ESSOAr | https://doi.org/10.1002/essoar.10500312.

1 | Non-exclusive | First posted online: Mon, 14 Jan 2019 08:00:24 | This content has not been peer reviewed.
Supersaturation Fluctuations from Scalar Transport in Moist Rayleigh-Bénard Convection:
One-Dimensional-Turbulence
1 1 2
Simulation
1 3
Kamal Kant Chandrakar , Will Cantrell , Steven Krueger , Raymond A. Shaw , and Scott Wunsch
1
Department of Physics, Michigan Technological University 2Department of Atmospheric Sciences, University of Utah 3Applied Physics Laboratory, Johns Hopkins University

Introduction Supersaturation Fluctuations Mixing Diagram

A careful characterization of moisture fluxes and supersatura-
101
tion statistics in atmospheric convection is significant for cloud Actual 4 X Dv
microphysical processes and dynamics. The small-scale super- D v = Dt 4 X Dt
10
0 100
saturation fluctuation could be an important mechanism for
droplet size-distribution broadening [1]. In an idealized sense,
atmospheric boundary-layer convection is equivalent to Rayleigh- 10-1
Bénard convection [2, 3]. Here, continuous plume eruption from
10-2
boundaries transports heat and moisture and produces the mixed

PDF

PDF
10-2
layer [4]. The presence of multiple driving scalars (water vapor
and temperature), with slightly different diffusivities, adds to the
degree of complexity. Additionally, in the supersaturation statis- 10-3
tics, a nonlinear coupling between water-vapor and temperature 10-4

fields makes the problem intriguing. Important parameters for 10-4

this problem are [5]:

g∆T H gϵ∆qv H 3 3
ν ν 10-6
Ramoist ≈ + , Pr ≡ , and Sc ≡ . 1.5 2 2.5 3 3.5 0 5 10
T0νDt νDt Dt Dv s [%] s [%]
Figure 4: Mixing diagram for different diffusivity cases at 20K applied temper-
Figure 2: Sample PDFs of supersaturation near the domain center for the different diffusivity cases (8 K applied temperature difference). ature difference.
Numerical Approach
We approach this problem using an idealized, one-dimensional- Scaling Analysis Conclusion
turbulence (ODT) model that faithfully represents the processes
of advection and diffusion in turbulent convection, including Moisture flux:
−1/3
• Scaling relations for heat and moisture fluxes are obtained as
fully-resolved boundary layers [6]. The range of Rayleigh number Sh ∝ Sc 1/2 1/3
Ramoist |P r ∼ 1 Sh ∝ Sc 1/2
Pr
1/3
Ramoist |P r ≪ 1
2.05 × 108 ≤ Ramoist ≤ 2.75 × 109 covered in simulations is rele- a function of Ramoist, P r, and Sc.
Sh : Sherwood Number; P r : Prandtl Number; Sc : Schmidt Number −1/6
vant to the Π-chamber. In order to understand the relative roles • In the bulk fluid, σT∗
and ∗
σqv
both follow a Ramoist
scaling
Supersaturation Fluctuations:
of the two diffusivities, we use the following four combinations. ( )2  relation. Moreover, the magnitude of scalar fluctuations
( )2
2  σev σT T ′e′v  increases with an increase in the respective scalar diffusivity.
• Actual Dv and Dt: actual diffusivities for the water vapor σs ≈ S
2
+ζ 2
− 2ζ
e¯v T̄ T̄ e¯v
and thermal fields at 283 K (Dv /Dt = 1.16). • PDFs of supersaturation become broader with an increase in
• Dv = Dt: Dv = the actual thermal diffusivity (Dv /Dt = 1).  ( )1/22 absolute value of diffusivity difference. Also, PDFs are slightly
( )2 √ ( )−1
σs 2 1 Pr −2 −1
 P r (1 + P r) Ra 5/3 z
• 4 × Dv : Dv = four times the actual value, and Dt = same as ∼ξ 1− moist
negatively skewed for cases with low diffusivity difference,
S S Sc H unlike the T and qv PDFs.
the actual value (Dv /Dt = 4.63).
• 4 × Dt: Dv = same as the actual value, and Dt = four times σT : temperature STD; σqv : water-vapor STD; z: vertical position; H : domain height; ξ ∝ H −3 • Both,scaling and numerical output suggests:
the actual value (Dv /Dt = 0.29). ∆T : applied temperature difference; ∆qv : applied water vapor mixing-ratio difference 5/3
σs ∝ Ramoist/H 6.
2
ν: kinematic viscosity; Dt: thermal diffusivity; Dv : water vapor diffusivity
Boundary Fluxes • The analysis of numerical output shows similar order
Supersaturation Fluctuations in Bulk: Contributions from Both Scalars contributions to the supersaturation variance from both
110
scalar variance and covariance.
Actual D
100 4 X Dv 0.035 • Distributionof points in a pressure-temperature mixing
90 4 X Dt 0.03 diagram deviate from the classical mixing line for isobaric
D v = Dt
100 mixing, when Dv ̸= Dt.
80 0.025
Sh X (Sc a Prb)

70 0.02
References
[%]
v
q

60
*

0.015 -1
s

10 [1] Chandrakar et al. Proc. Natl. Acad. Sci. (USA), 113(50).

50 [2] James W Deardorff. J. Atmos. Sci., 27(8):1211–1213, 1970.
Actual D Actual D [3] Douglas K Lilly. Quart. J. Roy. Meteor. Soc., 94(401):292–309, 1968.
0.01 4 X Dv 4 X Dv
[4] Kerry A Emanuel. Oxford University Press, 1994.
4 X Dt 4 X Dt
40
10 -2 [5] Niedermeier et al. Phys. Rev. Fluids, 3(8):083501, 2018.
9
10 D v = Dt D v = Dt [6] Wunsch et al. J. Fluid Mech., 528:173–205, 2005.
Ramoist
10 8
10 9
10 10
108 109
Figure 1: Variation of the scalar fluxes of water vapor (Sh) with moist Rayleigh Ramoist Ramoist
Acknowledgements
number (Ramoist). Fitting of the scaled Sh data produce Ramoist exponents
Figure 3: STDs of normalized water vapor mixing-ratio (left) and supersaturation fluctuation (right) at the domain center versus Ramoist, and their comparison
around 0.328 ± 0.006. −1/6 5/6 This work was supported by National Science Foundation Grant AGS-1754244
with scaling relations (dash-line) Ramoist and Ramoist. and by the NASA Earth and Space Science Fellowship Program.