ACPAtmospheric Chemistry and PhysicsACPAtmos. Chem. Phys.1680-7324Copernicus GmbHGöttingen, Germany10.5194/acp-15-10453-2015Ice phase in altocumulus clouds over Leipzig: remote sensing observations and detailed modelingSimmelM.simmel@tropos.deBühlJ.https://orcid.org/0000-0002-0354-3487AnsmannA.TegenI.https://orcid.org/0000-0003-3700-3232TROPOS, Leibniz Institute for Tropospheric Research, Permoser Str. 15, 04318 Leipzig, GermanyM. Simmel (simmel@tropos.de)24September20151518104531047019December201419January201524August20159September2015This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://acp.copernicus.org/articles/15/10453/2015/acp-15-10453-2015.htmlThe full text article is available as a PDF file from https://acp.copernicus.org/articles/15/10453/2015/acp-15-10453-2015.pdf
The present work combines remote sensing observations and detailed cloud
modeling to investigate two altocumulus cloud cases observed over Leipzig,
Germany. A suite of remote sensing instruments was able to detect primary ice
at rather high temperatures of -6 ∘C. For comparison, a second
mixed phase case at about -25 ∘C is introduced. To further look
into the details of cloud microphysical processes, a simple dynamics model of
the Asai-Kasahara (AK) type is combined with detailed spectral microphysics
(SPECS) forming the model system AK-SPECS. Vertical velocities are prescribed
to force the dynamics, as well as main cloud features, to be close to the
observations. Subsequently, sensitivity studies with respect to ice
microphysical parameters are carried out with the aim to quantify the most
important sensitivities for the cases investigated.
For the cases selected, the liquid phase is mainly determined by the model
dynamics (location and strength of vertical velocity), whereas the ice phase
is much more sensitive to the microphysical parameters (ice nucleating
particle (INP) number, ice particle shape). The choice of ice particle shape
may induce large uncertainties that are on the same order as those for the
temperature-dependent INP number distribution.
Introduction
According to altocumulus and altostratus clouds
together cover 22 % of the Earth's surface. For single-layered altocumulus
clouds, observations by show the typical feature with a
maximum of liquid water in the upper part of the cloud (increasing with
height) and an ice maximum in the lower part of the cloud, mostly below
liquid cloud base down in the virgae; this was previously reported from
and . also
emphasized a lack of significant temperature inversions or wind shears as a
major feature of these clouds. show that the ratio of
ice-containing clouds increases with decreasing temperature. However, the
numbers are different for different locations with similar dynamics but with
different aerosol burden, e.g., at northern and southern midlatitudes,
underlining the question for the influence of ice-nucleating particles (INPs).
The observations with the highest temperatures are close to the limit at
which the best atmospheric ice nuclei are known to nucleate ice in the
immersion mode. This can only be attributed to the aerosol particles that
are formed out of or at least contain biological material such as bacteria
, fungi, or pollen. This is corroborated by the review of
stating that only biological particles are known to form ice
above -15 ∘C. However, these observations are from laboratory
studies and it is still unclear whether or to what extent these extremely
efficient ice nuclei are abundant in the atmosphere, especially above the
boundary layer. One idea is that freezing is caused by soil dust with
biological particles dominating the freezing behavior ,
which could explain on the one hand the atmospheric abundance of biological
material and on the other hand the relatively high freezing temperatures
above -15 ∘C of ambient measurements. Seeding from ice clouds
above can be excluded for the cases presented, which means that ice has formed
at the cloud temperatures observed.
Ice nucleation still is a large source of uncertainty in cloud modeling.
Recently, several studies use combinations of vertically fine-resolved models
with rather detailed representation of the ice nucleation processes. Often,
wave clouds are used as comparison since they represent rather ideal
conditions when they are not influenced by ice seeding from layers above.
applied a 1-D kinematic model with bulk microphysics but
prognostic INPs. use a Langrangian parcel model for the
comparison of the ice nucleation schemes of and
under certain constraints. A 1-D column model with a very
detailed 2-D spectral description of liquid and ice phase is employed by
. used a 1.5-D model with spectral microphysics
for shallow convective clouds for a sensitivity study of immersion freezing
due to bacteria and its influence on precipitation formation.
Most ice microphysics descriptions are lacking in models from the fact that ice
nuclei are not represented as a prognostic variable. These models diagnose
the number of ice particles based on thermodynamical parameters such as
temperature and humidity and are, therefore, not able to
consider whether INPs were already activated at previous time steps in the
model.
However, despite its important contribution, ice nucleation does not
determine the entire microphysics of mixed-phase clouds alone. It is rather
the complex transfer between the three phases of water: water vapor, liquid
water and ice described by the Wegener–Bergeron–Findeisen (WBF) mechanism
. It is well-known that due to the
different saturation pressures of water vapor with respect to liquid water
and ice, a mixed-phase cloud is in a non-equilibrium state that,
nevertheless, may lead to a quasi-steady existence . The
main drivers for this phase transfer are vertical velocity (leading to
supersaturation and subsequent droplet formation) and ice particle formation
and growth (WBF starts) leading to sedimentation of the typically fast
growing ice particles (WBF ends due to removal of ice). The motivation of
this work is to shed more light on the relative contributions of the
different processes involved in these complex interactions.
The paper is structured as follows. Section describes the
remote sensing observations of two mixed-phase altocumulus cloud cases above
Leipzig. The dynamical model, as well as the process descriptions and initial
data used for this study, is specified in Sect. .
Section refers to changes in the dynamic parameters of the
model to identify base cases, which describe the observations sufficiently
well to perform sensitivity studies with respect to microphysical parameters.
The results for those sensitivity studies are presented in
Sect. and Sect. closes with a discussion of the
results.
Remote sensing observations
Altocumulus and altostratus clouds are regularly observed with the
Leipzig Aerosol and Cloud Remote Observations System (LACROS) at the Leibniz Institute for
Tropospheric Research (TROPOS).
LACROS combines the capabilities of Raman/depolarization lidar ,
a MIRA-35 cloud radar , a Doppler lidar ,
a microwave radiometer, a sun-photometer and a disdrometer to measure
height-resolved properties of aerosols and clouds. The Cloudnet framework
is used to derive microphysical parameters like
liquid-water content or ice-water content .
The following two cases have been selected to illustrate this variety
and to serve as examples to be compared to model results.
Case 1: warm mixed-phase cloud
One of the warmest mixed-phase clouds within the data set was observed on
17 September 2011 between 00:00 and 00:22 UTC (see
Fig. ). The liquid part of the cloud extends from about
4250 to 4450 m height at temperatures of about -6 ∘C according to
the GDAS (Global Data Assimilation System) reanalysis data for Leipzig.
Liquid water content (LWC) is between 0.1 to 1 g m-3, whereas ice
water content (IWC) is about 3–4 orders of magnitude smaller and reaches its
maximum value within the virgae (see Fig. ). Liquid water
path (LWP) measured by a microwave radiometer varies between 20 and
50 g m-2 (mostly about 25 g m-2), whereas ice water path (IWP)
is only slightly above the detection limit of about 0.01 g m-2
implying a rather large uncertainty with correspondingly large error bars.
Virgae (falling ice) are observed down to about 3000 m, which is close to
the 0∘C level. This is supported by Fig.
where the cloud radar (right panel) mainly shows particles falling from the
top layer. Therefore, particles are mainly moving downwards (green color) and
can be identified as ice particles by their size. Only at the very top (at
about 4300 m) are particles small enough to still be lifted upwards (yellow
colors). The Doppler lidar (left panel), however, shows the motion of small
cloud droplets at the predominantly liquid cloud top. Hence, in this plot the
cloud-top turbulence becomes visible. Vertical wind speeds range from about
-1.5 to 1.0 m s-1 with probability density function (pdf) maxima at -0.5 and
0.5 m s-1, respectively (Fig. ).
Lidar and radar observations on 17 September 2011 (case 1). Left:
lidar range-corrected 1064 nm signal (in logarithmic scale, arbitrary units
a. u.); right: radar reflectivity. The dashed box denotes the region for which
case 1 observations are shown in the following figures.
Cloudnet derived water contents for case 1. Left: liquid water
content; right: ice water content (both in logarithmic scale).
Vertical velocity for case 1. Left: derived from lidar (valid for
more numerous smaller droplets at cloud base); right: derived from radar
observations (valid for large particles; virgae).
Case 2: colder mixed-phase cloud
A much colder case was observed on 2 August 2012 between 21:00 and 21:40 UTC
(see Fig. ). Liquid water was measured around 7500 m at
about -25 ∘C with a LWP between 10 and 30 g m-2 and a LWC
of up to 0.1 g m-3 that is much smaller than case 1. As can be
expected due to the lower temperature, the ice phase was much more massive
than in case 1 and reached down to about 5500 m with an IWP of about
1–10 g m-2 and an IWC of up to 0.01 g m-3, which means that in
some parts of the cloud, ice and liquid water reach the same order of
magnitude (see Fig. ). Vertical wind speeds were in the
same range as in the warmer case described above
(Fig. ).
Accuracy of the IWC is ± 50 %. For the LWC calculated by the scaled
adiabatic approach, the same order of magnitude applies. Vertical wind speeds
are measured directly by evaluation from the recorded cloud radar and Doppler
lidar spectra. Errors are ± 0.15 m s-1 for the cloud radar and
± 0.05 m s-1 for the Doppler lidar. These errors are mainly due
to the pointing accuracy of the two systems.
Lidar and radar observations on 2 August 2012 (case 2). Left: 532 nm
attenuated backscatter coefficient; right: radar reflectivity.
Cloudnet derived water contents for case 2. Left: liquid water
content; right: ice water content (both in logarithmic scale).
Vertical velocity for case 2. Left: derived from lidar (valid for
more numerous smaller droplets at cloud base); right: derived from radar
observations (valid for large particles; virgae).
Model description and initialization
For the model studies, an Asai–Kasahara (AK) type model is used . The
model geometry is axisymmetric and consists of an inner and an outer cylinder
with radii of 100 and 1000 m, respectively, resulting in a radius ratio of
1:10, which is typical for this setup. However, the geometric configuration
of the model is not intended to match the geometry of the clouds (and the
cloud-free spaces between the clouds) but is rather meant to provide the
possibility of horizontal exchange between clouds and a cloud-free
background.
The vertical resolution is
constant with height and is chosen to be Δz=25 m to give a sufficient resolution of the
cloud layer and to roughly match the vertical resolution of the observations.
In contrast to a parcel model, the vertically resolved model grid allows for a
description of hydrometeor sedimentation. This is important especially for
the fast growing ice crystals to realistically describe their interaction
with the vapor and liquid phase (Wegener–Bergeron–Findeisen process).
A time step of 1 s was used for the dynamics as well as for the microphysics.
However, in contrast to other Asai–Kasahara model studies, updrafts are not initialized by
a heat and/or humidity pulse in certain layers for a given period of time.
Instead, vertical velocity (updrafts and downdrafts) in the inner cylinder is prescribed,
which is more similar to a kinematic model like the Kinematic Driver model (KiD) .
In that way dynamics can be controlled to make sure that it is close to the observations.
The cloud microphysics is described by the mixed-phase spectral microphysics module SPECS
. SPECS provides a joint spectrum for the liquid phase (soluble
wetted aerosol particles as well as cloud and rain drops) and one spectrum for the ice phase.
For this case study, collision processes between ice particles and drops
(riming) and between ice particles and ice particles (accretion) are not
taken into account. On the one hand, this is to exclude further
uncertainties that would be introduced by the collision/collection kernel
for those interactions; on the other hand, only small or neglectable effects
are expected. Clouds are shallow which means that there is not much time for
the ice particles to interact with droplets (especially when the ice is
preferentially formed near cloud base and sediments out soon). In addition,
for case 1 ice particle concentrations are low, which highly limits the
probability of collisions. At the low temperatures of case 2 sticking
efficiency is expected to be low. This assumption is corroborated by the
findings of stating that water vapor deposition (and
sublimation), balanced by sedimentation are more important than accretional
growth.
Description of ice microphysics
In the following, the differences in the description of the microphysics compared
to are described.
Immersion freezing
For this study, immersion freezing is assumed to be the only primary ice
formation process. Since during the above-mentioned observations no in situ
measurements of the INPs are available, the parameterization of
is used assuming that all INPs are active in the immersion
freezing mode. The parameterization of is based on an
empirical relation of INPs and the number of aerosol particles with radii >250 nm (NAP,r>250nm). To cover case 1, the
parameterization is extrapolated to -5 ∘C despite the fact that
the underlying measurements were only taken at -9 ∘C and below. As
base case NAP,r>250nm=105 kg-1 air is used as
input data for the parameterization resulting in about 0.01 active INPs per
liter for -6 ∘C (case 1) and about 0.5 INPs per liter for
-25 ∘C (case 2) at standard conditions.
This corresponds to a relatively low number of larger aerosol particles but is well within the range
observed by .
For the potential INPs a prognostic temperature resolved field with 20
temperature bins with a resolution of 1 K is introduced into SPECS. It
ranges from -5 to -25 ∘C to cover the temperature range for the
selected cases and can easily be changed for other case studies. This is a
simplified version of the method used by . The potential
INP field is initially defined in every grid cell (layer) and is transported
vertically with the given up-/downdrafts and horizontally exchanged between
inner and outer cylinder in the same way as the other hydrometeor fields
(drops and ice crystals). Immersion freezing occurs as soon as liquid drops
above a certain size limit are present and the temperature of a certain
potential INP bin is reached. Then the respective amount of drops freezes (if
available) instantaneously and is transferred from the liquid to the frozen
spectrum. If more drops larger than the size threshold of 10 µm
than active INPs are present, the INPs are distributed evenly over all drop
size bins above the threshold value. The drop size threshold was chosen to
restrict freezing to droplets and to prevent (large) non-activated aerosol
particles at high relative humidity (but subsaturated with respect to water) outside the
cloud from freezing. If ice crystals melt below the freezing level, they
contribute to the potential INP field at that level.
Ice particle shape
It is well known that ice particle shape highly influences water vapor
deposition (described by changing the capacitance of the particle) as well as
terminal fall velocity of the ice particle. Therefore, instead of the
previously chosen spherical ice particle shape, ice particles now can be
prescribed as hexagonal columns or plates. The aspect ratio can be either
constant for all size bins or be changed with size following the approach of
. Typically, with increasing particle size, the deviation
from an uniform aspect ratio increases. In our simulations, a constant
uniform aspect ratio (ar=1) is used as base case. From
the size-varying aspect ratios for plates (ranging from 15 to
3000 µm with a single description) and columns (for size ranges of
30 to 100 µm, 100 to 300 µm, and above 300 µm
in diameter) are calculated from the mass-dimension power laws and used for
sensitivity studies.
The (relative) capacitance needed for the calculation of deposition growth of the ice
crystals is modeled using
the method of for the aspect ratios given above.
Ice crystal terminal fall velocities are calculated according to
using the same aspect ratios.
Model initializationThermodynamics
The Asai–Kasahara model has to be initialized with vertical profiles of
temperature and dew point temperature either from reanalysis data (here GDAS)
or radiosonde (RS) profiles from nearby stations (here Meiningen, Thuringia).
Figure shows profiles of temperature and relative humidities with
respect to liquid water and to ice, respectively, for both cases. For case 1,
profiles from both methods show a similar general behavior but the
radiosonde profile of Meiningen measured at 00:00 UTC is used since it
provides a finer vertical resolution than the GDAS reanalysis data (cp.
Fig. ). However, for case 2 the Meiningen RS profile misses the
humidity layer at the level where the clouds were observed. This means that
the profile is not representative for the given meteorological situation.
Therefore, GDAS reanalysis data for Leipzig at 21:00 UTC were chosen.
Finally, both profiles used show a sufficiently humid layer where the clouds
were observed, so that the lifting of these layers leads to supersaturation and
subsequent cloud formation.
Vertical profiles of temperature (left) and relative humidity
(right) with respect to liquid water (full lines) and ice (dashed lines)
based on a radiosonde observation (Meiningen) for case 1 (black) and from
GDAS (grid point Leipzig) for case 2 (red).
As mentioned above, vertical velocity (updrafts and downdrafts) in the inner
cylinder is prescribed at cloud level ranging from hbot to
htop. The center of this interval is given by hmid=(htop+hbot)/2 and its half-depth by hdepth=(htop-hbot)/2. hbot ranges from 3800 to
4100 m for case 1 and from 7000 to 7300 m for case 2. The respective values
for htop are 4500 and 7700 m. The vertical dependency (cf.
Fig. , left) is given by
fh(h)=hdepth2-(h-hmid)2hdepth2forhbot≤h≤htop
resulting in the time- and height-dependent function
w(h,t)=wmid(t)fh(h)forhbot≤h≤htop
and w(h,t)=0 otherwise, defining wmid(t) as the updraft
velocity at hmid. In order to match the observed wind field
distributions rather closely, wmid(t) is chosen as a stochastic
function
wmid(t)=wave+fscalδ(t)3|δ(t)|,
where wave is the average (large-scale) updraft velocity at
hmid varying between 0.1 m s-1 and 0.4 m s-1,
fscal is the scaling factor determining the range of updraft velocities
(chosen as 4 m s-1 to obtain a difference of minimum and maximum
velocity of 2 m s-1), and δ(t) is a random number ranging from
-0.5 to +0.5 obtained from a linear stochastic process provided by FORTRAN.
After 30 s model time, a new δ(t) is created. Different realizations
of the stochastic process are tested (see below). For example, wmid(t)
ranges from -0.7 m s-1 to 1.3 m s-1 if
wave=0.3 m s-1 and fscal=4 m s-1 as it
is shown in the temporal evolution and the histogram in
Fig. .
Vertical velocity field of the inner cylinder for case 1. Left:
height dependence (red line) and temporal evolution of one realization of the
stochastic vertical velocity field (black line) for
wave=0.3 m s-1 at hmid. Right: histogram of
velocity field. Vertical velocity for case 2 is identical but for heights
between 7100 and 7700 m.
Due to the height dependent vertical velocity w, a horizontal transport
velocity uk (exchange between inner and outer cylinder) is induced in the
Asai–Kasahara formulation for a given model layer k.
uk=-wk+12ρk+12-wk-12ρk-12frΔzρk
Full indices k indicate values at level centers whereas half indices (k+12,
k+12) describe values at level interfaces. fr=2/ri is a geometry parameter with
the radius ri=100 m of the inner cylinder.
The prescribed velocity field leads to the following effects (all descriptions are related to
the inner cylinder if not stated otherwise explicitly):
In the updraft phase: in the upper part (between hmid and htop) of the updraft,
mixing occurs from the inner to the outer cylinder, whereas in the lower part
(between hlow and hmid) horizontal transport is from the
outer cylinder into the inner one.
For downdrafts it is the other way: this means that below hmid
drops and ice particles are transported from the inner cylinder to the
outer one and are therefore removed from the inner cylinder.
below hlow or above htop, no horizontal exchange takes place.
The question arises to which extent this dynamical behavior reflects
the real features of the observed clouds and whether this is critical for the topics
aimed at in this study.
Prescribing vertical velocity in any way also means that a feedback of microphysics
on dynamics due to phase changes
(e.g., release of latent heat for condensing water vapor or freezing/melting processes)
is not considered by the model.
Aerosol distribution
Since no in situ aerosol measurements are available, literature data are used.
The Raman lidar observations do not show any polluted layers for both cases;
therefore, data from LACE98 are used which should be
representative for the free troposphere over Leipzig. For case 1 values for
the lower free troposphere (M6), for case 2 those from the upper free
troposphere (M1), are used see Table 6.
Model results: dynamics
In a first step, the aim is to achieve a sufficient agreement concerning
macroscopic cloud features, as well as (liquid phase) microphysics, as far as
they were observed. The following parameters describing model dynamics
(updraft velocity) are varied to identify a “best case”, which in the second
step can be used to perform sensitivity studies with respect to (ice)
microphysics (see also Tables and ).
hlow: ranging from 3800 to 4100 m for the warmer and from 7000 to 7300 m for the colder case, this parameter influences the vertical cloud extent and,
therefore, liquid water content and liquid water path.
wave: ranging from 0.1 to 0.4 m s-1. Higher average updraft also leads to higher LWC.
Due to the lateral mixing processes the model setup requires a positive updraft velocity in average to form and maintain clouds.
δ: four different realizations of the stochastic process are used. This influences the timing of the cloud occurrence as well as LWC and LWP but not systematically.
All model results shown refer to the inner cylinder.
Overview of the model results for the dynamic sensitivity runs for
the warmer case 1 (maximum values of L/IWC: liquid/ice water content; L/IWP:
liquid/ice water path; CDN: cloud drop number; IPN: ice particle number).
RunParameter valueLWCIWCLWPIWPCDNIPNdiffering from base caseg m-310-3 g m-3g m-210-3 g m-2cm-3L-1W_base–0.3550.37941.3362.2746.890.0197W_h38hbot=3800 m0.4260.40857.0573.1148.630.0235W_h40hbot=4000 m0.2890.35728.5858.1261.480.0240W_h41hbot=4100 m0.2190.32418.2345.8159.530.0208W_w01wave=0.1 m s-10.1870.20017.4131.7343.360.0138W_w02wave=0.2 m s-10.2970.30032.8647.1854.570.0175W_w04wave=0.4 m s-10.3820.44844.4878.2552.660.0219W_r1stoch. realiz. r10.3360.31640.3254.8564.260.0163W_r3stoch. realiz. r30.3810.31442.8854.4843.030.0167W_r4stoch. realiz. r40.3460.24540.9146.9347.420.0151
Overview of the model results for the dynamic sensitivity runs for
the colder case 2 (maximum values of L/IWC: liquid/ice water content; L/IWP:
liquid/ice water path; CDN: cloud drop number; IPN: ice particle number).
RunParameter valueLWCIWCLWPIWPCDNIPNdiffering from base caseg m-3g m-3g m-2g m-2cm-3l-1C_base–0.3770.04129.3510.7170.560.462C_h70hbot=7000 m0.4520.04843.0611.3471.330.432C_h72hbot=7200 m0.2960.03518.7110.1190.510.396C_h73hbot=7300 m0.2150.02810.549.2777.610.337C_w01wave=0.1 m s-10.2190.04017.198.0176.980.292C_w02wave=0.2 m s-10.3160.04425.899.4274.400.415C_w04wave=0.4 m s-10.4020.04530.5811.8598.370.439C_r1stoch. realiz. r10.3660.02329.376.5786.640.257C_r3stoch. realiz. r30.3990.04630.229.9579.650.341C_r4stoch. realiz. r40.3730.04929.538.3395.890.419Case 1: warm mixed-phase cloud
Figures and show time-height plots of the
liquid- (contours, linear scale) and ice-water (colors, logarithmic scale)
content for case 1 illustrating the cloud sensitivity with respect to
variation of cloud base (hbot), average vertical updraft
(wave), and the realization of the stochastic process. Liquid clouds form in the updraft regions (cp.
Fig. ), whereas in the downdrafts the liquid phase vanishes
at least partly. If active INPs are available ice formation can take place
within the liquid part of the cloud. The INPs are partly already active near
liquid cloud base, which means that they trigger freezing as soon as the
droplets are formed. Less efficient INPs become active after further cooling
above cloud base. After ice formation rapid depositional growth takes place
and the ice particles almost immediately start to sediment. Due to the
supersaturation with respect to ice even below liquid cloud base, ice
particles still grow while sedimenting, reaching their maximum size before,
finally, subsaturated regions are reached and sublimation sets in.
Figures and show the time evolution of
liquid (lower panel) and ice water path (upper panel) for the same parameters
varied, reflecting the same temporal patterns. Table
summarizes the maximum values for LWC/IWC,
LWP/IWP as well as cloud droplet and ice particle
number concentration (CDN/IPN) for all dynamics sensitivity runs for case 1.
LWC (contours) and IWC (colors, logarithmic scale) for case 1.
Comparison of different values for hbot (upper left: W_base,
hbot=3900 m; upper right: W_h38, hbot=3800 m;
lower left: W_h40, hbot=4000 m; lower right: W_h41,
hbot=4100 m.)
Liquid (lower panel) and ice water paths (upper panel) for case 1.
Comparison of the different values for hbot.
LWC (contours) and IWC (colors, logarithmic scale) for case 1.
Comparison of different average updraft velocities wave (upper
panel: left: W_w01, wave=0.1 m s-1; right: W_w04,
wave=0.4 m s-1) and different stochastic realizations
(lower: left: W_r1, r1; right: W_r4, r4).
Liquid (lower panels) and ice water paths (upper panels) for case 1.
Comparison of the different values for wave (left) and the
different stochastic realizations (right).
One can clearly observe, that a lower hbot (Fig. )
results in a lower cloud base, larger vertical cloud extent as well as more
liquid water. The LWC maxima are within a factor of 2 for varying
hbot. A similar trend is observed for the ice phase (see also
Fig. ), but IWC maxima differ only by about 25 %. However,
the values of the two maxima of the condensed phase after about 15–20 min
and about 40 min model time are quite different. The first maximum is more
pronounced for the ice phase whereas the second one is larger for the liquid
phase. While the liquid phase is dominated by the updraft velocity (see
Fig. ) the ice phase additionally depends on INP supply. In
the first ice formation event at 15 min, all INPs active at the current
temperature actually form ice leading to an INP depletion. Due to the
horizontal exchange with the outer cylinder the INP reservoir is refilled,
but only to a certain extent when the second cloud event after 40 min sets
in. Due to the limited INP supply, the second ice maximum is weaker than the
first one. The stochastic velocity fluctuations cause fluctuations in
relative humidity, which are directly reflected by the liquid phase
parameters, whereas the ice phase generally reacts much slower. Sensitivity of
CDN and IPN with respect to change of hbot does not seem to be
systematic.
Increasing the average updraft velocity, wave leads to a similar
increase of liquid water and ice as lowering hbot (see
Figs. , upper panel and , left). This can
be expected since more water vapor flows through the cloud and is able to
condense. However, a certain limit seems to be reached for W_w04, since the
increase of LWP slows down (see maximum value at 40 min in
Fig. , left). This is due to the enhanced horizontal
exchanged following Eq. (). Additionally, the stronger updrafts
allow the ice particles to have a longer presence time in the vicinity of the cloud
and, therefore, an enhanced growth at comparably high supersaturation with
respect to ice before sedimentation sets in at larger sizes. This also leads
to an accumulation of ice particles and, therefore, to a higher IPN.
Surprisingly, CDN depends only weakly and not systematically on
wave, which is in contrast to the typical enhancement of CDN with
increasing updraft velocities.
Figures (lower panel) and (right) show
that different realizations of the stochastic process (as explained above in
Sect. ) lead to different temporal cloud evolutions. However,
differences in maximum LWP and LWC are much smaller than those discussed
above. Variations in maximum IWP and IWC, as well as CDN and IPN, are in the
range of about 30 %. This is also true for average LWP ranging from
18 g m-2 for W_r1 to 26 g m-2 for W_r3. However, despite the
different maxima and temporal evolutions of IWP, average IWP is almost
identical for the different stochastic realizations (0.023 g m-2).
This shows that changing the stochastic realization influences cloud
evolution in detail (timing) but does not change the overall picture.
With maximum values between 17 and 57 g m-2, the modeled liquid water
path is in the same range as the observed values (20–50 g m-2),
especially for the “wetter” runs (smaller hbot, larger
wave). Average LWP typically is about half (40–60 %) of the
maximum value for most of the runs, which also fits well into the
observations. Ice forms within the liquid layer and sediments to about
3800 m for most runs, which is less than for the observations. The (maximum)
modeled ice mixing ratio is in the same order of magnitude as the observed
one (about 10-7 kg m-3). The same holds for the ice water path
with values of about 0.01 g m-2 for both, model and observation. For
the other values, no observational data are available for comparison.
Case 2: colder mixed-phase cloud
Due to the lower temperatures of case 2 much more INPs are active and much
more ice is produced than in case 1 (see
Figs. – as well as
Table ). This also means that near the cloud base much more
active INPs are available and that a further cooling within the clouds only
slightly increases the number of active INPs leading again to a preferential
ice nucleation near the liquid cloud base. Due to the lower temperatures and the
more massive ice formation, the virgae reach down to more than 1500 m below
liquid cloud base, which is in concordance with the observations. The
principal behavior with respect to the sensitivity parameters is similar to case 1: the liquid phase is enhanced by either decreasing hbot
or increasing wave, showing the “saturation” effect slightly
more pronounced as in case 1. Different stochastic realizations only weakly
influence the maximum and average values of the liquid phase but change the
timing of occurrence. Generally, the variability of the ice phase is weaker
than in case 1. The different stochastic realizations show the highest
variability in IWC and IWP. Different variations of hbot show
almost identical IWPs,
whereas changing wave at least slightly influences maximum IWC and IWP,
which again can be attributed to the ice particle accumulation in the
updraft. Liquid water path is smaller than in case 1 and reaches maximum
values between 10 and 43 g m-2, which well covers the observed maximum
value of about 20 g m-2. Cloudnet observations show an IWC of
10-7–10-5 kg m-3, which is an increase by a factor of
10–100 compared to case 1. Similar values are obtained by the model results
underlining the strong temperature dependency of the ice nucleation process.
LWC (contours) and IWC (colors, logarithmic scale) for case 2.
Comparison of different values for hbot
(upper left: C_base, hbot=7100 m; upper right: C_h70,
hbot=7000 m;
lower left: C_h72, hbot=7200 m; lower right: C_h73,
hbot=7300 m).
Liquid (lower panel) and ice water paths (upper panel) for case 2.
Comparison of the different values for hbot.
Sensitivity studies
In the previous section it could be shown that dynamical parameters can be
chosen in a way that the model results (in terms of LWP, IWP as well as cloud
geometry) are in good agreement with the observations. This allows one to perform
sensitivity studies with respect to cloud microphysics. To cover the proper
sensitivities, we have to answer the question of which microphysical parameters
are expected to have a large influence on mixed phase microphysics and are
rather uncertain to be estimated. This leads to a (temperature-dependent) INP
number (NINP) that directly influences the ice particle number but
mostly is poorly known. To be consistent with the freezing parameterization
of the model, NINP is varied by changing
NAP,r>250nm, which additionally is easier to observe in
most cases. A second parameter is the shape of the ice particles that does
not influence the primary freezing process but the subsequent growth by water
vapor deposition onto existing ice particles and, therefore, the total ice
mass produced. Their relative importance shall be quantified and also be
compared to the influence of dynamics discussed above.
LWC (contours) and IWC (colors, logarithmic scale) for case 2.
Comparison of different average updraft velocities wave (upper left: C_w01, wave=0.1 m s-1, right: C_w04,
wave=0.4 m s-1) and the different stochastic realizations
(lower left: C_r1, r1, right: C_r4, r4).
INP number
Changing NAP,r>250nm leads to a temperature-dependent
change of INP number, which is relatively small for warmer conditions.
However, the effect increases with decreasing temperature. This is
illustrated by the following numbers. The parameterization of
gives about 0.009 active INPs per liter at standard
conditions (NINP) when
NAP,r>250nm=105 kg-1 at T=-5∘C. A
tenfold increase to NAP,r>250nm=106 kg-1 results
in about 0.012 active INPs per liter, which is a rise of only about 35 %. For
T=-7∘C, INP number rises by about 65 % for a tenfold increase
of NAP,r>250nm. This shows that for those rather high
temperatures considered for case 1, a massive change in
NAP,r>250nm leads to relatively small changes in
NINP and only a small effect on the ice phase can be expected. This
is confirmed by Fig. (left) showing liquid and ice water
contents for W_in6. IWC is enhanced by less than 60 % for W_in6 and by
about 160 % for W_in7, which is consistent for the given temperature range
(see Table ). Similar values are obtained for the change in
IPN. This directly leads to the conclusion that the individual ice particles
grow independently from each other. Their individual growth history is (in
contrast to drop growth) only influenced by thermodynamics as long as their
number is low enough, which seems to be the case here.
Overview of the model results for the microphysical sensitivity runs
for the warmer case 1 (maximum values of L/IWC: liquid/ice water content,
L/IWP: liquid/ice water path, CDN: cloud drop number, IPN: ice particle
number).
RunParameter valueLWCIWCLWPIWPCDNIPNdiffering from base caseg m-310-3 g m-3g m-2g m-2cm-3l-1W_in6NAP,r>250nm=106 kg-10.3540.61941.310.1046.690.0296W_in7NAP,r>250nm=107 kg-10.3541.00041.240.1741.610.0450W_colice shape: columns0.3531.83041.200.2742.900.0257W_plaice shape: plates0.3532.85041.130.4543.410.0267
This is confirmed by Fig. showing drop and ice particle size
distributions at the time when the maximum IWP is reached (16 min for
case 1, 17 min for case 2). For case 1 (upper panel), the liquid phase
(contours) is unaffected by the INP enhancement. Despite the increase of ice
particle number and mass, the shape of the ice particle size distribution
(colors) is not changed. The smallest ice particles can be observed at three
discrete height (and temperature) levels caused by the temperature resolved
parameterization of the potential INPs described in Sect. . In
reality this part of the spectrum showing rather freshly nucleated and fast
growing ice particles should be continuous over the height range from about
4100 to 4400 m. Nevertheless, the total number of ice particles formed is
described correctly.
Liquid (lower panels) and ice water paths (upper panels) for case 2.
Comparison of the different values for wave (left) and
the different stochastic realizations (right).
LWC (colors) and IWC (contours, logarithmic scale) for case 1
(W_in6, left) and case 2 (C_in6, right). Enhancing IN by increasing
NAP,r>250nm by a factor of 10.
LWC (contours) and IWC per bin (colors, both logarithmic scale) for
case 1 (upper panel) and case 2 (lower panel) for the respective base
case (left) and the case with enhanced IN number (right; in6) after 16 and 17
minutes model time, respectively, corresponding to the IWP maximum of the
base case runs.
One can conclude that increasing INP number therefore increases ice particle
number as well as ice mass proportionally. Generally, the ice mass remains
small and the liquid phase is not affected by the ice mass increase. Those
results are supported by Fig. (left) showing an unchanged
LWP and a proportionally growing IWP for increased INP numbers.
For the colder case 2 the parameters are varied in the same way. However, one
big difference is that a tenfold increase of
NAP,r>250nm at T=-25∘C results in a much
larger change in active INPs. Their number rises by 300 % from about 0.5 per
liter to about 2 per liter following the parameterization. This is reflected
by the IPN values in Table . Figure (right)
and Table show that ice mass increases in such a way that
liquid water is depleted partially (C_in6 by about 50 %) or almost totally
(C_in7) due to the Wegener–Bergeron–Findeisen process. Compared to C_base,
ice is enhanced by a factor of 3–4 for C_in6 and about 10 for C_in7
whereas IPN increases by a factor of 12. This can also be seen in the IWP
(Fig. , right, red lines) showing a limited increase for
C_in7, especially for the first maximum after 17 min. This means that the
results for C_in6 are still consistent with an independent growth of the
individual ice particles (as described above) despite the relatively high ice
occurrence.
Overview of the model results for the microphysical sensitivity runs
for the colder case 2 (maximum values of L/IWC: liquid/ice water content;
L/IWP: liquid/ice water path; CDN: cloud drop number; IPN: ice particle
number).
This is verified by the size distributions in Fig. (lower
panel). As in case 1 the ice particle size distributions only differ by the
number/mass, but not by shape. Additionally, the decrease in the liquid phase
is reflected also in the drop spectrum showing a more shallow liquid part of
cloud as well as droplet distribution shifted to smaller sizes.
Liquid (lower panel) and ice water paths (upper panel) for case 1
(left) and case 2 (right). Comparison of the sensitivities with respect to IN
number and ice particle shape.
However, for C_in7 the ice particles compete for water vapor, which becomes clear from
(i) the depletion of liquid water (resulting in a lower supersaturation with respect to ice)
and (ii) the ice mass enhancement factor being below the value expected from
the ice nucleation parameterization and below that of IPN.
This means that despite the higher number of INPs and, therefore, ice particles, the amount of ice
is limited by the thermodynamic conditions that result in the production of
more but smaller ice particles, similar to the Twomey effect for drop activation.
As mentioned earlier,
ice particle growth is not only restricted to the liquid part of the cloud but also occurs in the
layer below liquid cloud base, which is still supersaturated with respect to ice. This leads
to a decrease in relative humidity in this part of the cloud, which in turn weakens or suppresses
droplet formation by shifting liquid cloud base to higher altitudes. The lower LWC for the runs with higher
IWC therefore cannot only be attributed to the WBF processes but also to this indirect effect.
Ice particle shape
As discussed previously, for most of the cases (except for C_in7) changing
the parameters in the section above does neither influence the ice particles
themselves nor their individual growth. Additionally, due to their low
number, there is almost no competition between the ice particles for water vapor,
and, therefore, ice water content scales linearly with ice particle number.
In contrast to this, changing the ice particle shape from quasi-spherical
(ar=1) to columns or plates with size-dependent axis ratios deviating from
unity results in an increase of water vapor deposition on the individual ice
particles leading to enhanced ice water content due to larger individual
particles when ice particle numbers remain unchanged. This is due to (i)
enhanced relative capacitance resulting in faster water vapor deposition and
(ii) lower terminal velocities of the ice particles leading to longer
residence times in vicinity of conditions with supersaturation with respect
to ice.
Figure (left) shows the results for the runs using
hexagonal columns (W_col) as prescribed ice particle type. Compared to the
previous results (W_base, W_in6, W_in7) more ice mass is produced (see
Table ) but still the liquid part of the cloud remains
unaffected (compare also LWP and IWP in Fig. , left).
Similar results are obtained for the assumption of plate-like ice particles
(W_pla). The mass increase results from the larger ice particle size due to
the reasons discussed above, which can be seen from Fig.
showing the size distributions for W_col at different times. On the upper
left panel W_col is shown after 16 min corresponding to
Fig. . Compared to W_base, larger ice particles are produced
leading to more ice mass (equivalent radius up to 300 µm compared
to 189–238 µm for the base case). Additionally, due to the lower
fall speed of the columns (1.03 m s-1 vs. 1.75–2.24 m s-1),
the maximum of the ice is at about 4200 m compared to 4100 m for the base
case. On the upper right panel, size distributions after 21 min are shown
corresponding to the IWP maximum of W_col. Ice particles have grown larger
(equivalent radius up to 378 µm, length of the columns increases
from about 3 to 4.5 mm) and sedimentation has developed further with
increasing terminal velocity (1.13 m s-1). Similar results are
obtained for plates (W_pla) with terminal velocities of
0.89–1.21 m s-1, equivalent radii of 300–476 µm, and
maximum dimension of 1.8–3.2 mm.
LWC (contours) and IWC (colors, logarithmic scale). Results for
changing ice particle shape to hexagonal columns for case 1 (W_col, left)
and case 2 (C_col, right).
The lower terminal velocity of columns and plates despite
their larger size is leading to the stronger tilting of the virgae.
Additionally, the IPN is enhanced by about 30 % although ice nucleation
is identical to W_base. This can be attributed to the lower fall velocities, too, leading to an accumulation
of ice particles.
The differences between W_col and W_pla are caused by both, the higher relative capacitances of
and lower terminal fall velocities of plates compared to columns
(at least when their axis ratios are chosen following ).
LWC (contours) and ice water mass per bin (colors, both logarithmic
scale) for case 1 (upper panel) and case 2 (lower panel) assuming columns as
ice particle shape at IWP maximum of the respective base case (left) and at
IWP of the run (right).
For case 2 (C_col and C_pla), the liquid water reduction due to the
Bergeron–Findeisen process is similar to C_in6 (see Fig. ,
right, and Table ). In contrast to the respective case 1
runs, less ice is produced than for C_in7. The tilting of the virgae is not
as strong as in W_col, which is due to the larger ice particle sizes leading
to higher terminal fall velocities (1.43–1.60 m s-1). Additionally,
the lower air density leads to an increase of terminal velocity of more than
10 % independently from shape. Figure (lower panel) shows
the size distributions for C_col at different times. Due to the longer
growth time larger individual ice particles than in case 1 are produced
(equivalent radius up to 600 µm compared to 300 µm for
the base case).
To decide whether independent ice particle growth or competition occurs,
further runs with less INPs (C_col_in4 and C_pla_in4) are discussed (see
Fig. , right). IWC and IWP of these runs (in4) are about
one-third of the values of the respective runs with more INPs (in5). For ice
particle number, a factor of slightly more than 3 occurs, which means that
a weak competition for water vapor occurs for C_col and C_pla resulting in
slightly smaller individual ice particles compared to C_col_in4 and
C_pla_in4.
Conclusions
The model system AK–SPECS was applied to simulate dynamical and microphysical processes
within altocumulus clouds. Sensitivity studies on relative contributions on cloud evolution as well as
comparisons to observations were made.
Variation of the dynamic parameters as it was done in
Sect. leads to systematic differences mainly in the liquid
phase (LWC, LWP), which can easily be explained. More liquid water is produced
when either cloud base is lowered (corresponding to a larger vertical cloud
extent) or vertical wind velocity is increased. However, the effects of the
dynamics on the ice phase are surprisingly small, at least smaller than those
on the liquid phase. Increasing vertical velocity leads to an accumulation of
the smaller ice particles in the enhanced updraft.
On the other hand, much larger differences in terms of IWC and IWP were found when
microphysical parameters like INP number or ice particle shape were varied under identical
dynamic conditions. This is valid for both cases studied. However, at least for the
ice nucleation parameterization used, sensitivity of INP number strongly increased with decreasing temperature.
This means that relatively large differences concerning the ice phase can
only be reached when either INP number differs considerably or ice particle
shape is different (which should not be the case for relatively similar
thermodynamical conditions). After for case 1 with
temperatures of about -6 ∘C, column-like ice particles with
ar=0.1 could be expected (corresponding to W_col), whereas for case 2
(T<-24∘C) hexagonal particles with ar=1 are most likely (e.g.,
C_base). Those ice shapes were observed in laboratory studies at water
saturation, which was also valid for the observed cases when ice formed by
immersion freezing within the liquid layer of the cloud. However, below
liquid cloud base supersaturation with respect to ice decreases. These ice
shapes can also explain why a depletion of the liquid phase was not observed
in case 2 as it was predicted by the sensitivity studies using either columns
or plates as prescribed shape. Generally, the liquid phase is affected
considerably only when enough ice particles are present, which typically is
the case for cold conditions with a sufficient amount of INPs and fast growing
ice particle shapes (most effective for large deviations from spherical
shapes).
Acknowledgements
This study was supported by the Deutsche Forschungsgemeinschaft (DFG) under
grant AN 258/15. We also acknowledge funding from the EU FP7-ENV-2013
programme “impact of Biogenic vs. Anthropogenic emissions on Clouds and
Climate: towards a Holistic UnderStanding” (BACCHUS), project no. 603445.
Edited by: T. Garrett
References
Althausen, D., Engelmann, R., Baars, H., Heese, B., Ansmann, A., Müller,
D.,
and Komppula, M.: Portable Raman Lidar PollyXT for Automated Profiling of
Aerosol Backscatter, Extinction, and Depolarization, J. Atmos. Oceanic
Technol., 26, 2366–2378,
2009.
Asai, T. and Kasahara, A.: A theoretical study of compensating downward
motions associated with cumulus clouds, JOURNAL OF THE ATMOSPHERIC
SCIENCES, 24, 487–496,
1967.
Bauer-Pfundstein, M. R. and Görsdorf, U.: Target Separation and
Classification Using Cloud Radar Doppler-Spectra, in: Proceedings of the 33rd
Conference on Radar Meteorology, 2007.
Bergeron, T.: On the physics of clouds and precipitation, in: Proces verbaux de
l'association de Météorologie, pp. 156–178, International Union of
Geodesy and Geophysics, Lisboa, Portugal, 1935.
Bühl, J., Ansmann, A., Seifert, P., Baars, H., and Engelmann, R.: Toward
a
quantitative characterization of heterogeneous ice formation with
lidar/radar: Comparison of CALIPSO/CloudSat with ground-based observations,
Geophys. Res. Lett., 40, 4404–4408,
2013.
Bühl, J., Engelmann, R., and Ansmann, A.: Removing the Laser-Chirp
Influence
from Coherent Doppler Lidar Datasets by Two-Dimensional Deconvolution, J.
Atmos. Oceanic Technol., 29, 1042–1051,
2012.
Carey, L. D., Niu, J., Yang, P., Kankiewicz, J. A., Larson, V. E., and
Vonder Haar, T. H.: The Vertical Profile of Liquid and Ice Water Content in
Midlatitude Mixed-Phase Altocumulus Clouds, J. Appl. Meteorol. Climate,
47, 2487–2495, 2008.Dearden, C., Connolly, P. J., Choularton, T., Field, P. R., and Heymsfield,
A. J.: Factors influencing ice formation and growth in simulations of a
mixed-phase wave cloud, J. Adv. Model. Earth Syst.,
4, M10001, 10.1029/2012MS000163, 2012.
DeMott, P. J., Prenni, A. J., Liu, X., Kreidenweis, S. M., Petters, M. D.,
Twohy, C. H., Richardson, M. S., Eidhammer, T., and Rogers, D. C.:
Predicting global atmospheric ice nuclei distributions and their impacts on
climate, PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED
STATES OF AMERICA, 107, 11217–11222,
2010.Diehl, K., Simmel, M., and Wurzler, S.: Numerical sensitivity studies on the
impact of aerosol properties and drop freezing modes on the glaciation,
microphysics, and dynamics of clouds, J. Geophys.
Res.-Atmos., 111, D07202, 10.1029/2005JD005884, 2006.
Eidhammer, T., DeMott, P. J., Prenni, A. J., Petters, M. D., Twohy, C. H.,
Rogers, D. C., Stith, J., Heymsfield, A., Wang, Z., Pratt, K. A., Prather,
K. A., Murphy, S. M., Seinfeld, J. H., Subramanian, R., and Kreidenweis,
S. M.: Ice Initiation by Aerosol Particles: Measured and Predicted Ice
Nuclei Concentrations versus Measured Ice Crystal Concentrations in an
Orographic Wave Cloud, J. Atmos. Sci., 67,
2417–2436, 2010.
Field, P. R., Heymsfield, A. J., Shipway, B. J., DeMott, P. J., Pratt, K. A.,
Rogers, D. C., Stith, J., and Prather, K. A.: Ice in Clouds Experiment-Layer
Clouds, Part II: Testing Characteristics of Heterogeneous Ice Formation in
Lee Wave Clouds, J. Atmos. Sci., 69, 1066–1079,
2012.
Findeisen, W.: Kolloid-meteorologische Vorgänge bei Niederschlagsbildung,
Meteorolog. Z., 55, 121–133, 1938.
Fleishauer, R. P., Larson, V. E., and Vonder Haar, T. H.: Observed
Microphysical Structure of Midlevel, Mixed-Phase Clouds, J. Atmos. Sci., 59,
1779–1804, 2002.Fridlind, A. M., Ackerman, A. S., McFarquhar, G., Zhang, G., Poellot, M. R.,
DeMott, P. J., Prenni, A. J., and Heymsfield, A. J.: Ice properties of
single-layer stratocumulus during the Mixed-Phase Arctic Cloud Experiment: 2.
Model results, J. Geophys. Res.-Atmos., 112, D24202,
10.1029/2007JD008646,
2007.
Fukuta, N. and Takahashi, T.: The Growth of Atmospheric Ice Crystals: A
Summary of Findings in Vertical Supercooled Cloud Tunnel Studies, J. Atmos. Sci., 56, 1963–1979,
1999.Hartmann, S., Augustin, S., Clauss, T., Wex, H., Santl-Temkiv, T., Voigtlander,
J., Niedermeier, D., and Stratmann, F.: Immersion freezing of ice nucleation
active protein complexes, Atmos. Chem. Phys., 13,
5751–5766, doi10.5194/acp-13-5751-2013, 2013.
Heymsfield, A. J. and Westbrook, C. D.: Advances in the Estimation of Ice
Particle Fall Speeds Using Laboratory and Field Measurements, J.
Atmos. Sci., 67, 2469–2482,
2010.
Hogan, R. J., Mittermaier, M. P., and Illingworth, A. J.: The Retrieval of Ice
Water Content from Radar Reflectivity Factor and Temperature and Its Use in
Evaluating a Mesoscale Model, J. Appl. Meteor. Climatol., 45, 301–317,
2006.
Illingworth, A. J., Hogan, R. J., O'Connor, E. J., Bouniol, D., Delanoë, J.,
Pelon, J., Protat, A., Brooks, M. E., Gaussiat, N., Wilson, D. R., Donovan,
D. P., Baltink, H. K., van Zadelhoff, G.-J., Eastment, J. D., Goddard, J.
W. F., Wrench, C. L., Haeffelin, M., Krasnov, O. A., Russchenberg, H. W. J.,
Piriou, J.-M., Vinit, F., Seifert, A., Tompkins, A. M., and Willén, U.:
Cloudnet, Bull. Amer. Meteor. Soc., 88, 883–898,
2006.Kanitz, T., Seifert, P., Ansmann, A., Engelmann, R., Althausen, D., Casiccia,
C., and Rohwer, E. G.: Contrasting the impact of aerosols at northern and
southern midlatitudes on heterogeneous ice formation, Geophysical Res.
Letters, 38, L17802, doi10.1029/2011GL048532, 2011.
Korolev, A. and Field, P. R.: The Effect of Dynamics on Mixed-Phase Clouds:
Theoretical Considerations, J. Atmos. Sci., 65, 66–86, 2008.
Meyers, M. P., DeMott, P. J., and Cotton, W. R.: New primary ice-nucleation
parameterizations in an explicit cloud model, J. Appl.
Meteorol., 31, 708–721,
1992.
Mitchell, D.: Use of mass- and area-dimensional power laws for determining
precipitation particle terminal velocities, J. Atmos.
Sci., 53, 1710–1723,
1996.
Murray, B. J., O'Sullivan, D., Atkinson, J. D., and Webb, M. E.: Ice
nucleation by particles immersed in supercooled cloud droplets, Chem. Soc.
Rev., 41, 6519–6554, 2012.
O'Sullivan, D., Murray, B. J., Malkin, T. L., Whale, T. F., Umo, N. S.,
Atkinson, J. D., Price, H. C., Baustian, K. J., Browse, J., and Webb, M. E.:
Ice nucleation by fertile soil dusts: relative importance of mineral and
biogenic components, Atmos. Chem. Phys., 14,
1853–1867, 2014.Petzold, A., Fiebig, M., Flentje, H., Keil, A., Leiterer, U., Schroder, F.,
Stifter, A., Wendisch, M., and Wendling, P.: Vertical variability of aerosol
properties observed at a continental site during the Lindenberg Aerosol
Characterization Experiment (LACE 98), J. Geophys.
Res.-Atmos., 107, 8128, 10.1029/2001JD001043, 2002.
Phillips, V. T. J., DeMott, P. J., and Andronache, C.: An empirical
parameterization of heterogeneous ice nucleation for multiple chemical
species of aerosol, J. Atmos. Sci., 65, 2757–2783, 2008.
Pospichal, B., Kilian, P., and Seifert, P.: Performance of cloud liquid water
retrievals from ground-based remote sensing observations over Leipzig, in:
Proceedings of the 9th International Symposium on Tropospheric Profiling
(ISTP), 2012.Shipway, B. J. and Hill, A. A.: Diagnosis of systematic differences between
multiple parametrizations of warm rain microphysics using a kinematic
framework, Q. J. R. Meteorol. Soc., 138, 2196–2211,
2012.
Simmel, M. and Wurzler, S.: Condensation and activation in sectional cloud
microphysical models, Atmos. Res., 80, 218–236,
2006.
Smith, A. J., Larson, V. E., Niu, J., Kankiewicz, J. A., and Carey, L. D.:
Processes that generate and deplete liquid water and snow in thin midlevel
mixed-phase clouds, J. Geophys. Res., 114, D12203,
2009.
Sun, J., Ariya, P. A., Leighton, H. G., and Yau, M. K.: Modeling Study of Ice
Formation in Warm-Based Precipitating Shallow Cumulus Clouds, J.
Atmos. Sci., 69, 3315–3335,
2012.
Warren, S. G., Hahn, C. J., London, J., Chervin, R. M., and Jenne, R.: Global
distribution of total cloud cover and cloud type amount over land, Tech. Rep.
Tech. Note TN-317 STR, NCAR, 1998a.
Warren, S. G., Hahn, C. J., London, J., Chervin, R. M., and Jenne, R.: Global
distribution of total cloud cover and cloud type amount over the ocean, Tech.
Rep. Tech. Note TN-317 STR, NCAR, 1998b.
Wegener, A.: Thermodynamik der Atmosphäre, J. A. Barth Verlag, 1911.
Westbrook, C. D., Hogan, R. J., and Illingworth, A. J.: The capacitance of
pristine ice crystals and aggregate snowflakes, J. Atmos.
Sci., 65, 206–219, 2008.