ON THE RESPONSE OF HAILSTORMSTO ENHANCED CCN CONCENTRATIONS

William R. CottonDepartment of Atmospheric Science, Colorado State University

Approaches to hailstorm simulation

• 2 moment bulk microphysics• Hybrid approach: Bulk 1-M or 2-M

microphysics with bin representation of hail

• Full bin storm model• 3 moment bulk model

Bin approach

The full bin approach is the most realistic approach but is very computationally expensive and has only been done for hailstorms in two dimensions

Previous studies of CCN impacts on hail

To date, only two studies have examined the impacts of increasing CCN on hail (for same storm) with contrasting results• Noppel et al. (2010):used 2M microphysics in 3D storms and found

more numerous but smaller hailstones, reduction in hail amounts reaching surface for greater CCN concentrations

• Khain et al. (2011): used a full bin microphysics but in 2D and found an increase in total hail mass and sizes of hailstones, larger hail reaches surface for greater CCN amounts

• Both these studies were for the same storm

Adrian Loftus developed a 3-moment hail model in RAMS

Most bulk models are either 1-moment(predict on mass), or 2-moment(predict on mass and concentration) or a few are 3-moment(predict on mass, concentration, and the 6th moment which is proportional to radar reflectivity)

The 3-moment scheme is still computationally less demanding than a bin model but allows representation of the tail of the distribution where hail stones reside

Brief model description RAMS v.6 with bin-emulating two-moment microphysics (Saleeby & Cotton, 2004, 2008).

processes are weighted by dist’s function of Nx and Qx

For this study we added a triple-moment hail microphysics scheme (Loftus, 2012) that predicts of 6th and 0th moments so all three parameters of the distribution are variable:

Nh, Qh, and nh

11( ) exp

( )t

n n n

N D Dn Dv D D D

n

3MHAIL scheme development

Dm = 20 mm

•Spectral width parameter (nh)controls relative amounts of small versus large particles

•For 3MHAIL scheme, nh ranges from 1.0 to 10.0

•Hail mean mass diameter (Dmh) ranges from 0.8 to 40 mm (maximum Dmh is typically 10 mm for 2M scheme in RAMS)

•Actual hail diameters (Dh) range from 0.2 to 150 mm

Marshall-Palmer distribution

3MHAIL scheme predicts 6th moment [hail reflectivity factor (Zh)] in addition to Nth and rh in order to obtain nhand thereby allow for fully prognostic size distribution

3M vs 2M

• Predicting 6th, 3rd, and 0th moments of hail size spectrum allows all three gamma distribution parameters to vary freely- removes need to ‘tune’ parameters

• Variable nh improves realism in evolution of hail distribution during sedimentation and meltingbroadening/narrowing aloft, narrowing at low levelsmitigates artificial shifts towards large hail sizes

< 1 in1 in1.75 in

Hail reports

Adrian performed simulations for an actual case:

29 June 2000 STEPS supercell

Goodland, KS WSR-88D loop of lowest level radar reflectivity2025 to 0410 UTC

F1 tornado

Triple-moment hail microphysics

Left: Swath of column maximum reflectivity from KGLD radar. Surface fallout locations for hail ≥ 3 cm computed using particle growth model for 29 June 2000 supercell are overlaid (black dots) [from Tessendorf et al. 2005]

Right: Swath of column maximum equivalent reflectivity factor (Ze) [dBZ] with contours of surface hail concentrations [1x10-4, 0.01 m-3] having diameters ≥ 2 (blue), 3 (black) and 4 cm (purple) for simulation using 3MHAIL scheme. Times (min) of maximum Ze also shown.

Largest hail at surface following right turn

Largest hail collocated with largest Ze values

Sensitivity to CCN experiments

• Same model setup as in verification simulations (simulation time of 180 min)

• Model initialized with 5 different horizontally homogeneous CCN profiles

• Cloud droplet nucleation is CCN sink, no sources of CCN

• CCN assumed to be ammonium sulfate particles

• CCN < 300 cm-3 not typical of High Plains region (Hobbs et al. 1985)

Initial CCN profiles

Approximate cloud base height is 2 km

CCN sensitivity resultsz=982m

z=4947m

ccn100 ccn600 ccn3000

Ze[dBZ]

Increase in maximum Zeas CCN h

Similar storm structure and evolution in all cases

Sensitivity of hail size to CCNTime-height plots of maximum hail mixing ratio [g m-3] associated with hail having diameters ≥ 2 cm reveal a general trend of increases in large hail mass aloft as CCN increases, in agreement with Khain et al. (2011: bin microphysics) but in contrast with Noppel et al. (2009: two-moment bulk microphysics).

ccn100

ccn600

ccn3000

CCN numbers do not have appreciable impact on maximum vertical velocities in simulations, similar to results by Tao et al. (2007), Khain et al. (2011) and Lim et al. (2011) for strong convection.

Sensitivity of hail size to CCNTime series of surface maximum hail mixing ratio [g kg-1 ] associated with hail having diameters ≥ 2 and 4 cm exhibit a clear trend of increases in large hail mass at surface as CCN increases. A similar trend is seen in time series of maximum hail number concentrations associated with hail having diameters ≥ 2 and 4 cm (not shown) in general agreement with Khain et al. (2011).

D ≥ 4 cm

D ≥ 2 cm

Experiment Rain [kg m-2] Hail [kg m-2]

ccn100 17.3 27.6

ccn300 18.8 30.4

ccn600 18.5 38.6

ccn1500 16.4 44.4

ccn3000 14.4 46.0

Maximum accumulated amounts of hail increase with increasing CCN, in contrast to Noppel et al. (2009), whereas maximum accumulated amounts of rain generally decrease for CCN values greater than 300/cc.

Cold pool sensitivity to CCNAppears to be trend of increase in cold pool size and strength (greater negative qe’) as CCN increases from 100 to 600/cc, but then the opposite trend occurs as CCN increases from 600 to 3000/cc.

These trends generally follow the trends in maximum accumulated rain and hail, with less rain and more (and larger) hail leading to decreased cooling at low levels, similar to results from van den Heever and Cotton (2004).

Minimum surface qe’ [K] Mean surface qe’ [K] for grid points having qe’ ≤ -1 K

Surface cold pool area [km2]

A rather large number of hailstorm simulations were carried out varying both [CCN] and cloud base temperature.

Gustavo Carrio and I then exercised Adrian’s model

RAMS was set up as follows:

3-D domain: 100 X 100 X18 Km (Nx=Ny=200) Dx = Dy = 500m , Dt = 1s 1st level 40m, Dz stretching up to 500m Hot Bubble (2.5K), 30km upwind of the center

Benchmark case

Marginally severe single cell ordinary thunderstorm

Occurred over central Tennessee on 15 May 2009

2.5 cm diameter hail stones

CAPE = ~ 1500 J kg-1

[CCN]: 16 values between 150 and 2400 cm-3 constant at low levels, exponentially decreasing above 3km.

Cloud base: The sounding of the case study was systematically perturbed by varying low-level vapor contents from -12 to 12% (10 levels, ~ 500m).

Sensitivity experiments

Sensitivity experiments Green arrow denotes the profiles of May 15 2009

Blue arrow represents the simulations with the lowest [CCN] (150cm-3, “clean”).

In the following figures each point is a quantity that characterizes a run.

To isolate the dependence on [CCN] many figures show values normalized by each corresponding “clean” case (along the blue arrow).

Preliminary results Precipitation

For storms with higher cloud bases, significant CCN impact on: both total precipitation and hail fraction

For lower cloud bases → little impact

Preliminary resultsHail integral mass and rates

Large positive changes in both the integral mass of hail precipitation and maximum rates for runs with higher cloud bases.

Preliminary resultsHail concen and predominant size

Large positive changes for N’s for runs with higher cloud bases.

Positive although more modest impact on predominant diameters.

Peak updrafts and SCLW

Positive relative changes for storms with higher cloud bases.

Linked to the availability of supercooled liquid water.

Wmax Peak SCLW

Change in storm “size”

Increasing [CCN] tends to concentrate hail in smaller areas for the dries cases.

Larger accumulations concentrate in smaller areas of precipitation (as in Carrió & Cotton, 2011).

Downdrafts

CCN → downdraft area ↑ for runs with lower cloud bases.

Downdrafts

CCN → downdraft area ↑ for runs with lower cloud bases.

Larger impact occurs for run with higher rain fraction of precipitation

Summary:

We find significant sensitivity to CCN concentrations for storms with colder cloud base temperatures or higher cloud bases and with smaller distances between cloud base and freezing level(D)

For this regime we find larger predominant hail sizes and concentrations, greater integral masses of hail, and greater hail precipitation

For warmer cloud base temperatures and greater D

The response to varying CCN amounts is weak

Warm rain processes are much more active

Thank you!