Cirrus Parcel Model Comparison Project

4 Pages·2000·0.31 MB·English

by  Lin, Ruei-Fong

CIRRUS PARCEL MODEL COMPARISON PROJECT PHASE 1
Ruei-Fong Linl, *David O'C. St arr 2, Paul J. DeMott a
Richard Cotton 4, Eric Jensen 5, and Kenneth Sassen 6
1Universities Space Research Association, Greenbelt, MD, USA
2Laboratory of Atmosphere, NASA/GSFC, Greenbelt, MD, USA
aDept, of Atmospheric Sciences, Colorado State University, Fort Collins, CO, USA
4Meteorological Research Flight, Hants, UK
'_NASA/Ames Research Center, Moffett Field, CA, USA
6Dept. of Meteorology, University of Utah, UT, USA
1 INTRODUCTION all nucleation modes are switched on in the ALL-
MODE runs. Participants were asked to run the
The Cirrus Parcel Model Comparison (CPMC) is a HN-A-fixed runs by setting A = 2 (A is further dis-
project of the GEWEX Cloud System Study Work- cussed in section 2) or tailoring the nucleation rate
calculation in agreement with A = 2I. The depth
ing Group on Cirrus Cloud Systems (GCSS WG2).
The primary goal of this project is to identify cirrus of parcel lift (800 m) was set to assure that parcels
model sensitivities to the state of our knowledge of underwent complete transition through the nucle-
nucleation and microphysics. Furthermore, the com- ation regime to a stage of approximate equilibrium
mon ground of the findings may provide guidelines between ice mass growth and vapor supplied by the
for models with simpler cirrus microphysics modules. specified updrafts.
Table 1: Sinmlation identifiers. 2 MODEL DESCRIPTIONS
Five parcel modeling groups participated in the
W [m/s] 0.04f 0.2 1
CPMC (Table 2). Hereafter, we will refer to these
HN-ONLY Ch004 Ch020 Ch100
models as the C, D, J, L, and S models, respectively,
Wh004 Wh020 Whl00
as denoted in the table.
ALL-MODE Ca004 Ca020 Ca100
The estimate of the nucleation rate of ice in solu-
Wa004 Wa020 VVal00
tion droplets, Jh...... remains an active research area.
HN-A-fixed Ch020L
Jha_ was computed using either (1) the modified
Wh020L
classical theory approach (model J) or (2) the effec-
tive freezing temperature approach (hereafter, T¢ff
We focus on the nucleation regimes of the warm models, models C, D, L, S).
(parcel starting at -40°C and 340 hPa) and cold The TeIf models attempt to directly link mea-
(-60°C and 170 hPa) cases studied in the GCSS sured Jh_,¢ to nucleation rates of equivalent-sized
WG2 Idealized Cirrus Model Comparison Project pure water droplets J,, via the effective freezing tem-
[Start et al., 2000]. Nucleation and ice crystal perature, which is defined as
growth were forced through an externally imposed
7eyf = r + AAT, n, (1)
rate of lift. and consequent adiabatic cooling (Ta-
ble 1). The background haze particles are as- such that Jhaze = Jw(:lef f) as introduced by Sassen
sumed to be lognormally-distributed H2S04 parti- and Dodd [1988]. In (i), ATm is the equilib-
cles. Only the homogeneous nucleation mode is al- rium melting point depression (positive valued),
lowed to form ice crystals in the HN-ONLY runs; which depends on solute wt%, and A is an empir-
ical coefficient to account for additional suppres-
sion/enhancement of nucleation temperature due to
*Corresponding author's address: Ruei-Fong Lin, USRA,
NASA/GSFC, Code 913, Greenbelt, MD, 20771; E-Mail: lNote that A = 2 agrees approximately with data pre-
[email protected], na.sa.gov sented by Koop et al. [1998].
Table2:Participanctirrusparceml odels.
Organization UKMO CSU ARC GSFC U. Utah
Investigator Cotton (C) DeMott (19) Jensen (J) nin (L) Sassen (S)
Bin characteristic _ discrete continuous continuous continuous particle tracing
dr
Haze size b req or _ req req req req or _,
)_ 1.5 1.5 varying _ 1.0 1.7
deposition coef. _i 0.24 0.04 1 0.1 0.36
References Spice et al. DeMott Jensen et al. Lin [1997] Sassen and
[1999] et al. [1994] [1994] Dodd [1988]
DeMott Tabazadeh Sassen and
et al. [1998] et al. [19981 Benson [2000]
a Discrete vs continuous binning indicates if assuming that all particles have exactly the saule size in a given
size bin or a certain distribution of particle sizes is allowed in a bin.
breq vs. "_drdenotes either using the equlibrium-sized haze approximation or computing the diffusional
growth of haze particles explicitly.
See section 2 for detailed discussion.
non-ideal interaction between ions and condensed freezing haze size distributions. Nucleation rate data
water. Although Sassen and Dodd [1988] noted that over a wide range of values, e.g., data points beyond
an average A for different solutions was around 1.7, critical freezing conditions, are needed to diminish
values for specific solutions may range from i to 2.5. the inconsistency between the two approaches.
Little constraint was imposed on formulating het-
erogeneous nucleation because theoretical and ex-
perimental understanding are still quite poor. Mod-
sulfuric acid mass = l0 "13g
els C and L employ ice saturation ratio dependent
105
parameterizations of activated IN following Spice et
:
100 al. [1999] and Meyers et al. [1992], respectively.
These parameterizations are expected to represent a
10-5 maxinmm heterogeneous nucleation impact.
>
Haze particles of the given H2S04 aerosol distri-
10-10 \<:. \ ii..- bution are subject simultaneously to heterogeneous
and homogeneous nucleation in models D and S. The
10-15
number concentration of the activated IN in model D
-65 -60 -55 -50 -45 -40 -35
Temperature(C) is colnputed following DeMott et al. [1998] based on
field experiment data. This treatment was expected
to yield the most conservative estimate of IN in cir-
Figure 1: JhozeV vs. temperature for solute wt% rus. Model S computes the activated freezing nuclei
5, 15 and 25%. Solid, dashed, dash-dotted, dotted using Teff dependent Fletcher equation [Sassen and
curves denote models J, C, S, models D and L (same
Bcnson, 2000], where parameters were set to yield
curves), respectively, for X = 2. the most favorable conditions for heterogeneous nu-
cleation.
In model J, recent direct data on ice/solution sur-
Participants either assumed that haze particles
face tension was incorporated and activation energy
are in equilibrium with the enviromnent or com-
was inferred from recent laboratory measurements
puted the diffusional growth of haze partMes di-
of .]h_e for H2S04 particles following 7hbazadeh et
rectly (Table 2). The diffusional growth rate of haze
al. [1997] and Koop et al. [1998]. This approach to
particles more or less exponentially decreases with
determine Jhaze can be interpreted as a Tclf scheme
tenlperature ,as caused by water vapor saturation
with varying k (Figure 1). The intrinsic A varies in-
pressure. The response time scale to the deviation
versely with solute wt% and temperature. Also, the
from equilibrium can be considerably greater than
differences in the sensitivity of Jh,_eV (V is the vol-
one model time step in a swift, updraft in a cold
ume of the particle) to solute wt% between these two
environlnent. Therefore, large haze partMes may
approaches may lead to systematic differences in the
becomme oreconcentratetdhantilecorresponding
Cirrus initiation occurred over a narrower range
equilibrium-sizpearticlesinsuchconditions.This of altitude and RHi (relative humidity over ice) in
mayresultinconsiderabdleelayingofhazegrowth the warm HN-ONLY cases than in the cold cases
inmodels
C and S (Table 2) and affect ice particle (Fig. 3). The increasing sensitivity of the cloud base
formation rate. RHi as temperature decreases in the four T,-H mod-
els is primarily caused by ,_.
Heterogeneous nucleation is a possible explana-
3 RESULTS AND DISCUSSIONS
tion of the discrepancy between the observed thresh-
old RHw for cirrus formation and the theoreti-
As we proceed to describe the results and differences cally derived threshold RH_o (relative humidity over
between models, it must be noted that tile bench-
water) for homogeneous nucleation of H2S04 or
mark is not necessarily the median or the average of
(N H4)2S04 solution particles; e.g., [Heymsfield and
model results. The predicted N, (ice number concen-
Miloshevich, 1995]. Cirrus properties are affected
tration) at 800 m above the starting point is con> by the dominant nucleation mode in cloud initia-
pared (Fig. 2). In the HN-ONLY cases, to a first tion because of the distinct characteristics of the two
order approximation, the logarithm of N, increases modes.
quasi-linearly with the logarithm of updraft speed. The cloud base height, RHi and peak RH, in the
The predicted N, by models D, S and L are close; ALL-MODE cases (not shown) vary even more be-
N, by models J and C form the lowest and highest cause of our respective unbounded choices of hetero-
bounds in the six cases, respectively.
geneous nucleation. The impact of heterogeneous
nucleation on lowering Ni, peak RHi, and cloud for-
The cold cases motion altitude is extremely sensitive to the onset
conditions for nucleation and the subsequent ice par-
ticle formation rate. With heterogeneous nucleation,
Z [L| ,
O,Ol the peak RH, is lower in all but the case Wa100 by
o(x)l I
C D J L S C D J L S C D J L S model S. The predicted N, is reduced in all but the
W=i}.04 W=(}.21) W=l .(Hi
case Ca004 by model S.
The warm cases
We now discuss the results of the HN-A-ffxed sim-
I ulations. The nucleation regimes of Wh020L and
z- (I.1 Ch020L take place within the temperature range of
-43.2 to -44.2°C and -63.2 to -64.2°C, respectively.
UDJLS CDJLS CDJLS
w=004 W=(1211 W=l.OII The effect of temperature variation on mMeation
rates within this 1°C range is secondary, coml)ared
Figure 2: Ni predicted vs imposed updraft speed. to the evolution of haze solute wt%. Thus, it is jus-
The unfilled and flled bars denote HN-ONLY and tiffed to analyze and visualize results according to
ALL-MODE, respectively. the z - z_ coordinate (Fig. 4).
300 .525- 550 400 425 45I0
_ 5[" __ _ .* ..0.2o0,o4m_//ss
% oL_ " " ." _.o._°__.
140 150 60 170 I(}' IO' ll) IO' IO' II) I(l* IO 2 IO' M5 90 _5
IWC (Ing/nl _) N (I/L'_I d N/d1 (1t_I_} RH I'll
350 575 400 425 z_4_)--
•Cott_
lOf(b ) The worm cases •OeMott ]
,Jensen j
oSosse_ j
140 _50 60 '70 IO' Ill I I11 11)_ [111 Ill I11 _ I112 1111 145 911 95
RH (%) at zb IWC (ung/nl*l N, (I/_ ) d N,/dl (I1_c/_1 RH [',_
Figure 4: Ice water content (IWC), N_, ice particle
Figure 3: The RHi at cloud base Zb (Ni = 1liter -1) formation rate @t, and RH, oas functions of z - Zb.
and the corresponding ARHi, defined as the differ-
ence between peak RHi and RH, at zb (the HN- The triggering RHw range was reduced signifi-
ONLY cases). cantly, to less than 2% in Wh020L and 5% in Ch020L
in comparison to 3% and 8% in Wh020 and Ch020. CCN distribution. Phase 2 of the CPMC, now un-
The predicted Ni is only marginally affected. derway, examines the effects of varying aerosol dis-
At tile beginning of the nucleation stage ill tributions. Sensitivity of model results to CCN com-
Wh020L, ice particle fornlation rates by the four position is indirectly made by altering A.
T, lf nlodels are close. However, nlodels C and D
reach much larger RH_, that leads to larger instan- 5 ACKNOWLEDGMENT
taneous nucleation rates, and maintain the peak ice
formation rate longer than the other two models. The U.S. DOE ARM Program and NASA Radiation Sci-
Quite contrarily, the Ni curves of models D and L ences Program have provided direct support (RFL and
in Ch020L distinctly separate from those of models DS) for this GCSS project. We thank support from NSF-
C and S. This grouping incidentally coincides with ATM9632917 (PJD), NASA's Stratospheric Aerosol and
the grouping according to the haze size specifica- Gas Experiment (E J), NSF-ATM9528287 (KS) and DOE
grant DEF(30394ER61747 (KS).
tions. Large haze particles are more concentrated
than the corresponding equilibrium values in mod-
6 REFERENCES
els C and S. Yet, the nucleation regime in model S
was not sustained as long as in model C; a similar
DeMott, P. J., M. P. Meyers, and W. R. Cotton, Param-
finding is noted when comparing results of model D
eterization and impact of ice initiation processes rele-
and L. The results of model J feature slow ice par-
vant to numerical model simulations of cirrus clouds,
ticle formation rate, long nucleation duration, and
J. Atmos. Sci., 51, 77-90, 1994.
broader freezing haze number distribution. DeMott et al., The role of heterogeneous freezing nucle-
The above results indicate that nucleation dura-
ation in upper tropospheric clouds: Inferences from
tion time and the maximum nucleation rate achieved SUCCESS, Geophys. Res. Lett., 25, 1387-1390, 1998.
are the two key components in determining the final Heymsfield, A. J., and L. M. Miloshevich, Relative hu-
N,. These two factors are sensitive to the growth midity and temperature influences on cirrus forma-
rates of small ice crystals, which under the influence tion and evolution: Observations from wave clouds
of tile kinetic effect are sensitive to the deposition and FIRE lI, J. Atmo,s. Sci., 52, 4302-4326, 1995.
Jensen et al., Microphysical modeling of cirrus 1. Con>
coefficient, _3,. It weus found that varying _i from
parison with 1986 FIRE IF() measurements, ,1. Geo-
0.04 to 1 (Table 2) would result in about a factor of
phys. Res., 99, 10,421-10,442, 1994.
4,_5 (Wh020L) and 9_12 (Ch020L) variation in N,
Koop et al., A new optical technique to study aerosol
hy models C and L.
phase transitions: The nucleation of ice from H2S04
aerosols, J. Phys. Chem.A, 102, 8924-8931, 1998.
4 SUMMARY Lin, R.-F., 1997, A numerical study of the evolution of
nocturnal cirrus by atwo-dimensional model with ex-
plicit microphysics, Ph.D. thesis, 199 pp., Pennsylva-
Results of Phase 1 of CPMC projects show that the
nia State University, Aug. 1997.
predicted cloud properties strongly depend on up-
Meyers, M. P., P. J. DeMott, and W. R. Cotton. New pri-
draft speed. Significant differences are found in the
mary ice-nucleation parameterizations in an explicit
predicted N,. Detailed examination revealed that cloud model, J. Appl. Meteor., 31, 708-721, 1992.
the homogeneous nucleation formulation, aerosol Sassen, K., and (3. C. Dodd, Homogeneous nucleation
size specification, ice crystal growth (especially the rate for highly supercooled cirrus cloud droplets, .].
specification of the deposition coefficient for ice) and Atrnos. Sci., _5, 1357-1369, 1988.
water vapor uptake rate were the critical compo- Sassen, K., and S. Benson, Ice nucleation in cirrus
nents. These results highlight the need for new lab- clouds: A model study of the homogeneous and het-
oratory and field measurements to infer the correct erogeneous modes, Geophys. Res. Lett., 2Z 521-524,
2000.
values for critical quantities in tile cirrus regime.
Spice et al., Primary ice nucleation in orographic cirrus
No attempt was made to scrutinize the causes of
cloud: A numerical simulation of the microphysics,
differences in ALL-MODE simulations due to the
Quart. J. Royal. Met. Soc., 125, 1637-1667, 1999.
substantial differences in formulation of heteroge-
Starr et al., Comparison of Cirrus Cloud Models: A
neous nucleation. Nevertheless, it was confirmed
Project of the (3EWEX Cloud System Study (GCSS)
that the expected effect of a heterogeneous nucle- Working Group on Cirrus Cloud Syst.ems. this con-
ation process is to decrease AT,and the RH, required ference, 2000.
for cloud initiation. Clearly, new measurenmnts of Tabazadeh, A., E. J. Jensen, and O. B. Toon, A model
ice nuclei activation in cirrus conditions are war- description for cirrus cloud nucleation from homoge-
ranted. neous freezing of sulfate aerosols, J. Geophys. Res.,
CPMC Phase 1 was conducted based on a single 102, 23,845-23,850, 1997.