Aerosol - Cloud Interactions in the Arctic Boundary Layer

(1)

MASTER’S THESIS

KARIN RYBERG

Aerosol - Cloud Interactions in the Arctic Boundary Layer

Ny - Ålesund, Svalbard

MASTER OF SCIENCE PROGRAMME in Space Engineering

Luleå University of Technology Department of Space Science, Kiruna

2006:148 CIV • ISSN: 1402 - 1617 • ISRN: LTU - EX - - 06/148 - - SE

(2)

Aerosol – Cloud Interactions in the Arctic Boundary Layer, Ny-Ålesund, Svalbard

Master’s thesis

AEROSOL – CLOUD INTERACTIONS IN THE ARCTIC BOUNDARY LAYER

NY-ÅLESUND, SVALBARD

February 2006

KARIN RYBERG

Master of Science programme Space Engineering

Luleå University of Technology Department of Space Science

Stockholm University

Department of Applied Environmental Science

Air Pollution Laboratory

(3)

(4)

PREFACE

After four years of studying at Luleå University of Technology in Luleå and Kiruna and an additional year in Vienna at Vienna University of Technology my Master’s degree in Space Engineering was coming close. To get my degree, I had only to finish twenty weeks of thesis work. To do this I went to Stockholm, and the Department of Applied Environmental Science at Stockholm University.

I would first of all like to thank Johan Ström, my supervisor and guide at the Department of Applied Environmental Science, Stockholm University. Thank you for giving me such an interesting topic and for introducing me to the field of aerosol science. Thank you for all your help and patience.

I would also like to thank Victoria Barabash, my examiner at the Department of Space Science, Luleå University of Technology, in Kiruna, for help and comments during the work.

Special thanks goes to Gothenburg and my family who have supported me and helped me through the years. Additionally, I would like to thank all my friends from Luleå University of Technology for always being there and for all the time we spent together.

The content of this study has been presented by a poster at the NOSA symposium in Gothenburg, November 3 – 4, 2005.

Karin Ryberg

Stockholm, December 2005

(5)

(6)

ABSTRACT

Aerosols take part in the cloud formation process by acting as cloud condensation nuclei. An increase in anthropogenic aerosols in the atmosphere is believed to have significant effects on cloud properties and therefore weather and climate. In this study, aerosol – cloud interactions in the Arctic boundary layer are investigated, by using data from years 2003 and 2004, observed at Ny-Ålesund, Svalbard. In-situ measurements of aerosol particle size distributions from the Zeppelin station, 474m asl, are compared with lidar measurements of cloud and aerosol backscatter from Ny-Ålesund by statistical methods. From the lidar backscatter data three variables were educed: average backscatter from the boundary layer (0 - 2km), backscatter from 474m altitude and backscatter from 948m altitude. With data from the first variable a clear relationship between particle size and average backscatter can be seen, showing positive correlation for small aerosols, whereas negative correlation is observed for large aerosols (>100nm diameter) during times of high backscatter. The other two variables made it possible to study the aerosol – cloud interaction during periods when the station was in cloud free air whereas the level just above it was within clouds. This is an analogue to aircraft measurements where measurements are first made below clouds and then inside clouds. The result gave a rather complex picture, however the interpretation is that this complexity is due to a much stronger influence by the clouds on the aerosol properties than vice versa.

The research has been carried out at Stockholm University, Department of Applied Environmental Science (ITM), Stockholm, during summer and autumn 2005.

Keywords: Lidar measurements, Aerosol size distribution, Zeppelin station, Statistics.

(7)

(8)

1. INTRODUCTION

Clouds are fascinating weather phenomena and an important part of our climate system. Aerosols are another important part because they are the nuclei on which clouds form. This thesis focuses on aerosol-cloud interactions in the Arctic boundary layer at Ny-Ålesund, Svalbard during the years 2003 and 2004. In-situ measurements of aerosols and remote sensing of cloud properties are studied to see interactions between the two.

The hypothesis of the work is that a significant change in atmospheric aerosol properties will lead to significant change in the cloud properties in the Arctic boundary layer. This is often referred to as the so-called Twomey effect [Twomey, S., 1977], whereby an increase in aerosol particles lead to brighter clouds and hence an effect on climate. More about the Twomey effect will follow in chapter 1.4.1.

1.1. AEROSOLS

Aerosols are suspensions of solid or liquid particles in gas, commonly air. In atmospheric physics and meteorology, an aerosol is said to be a small droplet or particle suspended in the atmosphere.

Aerosol size can range from about 0.001µm up to 3µm or more. This study focuses on aerosols between 0.02 and 0.63µm.

1.1.1. SOURCES OF AEROSOLS

Aerosols are introduces into the atmosphere through both natural and anthropogenic processes.

Examples of natural sources are soil and deserts, animals and plants, oceans and fresh water and volcanoes and fires. Anthropogenic sources are for example biomass burning, cars and factories.

Not only particles are released from these sources, also different gases that will convert to aerosols in the atmosphere, referred to as secondarily produced particles. Clouds may also be a source of aerosols. The high humidity in and around clouds can promote particle formation and photons are scattered between cloud droplets, increasing their opportunity to take part in photochemical particle formation.

1.1.2. SINKS FOR AEROSOLS

There are essentially two ways to remove aerosol particles from the atmosphere: dry and wet deposition. Dry deposition is when aerosols by mainly gravity and inertial forces impact on surfaces; ground, ocean, trees, buildings etc. Wet deposition is when aerosols are incorporated into cloud droplets and removed with precipitation, so called in-cloud scavenging. It is also possible that aerosols located below a precipitating cloud become scavenged by impacting raindrops, so called below-cloud scavenging. Aerosols are small, which makes gravity less important in the removal process. Clouds therefore play a large roll for the aerosol residence time in the atmosphere.

(11)

4 (27)

1.2. CLOUD FORMATION AND ARCTIC CLOUDS

Clouds form when air cools, become saturated and condensate on available particles. These particles on which cloud drops form are called cloud condensation nuclei, CCN. Not all aerosols can act as CCN. Generally speaking the aerosol should be relatively big and hygroscopic. The amount of hygroscopic material on the particle is important for droplet formation. Droplets typically form on particles larger than approximately 100nm. Salt particles are good examples of efficient CCN. The drops grow bigger as condensation continues and also by collisions between droplets. The humidity necessary for drops to grow depends on the drops size and composition.

When the size of the cloud droplets increases, their falling velocity increase. As long as this velocity is smaller than the upward motion of the air, the droplets stay as a cloud. However, when the droplets are big enough, their velocity overcome the air motion and they fall down towards Earth as precipitation. The radius of a cloud droplet is normally less than 0.1mm, which could be compared with a typical raindrop with a radius of about 1mm.

Clouds interesting to this study are low-level clouds somewhere between the ground and two kilometres up in the atmosphere. Such clouds can be: Cumulus, which could be either optically thin or thick, mostly giving no precipitation, Stratocumulus, a continuous cloud cover that consists of water droplets and lastly, Stratus, an even, continuous cloud cover, almost always together with fog [Lejenäs, H., 1997]. However, in this study, no differentiation between cloud types will be done, only between clouds and clear sky. Therefore the names of the clouds will not be used. All clouds will simply be referred to as clouds.

1.3. THE ARCTIC

The Arctic region has got its name from the Greek word Arctos that means bear, from the stellar constellation Ursa major (the Great bear). A common definition of the Arctic is the area encircled by the Polar circle (66˚32’N). Another way to define the region is the area around the North Pole down to the tree line in the south. Figure 1 below shows the North Pole and its surroundings. The black square indicates the Svalbard islands, which have been enlarged in the right part of the figure. Svalbard and Ny-Ålesund (78° 55´ N, 11° 56´ E) will be a part of the Arctic in both definitions above.

Figure 1. The Arctic region and Svalbard [www.burgergames.com/Syndicate/kryo1.htm, 2005-10-31]. The black square shows the Svalbard area that is magnified in the right part of the figure

[ www.paulnoll.com/Locations/visiting-Svalbard.html, 2005-10-31].

(12)

5 (27) 1.3.1. ARCTIC CLIMATE CHANGE

The Arctic environment is one of the cleanest environments on Earth. This makes it extremely vulnerable to any kind of pollution. A recent report from ACIA (Arctic Climate Impact Assessment) [ACIA, 2004] states that Arctic climate change has already taken place, due to an increase of carbon dioxide and other greenhouse gases in the atmosphere. The Arctic is warming and has warmth more than twice as fast as the rest of the world during the last decades. This is alarming, not only for the Arctic areas but also for the rest of the world. Several studies have reported on thinning sea ice over the Arctic Ocean [Rothrock, D. A. et al., 1999; Chapman, W. L.

& Walsh, J. E., 1993], which will lead, among other effects to sea level rise around the world and maybe a shift in ocean currents due to more fresh water in the oceans. Regional and global climate will change; animals and plants will be affected, as well as land use and infrastructure.

1.3.2. EFFECTS OF THE POLAR FRONT ON ARCTIC CLIMATE

The Polar front separates air masses of polar and tropical origin. During winter and spring an intense high-pressure system over the Arctic presses the Polar front south. This makes areas with high pollution a part of the Arctic air mass. Winds carry pollutants from these areas towards the Arctic. The Arctic cold winter temperature and low precipitation rate gives the pollutants time to accumulate in the atmosphere. The lack of sunlight further promotes the high pollution rate.

Sunlight would help breaking down some contaminants. In the summer the high-pressure system over the Arctic breaks up, allowing the Polar front to retreat. Transportation of pollutants from lower latitudes therefore is much less important during the summer. The summer is also warmer, allowing more clouds and drizzle to form. This prevents the contaminants to be transported far from their sources.

1.3.2. ARCTIC AEROSOLS

Aerosols normally have a lifetime of 3-7 days. In the Arctic however, the lifetime differs a lot from summer to winter. The winter, which is dark due to lack of sunlight and stable with low precipitation rates, prolongs the summer lifetime of 3-7 days up to 3-7 weeks [Sirois, A & Barrie, L. A., 1999].

The size and number concentration of arctic aerosols also differs widely throughout the year. The summer is the cleanest period of the year even though it shows the largest number density of aerosols. These aerosols tend to be in Aitken mode (diameter between approximately 10 and 80nm) [Ström et al. 2003]. Due to their small size the source of these particles must be the Arctic it self. However, the exact mechanism is still a research topic. Recent studies suggest particle formation through photochemical reactions as a plausible pathway [NOSA, 2005]. The spring shows almost as many particles as the summer months, but they have different size and origin.

During this period the visibility reduction phenomenon Arctic haze can be seen, foremost between 0 – 2km up in the troposphere [Radke et al., 1989], which is the region on which this study focus. These events occur due to a strong east-west pressure gradient that exists during the winter, letting episodes of polluted air from Western and Central Europe or Siberia [Barrie, L. A., 1986] to reach the Arctic. Not until the polar front retreats during the spring, when moist turbulent air reaches the region do the aerosols disperse. The late winter and the early spring is thus the most polluted season of the year with a lot of large anthropogenic aerosols. Ström et al.

2003, gives more details about the Arctic aerosol size distribution.

(13)

6 (27)

1.4. AEROSOL – CLOUD INTERACTIONS

1.4.1. TWOMEY EFFECT

As mentioned, aerosols take part in the cloud formation process by acting as cloud condensation nuclei (CCN). When the water vapour in the atmosphere condenses to form clouds it condenses on particles available. If the water content is constant and the amount of aerosols is high, many small droplets will form instead of a few large [Twomey, S., 1977]. This is commonly referred to as the Twomey effect and it means that the cloud albedo changes (i.e. the clouds become brighter) and more radiation is reflected back to space. It also means that the lifetime of the cloud become longer (i.e. it will precipitate less) and the mean liquid water path (LWP) increase. These changes will affect the atmospheric thermodynamic process and thus weather and climate, for details see Garrett et al. (2002a).

1.5. AEROSOL – CLOUD – CLIMATE INTERACTIONS

The Earth receives energy from the Sun that warms the atmosphere and the planet and makes it inhabitable. The planet absorbs about 2/3 of this incoming radiation. The other 1/3 is reflected back to space by the atmosphere, aerosols, clouds or the Earth surface. For the Earth not to become warmer and warmer the planet emits, long-wave, thermal radiation. This thermal radiation is absorbed by so called greenhouse gases, which results in the mean temperature of the Earth being about 15ºC instead of -15ºC. Aerosols and clouds influence this radiation balance in different ways and by changing the number density of aerosols in the atmosphere several changes in our climate system may occur.

Figure 2. Anthropogenic and natural forcing of climate according to the IPPC (Intergovernmental panel on climate change). [www.ipcc.ch/present/graphics/2001syr/small/06.01.jpg, 2005-11-02]

(14)

7 (27)

Figure 2 shows the knowledge so far about aerosols and their impact on the climate. It also shows that there exist large uncertainties and that the understanding in many areas are low or very low.

Aerosol effects are commonly divided into direct and indirect effects. Reflection and absorption of solar radiation by aerosols are referred to as the direct effect. The aerosol influence on clouds is referred to as the indirect effect.

On clear days, aerosols such as soot absorb incoming solar radiation and warm the atmosphere.

Other particles reflect the radiation back to space, having a cooling effect. When clouds are present, incoming short-wave radiation is reflected back to space, but outgoing long-wave radiation is absorbed by clouds and thus has a warming effect on the atmosphere. The effect of the heating depends on the cloud height and temperature. Low-level clouds, with almost the same temperature as the Earth surface almost always cool the system [Twomey, S., 1991], but high clouds at cold temperatures may act as a greenhouse gas and warm the system.

During the Arctic summer, clouds are very common and short-wave radiation is reflected back to space, having a cooling effect on the area. But, because much of the Arctic surface already has a high surface albedo, this effect is relatively small compared to regions on lower latitude [Garrett, T. J. & Zhao, C. et al. 2004]. A bright surface over another bright surface can actually lower the ground albedo and somewhat warm the surface. During the winter, when the sun is not present, long-wave radiation is absorbed by clouds, which reduce the amount of outgoing energy to space and thus warming the planet. Warming is also the result of cloudy days during fall and spring when the sun only barely reaches over the horizon. Short-wave radiation is then reflected by the cloud base and absorbed or further reflected by the ground.

The overall effect of the indirect and the direct effect of aerosols on the Earth is thought to be cooling, however, for the Arctic region the opposite could be true. The Arctic cloud cover may act to warm the surface and slightly cool the lower atmosphere [Stone, R. S., 1997].

2. MEASUREMENTS AND METHOD

Aerosol measurements were made at the Zeppelin station (78º54´N, 11º53´E, 474m a s l), situated at the Zeppelin Mountain near Ny-Ålesund, Svalbard. It is an ideal place to study the atmosphere and influences of anthropogenic aerosols. It represents the boundary layer and air masses are almost unaffected by local sources of contamination. Lidar measurements were made from Ny-Ålesund. Figure 3 shows the both sampling sites.

(15)

8 (27)

Figure 3. Map of sampling sites. The DMPS was placed at the Zeppelin Mountain and the lidar was placed in Ny-Ålesund, both indicated on the map to the left.

The measurements made in this study observe the cloud and aerosol backscatter and aerosol size distribution at fixed places. The study must therefore be regarded as an Eulerian study. An air parcel under investigation in the atmosphere changes over time and there are two ways one can follow it: the Lagrangian or the Eulerian way. In the Lagrangian approach, the air parcel is followed completely over time. In the Eulerian approach, changes are measured at fixed points and one cannot be sure that the parcel measured at one point will be the same measured at the next point. Figure 4 illustrates the differences between Eulerian and Lagranian observations.

Figure 4. The Euler (left) and the Lagrange (right) principle respectively. The cartoon person represents the observer; the box represents the air parcel that is investigated.

(16)

9 (27)

2.1. AEROSOL MEASUREMENTS

Particle data from the Zeppelin station (78º54´N, 11º53´E, 474m a s l), Ny-Ålesund, Svalbard, was collected by the Department of Applied Environmental Science, Stockholm University and are freely available from the CREATE aerosol database at NILU, the Norwegian Institute for Air Research (www.nilu.no/projects/ccc/create/database.htm) in collaboration with the World Data Centre for Aerosols, which is a part of the Global Atmosphere Watch program of the World Meteorological Organization. Data was collected with a DMPS, Differential Mobility Particle Sizer. The instrument consists of several different units, working together to give information about the aerosol size distribution in an interval between 0.02 and 0.63µm. Examples of aerosol size distributions from two different months (January and August) are given in Figure 5 and 6.

The January distribution shows a maximum of particles larger than 100nm. These particles are referred to as Accumulation mode particles. The August distribution shows a maximum of small aerosols (<100nm). The small particles are referred to as Aitken mode particles. One size distribution is observed every second minute.

Figure 5. Aerosol size distribution for January (clear conditions).

Figure 6. Aerosol size distribution for August (clear conditions).

2.2. LIDAR MEASUREMENTS

To be able to get good ground based measurements of aerosol and cloud scattering a Micro Pulse Lidar (MPL) is used. A lidar gives long-term datasets, which make it possible to study aerosol cloud interactions over a long time (two years in this case). Data used in this study come from the German research station at Ny-Ålesund where NIPR (National Institute of Polar Research of Japan) with collaboration of the Norwegian Polar Institute (NP), Alfred-Wegener Institute for Polar and Marine Research of Germany (AWI) and other institutions have placed a cloud and aerosol observation system, including an MPL. An MPL sends out short pulses of laser light.

Particles in the atmosphere scatter this light and a part of the scattered light returns back to the instrument. The time it takes for the pulse to travel back and forth is a measure of the distance to the scattering object. To get a correct result the instrument takes into account how much of the outgoing pulse that has come back. Next time backscattered light reaches the instrument it knows how much light that was still out in the atmosphere and how much of that, that has come back.

An example of the cloud structure measured by the MPL can be seen in Figure 7. When using data from the MPL, it should be noted that the signal to noise ratio is lower during the day than during the night due to a higher background signal, including scattered solar radiation and data therefore contains more noise during the summer months.

0 50 100 150 200 250 300 350

10 100 1000

Size (nm) Number density (cm-3 )

0 50 100 150 200 250 300 350

10 100 1000

Size (nm) Number density (cm-3)

(17)

10 (27)

Figure 7.MPL measurement of cloud and aerosol

backscatter, Ny-Ålesund, Svalbard, October1, 2003.

2.3. DATA PREPARATION

The aerosol measurements contain data over the particle size distribution in 16 size classes between 0.02 and 0.63µm. The first eight of these (0.02 – 0.1µm) represents Aitken mode particles, and the other eight (0.1 – 0.63µm) represents Accumulation mode particles. From the size distribution, hourly averages of total number, area and volume densities were calculated as well as total length of the particles (as if particles where arranged in one row) and effective diameter. Effective diameter is essentially the total volume divided by the total surface area, see Appendix A for more information about how these variables are calculated.

From the lidar backscatter data, three variables were educed; average backscatter from the boundary layer (0 – 2km), average backscatter at 474m altitude and average backscatter at 948m altitude. The boundary layer is interesting because it is where most clouds are formed and where aerosol measurements are made (aerosol – cloud interactions are studied). The lower height represents the height where the aerosol measurements were made, i.e. the Zeppelin station. The upper height was chosen to still be in the boundary layer, not too far away from the measuring height but still far away enough that it is possible for the two heights to show different conditions in terms of cloudiness. If the two levels would be too far apart, the aerosols and clouds at the different levels might have little to do with each other. The set-up when cloud free on the lower level and cloudy on the upper level is analogues to aircraft measurements. During air borne measurements, one often starts by flying under a cloud cover to characterize the aerosol, then flying through the cloud cover to characterize the cloud. In this study it is assumed that the aerosol below the cloud is the same as the one that participate in the cloud formation.

The distance between the in-situ observations and the lidar site is about 2km. In order to smooth small-scale variability in cloud and aerosol properties, one-hour averages were calculated for all parameters. A total of 17544 data points was achieved during the two years investigated (2004 was a leap year).

(18)

11 (27)

The two datasets, one for lidar and one for aerosol data, had to be organized to match each other in time. Only data with concurrent observations was used in the analysis. No difference between the two years in question was made, which means that all data from January were mixed to represent January.

3. ANALYSIS

3.1. DISTINCTION BETWEEN CLEAR AND CLOUDY CONDITIONS

There is always a certain amount of backscatter from the atmosphere, but the interest here lies on clouds especially and it needs to be defined what is a cloudy data point and what is not. This was done by sorting the average backscatter values and identifying significant changes in the data series, see Figure 8. At a backscatter value of about 0.1 a sudden increase can be seen, which is interpreted as the threshold where significant amount of cloudiness is observed in the 2km layer.

Backscatter values below 0.1 were treated as cloud free, and backscatter values above 0.1 were treated as cloudy. Variations in backscatter can come from variable amount of cloud water but also from variations in the size of the hydrometeors, i.e. cloud droplets. In this study more cloudiness is referred to as more backscatter.

Occasionally the aerosol at the Zeppelin station is affected by clouds to a degree that leaves very few aerosols in the ambient air. To find these situations the integral number density was treated as the backscatter values, see Figure 9. A distinct reduction in the number density can be seen around 20cm^-3, why data points with lower number densities were sorted out.

Figure 8. Average backscatter values in the altitude range 0 – 2km.

Figure 9. Aerosol number density (cm^-3).

3.2. CORRELATION BETWEEN MONTHLY AVERAGE BACKSCATTER, 0 – 2 KM AND MONTHLY AVERAGE PARTICLE PROPERTIES

Particle number densities were plotted for both cloudy and non-cloudy conditions for particles larger than 20nm in diameter. These plots are shown in Figure 10a and b. The figure shows that summer and spring have the largest amount of particles, as mentioned above in the introduction.

Note also that the two figures are very similar, which indicates that on average the total number of particles are not very much influenced by grouping the data in cloudy and non-cloudy data.

0.01 0.1 1 10 100 1000 10000

0 5000 10000 15000 20000

Datapoints

Average number of aerosols

0.01 0.1 1 10

0 5000 10000 15000 20000

Datapoints

Average backscatter

(19)

12 (27)

Figure 10a. Monthly average particle number densities during cloudy conditions.

Figure 10b. Monthly average particle number densities during non-cloudy conditions.

For each month average values of particle number density, particle area, particle volume, particle length and effective diameter was calculated. The results are shown in Table 1a and b.

Table 1a. Monthly average values of particle number density, area, volume, length and effective diameter for cloudy conditions.

Number density

(cm^-3)

Particle area 10⁷(nm²)

Particle volume 10¹⁰ (nm³)

Particle length 10⁵(nm)

Effective diameter

(nm)

January 134.4 3.9 1.1 1.9 234.6

February 120.0 4.3 1.2 1.9 259.5

March 239.7 6.3 1.6 3.3 233.7

April 176.9 3.9 0.9 2.2 234.2

May 185.6 3.3 0.8 1.9 238.3

June 189.3 2.0 0.4 1.4 217.6

July 277.5 3.4 0.7 2.4 199.0

August 141.0 1.3 0.3 1.0 186.3

September 82.3 1.3 0.3 0.8 212.4

October 50.2 1.2 0.3 0.6 229.1

November 86.8 2.8 0.8 1.3 261.6

December 90.9 3.0 0.8 1.4 262.3

0 50 100 150 200 250 300

Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Number density (cm-3 )

0 50 100 150 200 250 300

Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

(20)

13 (27)

Table 1b. Monthly average values of particle number density, area. Volume, length and effective diameter for non-cloudy conditions.

Number density

(cm^-3)

Particle area 10⁷(nm²)

Particle volume 10¹⁰ (nm³)

Particle length 10⁵(nm)

Effective diameter

(nm)

January 197.2 6.4 1.8 3.0 243.8

February 180.4 7.2 2.1 3.1 279.7

March 273.0 7.3 1.9 3.7 238.2

April 188.9 4.4 1.1 2.4 241.9

May ^191.9 ^3.4 ^0.9 ^1.9 ^238.1

June ^153.1 ^1.7 ^0.4 ^1.2 ^223.9

July 328.7 3.7 0.8 2.7 191.1

August 192.3 1.4 0.3 1.2 188.4

September 72.1 1.2 0.3 0.7 225.7

October 57.3 1.1 0.3 0.7 220.4

November 93.5 3.0 0.8 1.5 258.2

December ^108.9 ^3.4 ^0.9 ^1.6 ^260.5

Monthly means of the variables in Table 1a were correlated with average backscatter for the altitude range 0-2km. Backscatter values above 0.1 were used. Correlation coefficients were calculated by equation 1 below,

( )

y x y

x

Y X Cov

σ ρ σ

= ⋅ ,

(1)

where ρ is the correlation coefficient, σ the standard deviation and Cov stands for covariance.

The Correlation coefficient has a value between –1 and 1 and shows the association between two variables. Confidence level is not calculated in this case, and will not be in following correlation calculations. In this study, focus lie on data trends and not specific correlation values. Also, to make calculations, data is approximated to be of Gaussian shape, which is not the case for aerosols.

Correlation coefficients obtained for correlation between monthly means of particle parameters and backscatter for 0-2km are presented in Table 2.

Table 2. Correlation coefficients between backscatter values above 0.1 and particle variables for cloudy conditions.

Variable Correlation coefficient

Number density 0.19

Particle area -0.59

Particle volume -0.65

Particle length -0.36

Effective diameter -0.68

(21)

14 (27)

In Table 2, only number density show positive correlation. Number density is also the variable that shows the lowest correlation. The highest magnitude of correlation is shown in effective diameter, but the value is negative.

Additionally to the correlation coefficient above, correlation coefficients for each month were calculated. The resulting correlation coefficients are presented in Table 3.

Table 3. Correlation coefficients for particle variables, cloudy conditions.

Number density

(cm^-3)

Particle area (nm²)

Particle volume (nm³)

Particle length

(nm)

Effective diameter

(nm)

January -0.150 -0.181 -0.197 -0.165 -0.160

February 0.069 0.119 0.121 0.109 0.024

March -0.082 -0.091 -0.076 -0.098 -0.092

April -0.016 -0.001 0.006 -0.010 -0.300

May 0.123 -0.002 -0.022 0.051 -0.116

June 0.055 0.017 0.019 0.030 -0.044

July -0.039 -0.078 -0.078 -0.071 -0.050

August -0.072 -0.060 -0.070 -0.062 -0.073

September 0.015 -0.042 -0.039 -0.027 -0.157

October 0.007 0.020 0.020 0.017 -0.069

November -0.146 -0.145 -0.144 -0.143 -0.031

December -0.171 -0.171 -0.176 -0.171 -0.141

Of the parameters presented in Table 3 none show high correlation. The effective diameter value for April shows the highest correlation, even if it is negative (-0.3). April also shows the lowest correlation value, –0.001 for particle area.

The calculations were repeated for backscatter values less than 0.1. The following correlation coefficients were obtained for the monthly mean variables:

Table 4. Correlation coefficients between backscatter values below 0.1 and particle variables.

Variable Correlation coefficient

Number density 0.34

Particle area -0.36

Particle volume -0.39

Particle length -0.19

Effective diameter -0.66

Table 4 show coefficients with the same signs as Table 2 do, but the magnitude is a bit different.

Here, number density show a value in the same size range as the other parameters. However, effective diameter still shows a high value of correlation. The lowest correlation can be seen in particle length.

(22)

15 (27)

Correlation parameters for every month are presented in Table 5.

January number density for the non-cloudy condition shows the largest correlation but no coefficients are high. The lowest correlation can be seen in December particle length.

Table 5. Correlation coefficients for particle variables, non-cloudy condition.

Number density

(cm^-3)

Particle area (nm²)

Particle volume (nm³)

Particle length

(nm)

Effective diameter

(nm)

January 0,184 0,159 0,151 0,170 0,117

February -0,103 -0,100 -0,099 -0,104 0,036

March -0,201 -0,065 -0,019 -0,131 0,101

April 0,045 -0,079 -0,085 -0,048 -0,077

May -0,105 -0,093 -0,092 -0,103 -0,062

June 0,034 -0,068 -0,091 -0,023 -0,134

July -0,061 0,057 0,064 0,024 0,096

August -0,022 0,061 0,086 0,008 0,168

September 0,106 0,097 0,086 0,113 -0,013

October -0,055 0,007 0,021 -0,018 0,100

November 0,037 0,024 0,043 0,019 0,050

December -0,027 0,027 0,048 0,002 0,109

None of the correlations are significant and the magnitude of correlation may be achieved by random numbers.

Note that even if it is referred to cloud free conditions, there is still variability in the backscatter signal below 0.1.

The analysis is enlarged by treating each size bin in the size distribution as an independent variable and correlate with backscatter. The resulting correlations over the year as function of particle size are plotted in Figures 11a and 11b for cloud free and cloudy conditions, respectively.

(23)

16 (27)

Figure 11. Correlation plots for average backscatter from 0 – 2km and aerosol size distribution. Monthly averages.

Figure a corresponds to the clear weather case, b the cloudy case.

A high positive correlation is seen for small aerosols, whereas larger aerosols and backscatter anti-correlate. This means that if the backscatter value increases, the number of small aerosols also increases, whereas the number of large aerosols decreases. The result educed is very similar for both cloudy and non-cloudy conditions. This indicates that no clear border between cloudy and clear weather can be made in the way that has been done in this study and that the data appear as a continuum.

The data presented in Figure 11 was averages for each month and the correlation was calculated over the whole year. The analysis was enlarged such that the correlations are performed for each month using hourly averages. The results are presented in Figure 12. The correlation drops drastically and each individual point is not significantly different from correlation between random values. However, the transition between different sizes is smooth which suggests that the data contains some information despite the low values. There is really no strong trend over the year. Some months stand out, for example March and May that shows the highest values of correlation. This could be result of Arctic haze events. Note that April shows very low values of correlation. This could be an effect of the episodic-ness of Arctic haze. For more thorough analyse, more data would be necessary. Some months does not contain data enough for reliable results.

a.

-1 -0,5 0 0,5 1

10 100 1000

Size (nm)

Correlation

b.

-1 -0,5 0 0,5 1

10 100 1000

Size (nm)

Correlation

(24)

17 (27)

Figure 12. Correlation between particle size classes and average backscatter for every month during cloudy conditions, arranged in the four seasons of the year, (a) corresponds to the winter season, (b) spring, (c) summer

and (d) autumn.

3.3. AVERAGE BACKSCATTER, 474M AND 948M

From the backscatter data two variables were educed, representing two slices in the atmosphere at 474m and 948m. Data for four different combinations were compiled: clouds present at both heights, no clouds present, clouds present only at the lower level and clouds present only at the higher level. Although the analysis above show that it might be difficult to separate between cloudy and non-cloudy data with respect to the correlations, the backscatter value 0.1 is again used to separate between the two. A total of 12199 data points were available, leading to 1242 points for 474m clear – 948m cloudy, 4807 for 474m cloudy – 948m cloudy, 3721 for 474m clear – 948m clear and 2429 points for 474m cloudy – 948m clear case. Monthly means of aerosol data were correlated with backscatter values for both levels. Data is represented in Figure 13. Each month was approximated to have 30 days. Correlation was also calculated for backscatter values and number of particles for each size class. The result is presented in Figure 14.

a.

-0,35 -0,25 -0,15 -0,05 0,05 0,15 0,25 0,35

10 100 1000

Size (nm)

Correlation

December January February

c.

-0,35 -0,25 -0,15 -0,05 0,05 0,15 0,25 0,35

10 100 1000

Size (nm)

Correlation

June July August

b.

-0,35 -0,25 -0,15 -0,05 0,05 0,15 0,25 0,35

10 100 1000

Size (nm)

Correlation

September October November

d.

-0,35 -0,25 -0,15 -0,05 0,05 0,15 0,25 0,35

10 100 1000

Size (nm)

Correlation

March April May

(25)

18 (27)

Figure 13. Correlation between monthly averages of aerosol size and average backscatter from the heights, 474m and 948m. (a) Corresponds to the 474m – clear, 948m – cloud case. (b) 474m – cloud, 948m – cloud.

(c) 474m – clear, 948m – clear. (d) 474m – cloud, 948m – clear.

a.

-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1

10 100 1000

Size (nm) Correlatio n

474 948

b.

-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1

10 100 1000

474 948

c.

-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1

10 100 1000

474 948

d.

-1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1

10 100 1000

474 948

(26)

19 (27)

Figure 14. Correlation between aerosol size and average backscatter from the heights, 474m and 948m.

(a) Corresponds to the 474m – clear, 948m – cloud case. (b) 474m – cloud, 948m – cloud.

(c) 474m – clear, 948m – clear. (d) 474m – cloud, 948m – clear.

The same trends are seen in the yearly dataset as in the monthly averaged dataset, but the monthly averages show slightly higher values of positive and negative correlation. The focus will lie on the monthly average dataset (Figure 13), to be able to compare with the result from the average backscatter from 0 – 2km dataset where monthly averages were used.

3.3.1. 474M – CLEAR, 948M – CLOUDY

The curve for clear conditions (474m) shows, as the curve for clear conditions at 0 – 2km, correlation for small particles and anti correlation for the larger ones. The curve for the cloudy region however, shows a curve with small correlation values with a minimum for sizes a little larger than 100nm diameter, which indicates that changes in backscatter i.e. cloudiness, and aerosol particles present little systematic interaction.

c.

-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5

10 100 1000

Size (nm)

Correlation ⁴⁷⁴

948 a.

-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5

10 100 1000

Size (nm)

Correlation ⁹⁴⁸

474

b.

-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5

10 100 1000

Size (nm)

948

d.

-0,5 -0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5

10 100 1000

Size (nm)

948

(27)

20 (27) 3.3.2. 474M – CLOUDY, 948M – CLOUDY

The plot for cloudy conditions shows negative correlations for all particle sizes. No real trends can be seen, but the lower height starts at a moderate level, become higher and then decreases again. The 948m curve starts at a low level and increases slightly with aerosol size. This is the only case for which the 474m curve does not show positive correlation for Aitken mode particles and negative correlation for Accumulation mode particles. In comparison to the 0-2km case this plot show very different result. However, it should be noted that this is the case where the lidar may have difficulties obtaining data, regarding that the signal must travel through one cloud cover to be scattered in a second one.

3.3.3. 474M – CLEAR, 948M - CLEAR

Plots over clear conditions on both heights show a similar curve as the correlation plot over size and backscatter from 0 – 2km. Small particles correlate positively with backscatter whereas large particles correlate negatively with backscatter. The curve for the upper height shows slightly higher correlation for small particles than the lower height. For large particles there is no difference between the two. If one compares this plot with the curve for clear conditions for backscatter values for 0 – 2km, it shows slightly lower correlation values for small particles but higher values for large particles.

3.3.4. 474M – CLOUDY, 948M – CLEAR

The curve of correlation between size and backscatter for cloudy conditions at the lower height but clear conditions at the higher height shows a very different result. The curve for 474m shows high correlation for small particles and high anti correlation for large particles. The 948m curve, however, shows the complete opposite feature, anti-correlation for small aerosols and correlation for large aerosols.

3.4. AEROSOL NUMBER DISTRIBUTION

To further analyse the results given for the average backscatter case for 0 – 2km and the average backscatter case at 474m and 948m, the logarithm of the average backscatter was divided into five intervals; -1.5 to -1, -1 to -0.5, -0.5 to 0, 0 to 0.5 and 0.5 to 1. For each interval, mean, median and quartiles of the number of aerosols were calculated and plotted. The plot over the upper quartile for average backscatter from 474m altitude is shown in Figure 15.

Figure 15. Aerosol upper quartile for each size class with backscatter intervals ranging from -1.5 to 0.5.

0 50 100 150 200 250 300

0 5 10 15 20

Size class

-1.5 -1 -0.5 0 0.5

(28)

21 (27)

Figure 15 shows the same result as the figure for size - backscatter correlation for the 0 - 2km backscatter case seen in Figure 10 and the non-cloudy case for backscatter from 474m and 948m seen in Figure 13c. They all show that for high backscatter values small aerosols dominate, whereas large aerosols dominate for low backscatter values, i.e. Aitken mode particles correlate with backscatter and Accumulation mode particles anti correlate with backscatter.

4. SUMMARY AND DISCUSSION

Data from the years 2003 and 2004 have been analyzed in order to study aerosol - cloud interactions in the Arctic boundary layer. Aerosol particle data and lidar backscatter measurements have been arranged and statistically analyzed.

Correlation values between lidar data and particle properties such as total concentration, area, volume, length and effective diameter are in general small, but a systematic difference between different particle sizes can be detected. Particle sizes smaller than about 100nm present positive values whereas particles larger than 100nm present negative values. Observations of processes of cloud activation and scavenging of CCN explain the negative correlation for larger particles. The positive correlation values for the smaller particles however, can be explained in two ways:

1. One, the number density of Aitken and Accumulation mode particles covariates. This means that in cloud-free air both small and large particles increase and decrease at the same time. If the larger particles control cloud brightness, according to the Twomey effect, but at the same time are incorporated in cloud drops as they act as CCN and removed, then an increase in small particles would correlate with backscatter values.

2. Two, new Aitken mode particles are formed in or in the vicinity of clouds.

In this study no correlation between the two aerosol modes (small and large particles) were found, which means that small and large particle variety do not necessarily have anything to do with each other. It is therefore believed that new particles form in connection to clouds.

It has been discussed that clouds may promote particle formation, but the exact nature of this process is not yet known. High relative humidity and enhanced actinic radiation flux near clouds are factors that would promote new particle formation. Garrett et al. 2002b discuss the topic of high amounts of Aitken mode particles over clouds in the Arctic. The study shows that during springtime high humidity above cloud tops often result in high numbers of Aitken mode particles.

Additionally, Clarke et al., 1999, reported about enhanced particle formation for warm clouds in marine environments.

During most of the year, the arctic air contains few CCN. Formation of boundary layer clouds then gives few, but rather big cloud droplets and the probability that drizzle will form is high.

Precipitating clouds will scavenge the CCN particles as the data used in this study indicates. The negative correlation observed for accumulation mode particles suggests that this process dominates the signal compared to the Twomey effect. A change in particle number density (i.e.

CCN) in the Arctic might then have a larger impact on the precipitation formation and the hydrological cycle rather than the brightness of clouds. According to Garrett et al., 2004, Arctic Stratus are particularly sensitive to haze aerosol concentrations, and cloud – aerosol effects on climate will therefore be strong over the Arctic.

(29)

22 (27)

The relations between the two different altitude datasets (Figures 13a-d and 14a-d) give mixed messages. However, the fact that a smooth relation between particle sizes and backscatter exists tells us that aerosols and clouds are coupled in the vertical.

For further research about the vertical coupling between aerosols and clouds, aircraft or balloon measurements can be used. Additionally, for further studies, records of weather, humidity and temperature can be useful. Another important improvement of this study is cloud observations.

The approximation made in this study, that backscatter values above 0.1 means that clouds are present might not give an accurate result. Furthermore, hourly means of cloud and backscatter are used. In one hour, weather and cloud conditions can change rapidly, giving a non-accurate value for the hour.

Even if the working hypothesis of this study, that more aerosols cause brighter clouds was not evident, other interesting results were obtained. It appears as if the effect on aerosols by clouds generates a stronger signal than the opposite relation that aerosols affect the clouds. This is important information in connection to the aerosol life cycle in the Arctic and the aerosol-cloud- climate interaction.

(30)

23 (27)

ACKNOWLEDGEMENTS

I would like to thank Masataka Shiobara at the National Institute of Polar Research, Japan and Masanori Yabuki at the Center for Environmental Remote Sensing, Chiba University, Japan for letting me use the lidar backscatter data retrieved from Ny-Ålesund during the years 2003 and 2004.

(31)

24 (27)

(32)

25 (27)

REFERENCES

1. ACIA, Impacts of a Warming Arctic, Arctic Climate Impact Assessment, Cambridge University Press, 2004.

www.acia.uaf.edu

2. Barrie, L. A., Arctic air pollution: An overview of the current knowledge, Atmos.

Environ., 20, 643 – 663, 1986.

3. Chapman, W. L., Walsh, J. E., Recent variations of sea ice and air temperature in high latitudes, Bull. Am. Meteorol. Soc., 74, 33 – 47, 1993.

4. Clarke, A. D., Kapustin, V. N., Eisele, F. L., Weber, R. J., McMurry, P.H., Particle production near marine clouds: Sulfuric acid and predictions from classical binary nucleation, Geophys. Res. Lett., 26, 2425 – 2428, 1999.

5. Garrett, T. J. & Zhao, C., Dong, X., Mace, G. G., Hobbs, P. V., Effects of varying aerosol regimes on low-level Arctic stratus, Geoph. Res. Lett., 31, L17105, 2004.

6. Garrett, T. J., Radke, L. F. & Hobbs, P. V., Aerosol Effects on Cloud Emissivity and Surface Longwave Heating in the Arctic, J. Atmos. Science 59, 769 – 778, 2002a

7. Garrett, T.J., & Hobbs, P.V., Radke, L.F., High Aitken nucleus concentrations above cloud tops in the Arctic, Notes & Correspondence, 779 – 783, 2002b.

8. Lejenäs, H., METEOROLOGI – En orientering om vädret och meteorologiska skeenden i lufthavet, Meteorologiska institutionen, Stockholms universitet, 1984, upplaga 9, 1997.

9. NOSA Aerosol Symposium, Göteborg 3-4/11 2005.

10. Radke, L. F., Brock, C. A., Lyons, J. H., Hobbs, P. V., Aerosol and lidar measurements of hazes in mid-latitude and polar airmasses, Atmos. Environ., 23, 2417-2430, 1989.

11. Rothrock, D. A., Yu, Y., Maykut, G. A., Thinning of the Arctic sea – ice cover, Geophys.

Res. Lett., 26, 3469 – 3472, 1999.

12. Sirois, A., Barrie, L. A., Arctic lower tropospheric aerosol trends and composition at Alert, Canada: 1980 – 1995, J. Geophys. Res., 104, 11599 – 11618, 1999.

13. Stone, R. S., Variations in western arctic temperatures in response to cloud radiative and synoptic-scale influences, J. Geophys. Res., 102, 21769 – 21776, 1997.

(33)

26 (27)

14. Ström, J., J. Umegård, K. Tøseth, P. Turnved, H. C. Hansson, K. Holmén, V. Wisman, A.

Herber, and G. König-Langlo, One year observation of particle size distribution and aerosol chemical composition at the Zeppelin Station, Svalbard, March 2000-March 2001, J. Phys. Chem. Earth 28, 1181-1190, 2003.

15. Twomey, S., Aerosols, Clouds and Radiation, Atmos. Env.25A, 2435 – 2442, 1991.

16. Twomey, S., The influence of pollution on the shortwave albedo of clouds, J. Atmos. Sci., 34, 1149 – 1152, 1977.

(34)

27 (27)

APPENDIX A

AEROSOL AREA (A)

The aerosol area is calculated by the following equation:

A =

∑

particlenumberof sizeclass∗averagesizeof sizeclass² AEROSOL VOLUME (V)

The aerosol volume is calculated by the following equation:

∑

^∗

= particlenumberof sizeclass averagesizeof sizeclass³ V

AEROSOL LENGTH (L)

The aerosol length is calculated by the following equation:

∑

^∗

= particle numberof sizeclass averagesizeof sizeclass L

AEROSOL EFFECTIVE DIAMETER (EffD)

The aerosol effective diameter is calculated by the following equation:

A EffD =V

(35)

28 (27)

Aerosol - Cloud Interactions in the Arctic Boundary Layer

MASTER’S THESIS

KARIN RYBERG

Aerosol - Cloud Interactions in the Arctic Boundary Layer

Ny - Ålesund, Svalbard

Master’s thesis

AEROSOL – CLOUD INTERACTIONS IN THE ARCTIC BOUNDARY LAYER

NY-ÅLESUND, SVALBARD

February 2006

KARIN RYBERG

Master of Science programme Space Engineering

Luleå University of Technology Department of Space Science

Stockholm University

Department of Applied Environmental Science

Air Pollution Laboratory

PREFACE

ABSTRACT

TABLE OF CONTENTS

1. INTRODUCTION

1.1. AEROSOLS

1.2. CLOUD FORMATION AND ARCTIC CLOUDS

1.3. THE ARCTIC

1.4. AEROSOL – CLOUD INTERACTIONS

1.5. AEROSOL – CLOUD – CLIMATE INTERACTIONS

2. MEASUREMENTS AND METHOD

2.1. AEROSOL MEASUREMENTS

2.2. LIDAR MEASUREMENTS

2.3. DATA PREPARATION

3. ANALYSIS

3.1. DISTINCTION BETWEEN CLEAR AND CLOUDY CONDITIONS

3.2. CORRELATION BETWEEN MONTHLY AVERAGE BACKSCATTER, 0 – 2 KM AND MONTHLY AVERAGE PARTICLE PROPERTIES

( )

3.3. AVERAGE BACKSCATTER, 474M AND 948M

3.4. AEROSOL NUMBER DISTRIBUTION

4. SUMMARY AND DISCUSSION

ACKNOWLEDGEMENTS

REFERENCES

APPENDIX A

∑

∑

∑