Polluting of winter convective clouds upon transition from ... · Polluting of winter convective clouds upon transition from ocean inland over central California: Contrasting case

Calhoun: The NPS Institutional Archive

Faculty and Researcher Publications Faculty and Researcher Publications

2014

Polluting of winter convective clouds

upon transition from ocean inland over

central California: contrasting case studies

Rosenfeld, Daniel

Elsevier

Atmospheric Research, v.135/136, 2014, pp. 112-127

http://hdl.handle.net/10945/43173

Atmospheric Research 135–136 (2014) 112–127

Contents lists available at ScienceDirect

Atmospheric Research

j ourna l homepage: www.e lsev ie r .com/ locate /atmos

Polluting of winter convective clouds upon transition fromocean inland over central California: Contrasting case studies

Daniel Rosenfeld a,⁎, Rei Chemke a, Kimberly Prather b, Kaitlyn Suski b, Jennifer M. Comstock c,Beat Schmid c, Jason Tomlinson c, Haflidi Jonsson d

a The Hebrew University of Jerusalem, Israelb University of California, San Diego, United Statesc Pacific Northwest National Laboratory, Richland, WA, United Statesd Naval Postgraduate School, Monterey, CA, United States

a r t i c l e i n f o

⁎ Corresponding author at: Institute of Earth SciencUniversity, Jerusalem 91904, Israel.

E-mail address: [email protected] (D. Ros

0169-8095/$ – see front matter © 2013 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.atmosres.2013.09.006

a b s t r a c t

Article history:Received 18 May 2013Received in revised form 4 September 2013Accepted 5 September 2013

In-situ aircraft measurements of aerosol chemical and cloud microphysical properties wereconducted during the CalWater campaign in February and March 2011 over the Sierra NevadaMountains and the coastal waters of central California. The main objective was to elucidate theimpacts of aerosol properties on clouds and precipitation forming processes. In order to accomplishthis, we compared contrasting cases of clouds that ingested aerosols from different sources. Theresults showed that clouds containing pristine oceanic air had low cloud drop concentrations andstarted to develop rain 500 m above their base. This occurred both over the ocean and over theSierraNevada,mainly in the earlymorningwhen the radiatively cooled stable continental boundarylayerwas decoupled from the cloud base. Supercooled rain dominated the precipitation that formedin growing convective clouds in the pristine air, up to the −21 °C isotherm level.A contrasting situation was documented in the afternoon over the foothills of the Sierra Nevada,when the clouds ingested high pollution aerosol concentrations produced in the Central Valley. Thisled to slow growth of the cloud drop effective radius with height and suppressed and evenprevented the initiation ofwarm rainwhile contributing to the development of ice hydrometeors inthe form of graupel. Our results show that cloud condensation and ice nuclei were the limitingfactors that controlled warm rain and ice processes, respectively, while the unpolluted clouds in thesame air mass produced precipitation quite efficiently. These findings provide the motivation fordeeper investigations into the nature of the aerosols seeding clouds.

© 2013 Elsevier B.V. All rights reserved.

Keywords:Cloud–aerosol interactionsPrecipitation suppression

1. Introduction

Orographic precipitation is an important water source,especially in semi-arid areas such as the western USA and theMiddle East. Anthropogenic air pollution has an important rolein determining the precipitation properties in such clouds.

Adding aerosols increases the number of CCN (CloudCondensation Nuclei) that nucleatemore numerous and smallercloud drops. This slows the drop coalescence and in turn the

es, The Hebrew

enfeld).

ll rights reserved.

conversion of cloud water into rain drops (Rosenfeld, 2000;Hudson and Yum, 2001; McFarquhar and Heymsfield, 2001;Yum and Hudson, 2002; Borys et al., 2003; Andreae et al., 2004;Hudson andMishra, 2007; Rosenfeld et al., 2008; Flossmann andWobrock, 2010). It also slows the mixed phase precipitationforming processes by decreasing the riming and growth rate ofice hydrometeors (Borys et al., 2003; Saleeby et al., 2008).Slowing the precipitation forming processes in shallow andshort lived orographic clouds is expected to cause a net decreasein precipitation amount in the upwind slope of the mountains(Griffith et al., 2005), often with some compensation at thedownwind slope (Givati and Rosenfeld, 2004, 2005; Jirak andCotton, 2005; Rosenfeld and Givati, 2006; Givati and Rosenfeld,

113D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

2007; Rosenfeld et al., 2007; Cotton et al., 2010; Yang et al.,2013).

These reports prompted investigation of the possibleeffects of aerosols on orographic precipitation by the meanson numerical simulations, which simulated the precipitationfrom clouds that develop in air mass that crosses topographicbarriers while incorporating different amounts of CCN and IN.The results of these model simulations support the hypothesisthat adding CCN suppresses orographic precipitation (Lynn andKhain, 2006; Muhlbauer and Lohmann, 2006; Saleeby et al.,2008, 2011). On the other hand, it is expected that adding icenuclei (IN) to supercooled clouds would increase precipitation(Creamean et al., 2013). Numerical simulations support thesegeneral trends (Muhlbauer and Lohmann, 2009; Lohmann,2002), by showing that the presence of aerosols that act asIN enhances mixed phase precipitation.

Two studies of Givati and Rosenfeld (2004) and Rosenfeldand Givati (2006) showed the decreasing pattern of the ratiobetween the precipitation amounts over the hills to theprecipitation amounts in the upwind lowland at the westcoast of the United States during the 20th century. Thispattern was associated with a decreasing trend of coarseaerosols,which act as giant CCN,whilemaintaining or increasingthe concentrations of the PM2.5 aerosols. No trends wereobserved in pristine areas. The emissions of air pollution peakedin the early 1980s and then decreased after regulations wereimposed to clean the air. This had a large effect on improving theair quality and reducing the amount of aerosols and theirprecursors. For example, according to the US EnvironmentalProtection Agency, the average amount of SO2 over the USAdecreased from 11.8 ppb in 1980 to 2.4 ppb in 2010. However,decreasing sulfur emissions does not necessarily decrease theCCN concentrations, and may even enhance the sources from“clean” power plants by three orders of magnitude as comparedto the CCN production by the old polluted technology because ofnucleation of huge concentrations of sulfuric acid particles atsizes of 1 to several nm inside the power plant stacks.These particles grow several hours after emission to CCNsizes, i.e., N50 nm (Junkermann et al., 2011). This might explainwhy the orographic enhancement factor continued to decreasedespite the efforts to clean the air. An alternative explanationbased on changing weather patterns could not be identified, atleast not for the western USA (Rosenfeld and Givati, 2006).

This question motivated two field campaigns called Sup-pression of Precipitation (SUPRECIP), which took place in thelate winters of 2005 and 2006. Aircraft measurements of theinteractions of clouds and aerosols weremade over the CentralValley and Sierra Nevada, CA. The objective was understandingthe relationships between aerosol sources and orographicprecipitation (Rosenfeld et al., 2008). One of the key findingswas that CN and CCN concentrations in the Central Valley aremuch higher than the concentrations in the coastal urbanareas, which implies that the Central Valley itself is a source ofhigh concentrations of pollution aerosols. As a result, losses oforographic precipitation in the Sierra Nevada could be ascribedto the pollution from the Central Valley. In addition, nonurbansources may play a major role in determining the properties ofthe clouds over the foothills of the Sierra Nevada (Rosenfeld etal., 2008). The SUPRECIP campaigns were followed in the latewinter of 2011 by field campaign called California Water(CalWater), aimed at better understanding of the impacts of

the various aerosols (local pollution coming from the boundarylayer, long range transport pollution coming aloft, and pristinemarine air) on the properties and precipitation of the waterand mixed phase clouds over central California.

Because the added pollution aerosols play a major role inchanging their microstructure upon transition from oceaninland, here we analyze three case studies in contrasting aerosolconditionswith the objective of obtaining a better understandingof the crucial impacts of the added aerosols on precipitationforming processes. The findings show surprisingly highlysupercooled rain in clouds that develop in pristine air, down totemperatures of−21 °C. A deeper investigation of this behavioris given by Rosenfeld et al. (2013). Precipitation did not occur inclouds with added continental aerosols that had the same depthas precipitating clouds with marine aerosols. The analysis ofthese few case studies sheds additional light on the possibleextent of aerosol impacts on precipitation processes.

2. The CalWater campaign and methodology

The CalWater campaign took place during February andearly March 2011. It included in-situ aircraft measurementsof cloud physical and aerosol chemical and physical character-istics. The campaign targeted mainly the orographic cloudsover the Sierra Nevada. The clouds were classified into:

a. Convective clouds triggered by orographic lifting over thefoothills and over the western slopes of the mountains.

b. Layer clouds formed by orographic lifting of moist stablelayers, mainly over the western slopes and cap clouds overthe crest line of the Sierra Nevada.

c. Convective clouds thatwere generated due to daytime solarsurface heating, mainly over the central valley and thefoothills.

d. Marine convective clouds formed over the ocean due tothe synoptic conditions that triggered them in unstableoceanic air masses.

Because the central Valley is documented to be a majorsource of CCN, it is important to note the meteorologicalconditions and time of day when the air from the valley floorcould have been ingested into the clouds. The first two cloudtypes were mostly decoupled from the boundary layer overthe Central Valley. Coupling occurred mainly in the after-noon, when the second and third cloud types merged.

The flights were aimed at obtaining vertical profiles of theaerosol and cloud properties for the four cloud types, andidentifying the aerosol types that they ingest. The objectiveswere:

a. Identify the aerosol sources based on their chemistry andthe meteorological context.

b. Detect the impacts of these aerosols on cloud microstruc-ture and precipitation forming processes for the differentcloud types as defined above.

In keeping with the objectives of documenting the processesof initiation of precipitation, the vertical evolution of aerosols andcloud properties were documented. The vertical span of themeasurements started from below cloud bases and ended abovetheir tops, when possible. The flight tracks were planned withpriority for obtaining large samples for the chemical analysis of

Table 1The subset of G-1 aircraft cloud physics instruments used in this study.

Variable Instrument Range Frequency

CN concentration TSI 3010TSI 3025

N10 nmN3 nm

1 Hz1 Hz

Aerosols spectra PCASPUHSAS

0.1 to 3 μm0.06 to 1 μm

1 Hz1 Hz

Aerosolsconcentration

PCASP calculatedUHSAS calculated

1 Hz1 Hz

Aerosols effectiveradius

PCASP calculatedUHSAS calculated

1 Hz1 Hz

Single particlechemistry

A-ATOFMS ~70 to 1200 nm

Cloud droplet spectra CDPCAS

2 to 50 μm1 to 50 μm

1 Hz1 Hz

Cloud dropletconcentration

CDP calculatedCAS calculated

1 Hz1 Hz

Cloud drop effectiveradius

GerberPVM-100aCDP calculatedCAS calculated

3 to 50 μm2 to 50 μm1 to 50 μm

1 Hz1 Hz1 Hz

Liquid water content SEA WCM-2000GerberPVM-100aCDP calculated

1 Hz1 Hz1 Hz

Total water content SEA WCM-2000 1 HzHydrometeorparticle spectra

2D-SCIP

5 to 3205 μm25 to 1550 μm

1 Hz1 Hz

114 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

the aerosol that formed the cloud particles. This required thecollection of asmany samples as possible in long horizontal flighttracks at the same height. However, these tracks are not ideal fordocumenting the vertical evolution of cloud microstructure.

The properties of the aerosols that feed the convectiveclouds are determined by the origin of the air masses.Possibilities include:

a. Local pollution coming from the boundary layer over theCentral Valley.

b. Local air pollution that comes from areas upwind of theCentral valley.

c. Marine air that has not interacted with the continentalboundary layer.

d. Long range transport of polluted air that has not interactedwith the continental boundary layer. This can contain desertdust and/or pollution that often comes across the Pacific fromEast Asia.

The flights were conducted on subsequent days andsometimes twice a day, in order to document the changingconditionswith the evolution of the synoptic condition and thediurnal cycle (e.g. decoupling or coupling of the boundary layerwith solar surface heating, and post- or pre-frontal clouds).

The DOE/PNNL Gulfstream-1 was used for this research. Itwas equipped with an extensive suite of cloud microphysicsand aerosol instruments. The instruments used in this studyare listed in Table 1.

An aircraft aerosol time-of-flight mass spectrometer(A-ATOFMS) was used to measure the chemical characteristicsof aerosols and cloud residues (Pratt et al., 2009a,b). Particleswere grouped into types based on similar mass spectralcharacteristics. Particles rich in ammonium (18NH4

+), amines(58C2H5NHCH2

+, 86(C2H5)2NCH2+) and nitrate (62NO3

−) alongwith organic carbon ion markers (27C2H3

+, 43C2H3O+) wereclassified as Central Valley pollution (Whiteaker et al., 2002).Sea salt was characterized by the presence of sodium (23Na+),sodium clusters (46Na2+, 62Na2O+, 63Na2OH+, 81/83Na2Cl+) andchloride (35/37Cl−) (Gard et al., 1998; Gaston et al., 2011). Dustwas identified by the presence of silicates (44SiO−, 60SiO2

−,76SiO3

−) and calcium (40Ca+, 96Ca2O+, 113(CaO)2H+) ionmarkers(Silva et al., 2000). Soot was characterized by carbon ionsubunits, Cn+ (12C+, 24C2+, 36C3+, 48C4+, 60C5+, etc.) (Shields et al.,2007; Sodeman et al., 2005; Toner et al., 2006). Urban pollutionwas identified as highly processed particles that contained onlynegative ion mass spectra containing sulfate (97HSO4

−) and/ornitrate (46NO2

−, 62NO3−) ion markers. Biomass burning particles

were characterized by a large potassium ion marker (39/41K+),potassium ion clusters (104K2CN+, 113/115K2Cl+, 213/215K3SO4

+)and elemental and organic carbon ions (Pratt et al., 2011; Silvaet al., 1999). Biological particles contained sodium (23Na+),potassium (39K+), organic nitrogen (26CN−, 42CNO−) andphosphate (79PO4

−) and sometimes metals (24Mg+, 40Ca+,56Fe+ or 52Cr+) (Fergenson et al., 2004; Pratt et al., 2009a,b;Russell, 2009). Organic carbon particles were characterizedby the presence of both carbon ion markers (12C+, 27C2H3

+)and oxidized organic carbon ion markers (43C2H3O+) (Qinet al., 2012).

The next section describes in detail several case studies, inwhich cloud and precipitation forming processes are shown tohave rather different behavior under similar dynamic conditions

butwith different aerosols. Quantification of the extent towhichthey represent the typical conditions or their importance in thecontribution of precipitation amounts for the region will beinvestigated for more flights in subsequent studies.

3. The case studies

3.1. The flight on 16 February 2011

Westerly post-frontal flow triggered embedded convec-tive clouds at the foothills and western slopes of the SierraNevada with cloud base height of about 600 m and tops near4 km, and cap layer clouds over the crest at heights between5 and 6 km. The situation is illustrated well by Fig. 5B inCreamean et al. (2013). The research flight took off fromSacramento at 17:01 UT (09:01 local time), climbed throughthe clouds towards the crest line to the north east of Sacramento,and profiled down back through the clouds and landed morethan 3 h later back in Sacramento, at 20:19 UT. Fig. 1 shows theflight track from this day.

Here we provide an overview of our understanding of thesituation, which will be supported by the observations in thenext section. Convective clouds were formed at the foothillsto the east of Sacramento in low level southwesterly flow of5.1–8.7 m·s−1. The base was decoupled from the boundarylayer in the early morning, but the diurnal solar heatingenhanced the convection over the foothills later when thesurface warmed air participated in the convection. The clouds inthe early morning formed in pristine maritime air and producedprecipitation mainly as supercooled rain at temperatures as coldas −21 °C, as already documented in Rosenfeld et al. (2013).Three hours later, with mixing of aerosols from the valley floor,the warm rain was less developed, and mostly replaced bygraupel.

Fig. 1. Flight track for the 16 February 2011, 17:00–20:20 UTC. The colorsrepresent the aircraft flight levels in msl.

115D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

Elevated orographic layer cap clouds formed in air thatapparently flowed from the Pacific Ocean with little surfacecontact, as shown in Section 4 by the aerosol concentrations andcomposition. These clouds produced mixed phase precipitationthat was likely nucleated by large amounts of desert dust thatcame with long range transport from Asia, as shown byCreamean et al. (2013). Convective clouds occasionally pene-trated the overlying layer clouds,which indicates somemixing ofthe local low-level air with the long range transported air.

3.2. The flights on 21 February 2011

The importance of the diurnal cycle and continental aerosolswas conspicuously evident in the two flights on the 21st ofFebruary 2011. The first flight documented convective cloudswith tops reaching height of 3 km that occurred in cyclonic coldsouthwesterly flow over the coastal waters of the Pacific Ocean.These clouds were contrasted with convective clouds of similarsize that formed over the foothills of the Sierra Nevada in theafternoon of the same day in the same air mass but with manylocally added aerosols. While the clouds over the oceanproduced precipitation very efficiently by warm rain, thistype of precipitation was completely suppressed in the cloudsover the foothills, which produced no precipitation or verylight graupel. Mid-level supercooled layer clouds with small

Fig. 2. Flight tracks for the 21 February 2011. The morning flight to the ocean took22:00–01:00 UTC. The colors represent the aircraft flight levels in msl.

amounts of ice were also documented in the morning flight.Fig. 2 shows the two different flight tracks.

3.3. Microphysical methodology

In trying to identify the microphysical impacts of aerosolson clouds, one has to keep in mind that aerosol effects oncloud drop size distribution are secondary to the impacts ofdepth above cloud base. This is overcome by presentation ofthe cloud properties as a function of cloud depth, D, which isthe vertical distance above cloud base in meters.

Aerosol effects on precipitation forming processes can bemasked by precipitation that falls from higher levels into themeasured cloud volume. This was overcome in previousstudies by trying to sample growing convective elements neartheir tops, where the clouds are young and no precipitation canfall from above.

These previous studies (Andreae et al., 2004; Rosenfeld etal., 2006, 2008; Freud and Rosenfeld, 2012) showed thatwarm rain is initiated when the cloud drop effective radius,re, exceeds 12–14 μm, or when the volume-weighted modalsize of the cloud drop size distribution, DL, exceeds 24 μm.The cloud drops grow with D until re or DL exceeds the valuesfor initiation of warm rain. Continued widening of the clouddrop size distribution with increasing height expands the tailof the drop size distribution into the drizzle and rain dropsizes. A smooth transition between the cloud and drizzle sizedrop between the diameters of 50 to 100 μm occurs whenthe drizzle forms in situ in the cloud volume. When this isobserved the measured precipitation can be considered ashaving formed in the measured cloud volume. This behaviorceases when the precipitation is not the result of the growthof drop size distribution in the measured cloud volume. Largedrizzle or rain drops that fall from above or the growth of icehydrometeors produces a separate peak in the spectrum. In theseconditions there is no smooth continuation of the spec-trum between 50 and 100 μm, but rather a second peak ofprecipitation appears. The continuation of conversion ofcloud into rain drops beyond the initiation stage is evidentby depletion of the cloud liquid water content (LWC). Thegrowth rate of the effective radius with height increaseswhen the CCN concentration at the cloud base decreases(Freud et al., 2011). More cloud drops are nucleated atcloud base for greater CCN concentrations. This causes the

place at 17:00–20:10 UTC. The afternoon flight to the Sierra Nevada was at

116 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

condensed cloud water to be distributed among larger numberof smaller drops. Therefore,more cloudwatermust condense forproducing a given cloud drop size. A greater vertical distanceabove cloud base is required for the added condensed cloudwater. Freud and Rosenfeld (2012) have shown that there is alinear relation between the number of activated CCN into dropsat cloud base andD for reaching a given re. Thismeans that if rainis initiatedwhen re exceeds a precipitation threshold value rep, Dfor rain initiation, Dp, is linearly dependent on the cloud basedrop concentrations.

We can see this behavior of the effective radius as a functionof height in clean and slightly polluted cases from the morningand noon of the flight of 16 February (Fig. 3) and for the pristinemorning and polluted afternoon flights of the 21st February(Fig. 4).

In a mixed phase cloud, the largest drops freeze or arecollected by the already existing ice hydrometeors. Therefore,when DL exceeds 24 μm in a mixed phase cloud, we still do nothavemuchwarm rain but rathermore efficient ice precipitation.

In a turbulent cloud, graupel grows faster than rain dropsof the same mass. This greater growth rate of the graupelincreases as the cloud contains smaller cloud drops for the

Fig. 3. CDP-measured cloud drops effective radius, re, as a function of heightwith cloud drops liquid water content, LWC, as color bar, in convectiveclouds, at the morning around 17:30 UT (panel A) and noon around 19:30UT (panel B), during the flights of 16 February 2011. Cloud base height and Dfor reaching re = 14 μm are marked. CDP concentration at the cloud base(600 m) reaches 30 cm−3 maximum in A and 130 cm−3 maximum at thecloud base (1000 m) in B.

same liquid water content (Pinsky et al., 1998). In order todevelop warm rain in more polluted clouds, the cloud mustgrow to higher altitudes and hence lower temperatures whichfavor graupel over rain drops. As a result of the smaller clouddrop size, warm rain will develop slower leaving the graupel tocapture the cloud drops at the expense of the rain drops.

4. Effects of incorporation of continental aerosols

In this section, we summarize the results of contrastingcases of convective clouds from the three flights mentionedin Section 3. As a result of incorporating continental aerosolin the clouds during the day in these clouds, we are able tocompare pairs of clouds at similar meteorological conditionsbut with different origins of the aerosols that were ingestedinto their bases.

4.1. Aerosol properties

In the early morning, the boundary layer is often decoupledfrom the free troposphere, so that the pollution from the Central

Fig. 4. CDP-measured cloud drops effective radius, re, as a function of height,in convective clouds, at the morning around 18 UTC (A) and afternoonaround 22 UTC (B) flights of 21 February 2011. The different growth of rewith height is marked. CDP concentration at the cloud base (600 m) reaches150 cm−3 maximum in A, 800 cm−3 maximum at the cloud base (1100 m)in B.

Fig. 5. CPC-3010 (magenta), PCASP (black) and cloud drops concentrationswith cloud drops water content, LWC, as color bar, as a function of height forthe flight of 16 February 2011. Panel A is around 17 UTC, when low PCASPconcentration at cloud base indicates that the cloud base is decoupled fromthe boundary layer. Panel B is around local noon (20 UTC), where highPCASP concentration at cloud base indicates coupling with the boundarylayer.

117D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

Valley does not reach the base of the clouds that form in themaritime air that overrides the boundary layer during westerlywinds. This situation occurred on the morning of 16 February2011. It was previously documented in the study area duringrainy conditions by Rosenfeld et al. (2008). Very low PCASP-measured aerosol concentrationswere observed near cloud base(6–30 cm−3, Fig. 5A). The composition of the cloud residueswith low concentrations during the upward convective cloudprofile is shown in Fig. 6A. They were composed of mainly seasalt with smaller contributions from organic carbon, dust,biological particles, urban pollution and Central Valley aerosols.The diurnal solar surface heating destroyed the surface inversionand caused some of the surface air to reach cloud bases andincrease there the PCASP-observed aerosol concentrations to30–150 cm−3 (Fig. 5B). The cloud residue composition duringthe downward profile at noon, shown in Fig. 6B, had lowerpercentages of sea salt and organic carbon with higherpercentages of urban pollution and biomass burning residues.The increase in pollution and biomass burning residues is inagreement with the conclusion that the boundary layer (BL)was no longer decoupled from the free troposphere.

The diurnal cycle of the aerosol concentration as a resultof the coupling and decoupling of the BL was also observedon the 21st of February 2011. Fig. 7 shows the increase ofaerosol concentration at 1000 m as the day progresses abovethe Central Valley near Sacramento. In the morning the BL isstill decoupled from the free troposphere. As a result thePCASP-measured aerosol concentration at 1000 m is only50 cm−3. At noon the boundary layer is no longer decoupledfrom the free troposphere, allowing polluted air from thevalley can reach the cloud base height. In the afternoon,aerosol concentrations at the cloud base reached values upto 630 cm−3.

The same diurnal behavior was observed by the A-ATOFMSinstrument. Fig. 8 shows the cloud residue composition fromthe 21st of February 2011. In the morning, mostly salt particleswith few pollution particles were measured in an elevatedcloud layer at around 4600 m (Fig. 8A). Some pollution cloudresidues of similar composition were also found in low cloudsover the sea (Fig. 8B), which suggests that this pollution did notlikely come from the California land area. This is supported bythe back trajectory of the air mass, which extends to thenorthwest into the Pacific Ocean, and curls back toWashingtonState 2.5 days back.

At noon, the air mass at 1000 m (20:05 UT, Fig. 8C) abovethe valley is no longer decoupled from the boundary layer asa result of the vertical mixing in the atmosphere. Fig. 8C showsthe aerosol composition out of clouds, because there were notenough clouds for measuring their drop residues. The aerosolsstill contained some sea salt, but it was dominated by biomassburning and Central Valley pollution. In the second flight inthe afternoon more air pollution from the boundary layer wasingested in the convective clouds that developed over the SierraNevada foothills (Fig. 8D–F), containing much more biomassburning, urban pollution and soot than in the morning flight,along with a high occurrence of Central Valley Pollution, whichis high in ammonium.

Large contrast was evident in the microstructure of themorning clouds over ocean and afternoon clouds over land.These convective clouds over ocean had relatively low aerosolconcentrations at their bases (max 300 cm−3, Fig. 9) compared

to N3000 cm−3 over land in the afternoon (Fig. 7C). The aerosolcomposition over ocean (Fig. 8B) shows large amount of sea saltaerosols and a low percentage of aerosol that can serve as goodIN, such as dust.

During the afternoon flight on the 21st of February thePCASP-measured aerosol concentrations increase with heightup to 2000 m (black dots, Fig. 7C). However, aerosol particleconcentrations are expected to decrease with height whentheir source of is the land surface. What could possibly causethe aerosol concentrations to increase so much with heightdespite the well vertical mixed conditions? In fact, the totalaerosol concentrations, as measured by the CPC (magenta dots,Fig. 7C) did decreasewith height. In order to explain the contrastwith the increasing PCASP with height, we hypothesize thatvolatile aerosols condense on the CNwhen the air rises and cools.The CPC instrument counts all the aerosol particles N~3 nm,whereas the PCASP minimum detectable size is 0.1 μm.When aCPC detected particle grows beyond 0.1 μm it gets detected bythe PCASP. If the aerosol particle size growswith height, more of

43%

16%

16% 5%

16%

3%

65%

8% 3%

16%

8%

A

Sea Salt Dust Biological Biomass Burning Soot Organic Carbon Central Valley Pollution Urban Pollution

B

Fig. 6. Chemical composition of cloud residues from February 16, 2011 during A) the upward convective clouds profile and B) the downward profile after the airmass had mixed with boundary layer air.

118 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

them are detected with height by the PCASP. Insights to thepossible cause of this behavior are provided by the analysis ofcloud residue composition as measured by the ATOFMS (Fig. 8).Indeed, Fig. 8 shows that in the afternoonmore air pollution fromthe boundary layer penetrated to the free troposphere and that itcontained high occurrence of Central Valley pollution up to aheight of 2500 m. Comparison with the flight on the 16th ofFebruary showed that the PCASP-measured concentrationsdecreased with height, and that the cloud residues containedlittle volatile constituents such as ammonium, sulfate andnitrate.Volatile aerosols that grow with height would also collectthe smaller aerosols and cause the decrease of total aerosolconcentrations with height. This was actually observed tooccur, as measured by the CPC concentrations with height,shown in Fig. 7C. The transition from many small particles atthe low level to fewer larger ones is shown by the verticalprofile of the aerosol size distributions measured by theUHSAS (Fig. 10). Fig. 8E shows the cloud residue chemistryabove 2500 m. This air mass is decoupled from the valleyfloor air. Therefore it contains less ammonium-rich aerosolsand more sea salt aerosols. The PCASP concentrations returnto the normal behavior of decreasing with heights (Fig. 7C).

The gradual transition from maritime to polluted air wasdocumented in the later part of the first flight, while flyingfrom the ocean eastward to landing in Sacramento (Fig. 2). Thishorizontal gradient is captured in the A-ATOFMS chemical data.The area to the west of Sacramento at noon (Fig. 8C) had lesspolluted air mass with more sea salt compared to the air masslater and further east over the foothills of the Sierra Nevada(Fig. 8D), which had more biomass burning and Central Valleypollution.

In summary, the origin of the aerosols that reached thecloud bases in these two days changed from a maritime airmass in the morning to a more polluted continental air massover the foothills of the Sierra Nevada in the afternoon. The

diurnal evolution of the pollution over the Central Valley nearSacramento was documented. The major differences in theaerosol properties between the convective clouds with marineand the continental aerosols, discussed above, are shown in thegraphs of the PCASP (Figs. 5 and 7) and aerosol chemistry(Figs. 6 and 8). While the clouds had little contribution fromlocal continental aerosols contained at their base some sea saltaerosols, the afternoon clouds over land contained at theirbases mainly biomass burning, urban pollution and in somecases Central Valley pollution aerosols.

4.2. Initiation of warm rain

These differences in the aerosol concentration at cloud baseand cloud residue composition corresponded to the largecontrast in the cloud physical properties as we show below.

In the morning of the 16th February, the low aerosolconcentration corresponded to low concentration of CCN andcloud droplets at cloud base (max 30 cm−3). This caused asteep growth of re with D, reaching the re threshold for warmrain initiation, 14 μm (Freud and Rosenfeld, 2012), at D =800 m (Fig. 3A). The vertical evolution of cloud drop sizedistributions is shown for the cloud passes that have thegreatest amount of liquid water content for a given height forthe morning profile (Fig. 11A). The DL in the morning cloudsis 800 m.

As we have mentioned already, this process of ingestingcontinental aerosols affects cloud microphysics properties. Inthe afternoon of the 16th February, the increase in aerosolconcentration near cloud base and hence also in cloud dropconcentration (max 130 cm−3) moderated the slope of rewith D. As mentioned in Section 3.3, DL and D for reaching there threshold for warm rain initiation of 14 μm (Freud andRosenfeld, 2012) are higher for clouds with larger cloud dropconcentration at their base. The D for reaching 14 μm increased

Fig. 7. CPC-3010 (magenta), PCASP (black) and cloud drop concentrationswith cloud drops water content, LWC, as color bar, as a function of height forthe flight of 21 February 2011, above the Central Valley. At 17:00 UTC(morning—panel A) the low aerosol concentration at 1000 m, 50 cm−3,indicates that the cloud base is decoupled from the boundary layer. At higheraltitudes we flew west away from the Central Valley. At 20:00 UTC (noon—panel B) aerosol concentration increases up to 80 cm−3 at 1000 m, whichindicates that the boundary layer is no longer decoupled from the freetroposphere. At higher altitudes we were flying east from the western rim ofthe Central Valley. At 22:00 UTC (afternoon—panel C) aerosol concentrationincreases up to 630 cm−3 at 1000 m. Panel C shows the contradictionbetween the CPC and PCASP behavior with height. In contrary to the CPC theaerosol concentration in the PCASP instrument decreases with height up to2000 m.

119D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

from 800 to 1700 m from the morning to the noon (Fig. 3,panel B). At the same time the DL increased from 800 to1350 m (Fig. 11). The smaller drops and larger D (DL) at noon

compared to the morning cloud profiles were associated withdifferences in the observed precipitation forming processes.

The same physical behavior of cloud drop size distribu-tions was documented at the convective clouds above theocean in the morning of the 21st of February. The relativelylow aerosol concentration at the cloud base of these maritimeconvective clouds (max 300 cm−3) caused low CDP concen-tration (max 150 cm−3) at the cloud base (Fig. 9). These lowconcentrations at cloud base led re to increase strongly withheight, which grew from 6 μm to the re threshold for warmrain initiation, 14 μm in only 500 m (Fig. 4A). The verticaldevelopment of the drop sizes is shown in Figs. 12 and 13A. DL

reached 24 μmat1100 m (500 mabove cloud base), indicatingthat rain was initiated at or above this height. Below that levelall the measured precipitation must have fallen from above orform as ice precipitation. This is evident by the separate peak ofprecipitation size particles.

The clouds over the foothills from the afternoon flight on the21st of February had high PCASP-measured aerosol concentra-tion at their bases (max 1000 cm−3) that causedhigh clouddropconcentration (max 800 cm−3) at their bases (Fig. 7C). Thedifferences in cloud base drop concentrations led to correspond-ing differences in the slopes of effective radiuswith height. The regrew from 4 μm to only 7.5 μm in a vertical distance of 1800 m(Fig. 4B) and did not even get close to the re threshold for warmrain initiation, 14 μm. This large difference in the re growthfrom the morning of this flight made a big difference in theprecipitation forming processes, as shown next. In contrast tothe morning clouds with marine aerosols over ocean, the highaerosol and clouddrop number concentrations in the afternoonflight limited DL to much less than the rain threshold of DL =24 μm (Fig. 13A). Respectively, no warm rain was observed,and not even graupel occurred in fresh growing convectivetowers.

Fig. 14 shows the gradient in cloud microphysical proper-ties, which correspond to the gradient in aerosol compositionfrom ocean to land, that was described at the end of theprevious section (Fig. 8C). As we headed east without changingheight from the ocean and towards landing in Sacramento inthe valley, the re of the convective clouds decreased along withan increase in cloud droplets concentrations. The sameoccurred as we kept heading east towards lower but morepolluted clouds, as a result of the transition from amaritime airmass with lots of sea salt to continental air mass with morebiomass burning and Central Valley pollution.

The microphysical analyses, shown above, show the influ-ence that the aerosol properties have on cloud microstructureand precipitation forming processes.

Lower DL and a steeper slope of re with height are the maincharacteristic of a highly pristine microstructure of clouds. Themore polluted clouds at noon have opposite microphysicalcharacteristics of higher DL and a moderate slope of re withheight.

4.3. Ice and warm rain domination processes

By analyzing the vertical evolution of volume weightedcloud drop size distributions one can elucidate the dominationof ice or warm rain processes in clouds from the isotherm 0 °Cup to their tops. As mentioned in the microphysical method-ology (Section 3.3), the smooth transitions from the cloud to

53%

11%

26%

11%

21%

1% 24%

2%

11%

40%

49%

4%

21%

2% 8%

17%

24%

3%

27%

7%

16%

22%

22%

%11%76

79%

4%

6%

10%

A B C

D

Sea Salt Dust Biological Biomass Burning Soot Organic Carbon Central Valley Pollution Urban Pollution

E F

Out of cloud

Fig. 8. Chemical composition of aerosols and cloud residues from two flights on February 21, 2011 during the following time periods: A) 17:20–17:27 UTC, cloudresidues in an elevated cloud layer at 4600 m to the west of the Central Valley; B) 18:45–19:07 UTC, cloud residues in low relatively clean clouds over the sea;C) 20:02–20:13 UTC, out of cloud aerosol while descending in the polluted air over the Central Valley; D) 21:58–23:02 cloud residues while climbing in pollutedconvective clouds at the foothills of the Sierra Nevada; UTC, F) 23:02–23:45 UTC, cloud residues in layer clouds above 2500 m that were partly mixed with thetops of the convective clouds.

120 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

rain drop sizes indicate cloud volumes in which warm rain isbeing produced. The distributionswith a separate precipitationpeak show cloud volumes towhich precipitation is falling fromabove, or in which ice precipitation is being produced.

Fig. 15 shows the particle size distribution from the flightof the 16th of February for the passes at the upper parts of the

Fig. 9. CPC-3010 (magenta), PCASP (black) and cloud drops concentrationswith cloud drops water content, LWC, as color bar, as a function of height forthe maritime convective clouds from the morning (around 18 UTC) flight of21 February 2011.

precipitating clouds that contained highly supercooled rain.Themorning clouds showmostly smooth distributions up to thecloud tops, indicating the dominance of warm rain processes(Fig. 15A). Warm rain at the coldest temperatures wasdocumented at 3600 m and −21 °C (Fig. 15C). At some pointthe supercooled rain was so intense at −8.5 °C that it melted

10

100

1000

104

105

106

10.1

UHSAS Concentration Vs. Diameter

215922.0481220619.1030222757.1221224243.1672224819.2053225942.2344230509.2909

UH

SA

S C

once

ntra

tion

[cm

-3]

Diameter [µm]

Fig. 10. This figure shows the increase in aerosol concentration with heightas measured from the UHSAS. We can see the change in behavior around0.1 μm where the PCASP starts to measure. While not taking into account thesmall aerosol below 0.1 μm, it seems that the aerosol concentration increaseswith height. At 2900 m this behavior stops, probably as a result of newairmass.

10-7

10-6

10-5

0.0001

0.001

0.01

0.1

10 100 1000

10 100 1000

172947.1408174516.2004174104.2040174729.2357

175114.3096175138.3173175622.3619

LWC

[g m

-3 µ

m-1

]LW

C [g

m-3

µm

-1]

Droplet Diameter [µm]

LWC CDP 2D-S Vs. Droplet DiameterA

10-7

10-6

10-5

0.0001

0.001

0.01

0.1

192718.3430

192807.3332

193058.2953

193915.2892

194747.2350

195212.1775

195814.1567

Droplet Diameter [µm]

LWC CDP 2D-S Vs. Droplet DiameterB

Fig. 11. The vertical evolution of volumeweighted cloud drop size distributionson themorning ascent (A) and noon descent (B) through the convective cloudson 16 Feb 2011. The cloud passes with the highest LWC for each height areshown.

LWC

[g m

-3 µ

m-1

]LW

C [g

m-3

µm

-1]

10-6

10-5

0.0001

0.001

0.01

0.1 185030.0604184854.0823

184759.0828

182919.1103

180442.1592

175136.2528

175019.2672

174024.2919

Droplet Diameter [µm]

LWC-CDP-2DS Vs. Droplet DiameterA

50

10-6

10-5

0.0001

0.001

0.01

0.1

10 100 1000

10 100 1000

222955.1248

223647.1364

224000.1443

224547.2044

225206.2257

225330.2344

230806.2915

Droplet Diameter [µm]

LWC-CDP-2DS Vs. Droplet DiameterB

Fig. 12. Panel A shows the LWC behavior of the convective clouds above theocean at the morning of the 21st. Panel B shows the LWC behavior of the“polluted” convective cloud at the afternoon. These clouds climb 1700 mabove their base without exceeding the diameter rain threshold of 24 μm.

121D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

away the layer of ice that formed previously on the aircraft'swindshield due to impaction of supercooled cloud drops. A fulldescription of the warm rain forming process of this case isgiven in Rosenfeld et al. (2013).

The noon passes show well separated cloud and precip-itation peaks, indicating the predominance of mixed phaseprecipitation (Fig. 15B). Warm rain still was formed, andreached the height of 3300 m and temperature of −19 °C.However, the dominant precipitation was graupel (Fig. 15D)whereas supercooled rain dominated the precipitation in themorning clouds.

The same behavior was seen on the 21st of February. Thevertical development of the drop sizes that were measuredduring the flights on the 21st of February is shown inFigs. 12A and 13A. At low altitudes warm rain that fell from

above created separate peaks. At the top of these convectiveclouds, these peaks were replaced with a smooth continua-tion from the CDP spectra to the 2D-S spectra between 50and 100 μm, where the spectra of the cloud drops becamewider towards the rain drop sizes. The rain was supercooled,with very few isolated ice particles observed in the cloudsover ocean at temperatures colder than −5 °C, despite theircloud top temperature reaching−12 °C. However, pockets ofhigh concentrations of columnar ice crystals were observednear the −4 °C isotherm (see Fig. 16). The crystals aggre-gated and formed snow and graupel at lower levels. Thesupercooled rain that fell from above froze when collidingwith the ice at the low levels. This created the strangesituation, where, as normally seen the rain changed to snow

1200 1600CDP Concentration and Altitude Vs. Longitude

CDP_conc[cm-3] Altitude (m)A

122 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

with height above the zero isotherm, but changed back tosupercooled rain at−4.5 °C, and remained so up to the cloudtops at −12 °C. How can this situation occur? Scarcity of INmight explain it. The cloud residue composition (Fig. 8B)shows large amount of sea salt aerosols and low amounts of

0

500

1000

1500

2000

2500

3000

3500

4000

CDP_DL 16 noonCDP_DL 21 morningCDP_DL 21 afternoonCDP_DL 16 morning

Hei

gh

t ab

ove

clo

ud

bas

e, m

Drop size of modal LWC, µm

Drop size of modal LWC, µm

24

Drop size of modal LWC Vs. Height above cload baseA

0

500

1000

1500

2000

2500

3000

3500

4000

CDP_DL 21 morning

CDP_DL 21 afternoon

California225

California1000

Hei

gh

t ab

ove

clo

ud

bas

e, m

Hei

gh

t ab

ove

clo

ud

bas

e, m

24

Drop size of modal LWC Vs. Height above cload baseB

0 10 20 30 40 50

0 10 20 30 40 50

0 10 20 30 40 500

1000

2000

3000

4000

5000

6000

7000

8000

Drop size of modal LWC Vs. Height above cload base

Amazon PyroAmazon PyroAmazon SmokyAmazon Trans 1005_2 GreenAmazon OceanThai SmokyThai CleanArgentina HailCalifornia PollutedCalifornia PristineCDP_DL 21 afternoonCDP_DL 21 morning

24

C

Drop size of modal LWC, µm

0

200

400

600

800

1000

200

400

600

800

1000

1200

1400

CD

P_c

onc[

cm-3

]

Altitude (m

)

Longitude (deg)

2

4

6

8

10

12

200

400

600

800

1000

1200

1400

1600CDP Re and Altitude Vs. Longitude

CDP_Re[um] Altitude (m)

CD

P_R

e[um

] Altitude (m

)

Longitude (deg)

B

Fig. 14. CDP concentration and altitude as a function of longitude (panel A).CDP-measured cloud drop effective radius, re, and altitude as a function oflongitude (panel B). As we flew east towards the valley, but without changingheight, the CDP increases (panel A, green arrow) and Re decreases (panel B,green arrow). Aswe kept flying east and descend towards the lower convectiveclouds above the valley the CDP increases (panel A, blue arrow) and Redecrease (panel B, blue arrow).

aerosol which can serve as effective IN, such as dust. This meansthat much ice can occur only when the small concentrations ofprimary ice crystals are greatly amplified by the ice multiplica-tion mechanism.

Fig. 13. The dependence of drop size on cloud depth in a global perspective.Panel A shows the modal LWC drop diameter, DL, as function of height abovecloud base from the flights on the 16th and 21st February 2011. The bold lineis a smoothing of the thin line that connects the individual measurements.Panel B shows the same for the morning and afternoon of the flight of 21stFebruary, and for the polluted and clean extremes from SUPRECIP 2008.Panel C shows the morning (black line) and afternoon (red line) flights from the21st February in a global context.Panel C was taken from Rosenfeld et al. (2008).

10-7

10-6

10-5

0.0001

0.001

0.01

0.1174807.2513174814.2534174819.2562175114.3096175121.3116175138.3173175256.3311175606.3606175615.3607175622.3619

Droplet Diameter [µm]

LWC CDP 2DS Vs. Droplet DiameterA

5010-7

10-6

10-5

0.0001

0.001

0.01

0.1

10 100 1000 10 100 1000

192948.3061192953.3040193004.2998

193038.2960193058.2953193103.2947

193218.2870193333.2863193636.2884

193642.2885

Droplet Diameter [µm]

LWC CDP 2DS Vs. Droplet DiameterB

50

D

C

LWC

[g m

-3 µ

m-1

]

LWC

[g m

-3 µ

m-1

]

Fig. 15. The highest altitude where warm rain was observed in the convective clouds according to the 2D-S instrument at the morning (A) 3619 m (brown line)and at noon (B) 3300 m. The smooth continuance around 50 μm is well seen at A whereas at B the graupel precipitation created a second peak. Panel C shows thelargest and coldest rain drops at the morning clouds at 17:56:26, 3615 m and −21.5 °C (brown line at panel A). Panel D shows the large abundance of graupelprecipitation in the noon clouds from the 2D-S at 19:29:54, 3035 m and −16.7 °C (red line at panel B).

123D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

The columnar habit of the ice crystals is compatible withthe observed temperature of −4 °C at the height of 1700 m.The temperature and large cloud drop size are consistentwith the hypothesis that the large concentrations of columnarcrystals was caused by ice multiplication (Hallett and Mossop,1974). The lack of ice from the convective cores and at coldertemperatures further supports the ice multiplication, as thisprocess requires time to advance. Therefore, the ice was foundonly in relatively old cloud elements that already lost most oftheir water into hydrometeors. These ice crystals aggregatedinto snowflakes and rimed into graupel while falling to lower

levels. The rain forming process in this case is also described ingreater detail in Rosenfeld et al. (2013).

The afternoon flight from the 21st of February climbedthrough the convective clouds that have developed in air thatcame from the valley with tops reaching 3 to 3.5 km (Fig. 2).The tops of these clouds expanded horizontally, forming layerclouds. The convective tops kept developing and penetratingthe layer to a short distance above it. The cloud drops at thelayer cloud remained as small as the drops of their parentconvective clouds, and reached the DL of only 12 μm (Figs. 12Band 17A). The cloud drops at the tops of the convective clouds

Fig. 16. 2D-S image from 17:59:29, 1708 m, −3.8 °C, shows ice columns.

124 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

grew larger simply because of being at greater depth abovecloud base (DL around 16 μm, Fig. 17, panel A) but still did notgrow enough to exceed the rain threshold initiation, DL of 24 μm.We can see from the 2D-S images (Fig. 17, panel B) that insidethese layer clouds there are large dendritic ice crystals, exceedingthe diameter of 1.5 mm. Apparently they grew to this large sizeby condensational growth as a result of the long lifetime of thelayer cloud near the tops of the convective clouds. The dendritichabit is compatible with the observed temperature between−12 °C and−15 °C at the height of 3000 m.When these iceprecipitation particles fell into convective clouds, they collectedthe cloud drops and became graupel. These are the precipitationparticles that are evident in Fig. 12B as the second peak.

As on the 16th of February, the clouds at the morning of the21st over ocean were also dominated by warm rain processesand developed large amounts of supercooled rain drops andsome ice precipitation by ice multiplication, while the afternoonclouds over the Sierra Nevada foothills could not develop anyprecipitation except for few graupel particles.

This domination of warm rain processes at the pristineclouds in themorning of both flights and the graupel formationat the more polluted clouds in the afternoons is once againrelated to the fact that these clouds ingested different aerosols(marine vs pollution aerosol).

5. Discussion and conclusions

The analysis of the contrasting conditions, encountered inthe three flights in CalWater presented here provided twomain results on the processes of transition from pristine topolluted convective clouds. The first result is the demonstratedgreat sensitivity of rain forming processes to the CCN aerosols.This result is compatible with previous studies showing similaraerosol effects (Andreae et al., 2004; Freud and Rosenfeld,2012; Rosenfeld et al., 2008). The second result was thesurprising dearth of ice in the growing convective cloudswhen formed in maritime air mass, evident in the high degreeof super cooling. Furthermore, some of the clouds producedless ice when developed to colder temperatures. Additional

documentation of the IN in some of these clouds is presented inCreamean et al. (2013) and Rosenfeld et al. (2013). Theseresults are new and deserve some additional discussion here.

Rosenfeld et al. (2008) concluded that the suppression ofwarm rain in central California occurs mainly in the afternoonorographic clouds over the foot hills of the Sierra Nevada. Theseclouds are mostly fed by local aerosol sources in the CentralValley. They are comprised of both urban and nonurban sources.The results presented here from the CalWater campaign fosterthe results of Rosenfeld et al., 2008. As we discussed above, themorning clouds had ample warm rain, but warm rain initiationwas suppressed in the afternoon orographic clouds as a result ofthe increase in aerosols entering the cloud base. These aerosols,as we saw earlier, are mostly pollution from the Central Valley(i.e. soot, biomass burning, urban pollution and ammonium-richaerosols).

5.1. The effects of aerosols on warm rain

Larger drop concentrations near cloud base lead to greaterdepth for onset of warm rain. Fig. 13A shows the drop size ofmodal LWC taken from Figs. 11(A, B) and 12(A, B) as a functionof D (height above cloud base). The DL of 24 μmwas reached atlower D when the clouds developed at earlier times of the day,with the BL still being decoupled from the cloud base, andtherefore smaller cloud drop concentrations were formed.

Rosenfeld et al. (2008) showed in their Fig. 14 the highcorrelation between CDP concentration and Dp over Californiaand in a global context. As the clouds contain higher dropconcentration the clouds reach the DL of 24 μm and start toprecipitate at higher altitudes. They measured pristine clouds inCaliforniawithdrop concentration of 150 cm−3which started toprecipitate at 0.5 km above cloud base. The same Dp occurred inour measurements at the maritime convective cloud in themorning of the 21st February, which had maximum CDPconcentration of 200 cm−3. Rosenfeld et al. (2008) alsoshowed that polluted clouds in California that had maxi-mum drop concentration of 400 cm−3 did not reach the DL

of 24 μm at a depth of 2 km. In the CalWater campaign, three

10-5

0.0001

0.001

0.01

0.1

1

0 10 20 30 40 50

230806.2915

230809.2924

231143.2943

230731.2945

231128.2952

230943.2986

Droplet Diameter [µm]

LWC-CDP Vs. Droplet DiameterA

Layer clouds that formed from the convective clouds - 12 µm

Tops of the convective clouds - 16 µm

B

LWC

[g m

-3 µ

m-1

]

Fig. 17. Convective clouds that spread horizontally and formed layer clouds at their sides develop different modal diameter for the convective tops and the expandedhorizontally layer clouds (panel A). 2D-S image, from 21 February 23:09:23 at 2979 and −11.7 °C, show dendrites that formed in these layer clouds (panel B).

125D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

years later, the polluted clouds in the afternoon of the 21sthad maximum CDP concentration of 800 cm−3 and showedrespectively a steeper slope of DL with height, which meansa greater height for rain initiation (Fig. 13B). Fig. 13C showsthat in a global context the clouds in California cover the entirerange of clouds. The pollution in the Central Valley caused theclouds to develop the same steep slope of DL with height as inthe polluted clouds over the Amazon and the clean air massover the ocean caused a lower Dp at maritime clouds than inpristine cases over the Amazon. Freud et al. (2011) showedthat, due to the nearly extreme inhomogeneous nature of cloudmixing with the environment, the re does not diverge muchfrom the adiabatic re. The linear relationship between adiabaticwater and cloud depth caused the ratio between Na, thenumber of activated clouddroplets near the cloud base, and thedepth where we reach a critical effective radius for precipita-tion, rep, to be almost linear. Freud and Rosenfeld (2012)showed that the rep is close to 14 μm. Their conclusions are ingood agreement with our results. The maritime clouds in the

morning of the 21st of February start to precipitate 500 m abovecloud base and reaches re of 14 μm 500 m above cloud base(Fig. 4A).

5.2. The effects of aerosols on ice precipitation

Ice processes did not contribute much to the precipitationin some of the convective clouds in these case studies, eventhough cloud top temperatures were colder than −21 °C andsupercooled rain was detected in most of their volume.Precipitation was initiated mostly as supercooled rain dropsin growing convective clouds in pristine maritime air up tothese cold temperatures. This was found both over the ocean,and in a maritime air mass that reached to the Sierra Nevadawhen not mixingwith the stable continental boundary layer inthe early morning. This is remarkable that such conditionspersist so extensively.

This apparent lack of ice nuclei underlines their impor-tance when they do occur. The convective clouds above the

126 D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

ocean on the 21st of February and above the Central Valley onthe 16th of February 2011 mostly developed warm rain. Noice was observed in the growing convective elements overthe ocean even at their tops (2900 m, where temperaturereached −12 °C), except for ice columns in maturing cloudsat the−4 °C isotherm. A deeper discussion of this behavior isgiven by Rosenfeld et al. (2013).

The difference that pollution aerosols can make insupercooled maritime clouds was evident in the flight of16 February. The ascent part of the flight documented convectiveclouds that were triggered by the Sierra Nevada, in maritime airthat was decoupled from the continental boundary layer. Warmrain developed in them almost as fast as in the maritime cloudsover ocean on the 21st of February (see the comparison ofDL(D) in Fig. 13). Similarly, no ice was formed in the growingconvective elements up to the coldest measurements in themat −21 °C. Graupel did form in the clouds when their topsbecame colder and more mature as they approached the ridgeline of the Sierra Nevada.

Upon the descent through similar clouds 2 h later, it wasfound that they ingested some air pollution which made theirdrops smaller andmore numerous, and also probably containedsome IN, as some pristine ice crystals were also observedbetween the−9 and−15 °C isotherms and dust was observedwith the A-ATOFMS. The clouds still initiated precipitation bywarm rain processes, but the supercooled rain drops frozeinto graupel that dominated the precipitation forms. Smallsupercooled rain drops were observed up to the −19 °Cisotherm. It appears that graupel formed in these clouds only athigh altitudes (3000 m and−17 °C) after the cloud developedwarm rain that was forced up by updrafts and froze at theseheights. The convective clouds that grew earlier in themorningdid not develop ice precipitation probably due to dearth ofaerosol that could serve as effective IN. At noon the cloudscontained larger amounts of dust, biological and biomassburning aerosols, and produced more ice.

Adding many more pollution aerosols, as was the case inthe afternoon flight of 21 February, suppressed the warm rainprocesses altogether along with the formation of graupel inthese clouds. Ice crystals did form in the clouds after maturingat temperatures of −12 °C or colder, which were capable ofproducing graupel when falling into younger parts of theclouds that still contained large amounts of supercooled water.This indicates that IN did exist in this polluted air, but the rateof growth of the ice hydrometeors is rather slow in clouds withvery small cloud drops, as already shown in previous studieselsewhere (Borys et al., 2003).

In summary, in amatter of only two days and three flightswedocumented quite contrasting conditions of cloud microstruc-ture and precipitation forming processes over central California,caused primarily by variability of the aerosols. The roles of bothCCN and IN are important. This study provides themotivation tolook into these roles more deeply, in the context of morecomprehensive study of the CalWater flights coupled with cloudsimulations.

Acknowledgments

The CalWater project was funded by the California EnergyCommission (CEC). The G-1 aircraft was operated by the PacificNorthwest National Laboratory with funds from CEC. The PI of

the CalWater project is Dr. Kimberly Prather. I thank the manypeople who worked hard to fund, plan and execute the fieldcampaign. Special thanks are due to the aircraft scientists andpilots. This study was partially supported by the U.S. Depart-ment of Energy (DOE) Office of Science (BER) AtmosphericSystem Research program.

References

Andreae, M.O., Rosenfeld, D., Artaxo, P., Costa, A.A., Frank, G.P., Longo, K.M.,Silva-Dias, M.A.F., 2004. Smoking rain clouds over the Amazon. Science303, 1337–1342.

Borys, R.D., Lowenthal, D.H., Cohn, S.A., Brown, W.O.J., 2003. Mountaintopand radar measurements of anthropogenic aerosol effects on snowgrowth and snowfall rate. Geophys. Res. Lett. 30, 1538.

Cotton, W.R., Ward, D., Saleeby, S., 2010. Aerosol impacts on wintertimeorographic precipitation. 13th conference on Clouds Physics.

Creamean, J.M., Suski, K.J., Rosenfeld, D., Cazorla, A., DeMott, P.J., Sullivan,R.C., White, A.B., Ralph, F.M., Minnis, P., Comstock, J.M., Tomlinson, J.M.,Prather, K.A., 2013. Dust and biological aerosols from the Sahara andAsia influence precipitation in the western US. Science 339 (6127),1572–1578. http://dx.doi.org/10.1126/science.1227279.

Fergenson, D.P., Pitesky, M.E., Tobias, H.J., Steele, P.T., Czerwieniec, G.A.,Russell, S.C., Lebrilla, C.B., Horn, J.M., Coffee, K.R., Srivastava, A., Pillai,S.P., Shih, M.T., Hall, H.L., Ramponi, A.J., Chang, J.T., Langlois, R.G., Estacio,P.L., Hadley, R.T., Frank, M., Gard, E.E., 2004. Reagentless detection andclassification of individual bioaerosol particles in seconds. Anal. Chem.76 (2), 373–378.

Flossmann, A.I., Wobrock, W., 2010. A review of our understanding of theaerosol–cloud interaction from the perspective of a bin resolved cloudscale modelling. Atmos. Res. 97 (4), 478–497.

Freud, E., Rosenfeld, D., 2012. Linear relation between convective cloud dropnumber concentration and depth for rain initiation. J. Geophys. Res. 117,D02207. http://dx.doi.org/10.1029/2011JD016457.

Freud, E., Rosenfeld, D., Kulkarni, J.R., 2011. Resolving both entrainment-mixing and number of activated CCN in deep convective clouds. Atmos.Chem. Phys. 11, 12887–12900. http://dx.doi.org/10.5194/acp-11-12887-2011.

Gard, E.E., Kleeman, M.J., Gross, D.S., Hughes, L.S., Allen, J.O., Morrical, B.D.,Fergenson, D.P., Dienes, T., Galli, M.E., Johnson, R.J., Cass, G.R., Prather,K.A., 1998. Direct observation of heterogeneous chemistry in the atmosphere.Science 279 (5354), 1184–1187.

Gaston, C.J., Furutani, H., Guazzotti, S.A., Coffee, K.R., Bates, T.S., Quinn, P.K.,Aluwihare, L.I., Mitchell, B.G., Prather, K.A., 2011. Unique ocean-derivedparticles serve as a proxy for changes in ocean chemistry. J. Geophys.Res.-Atmos. 116.

Givati, A., Rosenfeld, D., 2004. Quantifying precipitation suppression due toair pollution. J. Appl. Meteorol. 43, 1038–1056.

Givati, A., Rosenfeld, D., 2005. Separation between cloud seeding and airpollution effects. J. Appl. Meteorol. 44, 1298–1314.

Givati, A., Rosenfeld, D., 2007. Possible impacts of anthropogenic aerosols onwater resources of the Jordan River and the Sea of Galilee. Water Resour.Res. 43, W10419.

Griffith, D.A., Solak, M.E., Yorty, D.P., 2005. Is air pollution impacting winterorographic precipitation in Utah? Weather modification association.J. Weather Modif. 37, 14–20.

Hallett, J., Mossop, S.C., 1974. Production of secondary ice particles duringthe riming process. Nature 249, 26–28.

Hudson, J.G., Mishra, S., 2007. Relationships between CCN and cloudmicrophysicsvariations in clean maritime air. Geophys. Res. Lett. 34, L16804 (6 pp.).

Hudson, J.G., Yum, S.S., 2001. Maritime-continental drizzle contrasts in smallcumuli. J. Atmos. Sci. 58, 915–926.

Jirak, I.L., Cotton, W.R., 2005. Effect of air pollution on precipitation along thefront range of the rocky mountain. J. Appl. Meteorol. 45, 236–245.

Junkermann, W., Vogel, B., Sutton, M.A., 2011. The climate penalty for cleanfossil fuel combustion. Atmos. Chem. Phys. Discuss. 11, 24567–24589.http://dx.doi.org/10.5194/acpd-11-24567-2011.

Lohmann, U., 2002. A glaciation indirect aerosol effect caused by soot aerosols.Geophys. Res. Lett. 29 (4), 1052. http://dx.doi.org/10.1029/2001GL014357.

Lynn, B., Khain, A., 2006. The Influence of Aerosols on Precipitating Systems:Analysis of Results from the WRFSBM.

McFarquhar, G.M., Heymsfield, A.J., 2001. Parameterizations of INDOEXmicrophysical measurements and calculations of cloud susceptibility:applications for climate studies. J. Geophys. Res. 106, 28,675–28,698.

Muhlbauer, A., Lohmann, U., 2006. Dynamics vs. aerosol induced warm-phase microphysics and orographic precipitation. 12th Conference onMountain Meteorology.

127D. Rosenfeld et al. / Atmospheric Research 135–136 (2014) 112–127

Muhlbauer, A., Lohmann, U., 2009. Sensitivity studies of aerosol–cloudinteractions in mixed-phase orographic precipitation. J. Atmos. Sci. 66,2517–2538.

Pinsky, M., Khain, A., Rosenfeld, D., Pokrovsky, A., 1998. Comparison ofcollision velocity differences of drops and graupel particles in a veryturbulent cloud. Atmos. Res. 49 (2), 99–113 (September 1998).

Pratt, K.A., Mayer, J.E., Holecek, J.C., Moffet, R.C., Sanchez, R.O., Rebotier, T.P.,Furutani, H., Gonin, M., Fuhrer, K., Su, Y.X., Guazzotti, S.A., Prather, K.A.,2009a. Development and characterization of an aircraft aerosol time-of-flight mass spectrometer. Anal. Chem. 81 (5), 1792–1800.

Pratt, K.A., DeMott, P.J., French, J.R., Wang, Z., Westphal, D.L., Heymsfield, A.J.,Twohy, C.H., Prenni, A.J., Prather, K.A., 2009b. In situ detection ofbiological particles in cloud ice-crystals. Nat. Geosci. 2 (6), 398–401.

Pratt, K.A., Murphy, S.M., Subramanian, R., DeMott, P.J., Kok, G.L., Campos, T.,Rogers, D.C., Prenni, A.J., Heymsfield, A.J., Seinfeld, J.H., Prather, K.A.,2011. Flight-based chemical characterization of biomass burningaerosols within two prescribed burn smoke plumes. Atmos. Chem.Phys. 11 (24), 12549–12565.

Qin, X., Pratt, K.A., Shields, L.G., Toner, S.M., Prather, K.A., 2012. Seasonalcomparisons of single-particle chemical mixing state in Riverside, CA.Atmos. Environ. 59, 587–596.

Rosenfeld, D., 2000. Suppression of rain and snow by urban air pollution.Science 287, 1793–1796.

Rosenfeld, D., Givati, A., 2006. Evidence of orographic precipitation suppres-sion by air pollution-induced aerosols in the western United States. J. Appl.Meteorol. 45, 893–911.

Rosenfeld, D., Woodley, W.L., Krauss, T.W., Makitov, V., 2006. The structureof severe convective storms in Mendoza, Argentina. J. Appl. Meteorol. 45,1261–1281.

Rosenfeld, D., Dai, J., Yu, X., Yao, Z., Xu, X., Yang, X., Du, C., 2007. Inverserelations between amounts of air pollution and orographic precipitation.Science 315, 1396–1398.

Rosenfeld, D., Woodley, W.L., Axisa, D., Freud, E., Hudson, J.G., Givati, A.,2008. Aircraft measurements of the impacts of pollution aerosols onclouds and precipitation over the Sierra Nevada. J. Geophys. Res. 113,D15203. http://dx.doi.org/10.1029/2007JD009544 (23 pp.).

Rosenfeld, D., Chemke, R., DeMott, P., Sullivan, R.C., Rasmussen, R., McDonough, F.,Comstock, J., Schmid, B., Tomlinson, J., Jonsson, H., Suski, K., Prather, A.K.,2013. The Common Occurrence of Highly Supercooled Drizzle and Rain nearthe Coastal Regions of the Western United States. JGR. http://dx.doi.org/10.1002/jgrd.50529.

Russell, S.C., 2009. Microorganism characterization by single particle massspectrometry. Mass Spectrom. Rev. 28 (2), 376–387.

Saleeby, S.M., Cotton, W.R., Lowenthal, D., Borys, R., Wetzel, M.A., 2008.Influence of cloud condensation nuclei on orographic snowfall. J. Appl.Meteorol. 48, 903–922.

Saleeby, S.M., Cotton, W.R., Fuller, J.D., 2011. The cumulative impact of clouddroplet nucleating aerosol on orographic snowfall in Colorado. J. Appl.Meteorol. Climatol. 50, 604–625.

Shields, L.G., Suess, D.T., Prather, K.A., 2007. Determination of single particlemass spectral signatures from heavy-duty diesel vehicle emissions forPM2.5 source apportionment. Atmos. Environ. 41 (18), 3841–3852.

Silva, P.J., Liu, D.Y., Noble, C.A., Prather, K.A., 1999. Size and chemicalcharacterization of individual particles resulting from biomass burning oflocal southern California species. Environ. Sci. Technol. 33 (18), 3068–3076.

Silva, P.J., Carlin, R.A., Prather, K.A., 2000. Single particle analysis of suspendedsoil dust from southern California. Atmos. Environ. 34 (11), 1811–1820.

Sodeman, D.A., Toner, S.M., Prather, K.A., 2005. Determination of singleparticle mass spectral signatures from light-duty vehicle emissions.Environ. Sci. Technol. 39 (12), 4569–4580.

Toner, S.M., Sodeman, D.A., Prather, K.A., 2006. Single particle characterization ofultrafine and accumulation mode particles from heavy duty diesel vehiclesusing aerosol time-of-flight mass spectrometry. Environ. Sci. Technol. 40(12), 3912–3921.

Whiteaker, J.R., Suess, D.T., Prather, K.A., 2002. Effects of meteorologicalconditions on aerosol composition and mixing state in Bakersfield, CA.Environ. Sci. Technol. 36 (11), 2345–2353.

Yang, X., Ferrat, M., Li, Z., 2013. New evidence of orographic precipitationsuppression by aerosols in central China. Meteorol. Atmos. Phys. http://dx.doi.org/10.1007/s00703-012-0221-9.

Yum, S.S., Hudson, J.G., 2002. Maritime/continental microphysical contrastsin stratus. Tellus, Ser. A and Ser. B, 54B, pp. 61–73.