ACPAtmospheric Chemistry and PhysicsACPAtmos. Chem. Phys.1680-7324Copernicus PublicationsGöttingen, Germany10.5194/acp-16-11395-2016Precipitation susceptibility in marine stratocumulus and shallow cumulus
from airborne measurementsJungEunsileunsil.jung@gmail.comhttps://orcid.org/0000-0003-0970-2730AlbrechtBruce A.SorooshianArminhttps://orcid.org/0000-0002-2243-2264ZuidemaPaquitahttps://orcid.org/0000-0003-4719-372XJonssonHaflidi H.Department of Atmospheric Sciences, University of Miami, Miami, FL,
33149, USADepartment of Chemical and Environmental Engineering, University of
Arizona, Tucson, AZ, 85721, USADepartment of Hydrology and Atmospheric Sciences, University of
Arizona, Tucson, AZ, 85721, USANaval Postgraduate School, Monterey, CA, 93943, USAnow at: National Institute of Meteorological Sciences, Jeju, 63568, South KoreaEunsil Jung (eunsil.jung@gmail.com)14September20161617113951141322February201614March201627August201630August2016This 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/16/11395/2016/acp-16-11395-2016.htmlThe full text article is available as a PDF file from https://acp.copernicus.org/articles/16/11395/2016/acp-16-11395-2016.pdf
Precipitation tends to decrease as aerosol concentration
increases in warm marine boundary layer clouds at fixed liquid water path
(LWP). The quantitative nature of this relationship is captured using the
precipitation susceptibility (So) metric. Previously published
works disagree on the qualitative behavior of So in marine low
clouds: So decreases monotonically with increasing LWP or
cloud depth (H) in stratocumulus clouds (Sc), while it increases
and then decreases in shallow cumulus clouds (Cu). This study uses airborne
measurements from four field campaigns on Cu and Sc with similar instrument
packages and flight maneuvers to examine if and why So behavior varies as a function of cloud type. The findings show that
So increases with H and then decreases in both Sc and
Cu. Possible reasons for why these results differ from those in previous
studies of Sc are discussed.
Introduction
Cloud–aerosol interactions are considered to be one of the most important
forcing mechanisms in the climate system (IPCC, 2013). It is believed that
aerosols suppress precipitation in warm boundary layer clouds. However,
there is considerable disagreement on the magnitude and even on the sign of
how aerosol perturbations affect cloud fraction and lifetime (Stevens and
Feingold, 2009). Furthermore, aerosol effects on clouds and precipitation
are not readily separable from the effects of meteorology. The precipitation
susceptibility metric, So, quantifies how aerosol
perturbations alter the magnitude of the precipitation rate (R)
while minimizing the effects of macrophysical factors (i.e., meteorology)
(Feingold and Siebert, 2009). It is defined as
So=-dlnRdlnNd,
and is evaluated at fixed cloud macrophysical properties, such as cloud
thickness (H) or liquid water path (LWP). In Eq. (1), aerosol
effects are embedded in the cloud droplet number concentration
(Nd) variable since aerosols serve as cloud condensation
nuclei (e.g., as aerosol concentration increases, Nd
increases). The minus sign is used in Eq. (1) to achieve a positive value of
So due to the expectation that increasing aerosols reduce
precipitation (all else being fixed). Towards improving the representation of
precipitation in larger-scale models, the application of Eq. (1) has also
been studied using more highly resolved models and remote sensing (e.g.,
Feingold and Siebert, 2009; Sorooshian et al., 2009; Terai et al., 2015;
Hill et al., 2015). In the original work on So (Feingold and
Siebert, 2009), cloud-base R and Nd were used. Since
then, slightly different definitions of So have been applied.
For example, Sorooshian et al. (2009) used an aerosol proxy (e.g., Aerosol
Optical Depth and Aerosol Index) instead of Nd for their
satellite data analysis. Terai et al. (2012, 2015) further defined
precipitation susceptibility as the sum of the susceptibilities of drizzle
intensity (SI) and drizzle fraction (Sf),
SR=SI+Sf,
where SI is equivalent to So. The difference
between SI and So is how large a threshold of
precipitation is applied for calculating So or
SI. Other studies focus on the probability of precipitation
(POP), defined as the ratio of the number of precipitating events over the
total number of cloudy events. Spop is used in some studies of
precipitation susceptibility (e.g., Wang et al., 2012; Mann et al., 2014;
Terai et al., 2015), and is equivalent to the Sf used
within Terai et al. (2012). In addition to the different definitions of
precipitation susceptibility, various forms of R and
Nd (e.g., cloud-base, vertically integrated, or ground-based
values) with different data thresholds have been used for the calculation of
the precipitation susceptibility depending on the data available. In this
study, precipitation susceptibility indicates So as defined in
Eq. (1) unless otherwise stated.
In global climate models (GCMs), aerosol effects on rain rate are represented
by either a prognostic scheme or an empirical diagnostic scheme. When GCMs
consider aerosols, the rain rate R is often parameterized in terms
of LWP and Nd as Eq. (2)
R=LWPαNd-β.
Climate models typically assume a fixed value of the autoconversion
parameter (β in Eq. 2), ranging between approximately 0 and
2 (e.g., Rasch and Kristjansson, 1998; Khairoutdinov and Kogan, 2000; Jones
et al., 2001; Rotstayn and Liu, 2005; Takemura et al., 2005). Readers should
note that rain rates from liquid clouds are usually from two terms; one is
from autoconversion and the other is from accretion (see Sect. 3.3). Since
So in Eq. (1) includes contributions from both autoconversion
and accretion, in the case where accretion has little contribution to
rain rate, So may then be equivalent to the exponent
β in Eq. (2) at fixed LWP. Field studies of precipitating
stratocumulus (Sc) clouds have reported β values ranging
from 0.8 to 1.75 at fixed LWP (e.g., Pawlowska and Brenguier, 2003; Comstock
et al., 2004; vanZanten et al., 2005; Lu et al., 2009). Such single
power-law fits, however, do not capture the changes in So with
LWP or H, which is important since previous works have revealed
that the response of cloud rain rates to aerosol perturbations vary as a
function of LWP (or H).
The qualitative behavior of So has been studied for low clouds
using models, remote sensing data, and in situ measurements. For model
studies of warm cumulus clouds (e.g., the adiabatic parcel model of Feingold
and Siebert, 2009), So varies from 0.5 to 1.1 with
increasing LWP, and exhibits three regimes. At low LWP, not enough water is
available with which to initiate rain, and So is insensitive
to aerosol perturbations. At intermediate LWP, suppression of
collision–coalescence by the increased aerosols is most effective. We will
refer to this regime as the ascending branch of So following
Feingold et al. (2013). At high LWP, the precipitation rate is more strongly
influenced by the LWP, and So decreases with increasing LWP
(the descending branch of So). This LWP-dependent pattern of
So is supported by satellite observations (Sorooshian et al.,
2009, 2010) and large-eddy simulations (LESs) (Jiang et al., 2010) for warm
trade cumulus clouds. In contrast, Terai et al. (2012) showed that
SR monotonically decreased with increasing LWP and H
in Sc clouds based on in situ measurements acquired during the VAMOS
Ocean-Cloud-Atmosphere-Land Study Regional Experiment (VOCALS-REx) field
study, while their SI, similar to So in aforementioned
studies, did not reveal any significant change with H and
maintained a value of ∼ 0.6. These inconsistent results have
raised questions of how cloud type impacts behavior of So as a
function of either H or LWP.
To begin to unravel why differences in the various studies exist, Feingold
et al. (2013) showed in modeling studies that the time available for
collision–coalescence (tc) is critical for determining the
LWP-dependent behavior of So, and may be at least partly
responsible for some of the differences. Gettelman et al. (2013) also showed
how the microphysical process rates impact So in the NCAR
Community Atmosphere Model version 5 (CAM5) GCM. They showed that the
behavior of So with LWP differs between the GCM and the
steady-state model of Wood et al. (2009); the values of So
were constant or decreased with LWP in the steady state model (consistent
with Terai et al., 2012; Mann et al., 2014), whereas the GCM
So behavior was more consistent with Feingold and Siebert
(2009), Sorooshian et al. (2009, 2010), Jiang et al. (2010), Feingold et al. (2013), and Hill et al. (2015). In their study, altered microphysical process rates were
able to significantly change the magnitudes of So, but the
qualitative behavior of So with LWP remained unchanged (i.e.,
So increases with LWP, peaks at an intermediate LWP, then
decreases with LWP). More recently, Mann et al. (2014) analyzed 28 days of
data from the Azores Atmospheric Radiation Measurement (ARM) mobile facility
where the prevalent type of clouds are cumulus (20 %), cumulus under
stratocumulus (10–30 %), and single-layer stratocumulus (10 %). They
showed that Spop slightly decreased with LWP. Terai et al. (2015) estimated precipitation susceptibility
(SI+Spop) in low-level marine
stratiform clouds, which included stratus and stratocumulus clouds, using
satellite data. The values of So in their study generally
showed similar behavior to that reported by Mann et al. (2014). Hill et al. (2015) examined how the representation of cloud microphysics in climate
model contributes to the behavior of So. They found that
single-moment schemes produce the largest uncertainty in So.
Only through increasing the number of prognostic moments (i.e., multi-moment
schemes capable of prognosing the rain droplet number as well as mass) could
the dependence of So on a particular scheme be reduced.
The inconsistent behavior of So in previous studies for warm
boundary layer clouds motivates the current study. The focus of this paper
is to examine and compare the qualitative behavior of So in Cu
and Sc using similar airborne measurements encompassing four field
campaigns. Two were focused on Sc clouds (VOCALS-REx and the Eastern Pacific
Emitted Aerosol Cloud Experiment, Sect. 2.2) and two campaigns targeted Cu
clouds (Barbados and Key West Aerosol Cloud Experiments, Sect. 2.3). The
strength of these four field campaigns' airborne measurements is that the
same research aircraft was deployed with a similar flight strategy and
instrument packages, facilitating a comparative analysis. Each of the four
field experiments sampled an area of about 100×100 km, and
thus, the mean interrelationships examined are representative of the GCM
spatial resolution. Data and methods are discussed in Sect. 2, followed by
results and discussion in Sects. 3 and 4, respectively. The findings are
summarized in Sect. 5. Acronyms used in this study are listed in Table A1
of the Appendix.
Data and methodsTO aircraft
The Center for Interdisciplinary Remotely Piloted Aircraft Studies (CIRPAS)
Twin Otter (TO) research aircraft served as the principal platform from
which observations for these four experiments were made. During these four
deployments, the TO supported similar instrument packages, and performed
similar cloud sampling maneuvers, including vertical soundings and level-leg
flights below, inside, and above the clouds. Each research flight lasted
∼ 3–4 h. The TO included the following three
in situ probes for characterizing aerosol, cloud, and precipitation
size distributions: the Passive Cavity Aerosol Spectrometer Probe (PCASP),
Cloud Aerosol Spectrometer (CAS), and Cloud Imaging Probe (CIP), with each
resolving particles of diameters 0.1–2.5 µm, 0.6–60 µm, and
25–1550 µm, respectively. A zenith-pointing 95 GHz Doppler radar was
mounted on top of the aircraft and detected cloud and precipitation
structures above the aircraft. Detailed information of the instruments on
the TO and flight strategies is provided elsewhere (Zheng et al., 2011;
Jung, 2012). All the instruments were operational during the flights
analyzed in this study except for the cloud radar, which was not operational
during the VOCALS TO flights.
So is calculated from Eq. (1) within bins of the cloud
thickness H. H was estimated as the height difference
between cloud tops and bases. Cloud tops were determined by the cloud radar
with a time resolution of 3 Hz and vertical resolution of 24 m (5 m) in
height for Cu (Sc). Cloud bases of Cu were determined by the lifting
condensation level (LCL) calculated from the average thermodynamic
properties of the sub-cloud layer for a given day. The LCLs varied little for
Cu, for example, during the Barbados Aerosol Cloud Experiment (Sect. 2.3);
the LCLs were 653.9 ± 146 m on average from the aircraft measurements,
which agreed with the 2-year LCL climatology in this region (700 ± 150 m) as documented in Nuijens et al. (2014). Although it is not shown in
this study, So was also estimated by using the cloud-base
heights determined from the Cu cloud-base level-leg flights; these results
were similar to those shown in this study.
In stratocumulus clouds, cloud tops are well defined due to the strong
capping temperature inversion (see Zheng et al., 2011) and cloud bases vary
more than tops (e.g., Fig. 2 of Bretherton et al., 2010). As a result, the
way that the cloud base is determined may affect So since the
changes in cloud base alternatively can change the cloud thickness.
Therefore, we estimate So using three different definitions
for cloud base. The first method is with LCLs calculated from the average
thermodynamic properties of the sub-cloud layer (shown as cb-lcl in Fig. 4,
same as Cu). For the second and third definitions (cb-local and cb-mean),
cloud bases are determined from the lowest heights where the vertical
gradients of liquid water contents (LWCs) are the greatest from the LWC
profiles. The LWC profiles are obtained (i) when the aircraft enters the
cloud decks to conduct level legs (cb-local), and (ii) from the nearest one
or two soundings to the cloud-base level-leg flights. The average height of
these two lowest heights (cb-mean, the average of i and ii) is used in this
study, along with cb-lcl and cb-local (Fig. 4 later). In general, the
heights approximately corresponded to the lowest heights that the LWCs
exceeded by 0.01 g m-3. So was also
estimated by using the heights from the cloud-base level-leg flights as the
cloud bases as was done for Cu, and the qualitative behavior of So
was preserved (not shown).
Nd and R were calculated from the drop size
distribution (DSD), which is obtained from CAS (forward scattering) and CIP
probes during the cloud-base level-leg flights, respectively. The CAS probe
acquires data every 10 Hz and then the DSDs at each channel are averaged to
1 Hz. The CIP acquires data every 1 s. The cloud radar samples at 3 Hz
and then is averaged to 1 Hz to match the probe data. Therefore,
Nd, R, and H in Eq. (1) were calculated in 1 s resolution
(except for VOCALS-REx; see Sect. 2.4). The impact of using
1 s data on the So estimates will be discussed
later in Sect. 3.2. R is defined as
R=π6∫25µm1550µmN(D)D3u(D)dD,
where u(D) is the fall speed of a drop with diameter D.
Three fall speed formulations are used: (1)
u=k1r2 with k1≈1.19×106 cm-1 s-1
was used for cloud droplets up to 30 µm radius; (2)
u=k3r with k3≈8×103 s-1 was used for
the size range of 40 µm <r< 0.6 mm; and
(3) u=k2r1/2 with k2≈2.01×103 cm1/2 s-1 for droplets of 0.6 mm <r<2 mm.
Stratocumulus cloud field campaigns: VOCALS-REx and E-PEACE
The geographical location of each field campaign (blue
solid). E indicates E-PEACE, K indicates KWACEX, and B shows BACEX. The
entire domain of VOCALS-REx is displayed as a solid grey box with domains of
C-130 (dashed grey) and TO (solid blue) flights.
From October to November 2008, the VOCALS-REx took place over the southeast Pacific
(69–86∘ W, 12–31∘ S), an area
extending from the near coastal region of northern Chile and southern Peru
to the remote ocean (Zheng et al., 2011; Wood et al., 2011; also see Fig. 1).
Three aircraft were deployed during VOCALS from 14 October to 15 November
(NSF/NCAR C-130, DOE G-1, CIRPAS TO). The TO sampled more coastal
marine stratocumulus decks near 20∘ S, 72∘ W (Fig. 1)
than the other two planes. Readers should note that the data in Terai et al. (2012)
used for their SR calculations, were also obtained from
VOCALS. However, their results were based on NSF/NCAR C-130 flights that
sampled cloud decks away from the coastal area (Fig. 1). Wood et al. (2011)
provided a comprehensive description of VOCALS experiments and Zheng et al. (2011)
provided a description of TO aircraft data during the VOCALS. TO data
from flights with decoupled boundary layers, abnormally higher cloud bases,
and moist layers above cloud tops were excluded, reducing the total number
of flights analyzed to 13 from the original total of 18 (Table 1).
Scatter diagrams of cloud droplet number concentrations,
Nd, and precipitation, R, for four field campaigns.
Colors indicate cloud thickness H. The dashed line indicates an
R value of 0.14 mm d-1.
From July to August 2011, the Eastern Pacific Emitted Aerosol Cloud
Experiment (E-PEACE) took place off the coast of Monterey, California, to
better understand the response of marine stratocumulus to aerosol
perturbations (Russell et al., 2013). E-PEACE included sampling controlled
releases of (i) smoke from the deck of the research vessel Point Sur, and (ii) salt aerosol from the TO research aircraft, along with
sampling (iii) exhaust from container ships transiting across the study area
(see Fig. 2 from Russell et al., 2013). During 9 out of 30 E-PEACE
flights, salt powder (diameter of 1–10 µm) was directly introduced
into the cloud decks to examine the effects of giant cloud condensation
nuclei (GCCN) on the initiation of warm precipitation (Jung et al., 2015).
After excluding the seeding cases and the non-typical Sc decks, 13 flights
remained from which we analyzed data (Table 1). Detailed information about
E-PEACE and TO data can be found elsewhere (Russell et al., 2013;
Wonaschütz et al., 2013).
Marine cumulus cloud field campaigns: BACEX and KWACEX
Shallow marine cumulus clouds are by far the most frequently observed cloud
type over the Earth's oceans, yet remain poorly understood, and have not
been investigated as extensively as oceanic stratocumulus. The marine
environments in the Caribbean Sea and the Atlantic Ocean provide an
excellent area to sample shallow marine cumulus clouds with a high
propensity to precipitate. In addition, African dust is transported
westward off of Africa periodically over the North Atlantic, affecting
clouds in its path including around Barbados and Key West, and thus
providing an excellent opportunity to observe aerosol–cloud–precipitation
interactions. To better understand such interactions in these trade cumuli
regimes, the Barbados Aerosol Cloud Experiment (BACEX) was carried out off
the Caribbean island of Barbados during mid-March and mid-April 2010 (Jung
et al., 2013), and the Key West Aerosol Cloud Experiment (KWACEX) during May 2012
near Key West (Fig. 1). For the BACEX, we analyzed 12 flights (Table 1). Readers are referred to Jung et al. (2016) for detailed information
about the cloud and aerosol properties during the BACEX. The marine
atmosphere during KWACEX was dry overall. A total of 6 out of 21 flights sampled
shallow marine cumulus clouds, of which 4 had sufficient data for
analysis (Table 1).
So calculation details
The distribution of Nd and R, with the corresponding
H, is shown in Fig. 2 for each field campaign as scatter diagrams
of Nd and R. All data shown in Fig. 2 were obtained
during the cloud-base level-leg flights. The southeast Pacific (SEP) Sc
decks (VOCALS, Fig. 2a) were overall drier and more polluted than those in
the northeast Pacific (NEP) Sc decks (E-PEACE, Fig. 2c); R=0.03 mm day-1 (median) and Nd=232 cm-3 in VOCALS,
but R=1.04 mm day-1 and Nd=133 cm-3 in E-PEACE. During E-PEACE, high Nd was observed in
a few cases, (e.g., Nd>400 cm-3 in Fig. 2c),
and they were likely associated with the emitted aerosols from the ship
exhaust and smoke (Russell et al., 2013; Wang et al., 2014; Sorooshian et
al., 2015). The marine environments of the Caribbean Sea showed wide
variations of R (e.g., order of 10-2 to 102 mm day-1; Fig. 2b and d). The Barbados campaign sampled the most
pristine environment of the four campaigns (Nd<350 cm-3, Nd=61 cm-3 on average), reflecting the
isolated location of the island in the North Atlantic even though the
experiment period included the most intense dust events of 2010 (Jung et
al., 2013). The marine environment near Key West was more polluted than
Barbados throughout the KWACEX campaign (Fig. 2d, Nd=206 cm-3 on average).
So was about 0.62 for E-PEACE (linear regression correlation
coefficient r=0.34), if calculated using all the individual 1 Hz data
points shown in Fig. 2 where H ranges from ∼ 100 m to
500 m. However, So was about 0.42 (r=0.21) if one
rainy day (shown as double circles in Fig. 10 later) was excluded from the
analysis, suggesting the artifact of wet scavenging (see Sect. 4), a
different predominant cloud microphysical process (autoconversion vs.
accretion) or the influence of macrophysical properties other than
H. These E-PEACE So values agree with values
estimated in previous campaigns in the same NEP region for
H∼ 200–600 m: So∼ 0.46–0.48 using H, and So∼ 0.60–0.63
using LWP (Lu et al., 2009). So during VOCALS is about 1.07
(r=0.46) for H∼ 150–700 m. Overall,
So values in this study are within the range of
So from the previous field studies of precipitating
stratocumulus clouds (So∼ 0.8 to 1.75 for a
fixed LWP in the studies of Pawlowska and Brenguier, 2003; Comstock et al.,
2004; vanZanten et al., 2005). Values of So for BACEX and
KWACEX are about 0.89 (r=0.38) and 0.77 (r=0.39),
respectively.
Dates (in mm/dd format) used for this analysis during each experiment.
RF indicates the research flight. However, note that RFs from E-PEACE and
VOCALS are not the same as RFs from Russell et al. (2013) and Zheng et al. (2011), respectively.
The daily mean cloud thickness (mean ± 1σ) for VOCALS is
shown with the H category (the group number is shown in the parenthesis).
See the details in Sect. 2.4.
Numbers inside brackets indicate e-folding time (seconds) of
Nd and R.
NA = not available
Examples of scatterplots used to calculate precipitation
susceptibility So (i.e., the slope) for E-PEACE. Black dots
indicate data points for an H interval between 160 and 190 m.
Numbers on the bottom right (blue) indicate the total number of data used.
So and linear coefficient (r) values are
shown in the upper right corner. Precipitation, R, increases
downward in y ordinate, and Nd increases toward the right
direction in x abscissa.
Although single power-law fits for a given field campaign give the general
sense of So values, they do not show the qualitative behavior
of So with H, which reveals which thickness is most
susceptible to aerosol perturbations. To further examine this,
So is calculated by assigning R and Nd
into the given intervals of cloud thickness for each campaign. The width of
each H interval is taken to be 30 m for Sc and 50 m for Cu. The
H intervals are arbitrary, but chosen to contain a similar number
of data points within each interval and provide a robust So
regardless of the interval choice. Within each H interval, we
performed a linear regression to find a best fit for the natural log of the
precipitation rate against natural log of Nd, and the
So is the slope of the fit (see Figs. 3 and 6, for example).
Cloud data are included in the analysis if the given precipitation rate is
greater than a threshold of 0.001 mm day-1. The low R
threshold is chosen to include precipitating and very lightly precipitating
clouds. The 0.001 mm day-1 threshold is indeed very low; the uncertainty
in rain rate calculation is larger than 0.001 mm day-1 threshold. For all
intents and purposes, the 0.001 mm day-1 threshold is equivalent to no
precipitation. The impacts of the R threshold and H
intervals on the So estimates are discussed in Appendices B and
C, respectively. An example of So is shown in Fig. 3 from
E-PEACE using every 1 s cloud data point (i.e., Nd and
R) for H between 160 and 190 m. The slope (i.e., linear
fit) in Fig. 3 corresponds to an So value of 0.24. The value
of So (0.24) is then plotted in the corresponding H
on the H–So diagram (e.g., Fig. 4 at the H
of 174 m, which corresponds to the average H of the interval). The
same procedure is repeated for all H intervals to obtain the
complete pattern of So with H. We tested and applied
a few criteria in the So calculations, such as minimum
R thresholds, and the total number of cloud data points and spans
of Nd for a given H interval. Based on these
sensitivity tests, we calculated So exclusively if
Nd varied a sufficient amount (e.g., dln(Nd)
spans at least 2.2) for a given H interval since little variation
of Nd does not provide the proper perturbation of aerosols.
For example, in Fig. 3a, dln(Nd) spans about 3.5. Slightly
different and broader criteria were applied for Cu mainly due to the lower
number of data points sampled in Cu. However, the qualitative behavior of
So was robust as long as the variation of Nd was
sufficiently large, regardless of the other criteria, although the details
were different (e.g., Fig. B1). In Fig. 4, most of the slopes are statistically
significant at the 99 % confidence level (e.g., filled symbols).
The number of data points used to calculate So and the
linear correlations and the P values indicating the statistically
significant level of confidence for the fitted lines are summarized in Table A2
for given H intervals. Additionally, So is calculated by considering e-folding time and by randomly resampling the
flights (Sect. 3.2), and the results are robust. This will be discussed
later.
Precipitation susceptibility, So, estimated
with aircraft measurements for (a) Cu (12 flights of BACEX and 4 flights
of KWACEX) and (b) Sc (13 flights of E-PEACE and VOCALS-REx). E-PEACE
So is estimated from (i) the cloud-base height, which
is identified using LCLs (cb-lcl) and (ii) from the vertical structures of
LWCs (lowest height where the vertical gradient of LWC is the greatest) as
the aircraft enters the cloud deck to conduct the cloud-base level-leg
flight (cb-local), and (iii) from the averaged cloud-base heights from the
nearby soundings and cb-local (cb-mean). Filled circles are statistically
significant at 99 % confidence level. The number of data points used for
So estimates and their statistical significance are shown in
Table A2.
So during VOCALS is calculated in slightly different ways from
other experiments since the cloud radar failed. First, H is
estimated from the vertical structure of LWC for each day (daily mean
H). Once H is determined for each flight, it is assigned
to a certain H bin. For example, H values of 9 November (164 ± 18 m) and 10 November (194 ± 21 m) are similar and thus assigned to the
same H bin (i.e., group 1 in Table 1). VOCALS H is
classified into four distinct groups. Once Nd and R
are assigned to the corresponding H, So then is estimated by
using all the data points that are assigned to the H group (e.g., Fig. 6b–i, later on).
LWP is commonly used as the macrophysical factor when quantifying Eq. (1).
However, in this study, we use H as a macrophysical factor since we
aim to compare So for both Sc and Cu. H corresponds
well to LWP for adiabatic clouds, for which LWP ∼H2. The adiabatic assumption, which may be valid in Sc, is not
valid in Cu (Rauber et al., 2007; Jung et al., 2016) to calculate LWP.
Moreover, the TO did not carry an instrument that measures LWP directly such
as a G-band vapor radiometer (e.g., Zuidema et al., 2012). Consequently, the
direct comparison with previous results of So with LWP (e.g.,
quantitative) is not possible. We also note that LWC decreases as drizzle
rates increase (e.g., see Fig. 8d of Jung et al., 2015). Consequently,
clouds that are precipitating (higher R) may have a LWP that is
lower than the adiabatic value, and a cloud with a small R may have
a LWP close to the adiabatic value. It should be also noted that the ranges
of H (and possibly LWP) differ substantially between Cu and Sc. For
example, H of Cu in this study can be as high as 1700 m, whereas
H of Sc is generally less than 500 m (e.g., Fig. 4). Additionally,
H for clouds that begin to precipitate may differ in Sc and Cu.
Further, the LWP for clouds that precipitate would be sub-adiabatic and
would have a smaller value of LWP than the LWP for non-precipitating clouds.
Consequently, So that is calculated from cloud fields with
diverse cloud types (e.g., Mann et al., 2014; Terai et al., 2015) may be
complicated since LWP is shifted to smaller values for (heavily)
precipitating clouds, and the H at precipitation initiation may
differ between cloud types. In general, the results are used with caution
when comparing with other studies in quantifying So since the
dominating cloud process and the choices applied in how to calculate
parameters involved with Eq. (1) can differ widely (e.g., Duong et al.,
2011).
ResultsSo in Sc and Cu
In this section, we show So calculated in three different
ways. First, So is calculated with 1 s data
(Fig. 4) for BACEX, KWACEX, E-PEACE, and VOCALS. Second, So is
calculated with reduced data points that are averaged over the e-folding
time of Nd. We show the results for BACEX, E-PEACE, and VOCALS
(Figs. 5 and 6). Lastly, So is calculated with randomly
resampled E-PEACE flights (Figs. 8 and 9). We will show the results in
turns.
So estimated with aircraft measurements for
(a) BACEX (Cu) and (b) E-PEACE (Sc). The 1 s data of individual flights
are reduced by averaging over the e-folding time of Nd for
each flight prior to the calculation.
So for VOCALS TO flight is calculated with
1 s data (grey) and cloud data that are averaged over an e-folding time
for each day (blue). The ln(Nd) and -ln(R) diagram
is shown for each H interval. The horizontal bar in (a) indicates
±1σ. So is calculated for the cloud data in
groups with similar H (shown in Table 1).
Daily mean values of Nd and R for
the 13 E-PEACE flights. Numbers indicate the flight numbers shown in Table 1.
So as a function of cloud thickness for (a) 12
E-PEACE flights, for groups A and B shown in Fig. 7. (b)So calculated
with randomly resampled RFs within (b) group A and (c) group B. RFs indicate
research flights. Dates are indicated in mm/dd format.
So as a function of H is shown in Fig. 4a for Cu.
So is calculated from Eq. (1) with Nd and
R that are sampled during the cloud-base level-leg flights at
1 s resolution. Cloud level-leg flights usually last 7–15 min on
average, with an aircraft speed of 50–60 m s-1. In Fig. 4a,
So during BACEX is insensitive to H, fluctuating around zero for clouds shallower
than 500–600 m, above which So begins to increase rapidly with
a peak of ∼ 1.6 near H∼ 1400 m. After
that, So starts to decrease as H increases. The
So during KWACEX follows S0 from BACEX,
especially in the thicker cloud regime where the majority of KWACEX data
were sampled.
The qualitative behavior of So for Sc is shown in Fig. 4b.
So during E-PEACE shows H-dependent So
patterns that are similar to those from BACEX. In the small H
regime (H<240 m), So is almost constant at
∼ 0.2. For H>∼ 240 m,
So increases gradually with increasing H and peaks at
So∼ 1.0 near H∼ 350–400 m. After
that, So decreases with increasing H. Figure 4b
further shows that the overall pattern of So is similar
regardless of how the cloud bases were determined, although the H
at which So peaks changes slightly (cb-mean, cb-local,
cb-lcl).
During VOCALS, So increases with increasing H, from
So∼ 0.1 near 170 m to So∼ 0.5 near 300 m. A minimum So value is shown
near H∼ 640 m. The negative values of
So in the largest H regime possibly result from
uncertainties in the So estimation or in unaccounted-for
macrophysical properties such as cloud lifetime. The failure of the cloud
radar during VOCALS was responsible for the fewer (four) H groups
(Table 1), leading to a low number of So values. Additionally, no data were available for H∼ 350–600 m (Fig. 3). The results of VOCALS clearly show the disadvantage of no cloud radar
(i.e., high resolution of LWP or H) for the So
estimates.
So calculated with an e-folding time and randomly
resampled flights
The dependence of 1 s data (Nd, R) on each other
is tested two ways. First, we calculated So by considering the
e-folding timescale (Leith, 1973) in which an autocorrelation decreases by
a factor of e. Secondly, we calculated So by randomly
resampling the flights. The e-folding time of Nd during
E-PEACE was found to vary from 4 to 10 min, while the
e-folding time of R varied from a few seconds to 1 to 2 min. The e-folding time of Nd within the VOCALS TO flights
varied from 2 to 6 min, and for the cloud-base precipitation was
less than (or approximately) 1 min (for a horizontal distance of less
than 3 km, consistent with Terai et al., 2012). In the case of BACEX (Cu),
the overall e-folding times were much shorter, varying 1–2 min
for Nd and less than 1 min for R. The e-folding times of
Nd and R are summarized in Table 1 for VOCALS,
E-PEACE, and BACEX. KWACEX was not included since there were only four
flights.
We calculated So with data averaged over the upper bounds of
the e-folding time (i.e., e-folding time of Nd) for E-PEACE,
BACEX, and VOCALS flights, and the qualitative behavior of So
reported with 1 s data is unchanged: So increases with
H, then peaks before it decreases again (Fig. 5 for BACEX and
E-PEACE and Fig. 6 for VOCALS). However, it should be noted that the
H that So peaks at is shifted toward the lower
H consistent with the results of Duong et al. (2011). The shift of
H to the lower H is substantial in Sc where the overall
H is smaller than H of Cu. Additionally, the effect of the
H interval on the So estimates is discussed in
Appendix C. In general, the results are robust regardless of the H
interval. However, if the H interval is chosen across a cloud
thickness range in which So changes substantially, the pattern
of So can be changed, indicating that the finer H
interval provides a more accurate So variation.
Second, we estimated So by randomly resampling the flights of
E-PEACE to see whether the sequential 1 s samples are statistically
independent. So calculated with random flights, at first
glance, showed two distinctive types of behavior (not shown, but similar to
Fig. 8a shown later). One is a similar pattern to that of the current
So shown in Fig. 4 while the other is an almost constant
So near zero. The cloud data sampled during E-PEACE formed two
groups (denoted as A and B in Fig. 7). The So pattern
calculated with cloud data of group A is similar to So shown
in Fig. 4: So is constant at lower H, followed by an
increase then decrease (Fig. 8a). In contrast, So values
calculated from group B were relatively constant near zero (blue in Fig. 8a) or So
with the descending branch only (e.g., light blue in Fig. 8c). Further analysis revealed
that the two RFs (RF13 and RF03) that have relatively small Nd
with high R explain the differences in the So
patterns (Fig. 9). If So is calculated with cloud data that do
not include data from clean with heavy precipitating environments (i.e.,
RF13 and RF03), So shows a similar pattern as that in
Fig. 4.
The effects of high precipitation (RF03 and RF13) on
So estimates. (a)So calculated for 13 flights
during E-PEACE in addition to when either, or both, RF03 and RF13 are
excluded. RF03 and RF13 are the flights of high precipitation rates. (b)So is calculated from group A with and without RF03 and RF13.
R and Nd information for each flight is shown in Fig. 7.
The effect of autoconversion and accretion processes on So
For cloud droplets to become raindrops (typical diameters of cloud droplets
and drizzle drops are about 20 and 200 µm, respectively; Rogers and
Yau, 1989), they have to increase in size significantly by the
collision–coalescence process (autoconversion and accretion) Here,
autoconversion primarily refers to faster-falling large cloud droplets that
collect smaller cloud droplets in their paths as they fall through a cloud
and grow larger; accretion refers to precipitation embryos that collect
cloud droplets. In the intermediate LWP regime where So
increases with LWP or H (ascending branch of So) the
autoconversion process dominates. On the other hand, in the high LWP regime
where So decreases with LWP or H (descending branch
of So) the accretion process dominates (Feingold and Siebert,
2009; Feingold et al., 2013). The transition from the dominance of
autoconversion to accretion is reported to occur when De
exceeds ∼ 28 µm, and has been used as a rain initiation
threshold in Sc (e.g., Rosenfeld et al., 2012). Jung et al. (2015) also
showed that the precipitation embryos appeared (and initiated warm rain)
when the mean droplet diameters were slightly less than 30 µm from
the salt seeding experiments during E-PEACE, in the NEP Sc decks (e.g., see
Table 3, Figs. 6a and 7 in their study). Figure 10a shows that clouds
during VOCALS consisted of numerous small droplets (D<15µm in Fig. 10a), which primarily are involved with the autoconversion
process except for one flight (D∼ 37 µm,
RF09, 1 November). The dominance of smaller droplets during VOCALS TO flights
agree with the dominance of the ascending branch of So in Fig. 4b. On the other hand, E-PEACE Sc clouds are composed of larger-sized
droplets as well as small droplets (Fig. 10b).
Distribution of effective diameters (mean ± 1σ) for (a) VOCALS TO flights and for (b) E-PEACE. Cloud droplets on
11 August are shown as double circles in (b). Black and red indicate cloud droplets sampled from cloud-base and mid-cloud heights, respectively.
Discussion
This study shows the consistent behavior of So as a function
of a key macrophysical cloud property regardless of cloud type; i.e.,
So increases with increasing H (ascending branch) and
peaks at intermediate H before So decreases with
H (descending branch) in both Sc and Cu (Fig. 4). The results from
marine cumulus clouds (BACEX and KWACEX) are consistent with previous
modeling and observational studies of warm cumulus clouds (Sorooshian et
al., 2009, 2010; Jiang et al., 2010; Duong et al., 2011; Feingold and
Siebert, 2009; Feingold et al., 2013). However, So values
estimated from marine stratocumulus clouds (E-PEACE and VOCALS) are
inconsistent with previous in situ observations of warm stratocumulus clouds
(Terai et al., 2012; Mann et al., 2014), but are consistent with previous
satellite observations of weakly precipitating Sc (Sorooshian et al., 2010),
global climate model simulations (Gettelman et al., 2013; Hill et al.,
2015), and box and parcel model studies (Feingold et al., 2013) of Sc.
Possible reasons for why the current results differ from those in previous
studies of Sc are discussed here mainly by comparing results to those from
the Terai et al. (2012) study. The inconsistent behaviors of
So between our study and theirs may be due to a number
of factors. One of the most fundamental reasons could be in the differences
in the cloud fields that were sampled. In the SEP Sc decks, drizzle
intensity and frequency tend to increase westward from the coast (e.g.,
Bretherton et al., 2010) and their data set included several pockets of open
cells (POCs) with strong precipitation (personal communication with C.
Terai). It should be noted that the VOCALS C-130 flights (Terai et al., 2012)
sampled the cloud fields along 20∘ S (mainly over the open
ocean), whereas the VOCALS TO flights sampled the Sc decks near the
continents (Fig. 1). The westward increases in frequency and intensity of
drizzle coincident with the westward decrease in aerosols and
Nd, and also with larger LWP over the open ocean (e.g.,
Zuidema et al., 2012), suggest that the discrepancy possibly is
contributed to the different cloud microphysical process working on the
cloud field (autoconversion vs. accretion processes). Indeed, Gettelman
et al. (2013) showed that the accretion process dominated during VOCALS
C-130 flights; the accretion to autoconversion ratio was above 1 for all LWP
ranges during VOCALS observation (e.g., Fig. 5a in their studies).
Therefore, the enhanced (major) accretion process appears as a descending
branch of So predominantly. Hill et al. (2015) also showed
the monotonic decrease of So with LWP in the case that the
cloud data consist of exclusively larger particles (e.g.,
radius > 20 µm).
Second, the higher R threshold that Terai et al. (2012) used could
contribute to the discrepancies. Terai et al. (2012) used R= 0.14 mm d-1
as a minimum R threshold to estimate
So, where 0.14 mm d-1 corresponds to -15 dBz from
the Z–R relationship that they used (R=2.01Z0.77 from
Comstock et al., 2004). This R threshold is possibly too high to
capture the autoconversion processes that occur in more lightly
precipitating clouds such as clouds sampled during VOCALS TO flights. As a
result, the high value of minimum R threshold may primarily capture
the accretion process, which may contribute to the descending branch of
So in their study. As an example, this R threshold
rejects all the data in Fig. 2a (VOCALS TO flights) except for 1 day
(RF09, 1 November) when the mean effective diameter (at cloud base) is about 37 µm (Fig. 10a) and
the accretion process dominates for the day. Further, the impact of the
R threshold on the So estimates is evident in Fig. B2. Figure B2 shows that So decreases as the larger minimum
R threshold is used, in particular at larger H. Figure 9
also shows how clouds of low Nd with high R (e.g.,
RF03 and RF13 of E-PEACE) alter the behavior of So. The choice
of minimum R threshold can change the data set that will be used for
the estimates of So. The So metric is designed
to show the impact of aerosols on precipitation; as aerosol increases,
smaller sizes of numerous droplets form, and those droplets suppress the
collision–coalescence process, and in turn, precipitation. Therefore, to
study the extent that aerosols suppress precipitation, it would be more
appropriate to encompass the full range of weakly to heavily precipitating
clouds that include both autoconversion and accretion processes. It is also
noted that the framework of precipitation susceptibility is to measure the
impact of aerosol perturbations on the precipitation suppression, and thus,
the concept of So may not adequately apply to the clouds that
are already heavily precipitating since the accretion process has little
dependence on Nd. In addition to decreasing the LWP, the
precipitation itself can scavenge aerosols, leading to lower
Nd.
A visual description of (a) the effect of wet scavenging
and (b–c) the impact of an increase in rainfall rate for a given range of
Nd on the estimate of So. The solid line (with and without
black filled circles) represents true (to be expected) So, whereas the dashed
line (with and without gray filled circles)
indicates observed (responded) So. R increases downward on y
ordinate and Nd increase toward the right direction on x
abscissa. The direction of arrows inside the figure indicates the direction of So changes.
Third, the overall high values shown in Terai et al. (2012)
(So begins with around 3 near H∼ 50 m and ends with So∼ 0.8 near
H∼ 500 m) may reflect the effects of wet scavenging
(Fig. 11a; see also Duong et al., 2011), especially by considering that
their data set included several POCs with strong precipitation. We also noted
that So calculated from the 13 E-PEACE flights was about 0.62.
However, So calculated from 12 E-PEACE flights that excluded
one rainy day was about 0.42, which is consistent with larger
So in the presence of (heavy) precipitation possibly due to
the wet scavenging (but it is also possible the lower So is
due to the microphysical process). Consistently, So values
calculated from 9 BACEX flights (Cu), which excluded 3 heavy
precipitation cases, were also shifted to lower values than those estimated
from the entire 13 flights (not shown).
Fourth, Terai et al. (2012) used column-maximum Z and then
converted the Z to R by using a Z–R relationship for those
time periods when the lidar could not determine the cloud-base height due to
interference from heavy precipitation. This procedure can overestimate
precipitation for a given Nd. If the procedure (i.e.,
overestimates of R) occurs in a low Nd regime (e.g., left
half of the dotted line in Fig. 11b), the steeper slope (i.e., higher
So) would be obtained (Fig. 11b). If the procedure happens in
a high Nd regime, the lower slope would be attained (Fig. 11c). Based on Fig. 1 of their study, the former scenario
(Fig. 11b) would
occur, resulting in higher So than expected.
Fifth, the Z–R relationship that Terai et al. (2012) used
(R=2.01Z0.77, following Comstock et al. (2004)'s
Z=25R1.3) was derived for stratocumulus off of the coast of Peru,
using a shipboard scanning C-band radar. The Sc sampled during the VOCALS
C-130 flights may have a different microphysical process from which the
original Z–R relationship was derived. The microphysical processes are
responsible for the formation of DSD, and the variability of DSD determines
the theoretical limit of precipitation accuracy by radar via Z–R
relationship. That being said, changes in DSD imply different Z–R
relationships. The DSD variability (e.g., day to day, within a day, between
physical processes and within a physical process) causes about 30–50 % of
errors in R estimates with a single Z–R relationship (e.g., see Lee
and Zawadzki, 2005, and references therein). Besides, the Comstock et al. (2004) Z–R
relationship was derived from drop sizes ranging from 2 to 800 µm in diameter (for drops larger than 800 µm,
extrapolation was used). The Sc from VOCALS C-130 flights included several
POCs, while the clouds that the Z–R relationship was derived from were
characterized by persistent Sc, sometimes continuous and other times broken
with intermittent drizzle throughout. Therefore, using the Z–R relationship
of Comstock et al. (2004) may result in some additional uncertainties in
R estimates in Terai et al. (2012) as the error of the Z–R relationship
becomes larger in the bigger drop sizes (Z and R are
proportional to ∼D6 and ∼D4, respectively). Further, applying a Z–R relationship to
W-band (3 mm) radar returns is not valid if there are any droplets greater
than 1 mm since non-Rayleigh scattering (Mie effects) can dominate the radar
reflectivity. Note that the Terai et al. (2012) R retrievals were
made with a W-band radar. However, it is also true that the in situ sampling
of rain used in this study may miss a lot of raindrops because of the small
sample volume of the probe. The errors in R estimates with a single
Z–R relationship or R measured from probes, however, may not
critically affect the differences in So between studies as the
So metric is less sensitive to data uncertainty by
using the logarithmic form of the data (Eq. 1).
Lastly, Terai estimated Nd from the sub-cloud aerosols using
an empirical relationship, which may also contribute to the differences.
According to Jung (2012 in Fig. 4.5), the sub-cloud aerosols well represent
the cloud-base Nd in the updraft regime, although these
results are shown for the marine shallow cumulus clouds. Similarly, using
the aerosol proxy from the satellite data for the So
calculation also needs caution. Jung et al. (2016) showed that aerosol
optical depth (AOD) is not always a good indicator of the sub-cloud layer
aerosols especially when the fine particles from long-distance continental
pollution plumes reside above the boundary layer (e.g., Figs. 4–5 their
study). Mann et al. (2014) used a sub-cloud 10 m CCN (at 0.55 %
super-saturation) for the So calculation and showed a
decreasing trend of So with LWP as Terai et al. (2012) but
their overall So was smaller than those estimated from other
field studies. In cases where sub-cloud aerosols are used for the
So estimates, these estimates give a smaller
So than those using Nd due to the
decreasing fraction of aerosol activated with Na increasing, all else being equal (e.g., Lu et al., 2009).
Conclusions
The suppression of precipitation due to the enhanced aerosol concentrations
(Na) is a general feature of warm clouds. In this study, we
examined precipitation susceptibility So in marine low clouds
by using in situ data obtained from four field campaigns with similar
data sets; two of them focused on marine stratocumulus (Sc), and two targeted
shallow cumulus (Cu) clouds. We estimate So with 1 s
data, with data averaged over an e-folding timescale, and data subsampled
randomly from flights, with the key results preserved regardless of the
method used. This study is the first to show with airborne data that for
both Sc and Cu, So increases with increasing cloud thickness
H, and peaks at an intermediate H before decreasing. For
example, R is most susceptible for clouds of medium–deep depth,
such as H∼ 380 m for Sc in NEP where H
varies between 100 and 450 m, and H∼ 1200–1400 m for Cu in the
Caribbean Sea where H ranges from 200 to 1600 m. On the other hand,
R is less susceptible to Nd in both shallow
non-precipitating and deep heavily precipitating cloud regimes for both Sc
and Cu. The results are consistent with previous studies of warm cumulus
clouds, but inconsistent with those of warm marine stratocumulus clouds
in situ observations.
We suggest several possible reasons for why these results differ from those
in previous studies of Sc, for example, by comparing with in situ
measurements of Terai et al. (2012). The sources of these uncertainties
include the following: (i) geographical location of cloud decks that may be
related to the predominant cloud microphysical process at work (e.g.,
accretion process), (ii) R threshold differences, (iii) wet
scavenging effects (causing high values of So), (iv) the use
of maximum column Z to convert R under heavy rain
conditions where cloud base is not defined, (v) the use of the Z–R
relationship to estimate R, and (vi) the use of sub-cloud aerosols
to estimate cloud-base Nd.
We also found that the details of Nd (e.g., Fig. B1) or how
the cloud base is determined (Fig. 4) have little effect on both So
values and the qualitative H-dependent behavior. Further, here we
emphasize and caution that the choice of the R threshold for the
data analysis is important because the chosen threshold possibly can alter
the character of the data set used to calculate So by
subsampling the data. For example, if a high value of the minimum R
threshold is chosen in a data set where the majority of data have low
precipitation (e.g., VOCALS TO flights, Fig. 2a) and/or in the bimodal
population of precipitation, the threshold would, by chance,
eliminate/reduce the influence of the autoconversion process in favor of the
accretion process. Further, Fig. B2 shows that the So decreases as the minimum R threshold increases.
This study shows that the VOCALS C-130 flight data sets are likely dominated by
the accretion process occurring naturally (geographically remote ocean areas
where POCs are often observed) and by the choice of high R
thresholds.
The values of So in this study were calculated from in situ
measurements, and thus, no issues associated with the retrieval (e.g.,
satellite data), empirical relationships (e.g., Z–R relationship), or
assumptions (e.g., relations between sub-cloud aerosols and cloud-base
Nd) are encountered for the calculation of So. A
drawback, however, is the much smaller sampling volume of the in situ
microphysical probes compared to a radar volume, as this may generate an
underestimate of the rain rate. Further, we calculated So
separately for Cu and Sc to avoid any possible issues that may arise from
combining different cloud types (Sect. 2.4). The results, however, should be
used with caution when comparing to other studies in quantifying
So as the dominating cloud process and the choices applied to
calculate the parameters in So estimates (Eq. 1) can differ
widely.
The results of this work motivate future studies examining the same
relationships with a more direct measurement of cloud depth using a cloud
radar and/or LWP using a microwave radiometer, in addition to the
instruments/sensors that measure/retrieve R and Nd
(Na is also desirable). For the flight strategy, in-cloud
level legs at multiple altitudes (cloud base, mid-cloud, and cloud top) with
one sub-cloud level leg would be ideal to calculate So and
compare with other studies where So is calculated with
cloud base or vertically integrated variables. Level legs near the ocean
surface and sounding(s) to examine the background thermodynamic structures
on a given day are also recommended.
Data availability
The Twin Otter research aircraft data set is available from upon request by
email at balbrecht@rsmas.miami.edu or ejung@rsmas.miami.edu.
Data
Table of acronyms and symbols.
AcronymExpressionBACEXBarbados Aerosol Cloud ExperimentCAScloud aerosol spectrometerCIPcloud imaging probeCu(shallow marine) cumulus (cloud)DSDdrop size distributionE-PEACEEastern Pacific Emitted Aerosol Cloud ExperimentHcloud thicknessKWACEXKey West Aerosol Cloud ExperimentLCLlifting condensation levelLWCliquid water contentLWPliquid water pathNdcloud droplet number concentrationPCASPpassive cavity aerosol spectrometer probePOCspockets of open cellsRrainfall (precipitation) rateScstratocumulus (clouds)Soprecipitation susceptibilityTOTwin OtterVOCALS-RExVAMOS Ocean-Cloud-Atmosphere-Land Study Regional ExperimentZradar reflectivity
H interval and number of data points used in Fig. 4 for
each field study.
Numbers indicates that the total number of data points, followed by the
linear regression correlation coefficient (r) and P value
(two-tailed t test). Bold P values indicate that correlations are
statistically significant at the 99 % confidence level.
Sensitivity of R and Nd thresholds to So estimates
The sensitivity of So to Nd
threshold values. One standard deviation of mean thickness for given
H intervals is shown as horizontal bars. Dates are indicated in mm/dd format.
H-dependent precipitation susceptibility as a
function of R threshold values. Dates are indicated in mm/dd format.
The effect of H intervals on So estimates
So calculated with different H intervals can be seen
by comparing Figs. 4 and A1 as an example. H intervals in Fig. 4b are about 30 m, while H intervals in Fig. A1 are about 50 m.
The qualitative H-dependent behavior of So is robust
regardless of the chosen H intervals in case 1 s data are
used. However, the chosen H interval may have effect on the
estimate of So that is calculated with a fewer data points,
such as So that is calculated with data averaged over the
e-folding time.
The effect of H intervals on So estimates, which is estimated
with data averaged over the e-folding time, is shown in Fig. B1. In summary,
the results are robust regardless of H interval in general.
However, if the H interval is chosen across the cloud thickness
where the So changes substantially (such as in which the cloud
properties change substantially), the pattern of So can be
changed, indicating that the finer H interval would provide more
accurate So. This is shown in Figs. 7 and 8. In Fig. 7, an
H interval of 50 m hides the variation of So between
H 150 and 200 m. The ln (Nd) and -ln(R)
diagrams for H widths of 40 and 50 m are shown in Fig. 7. However,
in case that the So does not change substantially across the
H intervals, the So does not change even if the
larger H interval is used (e.g., Fig. 8d). For example,
So calculated with subsets of data (e.g., 220 ≤H< 250 m, 250 ≤H< 280 m, 280 ≤H< 310 m) are ∼ 0.24–0.25. If the
So is estimated with all the data that fall into the three
intervals (e.g., H>200 m), the value is about 0.28,
which is similar to three individual So values. The results
may indicate that the cloud properties, such as cloud thickness where the
cloud begins to precipitate, could be of importance for accurate estimates of
So by affecting the optimal H interval and/or ranges.
So is calculated with cloud data that are
averaged over an e-folding time for E-PEACE. So calculated
with three H intervals (Δ30 m, Δ40 m, and Δ50 m) is shown. Horizontal bar indicates ±1σ cloud
thickness for a given H interval.
The ln (Nd) and -ln (R) diagrams
with fixed H intervals: (left) ΔH=40 m, (right) ΔH=50 m.
The ln (Nd) and -ln (R) diagrams
with fixed H intervals (ΔH=30 m).
Acknowledgements
The authors gratefully acknowledge the crews of the CIRPAS Twin Otter for
their assistance during these field campaigns. EJ acknowledges Chris Terai
for his helpful discussion of the estimate of precipitation susceptibility.
This study was funded by ONR grants N000140810465, N00014-10-1-0811,
N00014-16-1-2567, and NSF grant AGS-1008848. We thank three anonymous
reviewers for thoughtful suggestions and constructive criticism that have
helped to improve the manuscript.Edited by: H. Wang
Reviewed by: three anonymous referees
ReferencesBretherton, C. S., Wood, R., George, R. C., Leon, D., Allen, G., and Zheng, X.: Southeast Pacific stratocumulus clouds, precipitation and
boundary layer structure sampled along 20∘ S during VOCALS-REx, Atmos. Chem. Phys., 10, 10639–10654, 10.5194/acp-10-10639-2010, 2010.Comstock, K. K., Wood, R., Yuter, S. E., and Bretherton, C. S.: Reflectivity
and rain rate in and below drizzling stratocumulus, Q. J. Roy. Meteor.
Soc., 130, 2891–2918, 2004.Duong, H. T., Sorooshian, A., and Feingold, G.: Investigating potential biases in observed and modeled metrics
of aerosol-cloud-precipitation interactions, Atmos. Chem. Phys., 11, 4027–4037, 10.5194/acp-11-4027-2011, 2011.Feingold, G. and Siebert, H.: Clouds in the Perturbed Climate System:
Their Relationship to Energy Balance, Atmospheric Dynamics, and
Precipitation, edited by: Heintzenberg, J. and Charlson, R. J., Strüngmann, Forum Reports, 2, Cambridge, MA: The MIT Press,
597 pp., 2009.Feingold, G., McComiskey, A., Rosenfeld, D., and Sorooshian, A.: On the
relationship between cloud contact time and precipitation susceptibility to
aerosol, J. Geophys. Res., 118, 10544–10554, 2013.Gerber, H., Arends, B. G., and Ackerman, A. S.: A new microphysics sensor
for aircraft use, Atmos. Res., 31, 235–252, 1994.Gettelman, A., Morrison, H., Terai, C. R., and Wood, R.: Microphysical process rates and global aerosol–cloud interactions, Atmos. Chem. Phys., 13, 9855–9867, 10.5194/acp-13-9855-2013, 2013.Hill, A. A., Shipway, B. J., and Boutle, I. A.: How sensitive are
aerosol-precipitation interactions to the warm rain representation?, J. Adv.
Model. Earth Syst., 7, 987–1004, 10.1002/2014MS000422, 2015.Intergovernmental Panel on Climate Change (IPCC): The Physical Science
Basis. Contribution of Working Group I to the Fifth Assessment Report of the
Intergovernmental Panel on Climate Change, edited by: Stocker, T. F., Qin, D., Plattner,
G.-K., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V.
and Midgley, P. M., Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA, 2013.Jiang, H., Feingold, G., and Sorooshian, A.: Effect of aerosol on the
susceptibility and efficiency of precipitation in trade cumulus clouds, J.
Atmos. Sci., 67, 3525–3540, 2010.Jones, A., Roberts, D. L., Woodage, M. J., and Johnson, C. E.: Indirect
sulphate aerosol forcing in a climate model with an interactive sulphur
cycle, J. Geophys. Res., 106, 20293–20310, 10.1029/2000JD000089,
2001.Jung, E.: Aerosol-Cloud-Precipitation Interactions in the Trade Wind
Boundary Layer, PhD dissertation, 184 pp., available at:
http://scholarlyrepository.miami.edu/oa_dissertations/900,
2012.Jung, E., Albrecht, B. A., Prospero, J. M., Jonsson, H. H., and Kreidenweis,
S. M.: Vertical structure of aerosols, temperature and moisture associated
with an intense African dust event observed over the Eastern Caribbean, J.
Geophys. Res., 118, 4623–4643, 10.1002/jgrd.50352, 2013.Jung, E., Albrecht, B. A., Jonsson, H. H., Chen, Y.-C., Seinfeld, J. H., Sorooshian, A., Metcalf, A. R., Song, S., Fang, M., and Russell, L. M.:
Precipitation effects of giant cloud condensation nuclei artificially introduced into stratocumulus clouds, Atmos. Chem. Phys., 15, 5645–5658, 10.5194/acp-15-5645-2015, 2015.Jung, E., Albrecht, B. A., Feingold, G., Jonsson, H. H., Chuang, P., and Donaher, S. L.: Aerosols, clouds, and precipitation in the North Atlantic trades observed
during the Barbados aerosol cloud experiment – Part 1: Distributions and variability, Atmos. Chem. Phys., 16, 8643–8666, 10.5194/acp-16-8643-2016, 2016.Khairoutdinov, M. and Kogan, Y.: A new cloud physics parameterization in
a large-eddy simulation model of marine stratocumulus, Mon. Weather Rev.,
128, 229–243, 2000.Lee, G. and I. Zawadzki, I.: Variability of drop size distributions:
time-scale dependence of the variability and its effects on rain estimation,
J. Appl. Meteor., 44, 241–255, 10.1175/JAM2183.1, 2005.Leith, C. E.: The Standard Error of Time-Average Estimates of Climatic
Means, J. Appl. Meteor., 12, 1066–1069,
10.1175/1520-0450(1973)012<1066:TSEOTA>2.0.CO;2, 1973.Lu, M. -L., Sorooshian, A., Jonsson, H. H., Feingold, G., Flagan, R. C., and
Seinfeld, J. H.: Marine stratocumulus aerosol-cloud relationships in the
MASE-II experiment: Precipitation susceptibility in eastern Pacific marine
stratocumulus, J. Geophys. Res., 114, D24203, 10.1029/2009JD012774,
2009.Mann, J. A. L, Chiu, J. C., Hogan, R. J., O'Connor, E. J., L'Ecuyer, T. S.,
Stein, T. H. M., and Jefferson, A.: Aerosol impacts on drizzle properties in
warm clouds from ARM Mobile Facility maritime and continental deployments,
J. Geophys. Res., 119, 4136–4148, 10.1002/2013JD021339, 2014.Nuijens, L., Serikow, I., Hirsch, L., Lonitz, K., and Stevens, B.: The
distribution and variability of low-level cloud in the North-Atlantic
trades, Q. J. Roy. Meteor. Soc., 140, 2364–2374, 2014.Painemal, D. and Zuidema, P.: Assessment of MODIS cloud effective radius
and optical thickness retrievals over the Southeast Pacific with VOCALS-REx
in situ measurements, J. Geophys. Res., 116, D24206,
10.1029/2011JD016155, 2011.Pawlowska, H. and Brenguier, J. L.: An observational study of drizzle
formation in stratocumulus clouds for general circulation model (GCM)
parameterizations, J. Geophys. Res., 108, 8630,
10.1029/2002JD002679, 2003.Rasch, P. J. and Kristjansson, J. E.: A comparison of the CCM3 model
climate using diagnosed and predicted condensate parameterizations, J.
Climate, 11, 1587–1614, 1998.Rauber, R. M., Ochs III, H. T., Di Girolamo, L., Göke,
S., Snod- grass, E., Stevens, B., Knight, C., Jensen, J. B., Lenschow, D.
H., Rilling, R. A., Rogers, D. C., Stith, J. L., Albrecht, B. A., Zuidema,
P., Blyth, A. M., Fairall, C. W., Brewer, W. A., Tucker, S., Lasher-Trapp,
S. G., Mayol-Bracero, O. L., Vali, G., Geerts, B., Anderson, J. R., Baker,
B. A., Lawson, R. P., Bandy, A. R., Thornton, D. C., Burnet, E., Brenguier,
J-L., Gomes, L., Brown, P. R. A., Chuang, P., Cotton, W. R., Gerber, H.,
Heikes, B. G., Hudson, J. G., Kollias, P., Krueger, S. K., Nuijens, L.,
O'Sullivan, D. W., Siebesma, A. P., and Twohy, C. H.: Rain in shallow cumulus over the ocean, B. Am. Meteorol. Soc., 88, 1912–1928, 2007.Rogers, R. R. and Yau, M. K.: A Short Course in Cloud Physics, 3rd
Edn., International Series in Natural Philosophy, 290 pp., 1989.Rosenfeld, D., Wang, H., and Rasch, P. J.: The roles of cloud drop
effective radius and LWP in determining rain properties in marine
stratocumulus, Geophys. Res. Lett., 39, L13801,
10.1029/2012GL052028, 2012.Rotstayn, L. D. and Liu, Y.: A smaller global estimate of the second
indirect aerosol effect, Geophys. Res. Lett., 32, L05708,
10.1029/2004GL021922, 2005.Russell, L. M., Sorooshian, A., Seinfeld, J. H., Albrecht, B. A., Nenes, A., Ahlm, L., Chen, Y.-C., Coggon, M., Craven, J. S., Flagan, R. C.,
Frossard, A. A., Jonsson, H., Jung, E., Lin, J. J., Metcalf, A. R., Modini, R., Muelmenstaedt, J., Roberts, G. C., Shingler, T., Song, S., Wang, Z., and Wonaschuetz, A.: Eastern Pacific Emitted Aerosol Cloud
Experiment, B. Am. Meteorol. Soc., 94, 709–729, 10.1175/BAMS-D-12-00015.1, 2013.Sorooshian, A., Feingold, G., Lebsock, M. D., Jiang, H., and Stephens, G.:
On the precipitation susceptibility of clouds to aerosol perturbations,
Geophys. Res. Lett., 36, L13803, 10.1029/2009GL038993, 2009.Sorooshian, A., Feingold, G., Lebsock, M. D., Jiang, H., and Stephens, G.:
Deconstructing the precipitation susceptibility construct: improving
methodology for aerosol-cloud-precipitation studies, J. Geophys. Res., 115,
D17201, 10.1029/2009JD013426, 2010.Sorooshian, A., Prabhakar, G., Jonsson, H., Woods, R., Flagan, R. C., and Seinfeld, J. H.: On the presence of giant particles downwind of ships in
the marine boundary layer, Geophys. Res. Lett., 42, 2024–2030,
10.1002/2015GL063179, 2015.Stevens, B. and Feingold, G.: Untangling aerosol effects on clouds and
precipitation in a buffered system, Nature, 461, 607–613, 2009.Takemura, T., Nozawa, T., Emori, S., Nakajima, T. Y., and Nakajima, T.:
Simulation of climate response to aerosol direct and indirect effects with
aerosol transport-radiation model, J. Geophys. Res., 110, D02202,
10.1029/2004JD005029, 2005.Terai, C. R., Wood, R., Leon, D. C., and Zuidema, P.:
Does precipitation susceptibility vary with increasing cloud thickness in marine stratocumulus?, Atmos. Chem. Phys., 12, 4567–4583, 10.5194/acp-12-4567-2012, 2012.
Terai, C. R., Wood, R., and Kubar, T. L.: Satellite estimates of
precipitation susceptibility in low-level marine stratiform clouds, J.
Geophys. Res., 120, 8878–8889, 2015.vanZanten, M. C., Stevens, B., Vali, G., and Lenschow, D. H.: Observations
of drizzle in nocturnal marine stratocumulus, J. Atmos. Sci., 62, 88–106,
2005.Wang, M., Ghan, S., Liu, X., L'Ecuyer, T. S., Zhang, K., Morrison, H.
M.,
Ovchinnikov, R. E., Marchand, R., Chand, D., Qian, Y., and Penner, J. E.:
Constraining cloud lifetime effects of aerosols using A-Train Satellite
observations, Geophys. Res. Lett., 39, L15709, 10.1029/2012GL052204,
2012.Wang, Z., Sorooshian, A., Prabhakar, G., Coggon, M. M., and Jonsson, H. H.:
Impact of emissions from shipping, land, and the ocean on stratocumulus
cloud water elemental composition during the 2011 E-PEACE Field Campaign,
Atmos. Environ., 89, 570–580,
doi10.1016/j.atmosenv.2014.01.020, 2014.Wonaschütz, A., Coggon, M., Sorooshian, A., Modini, R., Frossard, A. A., Ahlm, L., Mülmenstädt, J., Roberts, G. C., Russell, L. M., Dey, S.,
Brechtel, F. J., and Seinfeld, J. H.: Hygroscopic properties of smoke-generated organic aerosol particles emitted in the marine atmosphere, Atmos. Chem. Phys., 13, 9819–9835, 10.5194/acp-13-9819-2013, 2013.Wood, R., Kubar, T. L., and Hartmann, D. L.: Understanding the Importantce
of Microphysics and Macrophysics for Warm Rain in Marine Low Clouds. Part
II: Heuristic Models of Rain Formation, J. Atmos. Sci., 66, 2973–2990,
10.1175/2009JAS3072.1, 2009.Wood, R., Mechoso, C. R., Bretherton, C. S., Weller, R. A., Huebert, B., Straneo, F., Albrecht, B. A., Coe, H., Allen, G., Vaughan, G., Daum, P.,
Fairall, C., Chand, D., Gallardo Klenner, L., Garreaud, R., Grados, C., Covert, D. S., Bates, T. S., Krejci, R., Russell, L. M., de Szoeke, S.,
Brewer, A., Yuter, S. E., Springston, S. R., Chaigneau, A., Toniazzo, T., Minnis, P., Palikonda, R., Abel, S. J., Brown, W. O. J., Williams, S.,
Fochesatto, J., Brioude, J., and Bower, K. N.: The VAMOS Ocean-Cloud-Atmosphere-Land Study Regional Experiment (VOCALS-REx):
goals, platforms, and field operations, Atmos. Chem. Phys., 11, 627–654, 10.5194/acp-11-627-2011, 2011.Zheng, X., Albrecht, B., Jonsson, H. H., Khelif, D., Feingold, G., Minnis, P., Ayers, K., Chuang, P., Donaher, S., Rossiter, D., Ghate, V.,
Ruiz-Plancarte, J., and Sun-Mack, S.: Observations of the boundary layer, cloud, and
aerosol variability in the southeast Pacific near-coastal marine stratocumulus during VOCALS-REx, Atmos. Chem. Phys., 11, 9943–9959, 10.5194/acp-11-9943-2011, 2011.Zuidema, P., Leon, D., Pazmany, A., and Cadeddu, M.: Aircraft millimeter-wave passive sensing of cloud liquid water and water vapor during VOCALS-REx, Atmos. Chem. Phys., 12, 355–369, 10.5194/acp-12-355-2012, 2012.