid 2. The thick solid, thick dashed, thin solid, and thin dashed timelines represent the CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations, respectively.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Time series of grid-integrated cumulative (a) snow + pristine ice + aggregates mass above 10 km, (b) graupel + hail mass, and (c) rain mass below 5 km; (d) time series of maximum precipitation rates; and (e) grid-integrated accumulated precipitation on grid 2. The thick solid, thick dashed, thin solid, and thin dashed timelines represent the CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations, respectively.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Time series of grid-integrated cumulative (a) snow + pristine ice + aggregates mass above 10 km, (b) graupel + hail mass, and (c) rain mass below 5 km; (d) time series of maximum precipitation rates; and (e) grid-integrated accumulated precipitation on grid 2. The thick solid, thick dashed, thin solid, and thin dashed timelines represent the CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations, respectively.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Figure 7 shows mean profiles of rain and hail microphysical parameters at 85 min on grid 2. The microphysical profiles at 85 min were generally representative of those throughout the simulation period. With reduced collision efficiencies, the dusty simulations produced fewer raindrops than the clean simulations. The CLN-DRY and CLN-WET supercells produced maximum raindrop concentrations of approximately 6000 and 4000 m−3, respectively, while the dusty supercells produced concentrations near 1000 m−3 in updrafts (Fig. 7a). However, because of the abundance of cloud drops available and higher in-cloud trajectories by raindrops and thus longer net liquid water paths (Bowen 1950), the raindrops that did form were able to grow near double the size of those in the clean simulations (Fig. 7b). This result is consistent with previous modeling work (Storer et al. 2010; Lim and Hong 2010). Similar results were seen in the production of hail. The CLN-WET and CLN-DRY simulations led to net maximum hail concentrations of around 1800 and 1500 m−3, respectively, within updraft regions (Fig. 7e). The dusty simulations produced concentrations near 600 m−3, but the maximum sizes of the hailstones in these simulations were larger than in the clean cases, due to slightly stronger updrafts and more available cloud water for riming (Fig. 7f). This translated to greater hail number concentrations but of smaller sizes in downdraft regions in the clean cases (Figs. 7g,h). Differences between simulations in raindrop concentrations and sizes within downdraft regions were also similar to those seen in the updraft regions. The clean simulations produced the largest raindrop concentrations in downdrafts (1700 and 1000 m−3 for CLN-WET and CLN-DRY, respectively) compared to the dusty simulations (1000 and 600 m−3 for DST-WET and DST-DRY, respectively). The HM simulations also produced more numerous raindrops than their respective LM simulations. Again, however, the dusty simulations produced slightly larger raindrops compared to those in the clean simulations. As a result, the cleaner, higher-moisture simulations produced the most rain, graupel, and hail mass throughout the domain of grid 2 (Figs. 6b,c) and therefore the most accumulated precipitation, even though maximum precipitation rates were somewhat chaotic throughout the simulations (Figs. 6d,e). As expected, with more moisture available for precipitation processes, the HM cases tended to yield the highest precipitation rates throughout the simulation period. Figure 8 displays accumulated precipitation on grid 2 at 110 min for all simulations. It is apparent that the LM simulations produced the greatest localized amounts of precipitation in the rear flank, while the HM simulations yielded the most grid-2 cumulative precipitation. The clean simulations showed a relative maximum of accumulated precipitation further back within the FFD, while the dusty simulations produced more accumulated precipitation within the RFD, near the main updraft. As the clean supercells possessed significantly more raindrops and hailstones but of smaller sizes compared to the dusty supercells, more total surface area of precipitation particles was exposed to the air while falling through the downdrafts, leading to more net evaporative cooling and thus colder, stronger downdrafts in the clean simulations (Fig. 3d). This produced larger, colder cold pools at the surface in the clean cases compared to the dusty simulations (Fig. 5). Furthermore, the HM supercells produced significantly stronger cold pools compared to the LM supercells because the HM simulations contained more available moisture for producing precipitation and therefore produced the most rainfall.
Profiles of raindrop (a) mean concentration and (b) median diameter within updraft regions (w > 1 m s−1), and (c) concentration and (d) median diameter within downdrafts (w < −0.5 m s−1). Profiles of hail (e) mean concentrations and (f) median diameter within updraft regions (w > 1 m s−1), and (g) mean concentrations and (h) median diameter within downdrafts (w < −0.5 m s−1) at 85 min on grid 2.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Profiles of raindrop (a) mean concentration and (b) median diameter within updraft regions (w > 1 m s−1), and (c) concentration and (d) median diameter within downdrafts (w < −0.5 m s−1). Profiles of hail (e) mean concentrations and (f) median diameter within updraft regions (w > 1 m s−1), and (g) mean concentrations and (h) median diameter within downdrafts (w < −0.5 m s−1) at 85 min on grid 2.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Profiles of raindrop (a) mean concentration and (b) median diameter within updraft regions (w > 1 m s−1), and (c) concentration and (d) median diameter within downdrafts (w < −0.5 m s−1). Profiles of hail (e) mean concentrations and (f) median diameter within updraft regions (w > 1 m s−1), and (g) mean concentrations and (h) median diameter within downdrafts (w < −0.5 m s−1) at 85 min on grid 2.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Accumulated precipitation on grid 2 at 110 min for the (a) CLN-DRY, (b) CLN-WET, (c) DST-DRY, and (d) DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Accumulated precipitation on grid 2 at 110 min for the (a) CLN-DRY, (b) CLN-WET, (c) DST-DRY, and (d) DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Accumulated precipitation on grid 2 at 110 min for the (a) CLN-DRY, (b) CLN-WET, (c) DST-DRY, and (d) DST-WET simulations.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
c. Tornadogenesis on grid 3
Tornado-like vortices are produced in all four simulations, but at different times and of varying strength and longevity, suggesting that both low-level moisture and aerosol-indirect microphysical influences impacted vortex spinup. The CLN-DRY supercell produces a vortex at 66 min, which dissipates at 78 min, then reforms from 81 to 82 min, lasting a total of about 13 min. The CLN-WET supercell spawns a vortex for approximately 5 min, from 76 to 80 min. The DST-DRY simulation creates a vortex at 61 min, which lasts a duration of 9 min, dissipating at 70 min. The DST-WET supercell produces a weak cyclonic circulation from 77 to 79 min, lasting a total of 3 min (note that grid-3 model output was created at 1-min intervals because of computational limitations). This means that vortex development was delayed in the HM simulations compared to the LM simulations, by 10 min in the clean continental simulations and 16 min in the dusty simulations. The CLN-DRY simulation produced a vortex 5 min later than the DST-DRY simulation, while the CLN-WET and DST-WET simulations produced vortices near the same time (76 vs 77 min, respectively). Furthermore, the duration of the vortices was greater in the LM simulations compared to the HM simulations, by 8 and 6 min in the case of the clean continental and dusty simulations, respectively. Figure 9 displays the near-surface (~24 m) temperature, vertical vorticity, perturbation pressure, and ground-relative winds (storm-relative wind vectors overlaid) on grid 3 for all four simulations during maximum vortex intensity while a coherent, convergent cyclonic circulation exists. Vortex characteristics for each simulation are presented in Table 2, including maximum vertical vorticity and horizontal winds, pressure drop, and minimum near-vortex cold-pool temperature. At 75 min, the tornado-like vortex in the CLN-DRY simulation (Figs. 9a–d) coincides with maximum relative vertical vorticity of 0.215 s−1, a pressure drop of 7.2 mb relative to the neighboring flow (within the surrounding 12 × 12 km2 of the vortex center), and a strong cyclonic circulation as evident in the storm-relative wind vectors. Maximum ground-relative winds are at EF-2 intensity south and southwest of the vortex center, with maximum surface winds exceeding 50 m s−1 just south of the vortex. The circulation actually achieves higher vertical vorticity and a greater pressure drop at 81 min (0.4 s−1 and 9 mb, respectively). However, there was no clear cyclonic circulation at the surface at this time. The cold pool is relatively weak near the vortex, with minimum temperatures near 25°C immediately surrounding the vortex. Figures 9e–h display the surface vortex produced by the CLN-WET supercell at 77 min. The vortex achieves a maximum vertical vorticity value of 0.217 s−1, associated with a pressure drop of 5.4 mb. A cyclonic circulation exists at the surface, weaker than that seen in the CLN-DRY simulation. The CLN-WET vortex produces winds of EF-1 intensity south of the vortex center, with maximum winds reaching 42.4 m s−1. The cold pool west of the vortex is noticeably colder compared to that in the CLN-DRY case, as minimum temperatures around the CLN-WET vortex fall to 22°C, roughly 3°C lower than those temperatures produced in the CLN-DRY simulation. In the case of the CLN-DRY near-vortex environment, storm-relative winds north and northeast of the vortex center clearly contain an easterly and a northerly component within the vicinity of the FFD, while the winds to the west and northwest contain both a northerly and westerly component within the RFD. In the CLN-WET simulation, storm-relative winds north and northwest of the main vortex are due northerly, while the winds to the west are due westerly. Assessing storm-relative near-surface wind magnitudes (not shown) reveals that the CLN-WET FFD- and RFD-based gust fronts are associated with stronger wind speeds in the CLN-WET simulation compared to the CLN-DRY simulation, providing a situation where the warmer, less negatively buoyant air from the inflow region to the southeast is less able to surround and enter the vortex.
Near-surface (left to right) temperature, vertical vorticity, perturbation pressure, and ground-relative winds overlaid with storm-relative wind vectors for grid 3. These variables for the (top to bottom) CLN-DRY, CLN-WET, DST-DRY, and DST-WET cases at the times of maximum near-surface vortex intensity. Note that a storm-relative wind vector represents the wind at the location of the end of its tail.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Near-surface (left to right) temperature, vertical vorticity, perturbation pressure, and ground-relative winds overlaid with storm-relative wind vectors for grid 3. These variables for the (top to bottom) CLN-DRY, CLN-WET, DST-DRY, and DST-WET cases at the times of maximum near-surface vortex intensity. Note that a storm-relative wind vector represents the wind at the location of the end of its tail.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Near-surface (left to right) temperature, vertical vorticity, perturbation pressure, and ground-relative winds overlaid with storm-relative wind vectors for grid 3. These variables for the (top to bottom) CLN-DRY, CLN-WET, DST-DRY, and DST-WET cases at the times of maximum near-surface vortex intensity. Note that a storm-relative wind vector represents the wind at the location of the end of its tail.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Tabulation of various near-surface vortex characteristics associated with all four simulations during maximum vortex intensity while a coherent, convergent cyclonic circulation exists. Characteristics include minimum cold-pool temperature (min T), maximum vertical vorticity (max ζ), local pressure drop (Δp), maximum horizontal wind (max Vh), and model time (t).
The near-surface vortex in the DST-DRY supercell (Figs. 9i–l) reaches its strongest intensity at 69 min and is most similar to the vortex in the CLN-DRY supercell compared to that in the CLN-WET experiment. The maximum vertical vorticity is 0.236 s−1. A pressure drop of about 8 mb is achieved, and the storm-relative wind vectors indicate a strong cyclonic circulation with EF-1 ground-relative wind intensity immediately south of the vortex center, as maximum ground-relative winds reach 47.7 m s−1. The cold pool near the vortex is of a similar structure to that in the CLN-DRY simulation. Storm-relative winds north of the vortex contain an easterly component, while winds directly west contain a northerly component. The DST-DRY experiment also produces a larger region where air with temperatures just below 25°C protrudes into the western edge of the vortex circulation (Figs. 9e,i; light green). The DST-WET simulation (Figs. 9m–p) produces a near-surface relative vertical vorticity maximum near 0.3 s−1 at 79 min. This vorticity maximum is associated with a pressure drop of approximately 8 mb and EF-1 surface winds south of its center where a small region of maximum ground-relative winds reach 50.2 m s−1 immediately southwest of the vortex center. A more noticeable region of surface winds approximately 1 km wide surrounds this patch of EF-1 intensity, reaching 44 m s−1. The cold pool southwest of the vortex is colder than that of the CLN-WET simulation, with minimum temperatures reaching 21°C. However, minimum temperatures immediately surrounding the western and southern edges of the surface vortex are similar to those in the CLN-WET simulation (22°C). The storm-relative winds associated with the outflow from the FFD and RFD are most similar to the CLN-WET simulation.
The maximum ground-relative near-surface winds occur where the direction of primary vortex rotation and storm propagation coincide with the outflow from the RFD. It appears that the CLN-DRY and DST-DRY supercells produce similar vortices and similar cold-pool structures surrounding the developing vortices, while the CLN-WET and DST-WET simulations also compare favorably. This suggests that low-level moisture played the largest role in driving the cold-pool evolution within each storm. However, differences in the time of initiation, longevity, and intensity among all four simulated vortices suggest that aerosol effects did contribute to tornadogenesis, albeit at a secondary level.
(left) CAPE and (right) CIN calculated on grid 3 at 5 min before peak surface vortex intensity for the (top to bottom) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations. The × depicts the location of the developing surface vortex for each simulation. The 25°C near-surface (24-m) temperature contour is overlaid.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
(left) CAPE and (right) CIN calculated on grid 3 at 5 min before peak surface vortex intensity for the (top to bottom) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations. The × depicts the location of the developing surface vortex for each simulation. The 25°C near-surface (24-m) temperature contour is overlaid.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
(left) CAPE and (right) CIN calculated on grid 3 at 5 min before peak surface vortex intensity for the (top to bottom) CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations. The × depicts the location of the developing surface vortex for each simulation. The 25°C near-surface (24-m) temperature contour is overlaid.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
d. Isolating low-level moisture effects
The results of Markowski et al. (2002, 2003) indicated that higher ambient relative humidity at low levels was often linked to the coldness of the resulting RFD, as high boundary layer relative humidity was more conducive to higher-buoyancy environments for vortex development. The previous sections described how the HM simulations created the strongest, coldest RFDs, contrary to Markowski et al. (2002, 2003). However, the altered low-level moisture in the simulations presented herein also altered environmental CAPE, while the simulations discussed by Markowski et al. (2003) did not. They instead altered the background temperature profile in order to maintain similar CAPE values between their simulations. This makes it difficult to draw direct comparisons between the results of this study and those of Markowski et al. (2003), since CAPE is known to be a critical factor in storm intensity and resulting precipitation. Therefore, another nested grid simulation was conducted with the same three-grid setup as in the previous set of simulations, where the initial sounding used to generate convection contained the low-level moisture profile of the HM simulations but maintained the weaker CAPE of the LM simulations (see Table 1 for values; sounding not shown). This required increasing the temperatures of the initial sounding above 700 mb, which in turn slightly altered the ambient relative humidity profile aloft. The background aerosol concentrations were set to the clean continental values used in the CLN-DRY and CLN-WET simulations. The results from this simulation, hereafter referred to as CLN-WETb, were compared to those of the CLN-DRY and CLN-WET experiments in order to better understand how the aerosol–CCN effect compared to the influences of low-level moisture found previously by Markowski et al. (2003).
The CLN-WETb experiment produces a local maximum in vertical vorticity near the surface from 63 to 69 min. The vorticity center reaches a maximum value of 0.32 s−1 at 64 min. However, the vorticity center is only associated with a clear cyclonic circulation between 65 and 69 min. Figure 11 shows the near-surface (~24 m) temperature, vertical relative vorticity, perturbation pressure, and ground-relative winds (storm-relative wind vectors overlaid) for the CLN-WETb simulation on grid 3 at 66 min. The CLN-WETb vortex possesses a cyclonic circulation at the surface and a maximum vertical vorticity value of 0.18 s−1. The vortex is associated with a 5.5-mb pressure drop. Ground-relative winds are of EF-1 intensity, with winds approaching 40 m s−1 immediately south of the vortex. However, as with the CLN-WET and DST-WET simulations (Fig. 9), the maximum surface winds are actually located in the vicinity of the RFD-based outflow region southeast of the main vortex. Here, wind speeds exceed 44 m s−1. The near-surface winds north and northwest of the vortex are strongly northerly. The cold pool west of the vortex reaches a minimum temperature around 19°C. However, the region immediately surrounding the vortex contains higher temperatures, near 23°C. While all three simulations result in slightly different cold-pool strengths and positions relative to the developing vortex, the vortex that forms in the CLN-WETb simulation more closely resembles that produced by the CLN-WET simulation rather than that of the CLN-DRY experiment. These simulations yield similar FFD- and RFD-based cold-pool structures, ground-relative wind fields, and near-vortex pressure falls. Figure 12 displays precipitation rates on grid 2 at 65 min for the CLN-DRY, CLN-WET, and CLN-WETb experiments, overlaid with vertical velocities at 3.5 km. The CLN-DRY supercell produces relatively weak precipitation rates at this time with the core located back within the FFD, producing precipitation rates less than 75 mm h−1 (Fig. 12a). The CLN-WET supercell produces weak precipitation within its FFD. Precipitation rates greater than 150 mm h−1 exist within the RFD (Fig. 12b) approximately 4 km west of the main updraft core under which tornadogenesis occurs. Meanwhile, the CLN-WETb supercell creates the strongest precipitation core within the FFD (>100 mm h−1) but adjacent to the main updraft. Note that the CLN-WET supercell contained a noticeably stronger midlevel mesocyclone compared to that of the CLN-DRY and CLN-WETb simulations (not shown) due to the presence of significantly larger environmental CAPE, allowing for more precipitation to be wrapped cyclonically around the main updraft to the RFD, while more precipitation was able to fall within the FFD in the CLN-WETb supercell. The updraft is also noticeably weaker in the CLN-WETb case compared to the CLN-DRY and CLN-WET experiments because of the differences between the initial model soundings. Consequently, the cold pool in the CLN-WETb simulation was able to propagate through the developing surface vortex more easily compared to in the CLN-WET simulation (Fig. 11, column 1).
As in Fig. 9, but for the CLN-WETb simulation at 66 min.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
As in Fig. 9, but for the CLN-WETb simulation at 66 min.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
As in Fig. 9, but for the CLN-WETb simulation at 66 min.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Precipitation rates on grid 2 for the (a) CLN-DRY, (b) CLN-WET, and (c) CLN-WETb simulations at 65 min. Note that 3.5-km updrafts are overlaid, contoured at 10 and 20 m s−1.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Precipitation rates on grid 2 for the (a) CLN-DRY, (b) CLN-WET, and (c) CLN-WETb simulations at 65 min. Note that 3.5-km updrafts are overlaid, contoured at 10 and 20 m s−1.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
Precipitation rates on grid 2 for the (a) CLN-DRY, (b) CLN-WET, and (c) CLN-WETb simulations at 65 min. Note that 3.5-km updrafts are overlaid, contoured at 10 and 20 m s−1.
Citation: Journal of the Atmospheric Sciences 69, 3; 10.1175/JAS-D-11-043.1
4. Comparison with previous modeling studies
The results of the LM simulations differed somewhat from those of Lerach et al. (2008), which indicated that the CCN-polluted environment was more conducive for producing a tornado. The CLN-DRY near-vortex environment at low levels was actually more favorable for tornadogenesis than that of the DST-DRY simulation, as the CLN-DRY FFD- and RFD-associated outflow possessed warmer temperatures and less negative buoyancy (stronger CAPE and weaker CIN) than that of the DST-DRY simulation at the time of vortex development. There were some changes made to the model setup used in this study compared to that of Lerach et al. (2008). The Lerach et al. (2008) simulations did not include any IN in their simulations, whereas here we assumed some background profile (Fig. 2f) in the initialization. And while both sets of simulations assumed similar initial background profiles of CCN, the Lerach et al. (2008) simulations assumed a CCN source/sink scheme, where CCN were removed via droplet nucleation and replenished upon droplet evaporation. In this study we assumed no sinks of CCN, meaning that CCN concentrations were allowed to continuously diffuse and advect throughout the model domain. Then in a particular grid cell at a particular time, the number of cloud droplets that nucleated depended upon the CCN concentration associated with that grid cell. The simulated storms of Lerach et al. (2008) produced considerably higher precipitation rates and more evaporative cooling within the FFD and RFD, resulting in stronger cold pools and associated outflow. The clean continental simulation produced an FFD-based cold pool strong enough to undercut the storm’s core before a surface-based vortex could develop. This was not the case in the current set of simulations, and these differences between the two sets of model simulations might explain such differences in resulting precipitation.
The results from the LM simulations differed from those of Snook and Xue (2008) in that their simulations suggested that tornado potential increased as the cold pools weakened. The CLN-DRY simulation produced a stronger cold pool than the DST-DRY simulation overall. However, the cold pool in the DST-DRY simulation was closer to the developing vortex at the time of tornadogenesis than that in the CLN-DRY experiment, and thus colder, more negatively buoyant air surrounded the vortex. On the other hand, the HM simulations produced cold pools of comparable strength at the time of tornadogenesis, at least within the near-vortex environment. As a result, the DST-WET simulation actually produced a slightly stronger near-surface vortex than that of the CLN-WET experiment. It should be noted that no significant differences in surface convergence were found within the near-vortex environment among the CLN-DRY, CLN-WET, DST-DRY, and DST-WET simulations immediately prior to or during tornadogenesis.
The findings from the CLN-WETb simulation indicated that increasing the ambient relative humidity profile at low levels without affecting CAPE still resulted in a stronger cold pool that created a less favorable environment for tornadogenesis, contrary to the results of the idealized simulations by Markowski et al. (2003). This apparent contradiction was merely a result of the different microphysical parameterizations used in the Markowski et al. (2003) model setup compared to this study. In their simple axisymmetric setup, the ambient relative humidity was increased below about 720 mb between certain simulations. However, precipitation loading within their artificially imposed rain curtain was held constant. As a result, the increased relative humidity led to reduced evaporative cooling, a weaker cold pool, and a more favorable environment for tornadogenesis. In the less idealized simulations performed for this study, precipitation was not held fixed between simulations. The increased relative humidity at low levels acted as an increased moisture supply available for precipitation processes. Therefore, the simulation with increased relative humidity below 800 mb produced significantly stronger precipitation cores compared to the simulation initialized with lower relative humidity. The heavier precipitation resulted in greater net evaporative cooling, stronger downdrafts, and stronger cold pools, which provided an environment that hindered the tornadogenesis process. Nevertheless, the general results of Markowski et al. (2002, 2003) were consistent with those of this study. The strongest, longest-lived vortices were associated with warmer and weaker cold pools, higher CAPE, lower CIN, and thus less negative buoyancy in the near-vortex environment compared to those storms that produced shorter-lived, weaker vortices.
5. Summary
Five simulations were performed to compare the effects of CCN with those of low-level moisture and CAPE on tornadogenesis. While each simulation produced splitting supercells with the right mover producing a tornado-like vortex between 60 and 80 min, considerable differences were found between simulations with respect to storm microphysics and low- to midlevel dynamics. Increasing the ambient low-level moisture profile below 800 mb without altering the ambient temperature profile created significantly higher environmental CAPE. The combined effect was to produce spatially larger storms with stronger peak updrafts and stronger low-level downdrafts compared to the LM simulations. In addition, the dusty simulation produced slightly greater peak updraft velocities because of the presence of more supercooled liquid water aloft and the associated latent heat of freezing. The HM simulations produced storms with significantly stronger precipitation rates and higher accumulated precipitation, which resulted in greater evaporation and associated cooling, thus producing stronger cold pools at the surface associated with both the forward and rear flank downdrafts. The higher relative humidity at low levels in the HM simulations also delayed evaporation of falling precipitation, yielding the strongest peak downdrafts and evaporative cooling at lower levels compared to the LM simulations. The higher CCN concentrations in the dusty simulations (compared to the clean) reduced warm rain and yielded more supercooled water aloft, creating larger anvils with less ice mass available for precipitation. This resulted in lower hail number concentrations. However, raindrops and hailstones grew to larger sizes. As a result, the supercells from the dusty simulations underwent less evaporative cooling within downdrafts, and thus produced weaker low-level downdrafts and weaker, warmer cold pools compared to the clean simulations. However, with greater terminal fall velocities, the larger hydrometeors fell nearer to the storm’s core, which positioned the coldest region of the cold pool closer to the main updraft.
Tornadogenesis was related to the size, strength, and location of the FFD- and RFD-based cold pools. The combined influence of low-level moisture and CAPE played a noticeably larger role in the tornadogenesis process compared to the aerosol influence. However, the aerosol effect was still evident. Changing the low-level moisture profile resulted in changes to storm size and precipitation totals, but altering the background available CCN concentrations resulted in significant differences in storm microphysics and the location of precipitation cores. The combined effect determined the strength and location of the cold pool. The LM simulations produced the weakest cold pools and were most favorable for tornadogenesis, as these cold pools were associated with higher CAPE and lower CIN (less negative buoyancy) than those of the HM simulations.
It should be noted that aerosol–CCN effects on precipitation in deep convection have been shown to vary under different aerosol conditions and that the response of precipitation to the increase of aerosol concentration is nonmonotonic (Wang 2005; Tao et al. 2007; Li et al. 2008). Furthermore, the results from these simulations and those from Lerach et al. (2008) reveal that microphysical scheme selection can be important to simulating CCN impacts on supercell tornadogenesis, as different schemes and assumptions can significantly vary precipitation intensity within the FFD and RFD, and thus greatly alter the storm’s near-surface environment. In situations where altering the background CCN concentrations has a significant impact on precipitation rates and resulting cold-pool intensity, aerosol indirect microphysical effects on supercell storms and their cold pools can impact the likelihood of tornadogenesis via storm undercutting, a potential failure mechanism found by previous studies (e.g., Brooks et al. 1994). However, if conditions are such that storm undercutting is unlikely, aerosol effects are for more complex, dependent on other environmental factors, and will have a much smaller impact compared to the combined effect of low-level moisture and CAPE. Additional work is needed to identify which supercell environments are most likely to be impacted by aerosol indirect microphysical influences.
The magnitudes of maximum vertical vorticity associated with the tornado-like vortices created in our simulations were lower than have been observed and modeled by Snook and Xue (2008) (~0.2 vs ~1 s−1). One must note that friction was not taken into account at the lower boundary, a critical factor to the final spinup of a tornado (Rotunno 1979; Howells et al. 1988; Wicker and Wilhelmson 1993; Nolan and Farrell 1999; Trapp 2000) and likely one reason why maximum vertical vorticity in this study was smaller than values found in some previous work. The pressure deficits associated with our vortices were also smaller than those in some other studies, including Wicker and Wilhelmson (1995) and Grasso and Cotton (1995). The vortices simulated by these studies were accompanied by strong vertical motion near cloud base, possibly dynamically induced. This strong vertical motion might explain why their vortices, as well as many actual tornadoes, were stronger than those from this study (Gaudet et al. 2006). Finally, it is difficult to generalize the results of this study as they are based on a single thermodynamic sounding and an idealized supercell storm, which was initialized in an idealized model setup, albeit in three dimensions and at relatively fine grid spacing. Further investigation is required to address the robustness of the results presented. Nonetheless, these simulations provide insight toward understanding the possible role(s) of aerosol indirect microphysical in impacting supercell tornadogenesis.
Acknowledgments
We wish to express our sincerest thanks to Drs. Susan van den Heever of CSU and Brian Gaudet of PSU for their helpful insight regarding this work. We also thank Stephen Saleeby of CSU and Dr. Louie Grasso of CIRA for their assistance with the initial setup of the simulations as well as three reviewers for their constructive comments and suggestions. This work was funded by National Science Foundation (NSF) Grant ATM-0638910.
REFERENCES
Arakawa, A., and V. Lamb, 1981: A potential enstrophy and energy conserving scheme for the shallow water equations. Mon. Wea. Rev., 109, 18–36.
Bowen, E. G., 1950: The formation of rain by coalescence. Aust. J. Sci. Res., 3A, 193–213.
Braham, R. R., Jr., R. G. Semonin, A. H. Auer, S. A. Changnon Jr., and J. M. Hales, 1981. Summary of urban effects on clouds and rain. METROMEX: A Review and Summary, Meteor. Monogr., No. 40, Amer. Meteor. Soc., 141–152.
Brandes, E. A., 1978: Mesocyclone evolution and tornadogenesis: Some observations. Mon. Wea. Rev., 106, 995–1011.
Brooks, H. E., C. A. Doswell III, and R. B. Wilhelmson, 1994: The role of midtropospheric winds in the evolution and maintenance of low-level mesocyclones. Mon. Wea. Rev., 122, 126–136.
Burgess, D. W., R. A. Brown, L. R. Lemon, and C. R. Safford, 1977: Evolution of a tornadic thunderstorm. Preprints, 10th Conf. on Severe Local Storms, Omaha, NE, Amer. Meteor. Soc., 84–89.
Carrió, G. G., W. R. Cotton, and W. R. Cheng, 2010: Urban growth and aerosol effects on convection over Houston. Part I: The August 2000 case. Atmos. Res., 96, 560–574.
Clark, T. L., and R. D. Farley, 1984: Severe downslope windstorm calculations in two and three spatial dimensions using anelastic interactive grid nesting: A possible mechanism for gustiness. J. Atmos. Sci., 41, 329–350.
Cotton, W., and Coauthors, 2003: RAMS 2001: Current status and future directions. Meteor. Atmos. Phys., 82, 5–29.
Davies-Jones, R. P., 1982a: A new look at the vorticity equation with application to tornadogenesis. Preprints, 12th Conf. on Severe Local Storms, San Antonio, TX, Amer. Meteor. Soc., 249–252.
Davies-Jones, R. P., 1982b: Observational and theoretical aspects of tornadogenesis. Intense Atmospheric Vortices, L. Bengtsson and J. Lighthill, Eds., Springer-Verlag, 175–189.
Davies-Jones, R. P., 2000: Can the hook echo instigate tornadogenesis barotropically? Preprints, 20th Conf.on Severe Local Storms, Orlando, FL, Amer. Meteor. Soc., 269–272.
Davies-Jones, R. P., and H. E. Brooks, 1993: Mesocyclogenesis from a theoretical perspective. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 105–114.
Eagan, R. C., P. V. Hobbs, and L. F. Radke, 1974: Particle emissions from a large Kraft paper mill and their effects on the microstructure of warm clouds. J. Appl. Meteor., 13, 535–552.
Eskridge, R. E., and P. Das, 1976: Effect of a precipitation-driven downdraft on a rotating wind field: A possible trigger mechanism for tornadoes? J. Atmos. Sci., 33, 70–84.
Fan, J., R. Zhang, G. Li, W.-K. Tao, and X. Li, 2007: Simulations of cumulus clouds using a spectral microphysics cloud-resolving model. J. Geophys. Res., 112, D04201, doi:10.1029/2006JD007688.
Feingold, G., R. L. Walko, B. Stevens, and W. R. Cotton, 1998: Simulations of marine stratocumulus using a new microphysical parameterization scheme. Atmos. Res., 47–48, 505–528.
Foote, G. B., 1984: A study of hail growth utilizing observed storm conditions. J. Climate Appl. Meteor., 23, 84–101.
Gaudet, B. J., and W. R. Cotton, 2006: Low-level mesocyclonic concentration by nonaxisymmetric processes. Part I: Supercell and mesocyclone evolution. J. Atmos. Sci., 63, 1113–1133.
Gaudet, B. J., W. R. Cotton, and M. T. Montgomery, 2006: Low-level mesocyclonic concentration by nonaxisymmetric transport: II. Vorticity dynamics. J. Atmos. Sci., 63, 1134–1150.
Gilmore, M. S., J. M. Straka, and E. N. Rasmussen, 2004: Precipitation uncertainty due to variations in precipitation particle parameters within a simple microphysics scheme. Mon. Wea. Rev., 132, 2610–2627.
Grasso, L. D., 2000: The dissipation of a left-moving cell in a severe storm environment. Mon. Wea. Rev., 128, 2797–2815.
Grasso, L. D., and W. R. Cotton, 1995: Numerical simulation of a tornado vortex. J. Atmos. Sci., 52, 1192–1203.
Hill, G. E., 1974: Factors controlling the size and spacing of cumulus clouds as revealed by numerical experiments. J. Atmos. Sci., 31, 646–673.
Hobbs, P. V., L. F. Radke, and S. E. Shumway, 1970: Cloud condensation nuclei from industrial sources and their apparent influence on precipitation in Washington State. J. Atmos. Sci., 27, 81–89.
Howells, P. C., R. Rotunno, and R. K. Smith, 1988: A comparative study of atmospheric and laboratory-analogue numerical tornado-vortex models. Quart. J. Roy. Meteor. Soc., 114, 801–822.
Kaufman, Y. J., and T. Nakajima, 1993: Effect of Amazon smoke on cloud microphysics and albedo—Analysis from satellite imagery. J. Appl. Meteor., 32, 729–744.
Klemp, J. B., and R. B. Wilhelmson, 1978: The simulation of three-dimensional convective storm dynamics. J. Atmos. Sci., 35, 1070–1096.
Koehler, K. A., S. M. Kreidenweiss, P. J. DeMott, M. D. Petters, A. J. Prenni, and C. M. Carrico, 2009: Hygroscopicity and cloud droplet activation of mineral dust aerosol. Geophys. Res. Lett., 36, L08805, doi:10.1029/2009GL037348.
Lemon, L. R., 1974: Interaction of two convective scales within a severe thunderstorm: A case study. NOAA Tech. Memo. ERL NSSL-71, 43 pp.
Lerach, D. G., B. J. Gaudet, and W. R. Cotton, 2008: Idealized simulations of aerosol influences on tornadogenesis. Geophys. Res. Lett., 35, L23806, doi:10.1029/2008GL035617.
Leslie, L. M., and R. K. Smith, 1978: The effect of vertical stability on tornadogenesis. J. Atmos. Sci., 35, 1281–1288.
Li, G., Y. Wang, and R. Zhang, 2008: Implementation of a two-moment bulk microphysics scheme to the WRF model to investigate aerosol–cloud interaction. J. Geophys. Res., 113, D15211, doi:10.1029/2007JD009361.
Lilly, D. K., 1962: On the numerical simulation of buoyant convection. Tellus, 14, 148–172.
Lim, K.-S. S., and S.-Y. Hong, 2010: Development of an effective double-moment cloud microphysics scheme with prognostic cloud condensation nuclei (CCN) for weather and climate models. Mon. Wea. Rev., 138, 1587–1612.
Ludlam, F. H., 1963: Severe local storms: A review. Severe Local Storms, Meteor. Monogr., No. 27, Amer. Meteor. Soc., 1–30.
Markowski, P. M., 2002: Hook echoes and rear-flank downdrafts: A review. Mon. Wea. Rev., 130, 852–876.
Markowski, P. M., J. M. Straka, and E. N. Rasmussen, 2002: Direct surface thermodynamic observations within the rear-flank downdrafts of nontornadic and tornadic supercells. Mon. Wea. Rev., 130, 1692–1721.
Markowski, P. M., J. M. Straka, and E. N. Rasmussen, 2003: Tornadogenesis resulting from the transport of circulation by a downdraft: Idealized numerical simulations. J. Atmos. Sci., 60, 795–823.
Meyers, M. P., R. L. Walko, J. Y. Harrington, and W. R. Cotton, 1997: New RAMS cloud microphysics parameterization. Part II: The two-moment scheme. Atmos. Res., 45, 3–39.
Nelson, S. P., 1977: Rear flank downdraft: A hailstorm intensification mechanism. Preprints, 10th Conf. on Severe Local Storms, Omaha, NE, Amer. Meteor. Soc., 521–525.
Nolan, D. S., and B. F. Farrell, 1999: The structure and dynamics of tornado-like vortices. J. Atmos. Sci., 56, 2908–2936.
Pielke, R. A., and Coauthors, 1992: A comprehensive meteorological modeling system—RAMS. Meteor. Atmos. Phys., 49, 69–91.
Rosenfeld, D., 1999: TRMM observed first direct evidence of smoke from forest fires inhibiting rainfall. Geophys. Res. Lett., 26, 3105–3108.
Rosenfeld, D., 2000: Suppression of rain and snow by urban and industrial air pollution. Science, 287, 1793–1796.
Rotunno, R., 1979: A study in tornado-like vortex dynamics. J. Atmos. Sci., 36, 140–155.
Rotunno, R., and J. B. Klemp, 1982: The influence of the shear-induced pressure gradient on thunderstorm motion. Mon. Wea. Rev., 110, 136–151.
Saleeby, S. M., and W. R. Cotton, 2004: A large droplet mode and prognostic number concentration of cloud droplets in the Colorado State University Regional Atmospheric Modeling System (RAMS). Part I: Module descriptions and supercell test simulations. J. Appl. Meteor., 43, 182–195.
Seifert, A., and K. D. Beheng, 2006: A two-moment cloud microphysics parameterization for mixed-phase clouds. Part II: Maritime vs. continental deep convective storms. Meteor. Atmos. Phys., 92, 67–82.
Smagorinsky, J., 1963: General circulation experiments with the primitive equations. I. The basic experiment. Mon. Wea. Rev., 91, 99–164.
Snook, N., and M. Xue, 2008: Effects of microphysical drop size distribution on tornadogenesis in supercell thunderstorms. Geophys. Res. Lett., 35, L24803, doi:10.1029/2008GL035866.
Storer, R. L., S. van den Heever, and G. L. Stephens, 2010: Modeling aerosol impacts on convective storms in different environments. J. Atmos. Sci., 67, 3904–3915.
Tao, W.-K., X. Li, A. Khain, T. Matsui, S. Lang, and J. Simpson, 2007: Role of atmospheric aerosol concentration on deep convective precipitation: Cloud-resolving model simulations. J. Geophys. Res., 112, D24S18, doi:10.1029/2007JD008728.
Trapp, R. J., 2000: A clarification of vortex breakdown and tornadogenesis. Mon. Wea. Rev., 128, 888–895.
Trapp, R. J., and B. F. Fiedler, 1995: Tornado-like vortex genesis in a simplified numerical model. J. Atmos. Sci., 52, 3757–3778.
Tripoli, G. J., and W. R. Cotton, 1980: A numerical investigation of several factors leading to the observed variable intensity of deep convection over south Florida. J. Appl. Meteor., 19, 1037–1063.
Tripoli, G. J., and W. R. Cotton, 1986: An intense, quasi-steady thunderstorm over mountainous terrain. Part IV: Three-dimensional numerical simulation. J. Atmos. Sci., 43, 894–912.
van den Heever, S. C., and W. R. Cotton, 2004: The impact of hail size on simulated supercell storms. J. Atmos. Sci., 61, 1596–1609.
van den Heever, S. C., G. G. Carrió, W. R. Cotton, P. J. DeMott, and A. J. Prenni, 2006: Impacts of nucleating aerosol on Florida storms. Part I: Mesoscale simulations. J. Atmos. Sci., 63, 1752–1775.
Walko, R. L., 1993: Tornado spin-up beneath a convective cell: Required basic structure of the near-field boundary layer winds. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 89–95.
Wang, C., 2005: A modeling study of the response of tropical deep convection to the increase of cloud condensation nuclei concentration: 1. Dynamics and microphysics. J. Geophys. Res., 110, D21211, doi:10.1029/2004JD005720.
Wicker, L. J., and R. B. Wilhelmson, 1993: Numerical simulation of tornadogenesis within a supercell thunderstorm. The Tornado: Its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No. 79, Amer. Geophys. Union, 75–88.
Wicker, L. J., and R. B. Wilhelmson, 1995: Simulation and analysis of tornado development and decay within a three-dimensional supercell thunderstorm. J. Atmos. Sci., 52, 2675–2703.