Cirrus cloud occurrence as function of ambient relative humidity: a comparison of observations obtained during the INCA experiment

Abstract

Based on in-situ observations performed during the Interhemispheric differences in cirrus properties from anthropogenic emissions (INCA) experiment, we introduce and discuss the cloud presence fraction (CPF) defined as the ratio between the number of data points determined to represent cloud at a given ambient relative humidity over ice (RHI) divided by the total number of data points at that value of RHI. The CPFs are measured with four different cloud probes. Within similar ranges of detected particle sizes and concentrations, it is shown that different cloud probes yield results that are in good agreement with each other. The CPFs taken at Southern Hemisphere (SH) and Northern Hemisphere (NH) midlatitudes differ from each other. Above ice saturation, clouds occurred more frequently during the NH campaign. Local minima in the CPF as a function of RHI are interpreted as a systematic underestimation of cloud presence when cloud particles become invisible to cloud probes. Based on this interpretation, we find that clouds during the SH campaign formed preferentially at RHIs between 140 and 155%, whereas clouds in the NH campaign formed at RHIs somewhat below 130%. The data show that interstitial aerosol and ice particles coexist down to RHIs of 70-90%, demonstrating that the ability to distinguish between different particle types in cirrus conditions depends on the sensors used to probe the aerosol/cirrus system. Observed distributions of cloud water content differ only slightly between the NH and SH campaigns and seem to be only weakly, if at all, affected by the freezing aerosols.

Full text

Page 1

Atmos. Chem. Phys., 3, 1807–1816, 2003
www.atmos-chem-phys.org/acp/3/1807/
Atmospheric
Chemistry
and Physics
Cirrus cloud occurrence as function of ambient relative humidity: a
comparison of observations obtained during the INCA experiment
J. Str¨ om1, M. Seifert1, B. K¨ archer2, J. Ovarlez3, A. Minikin2, J.-F. Gayet4, R. Krejci1, A. Petzold2, F. Auriol4,
W. Haag2, R. Busen2, U. Schumann2, and H. C. Hansson1
1ITM, Air Pollution Laboratory, Stockholm University, Sweden
2DLR, Institut f¨ ur Physik der Atmosph¨ are, Oberpfaffenhofen, Germany
3LMD, Ecole Polytechnique, CNRS-IPSL, Palaiseau, France
4LAMP, Universit´ e Blaise Pascal, Aubiere, France
Received: 24 March 2003 – Published in Atmos. Chem. Phys. Discuss.: 30 June 2003
Revised: 30 September 2003 – Accepted: 15 October 2003 – Published: 27 October 2003
Abstract. Based on in-situ observations performed during
the Interhemispheric differences in cirrus properties from an-
thropogenic emissions (INCA) experiment, we introduce and
discuss the cloud presence fraction (CPF) defined as the ra-
tio between the number of data points determined to rep-
resent cloud at a given ambient relative humidity over ice
(RHI) divided by the total number of data points at that value
of RHI. The CPFs are measured with four different cloud
probes. Within similar ranges of detected particle sizes and
concentrations, it is shown that different cloud probes yield
results that are in good agreement with each other. The CPFs
taken at Southern Hemisphere (SH) and Northern Hemi-
sphere (NH) midlatitudes differ from each other. Above ice
saturation, clouds occurred more frequently during the NH
campaign. Local minima in the CPF as a function of RHI
are interpreted as a systematic underestimation of cloud pres-
ence when cloud particles become invisible to cloud probes.
Based on this interpretation, we find that clouds during the
SH campaign formed preferentially at RHIs between 140
and 155%, whereas clouds in the NH campaign formed at
RHIs somewhat below 130%. The data show that interstitial
aerosol and ice particles coexist down to RHIs of 70–90%,
demonstrating that the ability to distinguish between differ-
ent particle types in cirrus conditions depends on the sensors
used to probe the aerosol/cirrus system. Observed distribu-
tions of cloud water content differ only slightly between the
NH and SH campaigns and seem to be only weakly, if at all,
affected by the freezing aerosols.
Correspondence to: J. Str¨ om
(johan@itm.su.se)
1Introduction
There is a concern that anthropogenic emissions may change
the environment in a way that it could influence the fre-
quency of occurrence and microphysical properties of cir-
rus clouds. An obvious anthropogenic modification of the
cloud frequency of occurrence are contrails formed behind
aircraft nearthe tropopause. Based on analyzed satellite data,
Mannstein et al. (1999) and Meyer et al. (2002) deduced that
linear persistent contrails cover about 0.5–0.7% of the sky at
noon over Europe in the annual average. These results ex-
emplify the possible regional effect by contrails (IPCC 1999,
EC 2002).
Besides this direct contrail effect, observations indicate
that indirect aerosol effects may influence the properties of
high clouds as well (Str¨ om and Ohlsson, 1998; Boucher,
1999; Kristensson et al., 2000).
aerosol effect on cirrus is possible when at least two types
of freezing aerosol particles compete during cloud formation
(K¨ archer and Lohmann, 2003): adding efficient ice nuclei to
liquid aerosol particles can lead to a marked suppression of
pristine ice crystal number densities, the magnitude of this
effect depending on updraft speed, temperature, and number
and freezing properties of the ice nuclei.
Compared to our current level of understanding of warm
cloud microphysical processes, fundamental knowledge
about the formation and disappearance of cirrus clouds is
still lacking, as there are still open issues regarding the
preferred mode of ice nucleation in cirrus clouds (DeMott,
2002) and the evaporation kinetics of small ice crystals under
polluted conditions (Chen and Crutzen, 1994; Seifert et al.,
2003a). The situation is complicated by the fact that the effi-
ciency with which aerosol particles freeze is strongly linked
to the dynamical conditions prevailing during ice formation
(K¨ archer and Str¨ om, 2003).
A pronounced indirect
© European Geosciences Union 2003

Page 2

1808J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity
Studying a possible anthropogenic effect on clouds in-
volves the detection of subtle but systematic differences be-
tween the properties of clouds formed in a pristine environ-
ment and those formed in a perturbed environment. The pos-
sible anthropogenic influence on warm clouds has been stud-
ied extensively and covers a range of environmental condi-
tions in both hemispheres. However, until recently all in-
situ measurements of midlatitude cirrus had been performed
in the Northern Hemisphere. With the project INCA (In-
terhemispheric differences in cirrus properties from anthro-
pogenic emissions) the first observations of cirrus properties
in the Southern Hemisphere midlatitudes became available,
allowing clouds that formed under comparable meteorolog-
ical conditions in two very different regions of the world to
be compared with each other with an identical set of in situ
instruments.
Two aircraft campaigns were performed as part of INCA,
one in the Southern Hemisphere (SH) and one in the North-
ern Hemisphere (NH) midlatitudes. Here the abbreviations
SH and NH are used to identify the different campaigns and
do not refer to hemispherically averaged properties. The
first campaign based in Punta Arenas, Chile (54◦S) was per-
formed in March and April. The second campaign based
in Prestwick, Scotland (55◦N) was performed in Septem-
ber and October. Hence, the campaigns were performed in
equivalent seasons in the same year 2000. For more in-
formation about the INCA experiment we refer to: http:
//www.pa.op.dlr.de/inca/.
One of the first results published from INCA was a com-
parison of distributions of relative humidity over ice (RHI)
observed during the two campaigns (Ovarlez et al., 2002).
Both data sets presented a maximum in the frequency distri-
butionaround100%, butthedistributionwasskewedtowards
higher humidities in the SH data. In the present study, we in-
vestigate whether differences in the distributions of RHI are
related to differences in cirrus cloud properties by compar-
ing the occurrence of clouds as a function of RHI observed
during the two campaigns. For this purpose, we define cloud
occurrence as the ratio between the number of in-cloud data
points versus all data points at any given RHI and call this
ratio the cloud presence fraction (CPF). A substantial part of
the paper deals with the question of how to decide whether a
data point represents cloudy or cloud-free air.
2 Methodology
Although a cloud is something known to everyone, it may
sometimes be difficult or even impossible to provide a simple
definition for when a cloud is actually present or not. What is
the minimum crystal number density or horizontal and verti-
cal extent necessary for an ensemble of hydrometeors to be
called a cloud? Is a 1 m thick layer or a particle number den-
sity of 1m−3sufficient? We can raise similar questions for
observable parameters obtained by in-situ or remote sensing
methods alike. Because of theses difficulties, the presence or
non-presence of a cloud is usually determined by the detec-
tion limit of the particular sensor used to observe the cloud.
What is interpreted as a cloud by one sensor might be inter-
preted as cloud-free air by another.
2.1Description of the cloud probes
In this study we will make use of four different cloud sen-
sors to investigate the presence or non-presence of cirrus
clouds as function of ambient relative humidity. These in-
struments are the Counterflow Virtual Impactor (CVI), the
PMS FSSP-300, the PMS 2D-probe, and the Polar Neph-
elometer. The same instruments were used in both INCA
campaigns, which permits a direct comparison of the obser-
vations with respect to an unchanged payload configuration.
The afore-mentioned cloud probes were mounted on the re-
search aircraft Falcon operated by Deutsches Zentrum f¨ ur
Luft- und Raumfahrt (DLR). All probes have different ad-
vantages and limitations and provide information about dif-
ferent aspects of the cloud, as explained below. With the term
“cloud particles” used hereafter we mean “particles mea-
sured in the presence of cirrus clouds”.
The CVI (Ogren et al., 1985; Str¨ om et al., 1994) has its
inlet facing the direction of flight and operates by using an
internal flow opposite the flight direction to prevent ambient
air and small particles to enter the probe. In the upper tro-
posphere and at typical Falcon airspeeds of ∼180ms−1, this
lower cut-off is approximately 5µm aerodynamic diameter.
Cloud particles with these and larger sizes are mostly ice
crystals, especially at the high total number concentrations
(>0.1cm−3) typically found in young cirrus clouds. The
crystals entering the CVI are evaporated and their residues
are counted using condensation particle counters. A one-to-
one correspondence between the number of residual particles
and ice crystals is assumed, which is proven to be a valid as-
sumption (Seifert et al., 2003b). The total particle number
density captured by the CVI is an accurate measure of the to-
tal number density of ice crystals in young cirrus clouds. At
1Hz data resolution, one count registered by the CVI pay-
load corresponds to an ambient crystal number density of
∼0.0004cm−3(CVI detection limit).
The FSSP-300 probe uses the measured response from
scattered light coming from particles illuminated by laser
light (Baumgardner et al., 1992). To convert from the ob-
served scattered light signals to a particle size distribution,
a T-matrix calibration was used (Borrmann et al., 2000).
The method yields a size classification in 32 bins between
0.37µm to 15.77µm in diameter. Cloud particles in this
size range can be interstitial aerosol particles and ice crys-
tals, especially in the sub-µm size range. Due to noise in the
first 3 bins (smallest sizes), the minimum size class starts at
0.55µm. At 1Hz data resolution, one count registered by
the FSSP-300 corresponds to a particle number density of
∼0.2cm−3(FSSP detection limit).
Atmos. Chem. Phys., 3, 1807–1816, 2003www.atmos-chem-phys.org/acp/3/1807/

Page 3

J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity 1809
x2
Relative humidity over water
100
Cloud presence fraction
1
x1
Relative humidity over ice
100
Cloud presence fraction
1
x3
x2
x1
x3
a)
b)
Fig. 1. Schematic illustration of the cloud presence fraction as a function of the relative humidity for liquid clouds (a) and ice clouds (b). All
data points above X1are in-cloud data points. The point where clouds start to dissolve and where cloudy and non-cloudy air parcels coexist
is marked X2and the point where the cloud has completely disappeared is marked X3.
The PMS 2D-probe classifies the size of particles by the
shadow they create when passing in front of an array of de-
tectors illuminated by a laser. The particles are classified into
30 bins in the diameter range 25–800µm. Cloud particles in
this size range are exclusively water ice crystals, based on
their large condensed mass. Because such large ice crystals
are usually not spherical, a number of assumptions are intro-
duced to convert from raw data to size distributions (Gayet
et al., 1993). Most of the cloud water content (CWC) is lo-
cated in the size range of the 2D-probe. The ambient crystal
number density corresponding to one count registered by the
2D-probe depends on the crystal size, but typically the min-
imum detectable number density is of the order 10−6cm−3
(2D-probe detection limit).
The Polar Nephelometer measures the scattering phase
function of an ensemble of cloud particles in the size range
from a few µm to about 800µm diameter (Gayet et al.,
1997), i.e., essentially ice crystals. A laser beam at a wave-
length of 804nm illuminates the ice crystals near the focal
point of a paraboloidal mirror. The scattered light is de-
tected by 44 photodiodes in angles from ±3.49 to ±169◦.
The direct measurement of the scattering phase function pro-
vides the means to distinguish between solid and liquid (if
present) phase hydrometeors and to calculate the extinction
coefficient and asymmetry parameter (Auriol et al., 2001).
The Polar Nephelometer provides an optical response (vol-
ume extinction) that depends on the combined effect of ice
crystal size and number density.
2.2 Relative humidity and cloud presence fraction
Relative humidity was measured using a cryogenic frost
point mirror (Ovarlez and van Velthoven, 1997). The un-
heated inlet, a modified Rosemount-Goodrich temperature
housing, was located on the top of the fuselage. The relative
humidity is determined from the Sonntag saturation vapor
pressure formula (Sonntag, 1994), using the air temperature
data provided by the standard instrumentation on board the
Falcon. The relative uncertainty in observed relative humid-
ity is estimated to be better than 7% (2 standard deviations).
The expected CPF as function of relative humidity over
supercooled liquid water for warm clouds consisting of su-
percooled water droplets is illustrated in Fig. 1a. Such clouds
form and disappear at essentially the same relative humidity
marked X1and X2. Actually, a small supersaturation, typi-
cally a fraction of a percent, is necessary to activate aerosol
particles into cloud droplets. However, within measurement
capabilities of relative humidity, the CPF can be approxi-
mated as unity at and above water saturation and zero at hu-
midities below. Solution droplets may be in equilibrium with
ambient water vapor even at relative humidities well below
100%. These particles are not activated into cloud droplets
in the traditional sense.
Ice clouds are more complicated than their liquid counter-
parts since they may form at one RHI substantially (many
tens of percent) above saturation and dissolve completely at
muchlower(fewtensofpercent)valuesbelowsaturation; see
Sect. 3. The rate at which water molecules are transferred
between the different phases at cold temperatures is slower
than in warm clouds, which is why a cirrus cloud can per-
sist for some time in air subsaturated with respect to ice and
give rise to their often fuzzy appearance especially around
cloud edges. A possible scenario is illustrated in Fig. 1b.
At some relative humidity over ice, X1, ice crystals appear
in any given air parcel. In this case we do not care about
their size and number density, we simply acknowledge the
presence of cloud. The relative humidity may still increase
after crystals initially appear. This increase in relative hu-
midity continues until the formation phase is completed and
all crystals have formed. In our case this does not change
www.atmos-chem-phys.org/acp/3/1807/Atmos. Chem. Phys., 3, 1807–1816, 2003

Page 4

1810J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity
a)
b)
0 50100150
0.01
0.1
1
Fraction of in-cloud data points SH
RHI (%)
FSSP-300
CVI
Model
0 50100150
0.01
0.1
1
Fraction of in-clou datapoints NH
RHI (%)
FSSP-300
CVI
Model
Fig. 2. Observed cloud presence fractions as a function of relative humidity over ice using similar detection thresholds for the two probes
(CVI: red curves, FSSP-300: black curves) with respect to particle size (>4µm) and number density (>0.3cm−3). Data from the SH
campaign (a) and from the NH campaign (b). The blue curves are corresponding model results taken below ice saturation and exclusively
count ice crystals.
the presence or non-presence of the cloud and the CPF=1 at
X1and humidities above. At X2(near ice saturation), CPF
is less than unity since this point represents a mixture be-
tween air parcels that contain cloud particles and where the
humidity is perhaps decreasing, and air parcels that have not
yet formed a cloud and where the humidity might still be in-
creasing. To the left of X2(below ice saturation), ice crystals
begin to evaporate. Once below saturation, all clouds will
eventually have disappeared at some humidity X3. Because
of the slower evaporation processes in cold clouds the rela-
tive humidity at X3is different from X2. The details about
X1, X2and X3for cirrus clouds are not well known.
3 Observations
Our study is limited to observations performed above 6km
altitude and temperatures below 235K. Under these con-
ditions, clouds will consist of ice crystals and interstitial
aerosol particles (sometimes referred to as supercooled haze
droplets), as liquid water droplets would freeze sponta-
neously. Although the different cloud probes are based on
different working principles and have different detection lim-
its, as summarized in Sect. 2.1, it is interesting to compare
the consistency between the probes for a subset of the data
where there is an overlap in cloud detection. The FSSP-300
is able to detect clouds when particles are smaller than the
aerodynamic cut-off of the CVI, but the CVI is able to detect
clouds with a much lower particle number density than the
FSSP-300 is capable of. If a subset of the FSSP-300 data is
selected to emulate a CVI with respect to the size cut-off, and
a subset of the CVI data is selected to emulate the FSSP-300
with respect to the number density detection limit, the two
instruments should detect clouds with similar efficiency.
3.1Transition between clouds and cloud-free air
Figure 2 shows observed CPFs as a function of RHI from
both INCA campaigns (SH in Fig. 2a and NH in Fig. 2b)
taken with the CVI (red curves) and the FSSP-300 (black
curves). To enable a direct comparison, we have used sim-
ilar detection thresholds for both instruments: particle size
>4µm and particle concentration >0.3cm−3. Given these
thresholds, we expect the detected particles to be mostly ice
crystals, as such high number densities of large aerosol parti-
cles are hardly found in cirrus levels (K¨ archer and Solomon,
1999). For example, in situ measurements over continental
Europe revealed the presence of coarse-mode aerosol in the
tropopause region, but only at low concentrations (Schr¨ oder
et al., 2002, their Fig. 3b). The measured CPFs show more
structure than suggested by the schematic in Fig. 1b. We ob-
serve a plateau region between 90–130% (NH) and 90–150%
(SH), a transition region between 70–90%, and a dry region
<70%.
The comparison of CPF data from both campaigns shows
an excellent agreement between the CVI and FSSP-300 over
several orders of magnitude in CPF. Typically, the difference
between the two cloud probes is in the range of percent. The
CPF is plotted on a logarithmic scale to highlight the agree-
ment even down to RHI=70–80%. Before the CPFs rapidly
decrease around 80%, they stay elevated over a wide range
of RHI up to about 130% (NH) and 150% (SH). Recall that
the CVI counts residual particles from evaporated crystals,
whereas the FSSP-300 detects the scattered light from the
Atmos. Chem. Phys., 3, 1807–1816, 2003www.atmos-chem-phys.org/acp/3/1807/

Page 5

J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity1811
a)
b)
0 20 40 6080100120140160180
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of in-cloud data points SH
RHI(%)
PMS-2D 25 µm
PMS-2D 100 µm
Model
0 20 40 6080100120 140160180
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of in-cloud data pints NH
RHI (%)
PMS-2D 25 µm
PMS-2D 100 µm
Model
Fig. 3. Cloud presence fractions as a function of relative humidity over ice for different ice crystal size thresholds (black curves: 25µm; red
curves: 100µm) as measured by the PMS-2D probe (concentration threshold 10−6cm−3). Data from the SH campaign (a) and from the
NH campaign (b). Black and red curves in Fig. 5b are almost identical. The blue curves are corresponding model results taken below ice
saturation and exclusively count ice crystals >25µm and >10−6cm−3.
crystals in the ambient air. Although their working princi-
ples are completely different, Fig. 2 proves that the two in-
struments perform as expected within at least the overlapping
range in particle number density and size.
In Fig. 3 the CPFs derived from the 2D-probe are plot-
ted using two cloud presence size thresholds (25µm and
100µm, respectively), as obtained during the SH (Fig. 3a)
and NH campaign (Fig. 3b). In each data set, the two curves
are essentially on top of each other, indicating that whenever
ice crystals larger than 25µm are present in the cloud, crys-
tals of sizes larger than 100µm are present as well.
The CPF distributions shown in Fig. 3 share many fea-
tures of those shown in Fig. 2. As for the CVI and FSSP-
300 probes, we observe a plateau in CPF at RHI between
about 90–140% and a rapid decrease of CPF around 80%.
However, for RHI <70%, the CPFs drop almost to zero in
the case of the 2D-probe, strongly indicating that ice crystals
with sizes >25µm do not anymore exist under such subsat-
urations, while there seem to be values of CPF>0 in the case
of the CVI and FSSP-300 probes shown in Fig. 2, especially
in the NH data set. We reiterate this issue later in Sect. 3.2.
In what follows, we compare the findings shown in Figs. 2
and 3 with model simulations described in full detail by Haag
et al. (2003), whose principal goal is to infer freezing thresh-
olds and nucleation modes from the observed distributions
of RHI above ice saturation. Here, it is sufficient to note that
the microphysical trajectory model roughly captures the typ-
ical environmental conditions (temperatures, cooling rates,
mesoscale wave amplitudes and frequencies) that prevailed
during the NH and SH campaigns and uses a fairly detailed
microphysical scheme to predict the formation and disap-
pearence of cirrus clouds. The prescribed aerosol freezing
properties provide a consistent explanation of the SH data in
terms of homogeneous freezing, and of the NH data in terms
of homogeneous freezing competing with a small number of
efficient ice nuclei (cases HOM and MIX0.001 of Haag et
al. (2003), respectively).
We present model-derived CPFs computed from the dis-
tributions of RHI inside and outside of cirrus clouds. The
calculated CPFs do not include aerosol particles, in contrast
to the observations that make no distinction between these
two types of particles. They are plotted as blue curves in
Figs. 2 and 3, using the same criteria to define cloud as in
the observations with the CVI/FSSP-300 and 2D-probe, re-
spectively. The model curves show the same characteristic
transitionregimeasthemeasurementsatroughlycomparable
RHIs (see below). The reason for the strong decrease of CPF
in the model is that the evaporation of water molecules from
the ice crystals is kinetically limited. The time needed to (al-
most) completely evaporate water from ice crystals can be
longer than the time in which air parcels experience (signif-
icant) subsaturation due to warming, especially at low tem-
peratures and in the presence of temperature oscillations.
Although the model curves are very similar for the SH and
NH cases, they seem to be shifted to the right compared to
the observations. We offer four explanations to this apparent
discrepancy.
1. The calculated and measured CPF values may differ
from each other as, by definition, CPF depends on the
ratio between data points taken outside of clouds and
www.atmos-chem-phys.org/acp/3/1807/Atmos. Chem. Phys., 3, 1807–1816, 2003

Page 6

1812J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity
0 2040 60 80100120 140 160180
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of in-cloud data points NH
RHI (%)
0.001 cm
0.003
0.01
0.03
0.1
0.3
1
3
-3
020 406080100120140160180
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of in-cloud data points SH
RHI (%)
0.001 cm
0.003
0.01
0.03
0.1
0.3
1
3
-3
a)
b)
Fig. 4. Cloud presence fractions as a function of relative humidity over ice for different particle number density thresholds (color code given
in the legend) as measured by the CVI. Data from the SH campaign (a) and from the NH campaign (b).
those taken inside clouds. Hence, the shifts may in part
be artifacts, because this ratio will not be exactly repro-
duced by the model at any given RHI, and the model
CPFs need to be re-scaled as a function of RHI in order
to be directly comparable to the observations.
2. The physics of the ice-to-aerosol transition in the model
is incomplete. In the current version of the code, we
track the ice water mass and compare it with the equilib-
rium water mass aerosol particles with the same average
core (sulfuric acid) mass as contained in these crystals
would have at the local thermodynamic conditions. If
the ice water mass falls below that value, we remove the
corresponding ice particles and reinitialize the aerosol
size distribution. A single particle treatment and very
small time steps would improve the calculation; more
importantly, the chemistry of melting of the ice parti-
cles and their chemical composition besides water is not
known and does not justify a more detailed description.
3. Small-scale
ment/detrainment processes may affect the evaporation
time of small ice crystals but are not represented in the
model.
turbulentmixingandentrain-
4. The experimental concentration and size thresholds for
cloudpresenceprescribedinthecalculationsmaynotbe
accurate, as we find that the modeled transition region
is sensitive to changes in the concentration threshold.
While above the points have no impact on the results pre-
sented by Haag et al. (2003), they may at least in part help
reconcile data and model results, but important uncertainties
remain concerning the evaporation kinetics of small ice crys-
tals.
3.2Variation of cloud detection thresholds
K¨ archer and Solomon (1999) have emphasized the contin-
uum aspect of aerosol/cirrus cloud systems in the tropopause
region. Several particle types usually coexist, especially in
high relative humidity conditions. This notion is strongly
supported by the data shown in Figs. 4 and 5 described be-
low. In the size range detectable by the CVI (>5µm), we
expect the aerosol/cloud system to be predominantly com-
posed of ice crystals when CPF=1, with increasing contri-
butions from interstitial aerosol particles for CPF<1, espe-
cially at RHIs in the vicinity of saturation. (Of course, the
total number of aerosol particles regardless of size generally
exceeds the number of ice crystals by several orders of mag-
nitude.) Around the point where clouds start to dissolve, the
fraction of such large aerosol particles, if present in sufficient
concentrations, increases and eventually becomes dominant.
Owing to the poorly known kinetics of evaporation of small
ice crystals, as mentioned above, the RHI at which the last
ice crystals have disappeared is not well defined and may ex-
tend to values <70%. At very dry RHI, well below 70%, we
expect the aerosol/cloud system to be exclusively composed
of aerosol particles, some of which may exist in crystalline
or mixed states. Although the term “cloud” does not actually
apply in such very dry situations, we use the term CPF over
the entire range of RHI to support our claim that there is no
universal or well defined limit in RHI below which we al-
ways find “pure aerosol”. The only objective measure avail-
able to us is the detection limit of the sensors we use.
Atmos. Chem. Phys., 3, 1807–1816, 2003 www.atmos-chem-phys.org/acp/3/1807/

Page 7

J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity1813
020 4060 80100120140 160180
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of in-cloud data points SH
RHI (%)
0.6-15.8 µm
0.7-15.8
0.9-15.8
1.1-15.8
3.1-15.8
3.7-15.8
6.4-15.8
8.5-15.8
12.7-15.8
0 20 40 6080100 120 140160180
0.0
0.2
0.4
0.6
0.8
1.0
Fraction of in-cloud data points NH
RHI (%)
0.6-15.8 µm
0.7-15.8
0.9-15.8
1.1-15.8
3.1-15.8
3.7-15.8
6.4-15.8
8.5-15.8
12.7-15.8
a)
b)
Fig. 5. Cloud presence fractions as a function of relative humidity over ice for different particle size thresholds (color code given in the
legend) as measured by the FSSP-300. Data from the SH campaign (a) and from the NH campaign (b).
We will use data from the CVI to study in more detail the
presence or non-presence of clouds based on different num-
ber density thresholds, and the FSSP-300 using different size
thresholds. In Fig. 4 CPF is plotted versus RHI using dif-
ferent number density thresholds from the CVI for the SH
(Fig. 4a) and NH (Fig. 4b) campaign, respectively. For all but
the highest threshold (3cm−3) there is a sharp drop in CPF
between70–90%RHI,markingthetransitionregionbetween
cloudy and cloud-free air, as motivated in Sect. 3.1.
In the plateau region between 90% and 130% (NH) or
150% (SH), the CPF increases with increasing sensitivity
(decreasing concentration threshold) to measure cloud par-
ticles which are essentially ice crystals given the lower size
cut-off of the CVI. Note that the CPFs from both data sets
become rather small for the highest concentration threshold
(3cm−3) and do not even reach unity at the highest RHI. This
is explained by the fact that we now begin to miss most of the
ice crystals, because the threshold of 3cm−3is of the same
order as the average total number of ice crystals observed
in cirrus during both INCA campaigns (K¨ archer and Str¨ om,
2003).
In the dry region, RHI<70%, CPF is almost zero for con-
centration thresholds >0.3cm−3, indicating that cloud parti-
cles (i.e., either aerosol or ice particles) with sizes >5µm are
not present with such high number densities. As discussed in
Sect.3.1, it is not entirely clear from the data how far away
from the transition region towards low RHI ice crystals may
still exist (and form a “cloud”). To say at which RHI an ice
crystalbecomesanaerosolparticleisdifficult; atanyrate, the
fraction of aerosol particles will certainly increase with de-
creasing RHI. Consequently, as we relax the threshold con-
centrationtovalues<0.3cm−3, theCVIdetectslargeaerosol
particles down to very dry conditions in up to ∼40% of all
measurements at concentration thresholds <0.01cm−3.
In Figs. 5a and b CPF is plotted using different size thresh-
olds from the FSSP-300 for the SH and NH data set, re-
spectively. In selecting the size thresholds, we have used
the available size bin classifications and do not distinguish
between different particles types, i.e., dry particles, haze
droplets, or ice crystals.
The principal features of CPF are similar to Fig. 4. In
particular, the transition region is located around 80% RHI,
asbefore. However, giventhedetectionlimitof0.2cm−3, we
are mostly detecting ice particles with the FSSP-300 when
using the size thresholds >3.7–6.4µm. Using lower size cut-
offs results in the enhanced detection of (interstitial) aerosol
particles and ice crystals, and makes the interpretation of the
curves as true cloud presence fractions more difficult.
A notable difference between the NH and SH data sets oc-
curs in the dry region between the size thresholds 0.6µm and
0.7µm. During the NH campaign particles in the size range
0.6–0.7µm was present twice as often as during the SH cam-
paign. This perhaps indicates a difference in the presence of
ice nuclei between the two measurement locales or a differ-
ence in evaporation kinetics.
3.3 Onset of freezing in cirrus clouds
In Fig. 6, cloud data points are defined as having an extinc-
tion coefficient of 0.05km−1or larger as measured by the
Polar Nephelometer. In addition, this criterion had to be ful-
filled during at least four consecutive seconds in order to be
considered an in-cloud data point. A microphysical equiv-
alent of this optical threshold corresponds to ice crystals of
∼5µm diameter at a number density roughly between 0.05
and 0.1cm−3. The criteria used to define Polar Nephelome-
ter in-cloud data points presented in Fig. 6 are the same as
used by Ovarlez et al. (2002). Whereas all previous CPFs
www.atmos-chem-phys.org/acp/3/1807/Atmos. Chem. Phys., 3, 1807–1816, 2003

Page 8

1814J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity
0 2040 60 80 100 120 140 160 180
RHI (%)
0
20
40
60
80
100
Fraction of in-cloud data points
NH
SH
Fig. 6. Cloud presence fractions as a function of relative humidity
over ice measured by the Polar Nephelometer. If the probe, at a
given RHI, observes an extinction coefficient exceeding 0.05km−1
during four consecutive seconds, the data point is considered a
cloudy data point. This criterion roughly corresponds to particle
size and concentration thresholds of 5µm and 0.05–0.1cm−3, re-
spectively, indicating that mostly ice crystals are detected. Only
data during straight and level flight segments are included.
were calculated for each percent of RHI, the Polar Neph-
elometer data are averages over 10% increments.
Values of CPF near zero in the dry region below 70% sup-
port the notion that the Polar Nephelometer has exclusively
probed ice crystals. (The origin of the hump at RHI between
40–50% in the SH data set is not known.) In contrast to what
is expected from the schematic shown in Fig. 1b, however,
the CPF curves in Fig. 6 do not increase monotonically to
unity when RHI rises, but exhibit local minima centered at
120% (NH data) and 150% (SH data). This feature can also
be traced in the data of the other probes presented in Figs. 2–
5, but the extent of the minimum depends on the threshold
used to characterize in-cloud and out-of-cloud data points.
As the Polar Nephelometer detects ice crystals in cirrus
when they return an extinction signal >0.05km−1, a suffi-
cient number of crystals must have nucleated and grown to
certain sizes in order to produce such large extinction val-
ues. The very first, freshly nucleated crystals are certainly
not detectable by the Polar Nephelometer probe. When all
ice crystals have formed at RHI>130% (NH) and >150%
(SH), they grow and thereby reduce RHI and become de-
tectable by the Polar Nephelometer when crossing again the
RHI-regions containing the local minima. The same reason-
ing can be applied to the other probes.
This provides an explanation for the distinct minima ob-
served in the Polar Nephelometer data presented in Fig. 6.
Thus, the interpretation of the local minimum in observed
CPF above ice saturation is that it approximates the range in
RHI where the onset of cloud formation occurs. The compar-
ison of the NH and SH cases strongly suggests that the mode
of nucleation was different during the respective campaigns.
The study of Haag et al. (2003) investigates this difference
in ice nucleation in cirrus in more detail by analyzing the
distributions of RHI taken outside of and inside cloud with
the help of microphysical model simulations.
3.4 Cloud water content
Figure 7a shows the observed cloud water contents derived
from the sum of the CWC’s measured by the 2D-probe us-
ing an empirical formulation (Gayet et al., 2002) to convert
from size distribution to CWC, and the CWC derived from
the FSSP-300 size distribution assuming spherical particles
and using a particle mass density of 0.9gcm−3. The ob-
served distributions are normalized, which makes the areas
under the curves equal to unity.
Both NH (black curve) and SH (red curve) distributions of
CWC are rather similar and are skewed with a tail towards
lower CWCs. The calculated geometric mean CWCs are
7.3 and 10.4mgm−3for the SH and NH data, respectively.
The data points below 0.01mgm−3are almost entirely data
from the FSSP-300 when the 2D-probe did not register any
crystals, hence aerosol particles and perhaps the smallest ice
crystalscontributetosuchlowCWCs. Thewatervaporavail-
able to convert to CWC depends strongly on temperature, we
therefore derive a similar CWC distribution taken in a nar-
row temperature range between 225 and 227K, as presented
in Fig. 7b. The distributions in Figs. 7a and b essentially
exhibit the same features, but the mode or maximum of the
distributions for the more narrow temperature range is more
defined.
The rather small differences in the distributions of CWC
between the SH and NH campaigns along with the fact that
the distributions of temperatures, updraft speeds, and total
ice crystal concentrations were also quite similar suggests
that existing differences in the freezing properties of aerosol
particles between SH and NH data have hardly affected the
CWC. Rather the CWCs appear to be mainly determined by
dynamical factors such as the advection of water vapor con-
trolling the availability of condensable water.
4Conclusions
In this study we employed different cloud probes to deter-
mine the presence or non-presence of cirrus clouds as func-
tion of relative humidity over ice during the INCA cam-
paigns. The observed cloud presence fractions showed a
characteristic behavior when plotted as a function of rela-
tive humidity over ice: a flat, plateau-like region between
RHI=90% and values close to the cloud formation thresholds
above which CPF=1; a transition region between 70–90%
where the bulk of the ice crystals evaporate and the cloud
Atmos. Chem. Phys., 3, 1807–1816, 2003www.atmos-chem-phys.org/acp/3/1807/

Page 9

J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity 1815
a)
b)
1E-4
1E-4 1E-4
1E-3
1E-3 1E-3
0.01
0.010.01
0.1
0.1 0.1
1
11
10
10 10
100
100100
0.0
0.0 0.0
0.2
0.2 0.2
0.4
0.40.4
0.6
0.60.6
0.8
0.8 0.8
1.0
1.0 1.0
dN/dlog CWC (normalized)
CWC (mg m
-3)
0.1 1 10100
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
dN/dlog CWC (Normalized)
CWC (mg m-3)
Fig. 7. Observed distributions of cloud water content (SH: red curves, NH: black curves). In (a) all temperatures below 235K and relative
humidities between 95 and 105% RHI are used. In (b) the temperature only ranges between 225 and 227K. The observed CWC is based on
the sum of the FSSP-300 and 2D-probe data. Note the different scaling of the CWC axes.
dissolves; and a dry particle region <70% where eventually
all ice particles disappear.
The CPF derived from the Polar Nephelometer data ex-
hibits a distinct local minimum between 140–155% RHI in
the SH data set when plotted as a function of RHI. A less pro-
nounced local minimum in CPF is also suggested in the NH
data set between approximately 115–130% RHI. Our inter-
pretation of these features is that they correspond to the pre-
ferred ice nucleation thresholds during the respective cam-
paigns. The SH threshold is consistent with homogeneous
freezing, whereas the NH threshold indicates heterogeneous
ice formation. A comprehensive analysis of this finding is
presented in Haag et al. (2003).
Our data strongly support the notion that (interstitial)
aerosols and cirrus clouds form a continuum system. As we
have shown, how well we can experimentally distinguish be-
tween ice crystals, haze droplets, and other aerosol particles
as a function of RHI will largely depend on the sensors used
to probe the aerosol/cirrus system. The CPFs describe “pure
aerosol” below a poorly defined RHI <70%, but never really
describe “pure cloud” inasmuch as interstitial aerosol parti-
cles are not completely scavenged by the ice crystals during
thecloudlifetime. Theevaporationkineticsofsmallicecrys-
tals and the haze particle properties above ice saturation are
poorly constrained by observations and must remain under
scrutiny to fully understand how cirrus clouds form and dis-
appear in the atmosphere. In this regard, there is a need for a
measurement device combining the size detection sensitivity
of the FSSP-300 and the concentration detection sensitivity
of the CVI, and for size-resolved in situ chemical characteri-
zation of aerosol and ice particles.
Acknowledgements. This work was funded by the European Com-
mission through the projects INCA and PARTS. It also contributes
to the project “Particles and Cirrus Clouds” (PAZI) supported by
the Helmholtz-Gemeinschaft Deutscher Forschungszentren (HGF).
The Swedish Research Council is sponsoring this work by support-
ing ITM in airborne aerosol and cirrus activities. We thank the en-
tire INCA team for help in collecting the data.
References
Auriol, F., Gayet, J.-F., Febvre, G., Jourdan, O., Labonnotte, L.,
and Brogniez, G.: In situ observations of cirrus cloud scattering
phase function with 22◦and 46◦halos: Cloud field study on 19
February 1998, J. Atmos. Sci., 58, 3376–3390, 2000.
Baumgardner, D., Dye, J. E., Gandrud, B. W., and Knollenberg, R.
G.: Interpretation of measurements made by the forward scat-
tering spectrometer probe (FSSP-300) during the Airborne Arc-
tic Stratospheric Expedition, J. Geophys. Res., 97, 8035–8046,
1992.
Borrmann, S., Luo, B. P., and Mishchenko, M.: Application of
the T-matrix method to the measurement of aspherical (ellip-
soidal) particles with forward scattering optical particle counters,
J. Aerosol Sci., 31, 789–799, 2000.
Boucher, O.: Airtrafficmayincreasecirruscloudiness, Nature, 397,
30–31, 1999.
Chen, J.-P. and Crutzen, P. J.: Solute effects on the evaporation of
ice particles, J. Geophys. Res., 99, 18847–18859, 1994.
DeMott, P. J.: Laboratoty studies of cirrus cloud processes, in: Cir-
rus, edited by Lynch, D. K., et al., Oxford Univ. Press, New York,
pp.102–135, 2002.
European Commission (EC): European research in the stratosphere
1996–2000, Second assessment on stratospheric research, edited
by Amanatidis, G. T. and Harris, N. R. P., 257–307, 2002.
www.atmos-chem-phys.org/acp/3/1807/ Atmos. Chem. Phys., 3, 1807–1816, 2003

Page 10

1816J. Str¨ om et al.: Cirrus cloud occurrence as function of ambient relative humidity
Gayet, J.-F., Brown, P. R., and Albers, F.: A comparison of in-cloud
measurements obtained with six PMS 2D-C probes, J. Atmos.
Ocean. Technol., 10, 180–194, 1993.
Gayet, J.-F., Cr´ epel, O., Fournol, J. F., and Oshchepkov, S.: A new
airbornePolarNephelometerforthemeasurementsofopticaland
microphysical cloud properties. Part I: Theoretical design, Ann.
Geophys., 15, 451–459, 1997.
Gayet, J.-F., Auriol, F., Minikin, A., Str¨ om, J., Seifert, M., Kre-
jci, R., Petzold, A., Febvre, G., and Schumann,U.: Quantitative
measurement of the microphysical and optical properties of cir-
rus clouds with four different in situ probes: Evidence of small
ice crystals, Geophys. Res. Lett., 29, 2230–2234, 2002.
Haag, W., K¨ archer, B., Str¨ om, J., Minikin, A., Ovarlez, J.,
Lohmann, U., andStohl, A.: Freezingthresholdsandcirruscloud
formationmechanisms inferred from in situmeasurements ofrel-
ative humidity, Atmos. Chem. Phys., 3, 1791–1806, 2003.
Intergovernmental Panel on Climate Change (IPCC), 1999: Avia-
tion and the Global Atmosphere, edited by Penner, J. E., Lister,
D. H., Griggs, D. J., Dokken, D. J., and McFarland, M., Cam-
bridge Univ. Press, Cambridge, UK, pp. 373, 1999.
Kristensson, A., Gayet, J.-F., Str¨ om, J., and Auriol, F.: In situ ob-
servations of a reduction in effective crystal diameter in cirrus
clouds near flight corridors, Geophys. Res. Lett., 27, 681–684,
2000.
K¨ archer, B. and Solomon, S.: On the composition and optical ex-
tinction of particles in the tropopause region, J. Geophys. Res.,
104, 27441–27459, 1999.
K¨ archer, B. and Lohmann, U.: A parameterization of cirrus cloud
formation: Heterogeneous freezing, J. Geophys. Res., 108, 4402,
doi:10.1029/JD2002003220, 2003.
K¨ archer, B. and Str¨ om, J.: The roles of dynamical variability and
aerosols in cirrus cloud formation, Atmos. Chem. Phys., 3, 823–
838, 2003.
Mannstein, M., Meyer, R., and Wendling, P.: Operational detection
of contrails from NOAA-AVHRR-data, Int. J. Remote Sensing,
20, 1641–1660, 1999.
Meyer, R., Mannstein, H., Meerk¨ otter, R., Schumann, U., and
Wendling, P.: Regional radiative forcing by line-shaped con-
trails derived from satellite data, J. Geophys. Res., 107,
doi:10.1029/2001JD000426, 2002.
Ogren, J. A., Heintzenberg, J., and Charlson, R. J.: In-situ sampling
ofcloudswithadroplettoaerosolconverter, Geophys. Res.Lett.,
12, 121–124, 1985.
Ovarlez, J. and van Velthoven, P.: Comparison of water vapor mea-
surements with data retrieved from ECMWF analyses during
POLINAT experiment, J. Appl. Meteor., 105, 1329–1335, 1997.
Ovarlez, J., van Velthoven, P., Sachse, G., Vay, S., Schlager, H.,
and Ovarlez, H.: Comparison of water vapor measurements from
POLINAT2withECMWFanalysesinhigh-humidityconditions,
J. Geophys. Res., 105, 3737–3744, 2000.
Ovarlez, J., Gayet, J.-F., Gierens, K., Str¨ om, J., Ovarlez, H., Auriol,
F., Busen, R., and Schumann, U.: Water vapour measurements
inside cirrus clouds in Northern and Southern hemispheres dur-
ing INCA, Geophys. Res. Lett., 29, 1813–1817, 2002.
Schr¨ oder, F., K¨ archer, B., Fiebig, M., and Petzold, A.:Aerosol states
inthefreetroposphereatnorthernmidlatitudes, J.Geophys.Res.,
107, 8126, doi:10.1029/2000JD000194, 2002.
Seifert, M., Str¨ om, J., Krejci, R., Minikin, A., Petzold, A., Gayet,
J.-F., Schlager, H., Ziereis, H., Schumann, U. and Ovarlez, J.:
Aerosol-cirrus interactions: A number based phenomenon at
all?, Atmos. Chem. Phys., 3, in press, 2003a.
Seifert, M., Str¨ om, J., Krejci, R., Minikin, A., Petzold, A., Gayet, J.-
F., Schumann, U., and Ovarlez, J.: In situ observations of aerosol
particles remaining from evaporated cirrus crystals: Comparing
cleanandpollutedairmasses, Atmos.Chem.Phys., 3, 10371049,
2003b.
Sonntag, D.: Advancements in the field of hygrometry, Meteorol.
Z., 3, 51–66, 1994.
Str¨ om, J., Heintzenberg, J., Noone, K. J., Noone, K. B., Ogren, J. A,
Albers, F., and Quante, M.: Small crystals in cirrus clouds: their
residue size distribution, cloud water content, and related cloud
properties, J. Atmos. Res., 32, 125–141, 1994.
Str¨ om, J. and Ohlsson, S.: In situ measurements of enhanced crystal
number densities in cirrus clouds caused by aircraft exhaust, J.
Geophys. Res., 103, 11355–11361, 1998.
Atmos. Chem. Phys., 3, 1807–1816, 2003www.atmos-chem-phys.org/acp/3/1807/