doctor of philosophy היפוסוליפל רוטקוד · cloud fields; one that will be resilient...

102
Thesis for the degree Doctor of Philosophy By Rotem Bar-Or Advisor: Prof. Ilan Koren July, 2012 Submitted to the Scientific Council of the Weizmann Institute of Science Rehovot, Israel השפעתם של אירוסולים על התכונות האופטיות והמיקרו פיסיקליות של שדה ענניםThe effect of aerosols on the optical and microphysical properties of a cloud field עבודת גמר( תזה) לתואר דוקטור לפילוסופיה מאת רותם בר- אור תמוז, התשע" ב מוגשת למועצה המדעית של מכון ויצמן למדע רחובות, ישראל מנח ה: פרופ' אילן קורן

Upload: others

Post on 02-Sep-2019

1 views

Category:

Documents


0 download

TRANSCRIPT

Thesis for the degree

Doctor of Philosophy

By

Rotem Bar-Or

Name in English

Advisor: Prof. Ilan Koren

July, 2012

Submitted to the Scientific Council of the

Weizmann Institute of Science

Rehovot, Israel

השפעתם של אירוסולים על התכונות האופטיות

והמיקרו פיסיקליות של שדה עננים

The effect of aerosols on the optical and

microphysical properties of a cloud field

לתואר (תזה)עבודת גמר

דוקטור לפילוסופיה

מאת

אור-רותם בר

שם בעברית

ב"התשע, תמוז

מוגשת למועצה המדעית של

מכון ויצמן למדע

ישראל, רחובות

אילן קורן' פרופ :המנח

List of abbreviations

AOD Aerosol optical depth

Cb Cumulonimbus cloud

CCN Cloud condensation nuclei

CF Cloud fraction

CFF Cloud field fraction

Ci Cirrus cloud

Cu Cumulus cloud

DC Deep convective cloud

FMF Aerosol fine-mode fraction

GCM Global circulation model

IC Ice content

IGRA Integrated global radiosonde archive

IN Ice nuclei

LCL Lifted condensation level

LCT Lower cloudy troposphere

LES Large eddy simulation

LWC Liquid water content

MBL Marine boundary layer

MBP Moist bounding parameter

MODIS Moderate resolution imaging spectroradiometer

NASA National aeronautics and space administration

NOAA National oceanic and atmospheric administration

RH Relative humidity

SALR Saturated adiabatic lapse rate

Sc Stratocumulus cloud

SRH Sub-saturated relative humidity

WMO World meteorological organization

Abstract

Clouds and aerosols play key roles in the global climate system by altering Earth‟s energy

budget and by controlling the hydrological cycle. Both numerical simulations and observations are

used for studying processes related to clouds, aerosol and the interactions in between. A major

difficulty in observational analyses is the separation in space between clouds and aerosol, as the

border between these is not well defined and often not practically indictable. Such problem begs for

a different approach that will take into account spatial and spectral features of whole cloud fields,

including both clouds and aerosols. Thus, this study offers a new classification of the observed

atmosphere into cloud-fields (including clouds and their surrounding influenced zone) and cloud-

free.

For this purpose, a new morphological method for spatially determining cloud field

boundaries is presented and validated. Using this method, extensive analysis reveals global cloud

field coverage of 88%, while the corresponding cloud fraction is only 51%. It shows a strong

latitudinal dependence, pointing to a very small probability to sample a cloud field free atmosphere

in most regions over the globe. Furthermore, we find that cloud fields extend on average ~30 km

from cloud edges, as supported by independent studies.

Close examination of MODIS aerosol retrievals shows a clear trend linking the retrieved

aerosol properties with their distance from clouds, which is commonly believed to be mainly a

result of aerosol humidification. We study the optical significance of aerosol humidification using

large eddy simulations of cloud fields, for developing a parameterization of the relative humidity

(RH) as a function of the distance from clouds. Following this parameterization, atmospheric

radiative transfer simulations show that the optical significance of aerosol humidification in cloud

fields is limited to the area within 500 m from cloud edge, with a strong sensitivity of the retrieval

bias near clouds to the aerosol chemical and physical properties.

Finally, for studying the distribution of RH within cloud fields from measurements, we

develop a method using atmospheric sounding data for characterizing the sub-saturated RH vertical

profiles in cloudy atmosphere. Such method allows a calculation of the potential for aerosol

humidification directly from the best available RH measurements. It enables an estimation of the

impact of aerosol humidification on aerosol-cloud interaction analyses. In this study we show that

aerosol humidification can explain only a small fraction of the measured trends between aerosol and

convective clouds properties.

תקציר

א ובשל "לעננים ולאירוסולים ישנה השפעה מכרעת על האקלים בשל השפעתם על מאזן האנרגיה של כדה

, הן סימולציות נומריות והן תצפיות משמשות לחקר תהליכים הקשורים לעננים. תפקידם במחזור ההידרולוגי

הצורך להבחין בין עננים לאירוסולים מהווה כיום קושי עיקרי בניתוח . הגומלין ביניהם-לאירוסולים וליחסי

קושי זה דורש גישה . מכיוון שהגבול ביניהם לא מוגדר כהלכה ולעתים קרובות אף בלתי ניתן להערכה, תצפיות

. הכוללים עננים לצד אירוסולים, עננים שלמים-שונה שתתחשב במאפיינים המרחביים והספקטראליים של שדות

הכוללים את העננים עצמם )עננים -לשדות: במחקר זה אנו מציעים סיווג חדש של האטמוספרה הנצפית, ואכן

.ולאטמוספרה נקייה מעננים (אשר מושפע מנוכחותם, ואת האזור סביבם

שימוש בשיטה זו . מוצגת כאן שיטה מורפולוגית הקובעת את גבולותיהם המרחביים של שדות עננים, לצורך זה

-א מכוסים בשדות" משטח כדה88%-כ , בלבד51%בניתוח נתונים נרחב מגלה כי כאשר כיסוי העננים עומד על

אבחנה זו מראה כי ההסתברות . העננים בקו הרוחב-מאובחנת תלות ברורה של אחוז כיסוי שדות, בנוסף. עננים

-ניתוח התצפיות מראה ששדות, כן-כמו. ברוב העולם, לדגימת אטמוספרה נקייה משדות עננים הינה נמוכה מאד

.תלויים-כפי שעולה גם ממחקרים אחרים בלתי, מ מגבולות העננים" ק30עננים מתפרשים בממוצע למרחק של

מעלה קשר ברור בין תכונות האירוסולים המוערכות לבין MODISעיון מדוקדק בתצפיות אירוסולים מלווין

. קשר זה מיוחס ברובו לגידול האירוסולים עקב ספיחת מים בסביבה לחה, כיום. מרחק הדגימה מהענן הקרוב

תוך , עננים-אנו בודקים עד כמה משמעותית התרומה האופטית של גידול האירוסולים הלחים בשדות, במחקר זה

מנת לפתח פרמטריזציה של -על, (large eddy simulations)שימוש בסימולציות נומריות המדמות שדות עננים

המתבססות על , הדמיות של מעברי קרינה באטמוספרה. היחסית כתלות במרחק מהענן הקרוב-הלחות

המטרים 500-עננים מוגבלת ל-מראות שהתרומה האופטית של גידול אירוסולים לחים בשדות, פרמטריזציה זו

הדמיות אלה מראות בנוסף כי סטיית מדידות האירוסולים בקרבת הענן רגישה מאד . הקרובים ביותר לענן

.לתכונות הכימיות והפיסיקליות של האירוסולים אותם מודדים

פיתחנו שיטה המבוססת על , עננים-היחסית בשדות-בכדי להעריך את המאפיינים של התפלגות הלחות, לבסוף

. עננים-בשדות (מחוץ לענן)רוויה -היחסית בתת-אשר מחשבת את הפרופיל האנכי של הלחות, מדידות רדיוסונדה

ולהעריך את ההשפעות , שיטה זו מאפשרת לחשב את הפוטנציאל האופטי הגלום בגידול אירוסולים לחים

היחסית הישירות והטובות ביותר אשר -תוך שימוש במדידות הלחות, אירוסולים-האפשריות על חקר יחסי עננים

.זמינות היום

במחקר זה אנו מראים שגידול אירוסולים לחים יכול לספק הסבר רק לחלק קטן יחסית של המגמות הנמדדות

.בין אירוסולים ובין תכונותיהם של עננים קונבקטיביים

Table of Contents

1 Introduction ............................................................................................................ 1

1.1 Research objectives ............................................................................................ 1

1.2 Background ........................................................................................................ 2

1.3 Clouds ................................................................................................................ 4

1.3.1 Detection of clouds and cloud fields ........................................................... 5

1.3.2 Aerosol effects on clouds ............................................................................ 6

1.4 The twilight zone.............................................................................................. 10

1.5 Aerosol hygroscopic growth in cloud fields .................................................... 11

1.5.1 Humidification effects on aerosol size distribution .................................. 11

1.5.2 The relative humidity in cloud fields ........................................................ 14

2 Methods and data .................................................................................. 17

2.1 Cloud field masking algorithm ........................................................................ 17

2.1.1 Cloud field masking algorithm - description ............................................ 20

2.1.2 Cloud field masking algorithm - sensitivity study .................................... 22

2.2 Global cloud field coverage ............................................................................. 23

2.3 Aerosol retrieval inside cloud fields ................................................................ 24

2.4 Radiative effects of aerosol humidification on cloud fields............................. 25

2.4.1 Parameterization of the RH spatial distribution in cloud fields ................ 25

2.4.2 Atmospheric radiative transfer simulations of humidified aerosols ......... 26

2.5 Characterizing the RH in cloud fields .............................................................. 29

2.5.1 Locating the cloudy layer with radiosonde data ....................................... 29

2.5.2 Characterizing the SRH values in cloud fields ......................................... 32

2.5.3 The lower cloudy troposphere SRH and cloud development ................... 33

3 Results ............................................................................................................ 34

3.1 Global cloud field coverage ............................................................................. 34

3.2 Aerosol retrieved optical properties in cloud fields ......................................... 37

3.3 Radiative effects of aerosol humidification on cloud fields............................. 40

3.3.1 RH spatial distribution in cloud fields ...................................................... 40

3.3.2 Humidified aerosol properties in cloudy environment ............................. 42

3.3.3 Absorbing humidified aerosol properties in cloud fields .......................... 48

3.4 Upper-air measurements of the RH spatial distribution in cloud fields ........... 53

3.4.1 RH mean vertical profile in potentially cloudy layers .............................. 53

3.4.2 SRH profile and cloud development ......................................................... 56

4 Summary and discussion ...................................................................... 59

Tables ............................................................................................................ 64

5 References ............................................................................................................ 70

6 List of publications ............................................................................... 79

7 Declaration ............................................................................................................ 80

Appendix A ............................................................................................................ 81

Appendix B ............................................................................................................ 84

1

1 Introduction

1.1 Research objectives

The primary aim of this study is to better estimate (qualitatively and quantitatively)

the radiative effects in cloud fields, with the consideration of suspended aerosols.

Within this framework, the specific objectives are:

1.1.1 Presentation of an objective metric that defines the extent and coverage of

cloud fields; one that will be resilient to differences in the exact definition of

what is a cloud.

1.1.2 Development of a robust cloud field masking algorithm which can accurately

determine the cloud field boundaries, including the detectable clouds and the

transition zone between clouds and cloud free atmosphere (i.e. the twilight

zone).

1.1.3 Observation of the global cloud field coverage above land and oceans, with its

latitudinal dependence.

1.1.4 Assessment of the trends between remotely sensed aerosol retrievals and the

distance from clouds.

1.1.5 Estimation of the aerosol humidification contribution to the apparent aerosol

radiative features in cloud fields.

1.1.6 Acquiring better understanding of the spatial distribution of relative humidity

(RH) in cloud fields.

1.1.7 Evaluation of the relative humidity vertical profiles in cloud fields, based on

long-term atmospheric sounding measurement record, and estimation of the

humidified aerosol retrieval biases in cloud fields.

2

1.2 Background

The Earth's climate is changing. It has changed in the past, as observed in ice-core

data [Jouzel et al., 2007], and it is currently changing, as shown in many continuous

measurements taken during the last two centuries [Forster et al., 2007]. The main

driver for these climatic variations, as known today, is the divergence from

equilibrium of the Earth's global energy budget [Trenberth et al., 2009]. The energy

budget summarizes the incoming solar radiation, Earth's outgoing thermal emission,

and the interactions of radiation with atmospheric and surface features according to

their radiative properties, as demonstrated in Figure ‎1.1. Any perturbation from the

delicate equilibrium (namely “radiative forcing”) may directly alter the global

climate.

Figure ‎1.1 The global annual mean Earth‟s energy budget for the March 2000 to May

2004 period (W ⋅ m−2). The broad arrows indicate the schematic flow of energy in

proportion to their importance (Fig 1. in Trenberth et al., 2009).

3

During the last decades, many studies have tried to properly estimate the total

radiative forcing, including assessment of the anthropogenic contribution compared to

the natural one, in order to find the causes for global climate changes. The total

forcing is the sum of several components (anthropogenic - greenhouse gases, jet

contrails, and natural - solar flux fluctuations, volcanic activity, ocean thermo-

dynamical variances, and more), of which the aerosol radiative effects (directly and

through impact on clouds) have one of the greatest uncertainty [Forster et al., 2007].

Hence, the potentially crucial climatic signature of aerosols has been extensively

explored for decades, in attempt to better estimate the overall atmospheric radiative

effects of aerosols; either directly by scattering or absorbing radiation [Kaufman et al.,

2002], or indirectly by their effects on cloud spatial and physical properties (described

in detail in section ‎1.3.2).

Studying the variety, the non-linearity, and the intricate nature of these indirect

feedbacks requires accurate and reliable measurements and observations, which adds

difficulties to the assessment of their overall effect.

The regions of interest for studying aerosol-cloud interactions are difficult for

examination; as they are naturally contain both aerosols and clouds which spatially

and spectrally overlap, and therefore are difficult for separation. Aerosol and cloud

observations, commonly suffer from retrieval accuracy and quality uncertainties,

which may disqualify conclusions.

In this study, the radiative properties of whole cloud fields are addressed. A first

attempt to spatially bound cloud fields based on their morphological features is

presented and validated (Section ‎2.1), enabling the estimation of the global cloud-

field coverage (Section ‎3.1). The aerosol retrieval biases in the vicinity of clouds are

observed using satellite data (Section ‎3.2), and the net contribution of aerosol

hygroscopic growth (humidification) to these biases is examined (Section ‎3.3). As a

part of the assessment of aerosol humidification effect, the spatial distribution of the

relative humidity (RH) near clouds is simulated and parameterized (Section ‎3.3).

Finally, the RH values in the lower troposphere are measured for cloud fields, using

vast radiosonde data sets, and studied in respect to cloud vertical development

(Section ‎3.4).

4

1.3 Clouds

Clouds, in Earth‟s atmosphere, are commonly defined as “a detectable mass of

water droplets or ice crystals suspended in the atmosphere”. The formation of clouds

depends on the atmospheric thermo-dynamical conditions as super-saturation is

required for enabling the formation and existence of liquid or solid water particles.

The phase transition of water vapor into water droplets or ice particles, within an air

parcel, mostly occurs due to a vertical lift of the air parcel that is sufficient to form

super-saturation due to expansion and cooling.

There are several mechanisms that may contribute to air lifting. Meteorological

conditions may drive dynamic vertical lift (e.g. frontal systems), any source of

differential surface heating initiates vertical lifting, and physical geographic obstacles

efficiently serve as vertical lifting agents (namely aerographic lifting).

The phase transition of water vapor into liquid water or solid ice particles

releases latent heat and increases the buoyancy of the air parcel, which then increases

the parcel‟s vertical velocity. The developing stage of the cloud ends when the cloud

starts loosing water (or ice) by evaporation and precipitation [Rogers, 1989]. When

decaying, the cloud exhibits the stronger precipitation intensity. The maximal cloud

top height, during it's lifetime, is determined by the atmospheric thermo-dynamical

conditions; where a dry or stable layer stops supporting vertical movement and

hydrometeors existence. The size distribution of the droplets/ice particles varies in

time due to growth by diffusion, evaporation (mainly in the cloud top and edges),

collision-coalescence processes, breakup and sedimentation.

Cloud processes span the physical scales between 10−6 and 105 m, from aerosol

size through processes in clouds‟ turbulent eddy scales through cloud fields and up to

synoptic scale of hundreds of kilometers. The temporal scale of cloud lifetime may

spread between a few minutes (e.g. for a small cumulus in relatively dry environment)

to several days, as observed in marine decks of stratiform clouds.

Past studies showed that clouds have fractal features [Cahalan and Joseph, 1989;

Cahalan et al., 1994; Marshak et al., 1995; Borde and Isaka, 1996; Lovejoy and

Schertzer, 2006], characterized by a power-law size distribution [Kuo et al., 1993;

5

Koren et al., 2008b]. Cloud fields, are clusters of clouds that form in the same

environmental conditions and due to that have similar physical properties. Cloud field

is usually a cloud system with spatial characteristic scale larger than the typical scale

of its clouds. Cloud fields can be characterized by their total size, their cloud types

and spatial distribution and by their development in time. Recent studies suggested

that some cloud fields can be described as complex systems, that their organization

pattern is determined by interactions between the clouds, and influenced by the

environmental conditions such as aerosol loading [Koren and Feingold, 2011].

Constantly covering ~60% of Earth's surface, clouds serve as a primer

component in the climate system. Their effects on the radiation budget depend on

their type. Clouds reflect solar radiation back to space, causing cooling of the

atmosphere, and absorb the Earth‟s IR radiation, causing warming of the atmosphere.

Low-level shallow clouds cool the surface by reflecting solar radiation back to space

and emitting thermal radiation in a temperature similar to the earth surface beneath

them. High altitude clouds (e.g. Cirrus clouds) release less thermal energy to space,

and cause heating of the atmosphere [Koren et al., 2010] (see Figure ‎1.1). Clouds are

the main engine of the water cycle. Any change in precipitation patterns and amounts

would influence significantly the hydrological cycle as it determines how much is

absorbed by the ground and how much will flow over it.

1.3.1 Detection of clouds and cloud fields

The growing usage of space-borne platforms for remote sensing instruments for

cloud and aerosol research during the last decades [Kaufman et al., 2002; King et al.,

2003] raised the need to develop accurate methods to retrieve cloud and aerosol

properties [Platnick et al., 2003; Remer et al., 2005; Levy et al., 2007]. In order to do

so, it is common to classify the imaged domain into two: detectable cloud and

"assumed to be" cloud-free.

The commonly used method for detecting clouds and providing cloud mask

maps, using remote sensing instruments, involves the location of reflectance

signatures of suspended liquid water drops or ice particles, followed by the usage of a

6

reflectance threshold in order to separate the cloudy atmosphere from cloud-free

[Ackerman et al., 1998; Platnick et al., 2003; Dybbroe et al., 2005; Luo et al., 2008].

Wide swath imagers may add spatial analysis information to improve the threshold

choice using multiple pixel retrievals in a given domain [Martins et al., 2002].

In situ observations can use the sampled atmosphere and estimate the amount of

suspended liquid water or ice, which enables the separation between cloudy and

cloud-free samples. For these observations, the main challenges are the instrument

capabilities and accuracy [Verver et al., 2006].

Cloud masks may also be calculated for cloud resolving numerical simulation

output, which provide the liquid water content (LWC) or ice content (IC) for every

grid cell. However, recent study has shown that even simulated apparent cloud-free

regions within cloud fields contain some LWC [Jiang et al., 2009; Koren et al.,

2009].

While most previous studies that targeted cloud fields focused on cloud field

inner properties [Sengupta et al., 1990; Weger et al., 1992; Zhu et al., 1992; Weger et

al., 1993; Lee et al., 1994; Nair et al., 1998], our study presents a first attempt for

analytically defining the cloud field boundaries based on its inner cloud spatial

distribution (Section ‎2.1).

1.3.2 Aerosol effects on clouds

Aerosol may serve as cloud condensation nuclei (CCN) or ice nuclei (IN), which

produce cloud droplets or ice crystals in clouds. Therefore, changes in the aerosol

concentration affect the cloud droplet and ice particles concentration and size

distribution. An evidence for this connection was found in the observed effect of

aerosols on the cloud reflectance in the shortwave spectrum, namely „the Twomey

effect” [Twomey, 1977].

Twomey [1977] showed that an increase of the aerosol concentration (for a given

LWC) leads to an increase in the droplet number, which decreases their mean radius

and therefore enhances their optical cross section and optical depth. An example for

7

such observation is given in Figure ‎1.2, presenting satellite observed ship-tracks

[Coakley et al., 1987; Wang and Feingold, 2009] above the Atlantic Ocean,

demonstrating both radiative and microphysical effects of anthropogenic aerosols on

clouds. The particles from the ship engines act as CCNs, reducing the mean droplet

size (lower right panel), increasing the solar reflectance (upper panel) and the optical

thickness (lower left panel).

Figure ‎1.2 An unusually high number of ship tracks, as observed off of the coasts of

France and Spain in true-color (upper panel) and false-color images, representing the

retrievals of cloud optical thickness (lower left panel), and cloud effective particle

radius (lower right panel), from MODIS on the Aqua satellite on January 27, 2003

(source: MODIS Rapid Response website).

8

The effect of aerosol concentration values on the mean cloud droplet size drives

additional important dynamical and microphysical feedbacks. High aerosol loading,

which leads to smaller drops, reduces the efficiency of the collision-coalescence

process and due to that cause suppression of the warm rain. It may extend cloud

lifetime. This finding was observed and simulated for shallow clouds [Albrecht, 1989;

Jiang et al., 2006]. The warm rain suppression due to aerosol concentration increase

[Rosenfeld, 2000] was found also to be the first step in chain of events that cause

invigoration of convective clouds [Kaufman et al., 2005a; Koren et al., 2005]. This

chain of events includes smaller drops and smaller falling velocities, causing the

drops to reach higher altitudes, freeze at higher altitudes and release latent heat higher

in the atmosphere. All of this produces more vertically developed clouds with higher

tops, higher center of gravity of their rain column [Heiblum et al., 2012], bigger hail

particles, more intense electrical activity [Rosenfeld, 2000; Andreae et al., 2004;

Altaratz et al., 2010], and stronger rain rates [Koren et al., 2012].

Figure ‎1.3 schematically presents the observed differences between polluted

(high aerosol concentration) and pristine convective clouds, during their evolution in

time.

9

Figure ‎1.3 (Figure 2 of Rosenfeld et al., 2008): Evolution of deep convective clouds

developing in the pristine (top) and polluted (bottom) atmosphere. Cloud droplets

coalesce into raindrops that rain out from the pristine clouds. The smaller drops in the

polluted air do not precipitate before reaching the super-cooled levels, where they

freeze onto ice precipitation that falls and melts at lower levels. The additional release

of latent heat of freezing aloft and reabsorbed heat at lower levels by the melting ice

implies greater upward heat transport for the same amount of surface precipitation in

the more polluted atmosphere. This means consumption of more instability for the

same amount of rainfall. The inevitable result is invigoration of the convective clouds

and additional rainfall, despite the slower conversion of cloud droplets to raindrops

[Tao et al., 2007].

A different aerosol effect on clouds' processes is caused by aerosols which

absorb solar radiation and heat their surrounding, modifying the atmospheric

temperature and humidity profiles. [Hansen et al., 1997; Ackerman et al., 2000;

10

Koren et al., 2004; Koren et al., 2008a; Davidi et al., 2012]. By radiation absorption,

aerosols are able to heat their atmospheric layer, and influence the vertical

temperature gradient, stabilizing the underlining layers and destabilizing the layers

above their location. The result is weakening of shallow convective clouds [Koren et

al., 2004; Feingold et al., 2005].

On the other hand, in-cloud processes affect the aerosol properties. Those

processes affect the size distribution, spatial distribution, chemical composition and

optical properties, by transport, coalescence, and chemical processes [Feingold and

Morley, 2003].

1.4 The twilight zone

The regions for studying cloud-aerosol interactions by remote sensing methods

include the detectable clouds and their surrounding areas, where one can observe

clouds and aerosol properties in the vicinity of clouds. Those areas near clouds form a

continuous transition zone between cloud and cloud-free atmosphere (namely the

twilight zone). The twilight zone [Koren et al., 2007] has been recently recognized

and characterized as a wide belt surrounding detectable clouds, which contains small

clouds and cloud fragments, decaying and hesitant clouds, different levels of

humidified aerosols [Charlson et al., 2007; Koren et al., 2009]. The twilight zone‟s

content was shown to affect the apparent aerosol optical retrievals up to 30 𝑘𝑚 from

the nearest cloud edge [Charlson et al., 2007; Koren et al., 2007; Twohy et al., 2009a;

Varnai and Marshak, 2009; Bar-Or et al., 2010; Bar-Or et al., 2011].

In this region in the vicinity of clouds, the apparent aerosol optical and physical

retrievals from remote sensing instruments may be affected by the following

components: (1) aerosols may change their physical and optical properties due to

hygroscopic growth, where the relative humidity (RH) reaches high values [Twohy et

al., 2009a], (2) the apparent aerosol optical signal may be modified by contributions

from clouds which are too thin or too small to be detected as clouds by the instrument

or retrieval algorithm [Kaufman et al., 2005b], and (3) three-dimensional (3D)

radiative effects from multiple scattering inside clouds may be reflected into the

11

surrounding environment and contribute an additional signal to the apparent aerosol

retrievals [Wen et al., 2007; Marshak et al., 2008; Marshak et al., 2009].

The gaining interest in the twilight zone, accompanied by the recent observations

that show aerosol retrieval dependence on the distance from the nearest cloud, raised

the need to take this parameter into account when observing aerosols. Therefore, the

distance from the nearest cloud will be added to the new coming MODIS atmosphere

level 2 standard products (Robert Levy, personal communication).

The twilight zone was identified using remote sensing observations, but the

separation and the estimation of each of the three mentioned components contribution

to aerosol retrievals has not been accomplished yet. In the study described in Sections

‎2.4 and ‎3.3, a single component of the twilight zone is isolated (namely aerosol

humidification), and its contributions to the apparent aerosol optical depth (AOD) and

to the aerosol fine-mode fraction (FMF) are qualitatively estimated as a function of

the distance from the nearest cloud (dc).

In this study, the twilight zone is treated as a part of the cloud field. A cloud field

masking algorithm is presented in Section ‎2.1, and implemented in Section ‎3.1 for

estimating the global cloud field coverage and the average extent of the twilight zone

from the closest cloud. The unique features of aerosol retrievals in the twilight zone

are examined and described in Section ‎3.3.

1.5 Aerosol hygroscopic growth in cloud fields

Aerosol water uptake from their surrounding environment mainly depends on the

atmospheric relative humidity (RH) and on the aerosol chemical properties. Clouds,

being pockets of liquid water (or ice), naturally form in high RH regimes. Therefore,

one expects higher RH values near clouds, which may affect the suspended aerosol

size distributions and optical properties.

1.5.1 Humidification effects on aerosol size distribution

The water content of atmospheric aerosols determines their size, density, and

refractive index [Tang, 1996]. When dry aerosols are put in a humid environment,

12

they uptake water and grow. The growth rate as a function of the surrounding relative

humidity, namely the aerosol hygroscopic growth factor 𝑔 𝑅𝐻 , is defined in

Equation ‎1.1:

g RH =Dwet

Ddry=

Vwet

Vdry

13

= Vw + Va

Va

13

= 1 +Vw

Va

13

‎1.1

Where 𝐷𝑑𝑟𝑦 and 𝐷𝑤𝑒𝑡 are the diameters of the dry and the wet aerosols, respectively,

𝑉𝑑𝑟𝑦 and 𝑉𝑤𝑒𝑡 are the volumes of the dry and wet aerosol, respectively, 𝑉𝑎 is the

volume of the dry aerosol, and 𝑉𝑊 is the volume of the taken additional water content.

The amount of water that aerosols uptake from their surrounding environment

also depends on the aerosol hygroscopic properties, derived by the aerosol chemical

composition and microphysical shape. Rissler et al. [2006], followed by Petters and

Kreidenweis [2007], offered a single parameter representation of the hygroscopic

growth factor as a function of the water activity (𝑎𝑊), as presented in Equation ‎1.2:

1

aW= 1 + κ

Va

VW= 1 +

κ

g3 − 1 ‎1.2

Where 𝑎𝑊 is the water activity, 𝑉𝑎 and 𝑉𝑊 are the volume of the dry aerosol and of

the water addition, respectively, 𝑔 is the hygroscopic growth factor, and 𝜅 is defined

as the aerosol hygroscopicity parameter.

When neglecting the Kelvin effect, whose contribution to the growth of

aerosols with radii larger than 0.1 μm is smaller than 1% for a temperature range

relevant to this study [Hinds, 1999], the water activity is identical to the environment

RH value, and the hygroscopic growth factor as a function of RH may be evaluated as

follows (Equation ‎1.3):

g κ, RH = 1 + κ ∙RH

1 − RH

13

‎1.3

13

Where g is the hygroscopic growth factor, 𝜅 is the aerosol hygroscopicity parameter,

and RH is the relative humidity that the aerosol experience.

Using the kappa parameterization, the hygroscopicity (𝜅) values of a wide variety

of aerosol types and compositions were measured both in laboratory and in field

campaigns. Average marine aerosols were estimated to have kappa values of 0.7 ±

0.2 [Andreae and Rosenfeld, 2008], while coarse-mode sea salt is even more

hygroscopic and may reach 𝜅 = 1.2 [Kreidenweis et al., 2005; Andreae and

Rosenfeld, 2008; Sorooshian et al., 2008]. Dust particles are much less hygroscopic,

with values of 0.01 − 0.08 [Gasparini et al., 2006; Andreae and Rosenfeld, 2008;

Koehler et al., 2009; Twohy et al., 2009b; Yan et al., 2009]. Fine-mode biomass

burning aerosol 𝜅 values vary between 0.06 − 0.7 while the majority of the biomass

burning products obtain values of 0.06 − 0.33 [Andreae and Rosenfeld, 2008; Petters

et al., 2009; Carrico et al., 2010]. Pure black carbon is assumed to be absolutely non-

hygroscopic 𝜅 = 0 . Figure ‎1.4 shows the hygroscopic growth factor g for various

hygroscopicity parameter and relative humidity values, and Figure ‎1.5 presents the

change in the aerosol size distribution caused by a hygroscopic growth for a bimodal

lognormal distribution.

14

Figure ‎1.4 The hygroscopic growth factor g as a function of the relative humidity RH

and the hygroscopic growth factor 𝜅, following Equation ‎1.3. The g values around 2

are marked (black) to demonstrate the RH and 𝜅 values that will result in doubling of

the dry aerosol radius, by water uptake.

Figure ‎1.5 A demonstration of dry (blue line) and humidified (RH = 95%, red line)

aerosol bimodal log-normal size distributions, normalized to the total particle number.

The fine mode contains biomass burning particles rg = 0.08; σ = 0.7; κ = 0.3 , and

the coarse mode contains sea salt particles rg = 0.6; σ = 0.6; κ = 0.7 . The gray

dashed line represents the boundary between fine and coarse mode distributions

(when σ = 0.7). Note that the total particle number of the two presented aerosol

distributions is equal.

1.5.2 The relative humidity in cloud fields

As shown above in Section ‎1.5.1, the relative humidity (RH) may determine the

aerosol size distribution in cloud fields, as derived in Equation ‎1.3.

15

When estimating the aerosol humidification effect, it is necessary to evaluate the

dependence of RH in the distance from clouds. Very few studies had provided in-situ

measurements of this feature [Twohy et al., 2009b; Wang and Geerts, 2010],

encouraging further research in order to spatially characterize the RH in cloud fields.

According to these few recent studies, areas of high RH (RH ≳ 95%) which would

cause significant aerosol humidification effect are limited to the closest vicinity of

clouds, while most of the cloud field exhibits much lower RH values, whose average

is closer to the background RH value (far from clouds). Therefore, clouds' presence

efficiently marks areas of high RH values, as expected.

In Section ‎3.3.1 of this study, the RH in cloud fields is simulated using large eddy

simulation (LES) models, and a new parameterization of the RH as a function of the

distance from clouds is presented.

Cloud-aerosol interaction studies examine trends between the observed properties

of clouds and aerosols, trying to determine causality between aerosols and clouds.

These studies need to check all possible reasons for these trends before they can

conclude about causality. Other possible reasons for such observed trends can be

meteorology that drives them both or aerosol and clouds retrievals artifacts. Recent

studies identified an aerosol invigoration effect on convective clouds, showing that

increase in aerosol loading causes an increase in cloud horizontal and vertical

dimensions [Koren et al., 2005; Heiblum et al., 2012]. It was also shown that in

polluted environments with higher aerosol loading the rain rate is stronger [Koren et

al., 2012], and that the precipitation particles are located higher in the atmosphere

[Heiblum et al., 2012]. These studies examined carefully two factors that can

influence the observed trends: meteorological variance and cloud contamination, by

stratifying the observed data for different meteorological conditions, and filtering out

all AOD data that were suspected as “cloud contaminated”. However, the significance

of aerosol humidification (as an effect that influence the aerosol measured properties

in the vicinity of clouds) to the observed trends in these type of studies has not been

accurately evaluated yet.

Therefore, the significance of the aerosol humidification effect is studied in this

work. Several recent studies, using Global Circulation Models (GCM), had referred to

16

this question, concluding that aerosol humidification is a dominant generator of any

observed trend between aerosol and cloud properties [Quaas et al., 2008; Quaas et

al., 2010]. These findings were presented under the limitations derived by the non-

linearity features of this problem, and by the poor parameterizations that are available

in GCMs for this purpose, as stated in Quaas et al. [2008]. On the other hand, the few

in-situ measurements of RH and specific humidity near clouds [Twohy et al., 2009a;

Wang and Geerts, 2010] suggest that high values of RH exhibit only near clouds.

In this study, we wish to describe as accurately as possible the RH properties of

the cloudy atmosphere form the best available measurements and to estimate expected

AOD biases due to RH variations in the vicinity of clouds.

The most accurate and statistically valid RH profile measurement, to the best of

our knowledge, are atmospheric sounding (radiosonde) data sets. Analyzing the vast

number of measured atmospheric profiles, which include vertical RH profiles and

information about the extent of the cloudy layers, enables the study cloudy

atmosphere RH trends. Our research extends over 14 globally distributes atmospheric

sounding stations, and over 32 year long period.

First, we estimate the mean and the variance of the sub-saturated RH values in the

lower cloudy troposphere. The lower troposphere is likely to host most aerosol mass

(95% up to 2 km from surface, [Blanchard and Woodcock, 1980]), and was shown to

have the highest Kappa values, suggesting that most of humidification effect is likely

to be there.

Then we use the information that is folded within these profiles regarding the

thickness of the cloudy layer, and analyze the local differences per each station

between the mean sub-saturated RH values of the shallower subset and the more

developed (vertically thicker) clouds. This analysis examines the potential for

possible RH driven biases in the observed trends of cloud aerosol interactions.

17

2 Methods and data

2.1 Cloud field masking algorithm

Cloud field is defined in this study as the area that contains both detectable clouds

and the space around them, within a characteristic distance from each detectable

cloud. It is assumed that the likelihood to have an undetectable cloud (weak optical

signature and/or small relative to the sensor resolution) increases as one approach

detectable clouds [Koren et al., 2007; Koren et al., 2008b]. Therefore, this likelihood

is higher in a cloud field and decreases as moving away from it. Moreover, the same

behavior applies for pockets of high relative humidity and extra illumination coming

from the sides of clouds [Marshak et al., 2006; Wen et al., 2007; Marshak et al.,

2008; Zinner et al., 2008; Marshak et al., 2009; Varnai and Marshak, 2009].

A robust and simple to implement cloud field masking algorithm should comply

the following requirements: (1) the algorithm should use basic input data on cloud

distribution, like binary cloud mask, (2) the algorithm should be based on a spatial

analysis scheme, (3) the algorithm should be applicable to any informative input data

resolution (similar to characteristic cloud size), (4) the algorithm should be valid to all

cloud types.

All previous studies of cloud fields spatial structure have focused on the spatial

distributions of clouds inside cloud fields, and usually considered the entire examined

domain as a part of a cloud field (without any definition of the cloud field

boundaries). Therefore, this study focuses on developing a robust and easy to

implement cloud field boundary detection algorithm.

A few existing spatial analysis methods were considered for this research. For

example, one proposed a method used joint statistics and K-Nearest-Neighbor (kNN)

techniques [Sankaranarayanan et al., 2007] over binary cloud mask data, in order to

isolate the largest clusters in the examined domain, and then define the cloud field

boundaries as the cluster edges. This method lacks the ability to treat isolated small

cloud fields, and ignores these important fields. In addition, this method focuses on

clouds only and does not include the important twilight zone in the cloud field area.

Other methods that do mark a characteristic distance from the cloud field center lack

18

the ability to mark reasonable borders when the cloud field is not rounded shape (most

cloud fields are not rounded).

A cloud field masking algorithm requires a flexible method that will follow any

cloud field shape. Here, that requirement is defined as "locality", i.e. the algorithm

should be sensitive to scales which are higher than the scale of the whole cloud field

in order to mask fields with relatively complex shape. Therefore, the preferred metric

should rely on local properties of the clouds distribution. The best metric that meets

this requirement was found to be the distribution of the distance-from-nearest-cloud

[Koren et al., 2007], where each element represents the Euclidian distance of the

center of the pixel to the nearest cloud. Such metric is local by definition and it

considers nothing but the distances in the vicinity of each cloud.

Euclidian distance transform methods are being used in a wide range of spatial

analysis applications, such as linear and edge detection of objects in digital images

[Rosin, 2009], fractal dimension analysis of 2D objects [Adler and Hancock, 1994],

and recently – for cloud spatial and radiative properties [Koren et al., 2007; Marshak

et al., 2008; Bar-Or et al., 2010]. The last is gaining an increased interest, and a

dataset of the distance-from-the-nearest-cloud is planned to be added to the MODIS

atmosphere level 2 products. Deeper mathematical discussion and further synthetic

examples of the distance field distribution can be found in Ripley (1981).

In the presented algorithm, the probability distribution of the Euclidian distance

map is being used for distinguishing the inner cloud field area from the surrounding

cloud-free area. Examining the whole domain (including the cloud-free area), the

probability distribution shows two different regimes: (1) the intra cloud-field regime,

characterized by the distance probability distribution of clouds inside the field

(describing the cloud spatial distribution), and (2) the extra-field regime, which

asymptotically approximate distance probability distribution of a single giant cloud.

While the distance probability distribution inside the field has a maximum point,

representing the most common distance from a cloud inside the field, following by a

decrease in likelihood of larger distances, the distance probability distribution outside

the fields is monotonically increasing (with a slope that asymptotically goes to 2π,

away from the cloud field as the smoothed perimeter approximate a circle).

19

Having a local maximum in the intra cloud field distribution and monotonic

increase out of the cloud field defines a minimum in the transition between the

distributions.

The distance value corresponding to the local minimum is defined here as the field

distance parameter (R0), and it represents the largest distance-from-the-nearest-cloud

that is still considered to be part of the cloud field. The contour defined by R0 marks

the cloud field boundaries, distinguishing the cloud field from the surrounding cloud-

free area. Figure ‎2.1 shows a distance map and a distance from the nearest cloud

probability distribution for a synthetic cloud field. It clearly shows the different

distribution properties inside and outside of the cloud field. A true data example for

the extraction of the field distance parameter is presented in Figure ‎2.1.

Figure ‎2.1 (Fig. 1 in Bar-Or et al., 2011) Distance map of a synthetic cloud field,

composed of randomly distributed pixel-size clouds (a), and a zoom on its interior

distance map (b). The different distance probability distributions of the whole

synthetic field and of the inner field only (c and d, respectively) demonstrate the

transition point between the inner field distances and the complete field distances. The

20

decrease of the distance probability function in panel c (from distances of 100 pixels

and more) is due to the restricted domain size where larger distances are needed.

Figure ‎2.2 (Fig. 2 in Bar-Or et al., 2011) Analysis of an observed cloud field,

including the distance probability distribution (blue line), the filtered distance

probability distribution (green line), the distance cumulative probability (red line), and

the transition point, identified by the minimum of the filtered distance probability

distribution, and defining the field distance parameter (marked with orange arrow).

2.1.1 Cloud field masking algorithm - description

The complete cloud field bounding algorithm for the MODIS data is described

below:

Step 1. Data projection on an equal-area matrix

The large footprint of the MODIS instrument results in slight geometrical

distortion. The first step of the algorithm is a projection of all product granules (data

blocks) on an equal-area matrix, in order to avoid any errors due to geometrical

21

differences. The input data for this study is the MODIS cloud mask product

[Ackerman et al., 1998; Platnick et al., 2003]. The projection to 1 km equal-area

cloud mask and to 1 km ocean/land masks are done using MODIS Geolocation

product.

Step 2. Distance map extraction

The Euclidian distance from the nearest cloud is calculated, based on the 1

km/pixel equal-area cloud mask.

Step 3. Calculating the distance probability distribution

The distance cumulative distribution A(r) is calculated for varying distance

parameter values (r). A(r) is the total area that is closer than r from any cloud in the

observed domain. Then, the distance probability distribution is calculated as the

derivative d

dr(A r ).

Step 4. Noise corrections for the distance probability distribution

The generated distance probability function d

dr(A r ) may suffer from high noise

levels, mostly for low r values when the distances calculated for integer number of

pixels may create discontinuities in the distribution. In this step, a Gaussian filter is

applied on the distance probability distribution function in order to filter out high

frequency variations, and to enable the calculation of field distance parameter R0.

Figure ‎2.2 demonstrates the distance cumulative function A(r), the distance

probability distribution d

dr(A r ), and the smoothed distance probability function.

Step 5. Extracting the field distance parameter R0

The local minimum of the smoothed distance probability distribution is used to

determine the field distance parameter R0, as demonstrated in Figure ‎2.2.

Step 6. Calculating the cloud field boundaries and coverage

After determining R0, the Cloud Field Fraction (CFF) is calculated. The cloud

field fraction is a unit-less normalized ratio that represents the portion of cloud field

22

covered area in the whole examined domain and defined as: CFF =A(R0)

AD, where AD

is the domain‟s area. The CFF is analogue to the cloud fraction measure. CFF=0

represents an absolute cloud-free domain and CFF=1 a domain that contains only

cloud field area. Given that the domain‟s cloud fraction is A(r=0)

AD, the domain‟s CFF is

always equal or larger than the domain‟s cloud fraction.

The boundaries of the cloud field are represented by the contour r=R0 where all

the pixels whose distance from the nearest cloud is smaller than R0 are part of the

field.

2.1.2 Cloud field masking algorithm - sensitivity study

The proposed algorithm was examined by an extensive set of sensitivity tests,

verifying that the algorithm is not sensitive to the resolution of the input cloud mask

data or to the clouds' spatial distribution. The tests were conducted on both synthetic

and realistic (observed by MODIS) cloud fields. For these tests, the resolution of each

cloud mask data was reduced by a simple averaging of pixels, verifying that the cloud

fraction is constant. After reducing the data resolution (i.e. increasing the data pixel

size), the algorithm was used for calculating the field distance parameter. The

algorithm was found to be stable for varying data resolutions, provided that the data

pixel size is smaller than the characteristic length scale of the examined cloud field.

Figure ‎2.3 demonstrates the described resolution sensitivity test for a MODIS

observed cloud field (the same as was analyzed in Figure ‎2.3). In this case, the field

distance parameters calculated by the algorithm are in the range of 17.5-22.0 km, for

pixel size smaller than 7 km. The cloud fraction is stable in the range of 62.6%-

64.6%, as expected by a resolution reduction of a binary cloud mask.

23

Figure ‎2.3 (Fig. 3 in Bar-Or et al., 2011) The calculated field distance parameter R0

(blue line) and the relative cloud fraction (green line), as a function of MODIS data

resolution.

For the sake of completeness, the theoretical case of a cloud field that contains

only one circular cloud is also considered. In this case the distance probability

function has only one minimum at r=0, and therefore the calculated field distance

parameter is zero. These scenarios are easily identified by the algorithm and gain field

distance parameters that are defined by the mean field distance parameters of the

neighboring cloud fields. However, the probability to find such cloud fields in

realistic datasets is negligible.

2.2 Global cloud field coverage

The data analyzed in this section are MODIS-Terra atmosphere level 2 products

[Platnick et al., 2003; Remer et al., 2005; Levy et al., 2007], for one day at July 28th

,

2008. This study is based on analysis of 66 granules (dataset blocks), containing day-

24

light information, for this specific day, between latitudes 50°S-50°N. The cloud mask

input data are based on the MODIS cloud mask product [Ackerman et al., 1998;

Platnick et al., 2003], the aerosol properties data are based on the MODIS aerosol

product [Remer et al., 2005; Levy et al., 2007], and both sea/land mask and geo-

location data are based on MODIS Geolocation product.

The global CFF (50°S to 50°N) is estimated over land and ocean. In order to find

whether there is any dependence of the field distance parameter on the cloud type,

several cloud fields in each granule are manually classified. This classification is done

using the MODIS provided true-color images, based on the familiar spatial

morphology of the clouds type.

The classification includes four cloud types: Stratocumulus (Sc), shallow Cumulus

(Cu), Cirrus (Ci), and deep convective (DC). The field distance parameter of each of

the selected 170 cloud fields is calculated using the described algorithm; which treats

all cloud types in a similar way.

2.3 Aerosol retrieval inside cloud fields

A daily global MODIS dataset is analyzed in order to evaluate the total radiative

effect of the twilight zone on the aerosol retrieved physical and optical properties.

Both MODIS aerosol optical depth (AOD) and aerosol fine-mode fraction (FMF) are

examined as a function of the distance from the nearest cloud (dc). The distance map

generated for this analysis is based on the MODIS 1 km cloud mask product [Platnick

et al., 2003], and the aerosol retrievals are analyzed separately over Oceanic and land

surfaces, with latitudinal dependence. The surface type and geolocation data are given

by the MODIS geolocation product.

Furthermore, the sensitivity of the MODIS aerosol fine mode fraction (FMF) to

the distance from the nearest cloud is examined. The data for this sensitivity study is

selected to be only above oceans due to the high uncertainties in the aerosol fine-

mode fraction product retrieved above land.

25

2.4 Radiative effects of aerosol humidification on cloud fields

2.4.1 Parameterization of the RH spatial distribution in cloud fields

The current lack of high-resolution in-situ measurements of RH as a function of

the relative location of neighboring clouds (dc), urges the use of numerical simulation

based data for description of RH dc . For this purpose, the liquid water content

(LWC) and the RH values are calculated by the Regional Atmospheric Modeling

System (RAMS, Cotton et al., 2003) and by the Weather Research & Forcasting

model (WRF version 3, Skamarock et al., 2008) , for warm Cumulus cloud fields. The

cloud masking algorithm uses a LWC threshold of 0.01 g ∙ kg−1, and the RH field is

given by the model. The mean RH as a function of the distance to the nearest cloud

RH dc is calculated for every cloud-containing layer, and fitted to an exponential

function:

RH dc = RH0 + RHE − RH0 ∙ e−

dcδ

‎2.1

Where RH0 is the background relative humidity far from clouds, RHE is the relative

humidity on the cloud edge, and 𝐵 is the e-folding exponential distance scale of the

relative humidity growth near clouds. This fit represents the general averaged RH

field in the cloud layer.

The estimation of RH(dc) in this work is performed using outputs of two

different large eddy simulations. The first simulation (LES simulation A), was carried

out by the WRF (version 3) model, using the two-moment bulk microphysical scheme

(Morrison et al., 2009). This simulation was initialized based on August 21th

, 2007,

00:00 UTC sounding of temperature and moisture from Lihue, Hawaii (WMO station

number 91165). It covers a 15 × 15 km warm Cumulus field, with a horizontal

resolution of 100 m, and 149 terrain following vertical levels in the domain reaching

the model top at 8 km. The vertical resolution varies from 37 m at the lowest layer to

about 83 m at the highest layer. The second simulation used for this research (LES

simulation B), was carried out using the RAMS6 model, with the two-moment bulk

microphysical scheme [Walko et al., 1995; Meyers et al., 1997; Walko et al., 2000].

This simulation was initialized based on June 26th

2010, 12:00 UTC sounding of

26

temperature and moisture, measured in Bet Dagan, Israel (WMO station number

40179). It covers a 12.4 × 12.4 km warm Cumulus field, with a horizontal resolution

of 100 m and vertical resolution of 50 m below 4700 m (the total domain height is 9

km). The total duration of the input data is 1.5 hours, with a 5 minute average (for

simulation B), in order to include a variety of cloud sizes and ages.

2.4.2 Atmospheric radiative transfer simulations of humidified aerosols

When dry aerosol population experiences an increase in the environmental RH, it

is assumes that the water uptake by aerosols does not reduce the RH, and that all

particles are simultaneously growing with the same hygroscopic growth factor, as

shown in Equation ‎2.2. These assumptions are appropriate for aged, internally-mixed

aerosols.

r RH = rdry ∙ g RH ‎2.2

Where r(RH) is the humidified aerosol radius, rdry is the initial dry aerosol radius,

and g RH is the hygroscopic growth factor, as presented in Equation ‎1.3.

Similar to the aerosol size distribution, the aerosol mass distribution is also

modified by the gain of additional mass due to uptake of water content in a varying

RH environment, as expressed in Equation ‎2.3:

M RH = Mdry ∙ 1 +ρ

w

ρa

g3 − 1 ‎2.3

Where M(RH) is the humidified aerosol mass, Mdry is the initial dry aerosol mass,

ρw

and ρa are the bulk densities of pure water and of the dry aerosol, respectively, and

g is the hygroscopic growth factor, as presented in Equation ‎1.3.

By adding water mass content to the aerosols, the humidification process also

impacts both bulk density and refractive index of the humidified aerosol. These

27

properties are calculated using volume weighted means of the dry aerosol and pure

water properties (Equations ‎2.4-‎2.5):

ρ RH = ρw

+1

g3 ρ

a− ρ

w ‎2.4

Ref RH = nwet − i ∙ kwet =

= nw +1

g3 na − nw − i ∙ kw +

1

g3 ka − kw

‎2.5

Where Ref RH is the humidified aerosol refractive index, comprising the real

component nwet , and the imaginary component kwet . nw , kw , na , and ka are the real

and imaginary components of pure water, and dry aerosol refractive indices,

respectively, and g is the aerosol hygroscopic growth factor.

The volume weighted mean for both density and refractive index calculations is

commonly used in chemical, cloud and radiation models [Tang, 1996; Levoni et al.,

1997], and satisfies accuracy standards in recent experimental research [Levoni et al.,

1997; Michel Flores et al., 2012]. In this study, concentrated on the wavelength of

0.550 micron, pure water parameters are set to be: ρw

= 1 g ∙ cm−3; nw =

1.33; kw = 0.

For simulating the MODIS retrieval of the humidified aerosol properties the

Spherical Harmonic Discrete Ordinate Method – SHDOM [Evans, 1998] is used. In

these simulations, all suspended aerosols are assumed to be uniformly distributed

between heights of 1-2.7 km, which represent the cloud layer height in the LES

simulated cloud field in case A1, described in Section ‎3.3.1.

This assumption about aerosol location in the cloud layer only represents the

maximum humidification effect as in nature the aerosol is distributed from the ground

up, throughout the boundary layer to the free atmosphere, where the humidity may be

lower. The initial dry aerosol size distribution is set to be a bimodal log-normal

distribution, comprising fine-mode and coarse mode aerosol distributions. In order to

better simulate MODIS instruments retrievals, all radiative transfer simulations are

conducted at a wavelength Gaussian band of 545-565 μm , similar to MODIS band 4

28

[Remer et al., 2005]. The aerosol fine-mode fraction (FMF) is calculated using two

separate simulations: the first contains only a single fine-mode aerosol distribution,

and the second contains a bi-modal distribution. After acquiring the aerosol optical

depth for each simulation, the FMF is directly from its definition, using Equation ‎2.6:

𝐹𝑀𝐹 =𝐴𝑂𝐷𝑓𝑖 (𝜆 = 550𝑛𝑚)

𝐴𝑂𝐷𝑡𝑜𝑡 (𝜆 = 550 𝑛𝑚) ‎2.6

Where 𝐴𝑂𝐷𝑓𝑖 and 𝐴𝑂𝐷𝑡𝑜𝑡 is the aerosol optical depth of the fine mode aerosol and

the total aerosol distribution, respectively, and 𝜆 is the used wavelength.

The above scheme, which its conceptual model is demonstrated in Figure ‎2.4,

where as the outcome, calculates both AOD(dc) and FMF(dc) for any given aerosol

distribution. The sensitivity of both AOD and FMF to the varying RH is estimated by

running an extensive set of radiative transfer simulations for different size

distributions.

Figure ‎2.4 (Fig. 1 in Bar-Or et al.[2012]) The conceptual model operated in Section

‎3.3. Input data (gray circles): the cloud mask field, the relative humidity field, the

aerosol hygroscopic growth parameterization (growth param.), and the dry aerosol

physical and optical properties. Analyzed data (green rectangles): the field of

distances from the nearest cloud - dc, the hygroscopic growth factor as a function of

relative humidity - g RH , the relative humidity as a function of dc - RH dc , the

29

hygroscopic growth factor as a function of dc - g dc , and the humidified aerosol

optical and physical properties due to hygroscopic growth. Simulated data based on

the modified aerosol properties as a function of dc (blue rectangles): the humidified

aerosol optical depth - AOD dc , and aerosol fine-mode fraction - FMF dc .

2.5 Characterizing the RH in cloud fields

The input datasets for this analysis are the records of the World Meteorological

Organization (WMO) registered upper-air measurement stations, that are freely

available at the University of Wyoming Atmospheric Soundings database

(http://weather.uwyo.edu/upperair/sounding.html), or at the website of NOAA‟s

Integrated Global Radiosonde Archive (IGRA, Durre et al., 2006). These vast datasets

provide a continuous record of upper-air vertical profiles of pressure, temperature,

relative humidity, wind speed, and wind direction. In most stations, radiosondes are

released at least twice a day, at 00:00 UTC and at 12:00 UTC (i.e. 00Z and 12Z),

without any consideration of the local atmospheric conditions. This independent

sampling procedure enables the analogy of a long radiosonde record to a pseudo

Monte-Carlo experiment, which examines the statistical properties of several

atmospheric measures in a specific location.

In addition to the estimation of the RH vertical profile for the selected locations

(and seasons), it is possible to extract the height and thickness of the lowest (by

altitude) potentially cloudy layer the radiosonde sampled, enabling the assessment of

RH vertical profiles inside cloud fields (see Section ‎1.5.2). This potentially cloudy

layer is characterized by suitable conditions for cloud formation.

2.5.1 Locating the cloudy layer with radiosonde data

The lowest (by altitude) atmospheric cloudy layer is expected to extend from the

Lifted Condensation Level (LCL), which is calculated using Equation ‎2.7 [Bolton,

1980]. This equation is an approximated iteratively solved solution of the equations

that describe the dry adiabatic lapse rate and the dew point lapse rate (i.e. the point of

which an adiabatic parcel‟s temperature equals the dew point temperature).

30

𝑇𝐿𝐶𝐿 =1

1𝑇𝑑 − 56

+𝑙𝑛 𝑇/𝑇𝑑

800

+ 56 ‎2.7

Where TLCL is the temperature at the LCL (K), and T and Td are the temperature and

the dew-point temperature at the surface or at any chosen altitude below the LCL.

For each radiosonde profile, the cloudy layer base is set as the Lifted

Condensation Level (LCL [Bolton, 1980] ), which is defined as the height of

saturation for a rising air parcel based on the average humidity conditions at the

lowest 500 m above surface. In cases where the surface initiated LCL computation

results in an LCL that is higher than an atmospheric stable layer (where the observed

temperature lapse rate is lower than 3 𝐾 ∙ 𝑘𝑚−1), the calculation is considered

incorrect, and a corrected LCL is calculated using T and Td values at the top of this

stable layer (using Equation ‎2.7). The calculated LCL is set as the base of the cloudy

layer.

The upper limit of the cloudy layer is defined by two different methods. The

Saturated Adiabatic Lapse Rate (SALR) limit is the bound of the unstable layer for

saturated air above the LCL, a layer that enables cloud formation. The cloudy layer,

as defined by the SALR method describes the atmospheric layer that is conditionally

unstable for wet air and enables cloud formation (computed from the LCL height

using Equations ‎2.8-‎2.9):

𝑇𝑆𝐴𝐿𝑅 𝑕 = 𝑇 𝐿𝐶𝐿 + ΓSALR h 𝑕−𝐿𝐶𝐿

0

𝑑𝑕 ‎2.8

ΓSALR r, T = g ∙ 1 +Hvr

Rsd T ∙ Cpd +

Hv2rε

Rsd T2

−1

‎2.9

Where 𝑇𝑆𝐴𝐿𝑅 is the temperature following the saturated adiabatic lapse rate from the

LCL (𝑑𝑒𝑔𝐾), T(h) is the temperature vertical profile (𝑑𝑒𝑔𝐾), 𝑟(𝑕) is the water vapor

mixing ratio vertical profile (𝑔 ∙ 𝑘𝑔−1), ΓSALR is the saturated adiabatic lapse rate

(𝐾 ∙ 𝑚−1), g is the Earth‟s gravitational acceleration (𝑚 ∙ 𝑠−2), 𝐻𝑣 is the heat of

evaporation of water (𝐽 ∙ 𝑘𝑔−1), 𝑅𝑠𝑑 is the specific gas constant of dry air (𝐽 ∙

𝑘𝑔−1𝐾−1), 𝑅𝑠𝑤 is the specific gas constant for water vapor (𝐽 ∙ 𝑘𝑔−1𝐾−1), 𝐶𝑝𝑑 is the

specific heat of dry air at constant pressure (𝐽 ∙ 𝑘𝑔−1𝐾−1), and 𝜀 is the ratio of the

31

specific gas constant of dry air to the specific gas constant for water vapor

(dimensionless).

The second measure takes in account that often convective clouds overshoot

the conditionally unstable layer as long as the conditions of the layer above the SALR

are close to saturation. For that purpose, we define here a new measure namely the

Moist Boundary Parameterization (MBP, Equation ‎2.10). Convective clouds

occasionally exceed the SALR layer and penetrate the layer above, as long as it is

humid enough to keep the clouds from complete evaporation. The MBP is calculated

for each altitude using the temperature difference between the measured temperature

(T) and the dew point temperature (Td), divided by the measured relative humidity

(RH). This method is sensitive to both reduction in Td that will increase the numerator

and the reduction in RH that will decrease the denominator and therefore found to

detect small shifts in the profile well. The likelihood for cloud existence decreases

rapidly as the MBP increases. After inspecting the sensitivity of the profiles to the

MBP values were tested, a threshold value of 5 ℃/% was found to well define the

MBP cloudy layer upper limit for this study.

𝑀𝐵𝑃 𝑇, 𝑇𝑑 , 𝑅𝐻 = 𝑇 − 𝑇𝑑 /𝑅𝐻 ‎2.10

The Moist Bounding Parameter (MBP) as a function of the temperature (T), the dew

point temperature (Td), and the relative humidity (RH).

An example of one analyzed radiosonde sample is presented in Figure ‎2.5. The

two different methods for identifying potentially cloudy layers enable the treatment of

all convective cloud types. While most convective cloud layers have a similar top

boundary using the two methods, the temperature profile of deep convective cloud

layers in humid environment (e.g. tropical Cb cloud fields in the Amazon), is very

similar to the saturated adiabatic lapse rate. The similarity between the saturated

adiabatic lapse rate and the temperature profile in these clouds will cause the

algorithm to bound the layer lower than necessary, and therefore only the moist

bounding parameter threshold can find its physical limit.

32

Figure ‎2.5 A single radiosonde sample, obtained in Lihue, Hawaii (see Table 4), on

November 25th

, 1993, 00:00:00 UTC. The shown vertical profiles are the measured

relative humidity (left panel), the calculated moist bounding parameter (MBP, middle

panel), the measured temperature (right panel, red line), the measured dew-point

temperature (right panel, black dashed line), the saturated adiabatic lapse rate from the

estimated LCL (right panel, blue line), the LCL as calculated from the lowest 500m of

the sample (rigt panel, black horizontal line), and the upper boundaries of the cloud

layer as calculated by the SALR method and by the moist bounding parameter

threshold (right panel, brown and green horizontal lines).

2.5.2 Characterizing the SRH values in cloud fields

Integrating 32 year record of AS profiles for a given station, at a given

measurement time, during a specific season (presuming low meteorological variance),

is analogue to a Monte Carlo experiment which samples the atmosphere in a specific

location and time, uninfluenced by the atmospheric conditions. For a long 32 year

sample period this pseudo Monte Carlo experiment enables the calculation of the

33

mean sub-saturated (not in cloud) RH profiles - 𝑆𝑅𝐻 (𝑧), and of the RH standard

deviation profiles 𝜎𝑆𝑅𝐻(𝑧). These RH measurements in cloud fields, yet not in clouds,

provide the mean RH values that retrieved aerosols experience in the vicinity of

clouds.

2.5.3 The lower cloudy troposphere SRH and cloud development

To further explore the possible effects of SRH in the lower cloudy troposphere

on the observation of cloud-aerosol interactions, we keep following the same line of

reasoning. We estimate the SRH values as a function of the level of the cloud

development. Since a positive correlation in expected between cloud horizontal and

vertical dimensions [Koren et al., 2010] and both are correlated with the cloudy layer

thickness (between the LCL and the SALR), the data subset with the thicker cloudy

layer is likely to represent cases of vertically developed clouds and high cloud

fraction.

For each station, all radiosonde profiles are sorted by their cloudy layer thickness

(measured from the LCL to the SALR upper limit). The profiles are classified for

more developed cloud profiles, and shallower cloud profiles, separated by median of

the cloudy layer thickness, for each station.

A calculation shows that hygroscopic aerosols (e.g. sea salt aerosols with 𝜅 = 0.7

experience geometric cross section difference of 19.5% when their environment RH

value changes from 80% to 85%, and experience AOD change of 18.2% when the RH

value changes from 60% to 70%. Non-hygroscopic particles (e.g. biomass burning

aerosol – 𝜅 = 0.1) gain geometric cross section difference of 7.8% for RH increase of

80% to 85%, and AOD change of 4.8% for RH increase of 60% to 70%. The

geometric cross section difference is a first order approximation to the total expected

AOD change, neglecting the modification of the refractive indices due to water

uptake, and the aerosol size parameters.

34

3 Results

The results of Sections ‎3.1-‎3.4 below are included in the following manuscripts:

Bar-Or, R. Z., I. Koren, and O. Altaratz (2010) Estimating cloud field

coverage using morphological analysis, Environ. Res. Lett., 5, 014022,

doi:10.1088/1748-9326/5/1/014022.

Bar-Or, R. Z., O. Altaratz, and I. Koren (2011) Global analysis of cloud field

coverage and radiative properties, using morphological methods and MODIS

observations, Atmos. Chem. Phys., 11(1), 191, doi:10.5194/acp-11-191-2011.

Flores, J. M., R. Z. Bar-Or, N. Bluvshtein, A. Abu-Riziq, A. Kostinski, S.

Borrmann, I. Koren, and Y. Rudich (2012) Absorbing aerosols at high relative

humidity: closure between hygroscopic growth and optical properties, Atmos.

Chem. Phys., 12, 5511-5521, doi:10.5194/acp-12-5511-2012.

Bar-Or, R. Z., I. Koren, O. Altaratz, and E. Fredj: Humidified aerosol

properties in cloudy environment, Atmos. Res., 118, 280-294, doi:

10.1016/j.atmosres.2012.07.014.

Bar-Or, R. Z., I. Koren, O. Altaratz: Characterizing the relative humidity in

the lower cloudy troposphere, Geophys. Res. Lett., submitted.

3.1 Global cloud field coverage

The results (Table 1), which are based on the selected 170 cloud fields, show a

mean global field distance parameter of 29±1 km, in agreement with the results of

Bar-Or et al. (2010), that found a monthly averaged field distance parameter of 30 km

over the Atlantic Ocean during July 2008 (the error is calculated as a standard mean

error). The results also agree with Koren et al. [2007], that found a reflectance signal

effect up to 30 km from the nearest cloud edges. The comparison between the field

distance parameters of different cloud field types shows that Sc fields have the

smallest distance parameter, probably due to their sharp transition to cloud-free

atmosphere at their edges. The calculated field distance parameter for Cirrus cloud

fields is very close to the mean value of all cloud type, possibly because Cirrus field

may be located above a wider field of different cloud type. Such setting would lead

35

the algorithm to be sensitive to the clouds that appear on the border of the fields. All

calculated field distance parameter values are in the same order of magnitude of 29±1

km, with the extreme results being 17 and 39 km.

The global cloud field fraction for July 28th

, 2008, is 88%, calculated using the

algorithm generated field distance parameter R0, and 81% when using the constant R0

= 10 km for the whole dataset. These results are in line with Twohy et al., [2009a]

that found that only 8% of the detected cloud-free area above oceans is located in a

distance larger than 20km from the nearest detected cloud. This suggests that the CFF

over oceans, using a constant R0 of 20 km, is approximately 82%. Koren et al., (2007)

showed that an addition of 30 km belt around all clouds when the global cloud

fraction is 51%, will cover 81% of Earth‟s surface. It means a global CFF of

approximately 82% when using a constant R0 of 30 km.

Next, the calculated global CFF and the distance-from-the-nearest-cloud above

land and above ocean were separated and compared for different latitudes (with a

latitudinal resolution of 1°). The calculated latitudinal mean CFF, and the latitudinal

mean distance-from-the-nearest-cloud are presented in Figure ‎3.1.

36

Figure ‎3.1 (Fig. 4 in Bar-Or et al., 2011) Latitudinal mean cloud fraction (left panel,

lines), cloud field fraction (left panel, dots) and distance from the nearest cloud (right

panel), above land (red) and ocean (blue), based on MODIS Terra observations for

July 28th

, 2008.

The results show a clear difference between cloud fields above oceans and lands.

While the CFF above oceans is 80% or higher in all latitudes, the CFF above land

carries a strong signature of the global atmospheric circulation. The CFF above land

significantly decreases in the Hadley subsidence (desert belt) latitudes (10°S-25°S and

20°N-35°N). The northward shift of the Hadley cells is expected considering the date

of observed data, during the boreal summer (July 28th

, 2008). Moreover, a closer

examination of the differences between the CF and the CFF curves shows that while

the CFF over land has a similar trend to the CF, the CFF over oceans is uncorrelated

to its corresponding CF; this indicates that the spatial cloud field structures over land

significantly differ from the structures over oceans.

37

In spite of the low latitudinal variance of the distance-from-the-nearest-cloud

parameter over oceans, the mean latitudinal CFF reveals a clear difference between

the Hadley subsidence latitudes (10°S-25°S and 20°N-35°N) and the Inter Tropical

Convergence Zone (ITCZ, 5°S-15°N). This transition is probably a result of the

difference between the spatial properties of the marine cloud fields over the ITCZ

(mostly deep convective), and the cloud fields in the subsidence zone latitudes

(mostly shallow clouds). The ITCZ‟s distinguished marine CFF behavior

demonstrates the ability of the algorithm to indentify different cloud fields by their

distance probability distribution, in spite of their similar mean distance-from-the-

nearest-cloud values.

Furthermore, these results indicate that while the likelihood to sample a total

cloud-free (away from cloud fields) pixel above oceans is ~12%, the likelihood to

sample such a pixel above land varies between ~10% in the ITCZ latitudes, and ~80%

in the central latitudes of the Southern subsidence zone (around 20°S, in the observed

day).

3.2 Aerosol retrieved optical properties in cloud fields

A daily global dataset is used to show these effects, separately above land and

ocean. Figure ‎3.2 presents the mean aerosol optical depth (AOD) as a function of the

distance-from-the-nearest-cloud, showing a clear exponential decay. The higher mean

AOD values above land (compared to the oceans) are expected, because of the higher

aerosol concentrations observed above land, and agree with past research [Remer et

al., 2008].

In addition to the analysis presented in Bar-Or et al., [2010], where Cirrus cloud

fields are excluded from the input data, a global analysis of all cloud types is

included. The optical contribution of the twilight zone around Cirrus clouds is

considered in this global analysis because undetectable Cirrus clouds often appear in

the vicinity of detectable clouds and affect the aerosol optical retrieval.

The global mean AOD as a function of the distance-from-the-nearest-cloud is

found to be monotonically decreasing (Figure ‎3.2). An interesting finding is the clear

38

change in the decrease rate, around 30 km distance from the nearest cloud. Given the

large size of the dataset (more than 1 million pixels), this sharp rate change may point

to a characteristic influence scale of cloud fields, supporting the findings presented in

Section ‎3.1, and previous studies [Koren et al., 2007; Remer et al., 2008; Twohy et al.,

2009a]. The exponential decay found here agrees with previous findings and the

coefficients presented here agree with the values calculated by Bar-Or et al. [2010].

The global mean AOD values found here are also supported by the MODIS-Terra

long-term global analysis of Remer et al. [2008] that found a global mean AOD of

0.19 above land, and of 0.13 above Ocean.

Figure ‎3.2 (Fig. 5 in Bar-Or et al., 2011) Global mean aerosol optical depth (AOD)

as a function of the distance from the nearest cloud, retrieved above land (red dots)

and above ocean (blue dots), and the matching exponential fitting functions (purple

and green lines, respectively), for distances of 0-30 km from the nearest cloud. The

error bars represent the standard mean error. All data are based on MODIS Terra

observation for July 28th

, 2008. The constant distance lines highlight the values for

R0=10 km and for the calculated global characteristic value (R0=30 km, see Section

‎3.1).

39

The mean FMF values retrieved above ocean as a function of the distance-from-

the-nearest-cloud is presented in Figure ‎3.3. The results clearly show two regimes in

the graph. The first, in the range of distances between 0~33 km from the nearest

cloud, that shows an exponential increase of the FMF with the distance from the

nearest cloud, varying from 37.1% at 0-1 km to 53.4%, 32-33 km from the nearest

cloud. The calculated exponential fit in this regime saturates at FMF of ~54%. The

second regime in the graph is for distances that are larger than ~33 km that shows a

slow decrease of the FMF when increasing the distance from the nearest cloud (in the

range of 53.4%-51.7%).

Figure ‎3.3 (Fig. 6 in Bar-Or et al., 2011) Global mean aerosol fine-mode fraction

(FMF) as a function of the distance from the nearest cloud, retrieved above ocean

(blue dots), and its exponential fitting function (green line), for distances of 0-30 km

from the nearest cloud. The error bars represent the standard mean error. All data are

based on MODIS Terra observation for July 28th

, 2008. Similarly to Figure ‎3.2, the

constant distance lines highlight the values for R0=10 km and for the calculated global

characteristic value.

40

The results of the first regime in the graph can be explained by the theoretical

superposition of two effects. The first is the aerosol swelling process that produces

sharp exponential decay in the aerosol size, as the distance to the nearest cloud

increases [Pahlow et al., 2006]. The second is the effect of undetectable clouds that

increase the retrieved aerosol apparent size.

Both AOD and FMF trends in cloud fields strengthen the need in distinguishing

between measurements of aerosols inside and outside cloud fields, because of the

significant difference in aerosol properties and its measured optical characteristics

near detectable clouds.

3.3 Radiative effects of aerosol humidification on cloud fields

3.3.1 RH spatial distribution in cloud fields

The RH simulated data of cases A1 and B (see Section ‎2.4.1) are fitted to

exponential RH(dc) functions, as described by Equation ‎2.1. The extracted mean

background relative humidity (RH0) values, based on all cloud containing layers in

cases A and B respectively, are 64.9% and 59.7%, and the extracted mean exponential

distance scale (δ) values are 85 m and 93 m.

In both cases, the lower RH0 values are mostly attributed by calculations of RH

change around tops of clouds within the inversion layer. In this layer the background

RH is relatively low (due to the sharp decrease in RH with altitude), and increases

rapidly when approaching a cloud that happened to break into the inversion layer.

This may result in the relative low RH0 values found. The higher atmospheric layers

drag the mean RH0 function to lower RH0 values, and shorter exponential distance

scale (δ) values. A counter example is shown when the analysis in case A1 is done

only with the lower atmosphere (below 2.6 km), which includes marine boundary

layer cumulus clouds in a moist environment. The mean RH0 value is higher (78.4%),

and the derived exponential distance scale (δ) is longer (87.5 m). Figure ‎3.4 presents

the derived RH(dc) functions based on filtered case A1, and on case B.

41

Figure ‎3.4 Parameterization of the mean relative humidity as a function of the

distance from the nearest cloud - RH dc (thick lines) and the corresponding standard

deviation (shadowed area), as estimated by LES simulation A1 (pink), and B

(yellow), as described in Section ‎3.3.1 and Equation ‎2.1. The presented data for

simulation A includes only the lower atmosphere (below 2.6 km), while simulation B

includes all altitudes.

The simulated results agree with in-situ measured relative humidity values in the

vicinity of clouds (RH(dc)). Twohy et al. [2009a] presented RH(dc) air borne

measurements sampled in warm trade cumulus cloud field, during INDOEX

campaign [Clarke et al., 2002]. Fitting the results of Twohy et al. [2009a] to Equation

‎2.1 finds exponential distance scale (δ) values of 98 m – 278 m , and background

relative humidity (RH0) values of 87%-89%. Additional results, from Wang and

Geerts [2010], averaged the specific humidity as a function of the distance from cloud

edge qv(dc), sampled in warm trade Cumulus field, during the RICO campaign

42

[Rauber et al., 2007]. Fitting these results to Equation ‎2.1 (assuming curve trend

similarity of qv (dc) and RH dc functions) extracts B values of 106 m – 120 m.

3.3.2 Humidified aerosol properties in cloudy environment

Using SHDOM radiative transfer model [Evans, 1998], extensive sensitivity

simulations calculate both aerosol optical depth (AOD) and aerosol fine-mode

fraction (FMF), as a function of the distance from the nearest cloud (dc). The

simulations are performed for bimodal log-normal aerosol distributions, and the

varying parameters are the physical and optical properties of the fine-mode aerosols

(see Table 2). All sensitivity simulations used coarse-mode of sea salt aerosols (sets

S1-S4 in Table 2), or of desert dust aerosols (sets D1-D4 in Table 2).

Figure ‎3.5 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (dc),

for relative humidity exponential parameterization RH dc that vary between sloped

(lowest background values, green), mean (red), and mild (highest background values,

43

blue), as described in Section ‎3.3.1. The simulated bimodal lognormal distribution

contains coarse mode NaCl, and fine-mode biomass burning aerosols (set R1 in Table

2).

Figure ‎3.6 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (dc),

for fine-mode hygroscopicity parameter κfi that vary between 0-1.1. The simulated

bimodal lognormal distribution contains coarse-mode NaCl, and fine-mode biomass

burning aerosols (set S1 in Table 2).

44

Figure ‎3.7 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (dc),

for fine-mode mass content mfit values that vary between 5 − 50 μg ∙ m−3. The

simulated bimodal lognormal distribution contains coarse-mode NaCl, and fine-mode

biomass burning aerosols (set S2 in Table 2).

45

Figure ‎3.8 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (dc),

for fine-mode mass content mfit values that vary between 5 − 50 μg ∙ m−3. The

simulated bimodal lognormal distribution contains coarse-mode NaCl, and fine-mode

biomass burning aerosols (set S3 in Table 2).

46

Figure ‎3.9 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (dc),

for fine-mode imaginary part of the refractive index Im Reffi values that vary

between 0 and 1. The simulated bimodal lognormal distribution contains coarse-

mode NaCl, and fine-mode biomass burning aerosols (set S4 in Table 2).

The simulation results show that aerosol optical depth and fine-mode fraction

trends near clouds are more sensitive to the hygroscopicity of the participating

aerosols (simulation sets S1 and D1 in Table 2) than to the mass content ratio between

fine-mode and coarse-mode aerosols (mfi mco ). The second most effective parameter

is the fine-mode aerosol geometric radius (rg,fi, sets S3 and D3 in Table 2); this

sensitivity is mainly due to the characteristic size scale of the fine-mode aerosol,

which its interaction with light at the wavelength of 0.550 μm is more efficient than

the coarse-mode's interaction.

The curve trend of FMF(dc) is found to significantly vary with the suspended

aerosol properties. The simulation results (summarized in Table 3) point to a delicate

47

balance between the two represented aerosol modes, which may be affected by their

hygroscopicity properties, and by their radiative interaction in the given wavelength.

The results of simulation sets D1-D4 (Table 2), as described in Section ‎3.3.2, are

presented in Appendix A. This set of sensitivity simulations is performed for different

fine-mode aerosol properties, while the coarse-mode aerosol properties are kept

constant with the values commonly used for desert dust aerosols (Dubovik et al.,

2002). Figure ‎0.1 presents the sensitivity of both 𝐴𝑂𝐷(𝑑𝑐) and 𝐹𝑀𝐹(𝑑𝑐) to the fine-

mode aerosol hygroscopicity parameter 𝜅𝑓𝑖 , for the value range between 0-0.7 (set D1

in Table 2), which is a possible hygroscopicity range for different biomass burning

aerosols (Petters et al., 2009; Carrico et al., 2010). The results agree with the findings

presented in Figure ‎3.6 (set S1 in Table 2), but show that 𝐹𝑀𝐹(𝑑𝑐) switches shape

only for less hygroscopic fine-mode aerosols (𝜅𝑓𝑖 ≲ 0.03). The 𝐹𝑀𝐹 decreases near

cloud edges, unlike the findings of set S1, when using sea-salt aerosols as coarse-

mode. A possible reason for the lower 𝜅𝑓𝑖 threshold for the 𝐹𝑀𝐹 𝑑𝑐 shape switch

may be the lower hygroscopicity of the coarse mode in this set (𝜅𝑐𝑜=0.03), which

cancels the coarse-mode 𝐴𝑂𝐷 increase contribution in the vicinity of clouds, unless

the fine-mode's hygroscopicity effect is smaller.

The sensitivity of the simulated AOD(dc) and FMF(dc) curve shapes to the fine-

mode aerosol mass content mfi, combined with coarse-mode desert dust (set D2 in

Table 2), agrees with the results of set S2 (Figure ‎3.5), with sea-salt aerosol. We find

no sensitivity of AOD(dc) and FMF(dc) curve shapes to the fine-mode mass content.

Figure ‎0.2 presents the sensitivity of AOD(dc) and FMF(dc) to the fine-mode

aerosol geometric mean radius rg , while considering coarse-mode desert dust aerosols

(set D3 in Table 2). We find that unlike the equivalent sensitivity simulation done

with sea-salt (set S3 in Table 2), there is no sensitivity of AOD(dc) and FMF(dc)

curve shapes to changes in the fine-mode geometric mean radius rg . This behavior is

most likely due to the low hygroscopicity of desert dust (κco = 0.03). The low

hygroscopicity of the coarse-mode aerosols, compared to that of the fine-mode

aerosols, increases the relative contribution of the fine-mode aerosols to any optical or

physical change of the whole aerosol distribution in a high RH environment. Hence,

48

when the coarse-mode aerosols are non-hygroscopic, the fine-mode aerosols are more

dominant in changing the total aerosol optical contribution near clouds.

The simulation set results presented in Figure ‎0.3 show the sensitivity of

AOD(dc) and FMF(dc) to the changes in the values of the fine-mode aerosol

absorption ability, via the imaginary part of the refractive index Im Reffi , with

coarse-mode desert dust aerosols (set D4 in Table 2). Unlike the results of the

equivalent simulation, done with coarse-mode sea salt aerosols (set S4 in Table 2),

the AOD(dc) and FMF(dc) curve shapes appear to be insensitive to changes

in Im Reffi . This finding may be explained by the low hygroscopicity of desert dust

(κco = 0.03), which decreases the relative influence of the coarse-mode aerosols in

high relative humidity environments, enabling the fine-mode aerosol hygroscopic

growth to drive both AOD(dc) and FMF(dc) in the vicinity of clouds.

3.3.3 Absorbing humidified aerosol properties in cloud fields

The optical behavior of different absorbing aerosol types is explored in vicinity of

clouds, using the Spherical Harmonic Discrete Ordinate Method model (SHDOM,

Evans, 1998) for atmospheric radiative transfer. The simulations focus on the change

in total extinction (in units of 𝑘𝑚−1), the single scattering albedo (ω), and the

asymmetry parameter (g) as a function of distance from the nearest cloud (dc) at

wavelengths of 532 nm and 355 nm.

In each simulation the hygroscopicity parameter κ [Petters and Kreidenweis,

2007], the real part of the dry aerosol complex refractive index, the dry effective

diameter (Deff-dry), assuming a single-mode log-normal distribution with a ln(σ) = 0.7,

and the mass of the aerosol layer, set at 5 μg m-3, are kept constant while varying the

imaginary part of the dry aerosol RI. Moreover, two different RH fields are used in

the model: one typical for the marine boundary layer (MBL), where the background

RH reaches ~88%, and the other describing continental cumulus cloud field (CCF)

where the background RH reaches 60%. Both RH fields are extracted from large-

eddy-simulation model results (UCLA-LES, Xue and Feingold, JAS, 2006, and

RAMS6, [Cotton et al., 2003], and supported by in-situ measurements [Twohy et al.,

49

2009a; Wang and Geerts, 2010]. The κ values used are κ=0.6, typical of ammonium

sulfate [Petters and Kreidenweis, 2007] and κ=0.15, typical of organic aerosols

[Petters et al., 2009]. The real part of the refractive index (RI) and the dry effective

diameters used are n = 1.504 and n = 1.626, and Deff-dry=0.1 μm and Deff-dry=0.2 μm.

The imaginary component of the RI for the dry aerosols was varied from 0.0 to 0.4 in

0.05 steps. The results for the extinction as a function of distance from the nearest

cloud (dc) are shown in Figure ‎3.10, for SSA (ω) vs. dc in Figure ‎3.11, and for the

asymmetry parameter vs. dc in Figure ‎3.12. All figures show 4 pairs of graphs, where

within each pair all variables are the same except the RH field; left side MBL and

right side CCF. The top graphs (a-MBL, a-CCF, b-MBL, b-CCF) show the

simulations for a Deff-dry = 0.1 μm, where the lower graphs (c-MBL, c-CCF, d-MBL,

d-CCF) for a Deff-dry = 0.2 μm. The differences within the top and lower graphs are the

κ and n values. The left graphs have a κ = 0.6 and n=1.504, where the right graphs

have a κ = 0.15 and n = 1.626.

Figure ‎3.10 shows that the extinction is practically independent from the

imaginary component of the complex refractive index, and that for a Deff-dry = 0.2 μm

there is even no spectral dependence. Furthermore, the greatest differences in

extinction occur in the first 50 m near the cloud edge, with a steep exponential

increase. Further away, the extinction remains practically constant for all of the CCF

cases, and it decreases slightly for the MBL cases; a maximum of 0.07 km-1 for the a-

MBL case. For the MBL case study the differences in extinction are dominated more

by the size of the effective dry diameter than by differences in κ or n, the results show

that between a-MBL and b-MBL the shape and value of the extinction are basically

the same (as well as between c-MBL and d-MBL). This suggests that the dry size

distribution of the aerosols present in the twilight zone is the dominant factor in the

total extinction. For the CCF cases, a greater difference is observed in the extinction

between the constant Deff-dry with different κ and n; where below an RH of 80% the

real part of the RI dominates. For example, there is around a 0.01 km-1 difference in

extinction between a-CCF and b-CCF, which means that for an aerosol layer of 5 km

there is an optical depth difference of 0.05 which is small but not negligible.

On the contrary to the extinction as a function of dc, a clear difference is shown

in the single scattering albedo (Figure ‎3.11) and asymmetry parameter (Figure ‎3.12)

50

vs. dc, for different degrees of absorption of the present aerosols. Figure ‎3.11 shows

that in the first 50 m from the cloud edge there are significant differences between the

highly absorbing (k=0.4) and lightly absorbing (k = 0.05) aerosol. Within this

distance, the single scattering albedo may decrease down to ω = 0.45 for an imaginary

component of k = 0.4, and down to ω = 0.75 for k = 0.05 at a wavelength of 355 nm

(see d-CCF) and 532 nm (see b-CCF). The value of ω for a specific imaginary

component at different dc‟s varies for each of the 8 cases presented. For the MBL

cases, there is a constant decrease in ω as the dc increases, whereas for the CCF cases

after the first 100 m the single scattering albedo remains constant. Another distinct

feature in the behavior of ω is that it is always lower at 355 nm than at 532 nm for the

same k value, with the exception of the k = 0.4 value of the b-CCF case. For the Deff-

dry = 0.2μm cases the differences between the wavelengths is greater than at Deff-dry =

0.1μm.

Figure ‎3.10 Extinction as a function of distance from the nearest cloud (dc), for four

different scenarios: a) the hygroscopicity parameter (κ) set at 0.6, the real part of the

refractive index (n) set at 1.504, and the dry effective diameter (Deff-dry) set at 0.1 μm;

51

b) κ=0.15, n=1.626, and Deff-dry= 0.1 μm; c) κ=0.6, n=1.504, and Deff-dry= 0.2 μm; d)

κ=0.15, n=1.626, and Deff-dry= 0.2 μm. For each scenario two different relative

humidity fields were used: one typical for the marine boundary layer (MBL), and the

other describing a continental Cumulus cloud field (CCF). Two wavelengths: 355 nm

(solid lines) and 532 nm (dashed lines), are shown. The imaginary component (color

scale) was varied in each case from 0 to 0.4.

Figure ‎3.11 Single scattering albedo (ω) as a function of distance from the nearest

cloud (dc), for the same scenarios as in Figure ‎3.10. See Figure ‎3.10 caption for full

description.

52

Figure ‎3.12 Asymmetry parameter (g) as a function of distance from the nearest

cloud (dc), for the same scenarios as in Figure ‎3.10 and in Figure ‎3.11. See Figure

‎3.10 caption for full description.

The asymmetry parameter results (Figure ‎3.12) suggest that the light is

predominantly scattered to the forward direction for both wavelengths, with generally

being larger for 355 nm than for 532 nm for the same k value (see c-MBL for the few

exceptions). Furthermore, the asymmetry parameter goes from a minimum value for

purely scattering aerosols, with values as low as g = 0.66 at 532 nm (b-CCF), to a

maximum value for the highly absorbing aerosols; i.e., as the imaginary component is

increased the scattered light is directed more in the forward direction. Furthermore,

there is no clear pattern that describes the changes of the asymmetry parameter with

dc.

53

3.4 Upper-air measurements of the RH spatial distribution in cloud fields

3.4.1 RH mean vertical profile in potentially cloudy layers

A long record of upper-air measurements between 1980 and 2011 is examined,

for the stations listed in Table 4, using the methods described in Section ‎2.5. The

chosen season for this analysis are the month June-July-August. We select a short

season in order to avoid biases due to sharp changes in the meteorological

characteristics. Moreover, June-July-August are tested to be meteorologically stable

for most station, showing small variances in the cloudy layer heights, the 500 hPa

height, and the 850 hPa and 500 hPa temperatures.

Figure ‎3.13 shows the daily mean heights of the calculated cloudy layer (using

the SALR bounding method, described in Section ‎2.5), as observed in Hilo, Hawaii

(PHTO, see Table 4) between 1980 and 2011. The seasonal variability in this station

is clearly small, as expected by a tropical oceanic station, so this station is found

suitable for the initial assessment of the RH mean vertical profile in cloud layers.

54

Figure ‎3.13 Daily mean cloud layer base (red line) and top (blue line), calculated

from 32 year long upper-air measurement record over Lihue, Hawaii (see Table 4), at

00:00 UTC.

The mean vertical profiles for the Hawaiian station (PHTO, see Table 4),

between June and August, are presented in detail in Figure ‎3.14. First, the observed

vertical probability distribution of cloudy layer sampling (Figure ‎3.14, left panel)

shows a maximum of 87% at the altitude of 1700 m above sea level. This value

indicates a cloud-rich lower atmosphere, as predicted for a humid maritime tropical

station.

Figure ‎3.14 Vertical profiles of the sampled cloud layer fraction (left panel), the

mean SRH values (right panel, black line), and the mean SRH values that are in cloud

layers, but not inside clouds (right panel, blue line). The gap between each dashed line

pair represents two standard deviations (one for each direction). The analyzed data are

55

all 00:00 UTC radiosonde observations of Hilo, Hawaii (see Table 4), between June

and August, from 1980 to 2011.

The observed mean SRH vertical profile of this specific station (Figure ‎3.14,

right panel, black line) shows a significant decrease in the SRH values in the free

atmosphere, suggesting that the relative humidity drops rapidly and reaches extremely

low values above ~2500 m above sea level, even above tropical oceanic areas. The

RH mean vertical profiles within cloudy layers (Figure ‎3.14, right panel, blue line) is

always significantly more humid than the general mean SRH profile, but still shows a

sharp decrease above ~2500, suggesting that the mean SRH values within the lower

(by altitude) cloudy layers (outside clouds), is not sufficient for aerosol radiative

effects due to hygroscopic growth, according to our findings in section ‎3.3.

Similar analyses are conducted on all 14 station data (Appendix B), showing

various SRH vertical profiles with different cloudy layer statistical and seasonal

features.

Since most aerosols are concentrated in the lower troposphere (95% up to 2 km

from surface [Blanchard and Woodcock, 1980], with the exception of long-range

aerosol transport, further analysis focuses on the lower cloudy troposphere (LCT),

which is introduced in this study, for the first time, as the atmospheric layer that

inhibits most aerosols and has the largest contribution to cloud-aerosol interactions.

Figure ‎3.15 summarizes the SRH standard deviation vs. the mean SRH, in the

LCT of all 14 stations during June-July-August for LCT upper height limits of 1 km,

2 km and 3 km above surface. A clear negative trend between the SRH and its

standard deviation is shown for all cases, with a slope that ranges between -0.317

(lowest 2 km) and –0.372 (lowest 1 km). Specifically, the results show that most

maritime stations are characterized by relatively high mean SRH values in the LCT,

accompanied by low SRH standard deviation values, and the vice versa for most of

the continental stations. This trend is kept along the lowest 1 km, 2 km and 3 km, with

lower mean SRH values, and higher standard deviation values for increased LCT

layer thickness.

56

Figure ‎3.15 The mean SRH standard deviation as a function of the mean SRH, as

calculated for the lower cloudy troposphere (LCT) of 14 globally distributed

continental (circles) and maritime (square) stations, using LCT upper height limits of

1 km (upper left), 2 km (lower left), and 3 km (lower right). The red dashed lines of

each panel represent the linear fit. All presented data are based on day time

measurements during the months June-July-August.

3.4.2 SRH profile and cloud development

In order to further explore the possible effects of SRH in the lower cloudy

troposphere on the observation of cloud-aerosol interactions, we estimate the SRH

values as a function of the level of the cloud development. Since a positive correlation

in expected between cloud horizontal and vertical dimensions [Wang and Rossow,

1995; Koren et al., 2010] and both of these are correlated with the cloudy layer

thickness (between the LCL and the SALR), a thicker cloudy layer is likely to

represent cases of vertically developed clouds and high cloud fraction.

For each station, all radiosonde profiles are sorted by their cloudy layer

thickness (measured from the LCL to the SALR upper limit). The profiles are

57

separated into two equally number of samples of thicker and shallower sets for each

station, classified by the cloudy layer thickness median. The cloudy layer thickness

values range from 859 m to 2266 m in the maritime stations with median values

between 481 m and 1900 m, and from 829 m to 2789 m in the continental stations

with median values of 500 m and 2929 m (Table 5). Note that especially in case of

convective clouds, these thicknesses do not capture the whole extent of the clouds but

the conditionally unstable layer of their profiles.

Figure ‎3.16 shows the difference between the mean SRH values in the lower

cloudy troposphere of developed and shallow cloud profiles (Δ(𝑆𝑅𝐻)𝐿𝐶𝑇), as a

function of the mean SRH values in the lower cloudy atmosphere (𝑆𝑅𝐻𝐿𝐶𝑇), for LCT

upper height limits of 1 km, 2 km and 3 km above surface.

Figure ‎3.16 The difference between the mean SRH values in the lower cloudy

troposphere of developed and shallow profiles (Δ(𝑆𝑅𝐻)𝐿𝐶𝑇), as a function of the

58

mean SRH values in the lower cloudy atmosphere (𝑆𝑅𝐻𝐿𝐶𝑇), for 14 stations, in the

months of June-July-August, during 1980-2011, for LCT upper height limits of 1 km

(upper panel), 2 km (middle panel) and 3 km (lower panel) above surface.

The results show that the differences of the mean SRH in the lower cloudy

troposphere between developed and shallower cases are around 5% (Figure ‎3.16, and

Table 5 in the supplementary material). A calculation shows that hygroscopic aerosols

(e.g. sea salt aerosols with 𝜅 = 0.7 experience geometric cross section difference of

19.5% when their environment RH value changes from 80% to 85%, and experience

AOD change of 18.2% when the RH value changes from 60% to 70%. Non-

hygroscopic particles (e.g. biomass burning aerosol – 𝜅 = 0.1) gain geometric cross

section difference of 7.8% for RH increase of 80% to 85%, and AOD change of 4.8%

for RH increase of 60% to 70%. The geometric cross section difference is a first order

approximation to the total expected AOD change, neglecting the modification of the

refractive indices due to water uptake, and the aerosol size parameters.

59

4 Summary and discussion

The recent recognition of the importance of the transition zone between detected

clouds and cloud-free atmosphere (the twilight zone, see Section ‎1.4) upraised

questions regarding its nature and its climatic implications. Some of these questions

deal with issues as How to determine a cloud field that includes the twilight zone?

What are the optical properties of the whole cloud field and of its twilight zone? and

how are the aerosol properties and retrievals affected when located in the vicinity of

clouds? This research was aimed at the determination and exploration of a cloud field

as an entity, including clouds and aerosol. We studied clouds and aerosols spatial

features and radiative properties related to their location in the field and with respect

to the RH spatial variation.

In this study, the twilight zone was considered as an integral part of a cloud field,

and the observed atmosphere was classified into cloud fields (detectable clouds

surrounded by twilight zone) and cloud-free. The new offered definition had required

a suitable analytical tool that enables quantitative research of the atmosphere under

the new classification. This tool was presented in this study, namely the cloud field

masking algorithm (Section ‎2.1), and was validated as a robust method for defining

the boundaries of any cloud field, based on its detectable clouds' spatial distribution.

The core assumption behind the cloud field masking algorithm is that every cloud

field has a characteristic length scale that represents the maximal distance between

clouds, defining the cloud spatial distribution inside the field. This characteristic scale

was named here “the distance parameter (𝑟0)”, and its calculation was done by

operating the Euclidian distance transform on the detected cloud mask, followed by a

discrimination between the inner field and extra-field cloud distributions. The exact

distance that marked the boundary between the distributions was set as 𝑟0, and

enabled the spatial bounding of the observed cloud field. Detailed graphic description

is found in Figure ‎2.1.

Analytically masking cloud fields also enabled the presentation of a new metric to

evaluate cloud field coverage. We have named this metric “the cloud field fraction”

60

(CFF), and similar to the well-known cloud fraction (CF) metric, it was defined as the

ratio between the cloud fields covered area to the area of the whole observed domain.

The cloud field masking algorithm was used for the estimation of the global cloud

field coverage for a single selected day. A global mosaic of 1 km cloud mask data, as

observed by MODIS Terra satellite was analyzed in order to extract the global CFF,

and the mean 𝑟0values for different cloud field types. The total observed global cloud

field coverage was 87%, while the observed cloud fraction for that day was only 51%

(Section ‎3.1). This finding emphasizes the low likelihood to randomly sample a cloud

field free atmosphere above Earth. It also stresses out the need to better quantify the

optical and the physical properties of aerosols which are located in cloud fields as

they are affected by the humid environment there.

It was also found that on a global average cloud fields extend ~30 km from the

detectable cloud edge (𝑟0 ≅ 30 𝑘𝑚), while different types of cloud fields experience

different 𝑟0 values in accordance to their typical spatial distribution of clouds (Table

1). The 30 km scale, found here using a morphological analysis, supports previous

studies that found similar distance scales using different types of analysis on remote-

sensing data [Koren et al., 2007; Bar-Or et al., 2010]. A closer examination of the

calculated data showed a clear latitudinal dependency of the mean CF, the mean CFF

and the mean distance from clouds (𝑑𝑐), above lands and oceans. It was found that a

high probability to sample a cloud-free pixel is limited only to some desert areas

under the Hadley subsidence zone (Figure ‎3.1).

The global MODIS 1 km resolution data set of distance from cloud (𝑑𝑐) was then

used in order to observe the trends in the aerosol optical properties in the vicinity of

clouds. Based on global MODIS aerosol product for the matching day, it was shown

that both aerosol optical depth (AOD) and the aerosol fine-mode fraction (FMF)

strongly depend on the sample‟s distance from the nearest detectable cloud (𝑑𝑐).

While the AOD and 𝑑𝑐 has a negative exponential trend (Figure ‎3.2), the FMF and 𝑑𝑐

has a positive exponential trend (Figure ‎3.3, Section ‎3.2). Moreover, both AOD and

FMF retrievals showed a scale break in the distance of ~30 km from the nearest cloud,

which may point to a natural distance scale of cloud fields. The consideration of this

finding, with the observed vast global cloud field coverage, may be used for better

61

future assessment of cloud field radiation budgets. The addition of cloud distance

dependent aerosol optical depth into large scale (coarse resolution) numerical models

following the findings presented here, is able to improve the estimation of aerosol

radiative forcing, which is currently poorly understood [Forster et al., 2007],

An attempt to explain the aerosol retrieval biases in clouds' vicinity raises three

main features that affect the optical and physical properties of aerosols in this zone:

(1) aerosol hygroscopic growth (aerosol humidification) due to high RH values near

clouds [Feingold and Morley, 2003; Twohy et al., 2009a; Quaas et al., 2010], (2)

signal contribution by clouds which are too small or too optically thin to be identified

as clouds, and (3) 3D radiative effects from light scattered by neighboring clouds

[Marshak et al., 2006; Varnai and Marshak, 2009]. For better understanding of the

physical mechanisms that cause the twilight zone‟s radiative effects, we isolated the

aerosol humidification component and investigated its net contribution.

For this purpose, a new detailed parameterization of the relative humidity (RH)

values as a function of the distance from clouds (𝑑𝑐) was presented (Section ‎3.3.1) for

warm Cumulus cloud fields (Sections ‎2.4.1 and ‎3.3.1). This parameterization was

developed using two different large eddy simulation models (LES), and supported by

the few in-situ observations that measured RH in respect to the closest cloud location

[Korolev and Isaac, 2006; Twohy et al., 2009a; Wang and Geerts, 2010]. In these

cloud fields, the RH was found to exponentially decrease as the distance from cloud

increases, with an e-fold distance scale of 98-278 m. Therefore, we conclude that the

RH high values range, that is relevant to significant aerosol humidification effect is

estimated to extend up to 0.5 km from the nearest cloud (Section ‎3.3.1), suggesting

that aerosol humidification contribution to biases in aerosol retrievals in the twilight

zone is not the dominant one in distances of 0.5-30 km from clouds.

Additionally to the aerosol humidification research implications, this RH

parameterization can be used to improve the representation of cloud fields‟ internal

properties in large scale numerical simulations, like in Global Circulation Models

(GCM). The lack of similar parameterizations forces GCM models to assume simple

spatial relative humidity distribution within grid cells, leading to unavoidable errors in

estimations of radiative properties of clouds and humidified aerosols. Although the

62

weaknesses of the present RH distribution were acknowledged by GCM studies

[Quaas et al., 2008; Quaas et al., 2010], it is still used [Quaas, 2012]. Therefore, the

new RH parameterization presented here may improve the representation of sub-grid

spatial humidity and cloud distributions in GCM's.

Using the new RH parameterization in a cloud field, a close examination of

aerosol humidification radiative effect near clouds, as a function of aerosol physical

and chemical properties was conducted (Section ‎3.3.2 and Section ‎3.3.3). This study

was carried out using an atmospheric radiative transfer model (SHDOM, Evans,

1998), to simulate the MODIS aerosol retrieval variance near clouds, due to aerosol

humidification. The results showed variant radiative behaviors of both AOD and FMF

as a function of the distance from clouds, suggesting that the net contribution of

aerosol humidification depends strongly on the aerosol properties, and not only on the

environmental conditions. This finding suggests that the overall trends of aerosol

optical properties near clouds, even in small distance from clouds (Section ‎3.2), are

not necessarily due to aerosol humidification. Further possible future application of

this result may be the retrieval of aerosol properties based on their observed optical

signatures and on their location in respect to clouds.

The next part of this research studied the RH characteristics inside cloud fields

(yet outside clouds), for better estimations of the radiative contribution of humidified

aerosols that will enable also a better representation of aerosol humidification effects

in climate models. For that purpose, an extensive 32 year long atmospheric sounding

record was analyzed and the sub-saturated relative humidity (SRH) vertical profiles

inside cloud fields were examined (Section ‎3.4). The results showed that the mean

sub-saturated (out of clouds) relative humidity value range (SRH), in the lower

troposphere of profiles that exhibit cloudy layer, is ~80% accompanied by a standard

deviation of ~7% for most maritime stations, and ~67% with standard deviation of

~14% for most continental stations. Furthermore, it was found that the mean SRH and

the SRH variance have a clear negative trend (the stations whose mean SRH is higher

are characterized by low SRH variance). The findings presented in Section ‎3.4.2

suggest that the measured SRH values in cloud fields, by their mean values, cannot

result in significant radiative effect due to aerosol humidification.

63

Finally, the dependence of SRH values in the cloud vertical development was

explored, for estimating the likelihood for biases of aerosol-cloud interaction study

conclusion due to aerosol humidification radiative effects. In this part of our study we

examined the probability to wrongly misinterpret trends between cloud and aerosol

properties as products of aerosol-cloud physical feedbacks, due to aerosol

humidification near clouds. We divided each station‟s data into developed and

shallow cloudy layers, using the median cloudy layer thickness, in order to examine

the dependence of the RH properties in cloud vertical development. The results

showed that the difference in range, in the mean SRH of the lower cloudy troposphere

between profiles with shallow and developed cloudy layers for all stations, is 1%-9%

for LCT of 1 km above surface, 3%-10% for LCT of 2 km above surface and 6%-14%

for LCT of 3 km above surface. This finding suggests that the mean SRH values in

the lower cloudy troposphere are relatively similar for different cloudy layer

thicknesses. Therefore, we conclude that observed trends between aerosol and cloud

properties are not likely to be a result of aerosol humidification near clouds. Thus, this

result supports past studies which examined physical aerosol-cloud feedback using

observed trends between aerosol and cloud properties in cloud fields [Koren et al.,

2005; Koren et al., 2008a; Heiblum et al., 2012; Koren et al., 2012].

This thesis explored some of the twilight zone‟s core questions; beginning with

the fundamentals of cloud and cloud field spatial definitions, through the observed

optical trends of aerosols near clouds, continuing with the isolation and assessment of

the radiative contribution of aerosol humidification in cloud fields. The outcomes of

this research are new tools for future study of cloud fields and aerosol-cloud

interactions, using simulations, observations and measurements. We believe that the

findings presented above improve our understanding of cloud field spatial and

radiative properties, enabling better evaluation of aerosol and cloud properties and

aerosol-cloud interactions. There is great potential in implementation of the new

methods and knowledge presented here in future climate research.

64

Tables

Table 1 (Table 1 in Bar-Or et al., 2011) Granule cloud fraction (CF), cloud field

fraction (CFF), and field distance parameters (R0) as calculated globally for July 28th

,

2008. The cloud fields are classified to: Cirrus (Ci), Stratocumulus (Sc), Cumulus

(Cu), and Deep Convective (DC). All values refer both to fields and granules,

respectively.

Mean Standard

deviation

Number

of

Samples

Granule Cloud Fraction 51% 23% 66

Granule CFF (Calculated R0) 88% 66

Granule CFF (R0=10km) 80% 23% 66

R0 – All 29 km 9 km 170

R0 – Ci 30 km 9 km 38

R0 – Sc 25 km 8 km 21

R0 – Cu 29 km 9 km 80

R0 – DC 31 km 8 km 31

65

Table 2 The input physical and optical aerosol properties, used for the results described in Section ‎3.3.2. Both fine-mode and coarse-mode

aerosol are assumed to have a log-normal distribution, characterized by the geometric mean radius 𝑟𝑔 , the log-standard deviation 𝜎, and the total

mass content 𝑚. The aerosol physical properties listed are the aerosol hygroscopicity parameter 𝜅, the aerosol complex refractive index 𝑅𝑒𝑓, and

the aerosol bulk density 𝜌. All values describe dry aerosol.

fine-mode aerosol coarse-mode aerosol

Set

𝑟𝑔 ,𝑓𝑖

𝜇𝑚

𝜎𝑓𝑖 𝜅𝑓𝑖 𝑅𝑒𝑓𝑓𝑖 𝜌𝑓𝑖

𝑔

𝑐𝑚3

𝑚𝑓𝑖

𝜇𝑔

𝑚3

𝑟𝑔 ,𝑐𝑜

𝜇𝑚

𝜎𝑐𝑜 𝜅𝑐𝑜 𝑅𝑒𝑓𝑐𝑜 𝜌𝑐𝑜

𝑔

𝑐𝑚3

𝑚𝑐𝑜

𝜇𝑔

𝑚3

R1 0.06 0.7 0.3 1.510-i·0.021 1.5 5 0.6 0.6 0.7 1.546-i·0.003 2.17 50

S1 0.06 0.7 0-1.1 1.510-i·0.021 1.5 5 0.6 0.6 0.7 1.546-i·0.003 2.17 50

S2 0.06 0.7 0.3 1.510-i·0.021 1.5 5-50 0.6 0.6 0.7 1.546-i·0.003 2.17 50

S3 0.06-0.3 0.7 0.3 1.510-i·0.021 1.5 5 0.6 0.6 0.7 1.546-i·0.003 2.17 50

S4 0.06 0.7 0.3 1.510-i·(0--1) 1.5 5 0.6 0.6 0.7 1.546-i·0.003 2.17 50

D1 0.06 0.7 0-0.7 1.510-i·0.021 1.5 5 2.3 0.6 0.03 1.550-i·0.002 2 50

D2 0.06 0.7 0.3 1.510-i·0.021 1.5 1-25 2.3 0.6 0.03 1.550-i·0.002 2 50

D3 0.04-0.3 0.7 0.3 1.510-i·0.021 1.5 5 2.3 0.6 0.03 1.550-i·0.002 2 50

D4 0.06 0.7 0.3 1.510-i·(0--1) 1.5 5 2.3 0.6 0.03 1.550-i·0.002 2 50

66

Table 3 Result summary for Section ‎3.3.2

Set examined

parameter

coarse-

mode

aerosol

Curve shape sensitivity of 𝑨𝑶𝑫(𝒅𝒄) and 𝑭𝑴𝑭(𝒅𝒄)

R1 𝑅𝐻( 𝑑𝑐 sea salt

𝐴𝑂𝐷(𝑑𝑐) and 𝐹𝑀𝐹(𝑑𝑐) curve shapes and values are

sensitive to the slope of the exponential

𝑅𝐻 𝑑𝑐 parameterization far from clouds 𝑑𝑐 >0.1 𝑘𝑚). In the close vicinity of clouds there is no

sensitivity to the 𝑅𝐻 𝑑𝑐 parameterization (Figure ‎3.5).

S1 𝜅𝑓𝑖 sea salt

𝐹𝑀𝐹(𝑑𝑐) curve shape is sensitive to the fine-mode

hygroscopicity. 𝜅𝑓𝑖 > 0.3 leads to increasing FMF near

clouds, while 𝜅𝑓𝑖 < 0.3 leads to decreasing FMF near

clouds (Figure ‎3.6).

S2 𝑚𝑓𝑖 sea salt

Both 𝐹𝑀𝐹(𝑑𝑐) and 𝐴𝑂𝐷(𝑑𝑐) curve shapes are not

sensitive to the fine-mode aerosol mass content 𝑚𝑓𝑖

(Figure ‎3.7).

S3 𝑟𝑔 ,𝑓𝑖 sea salt

The curve behavior of 𝐹𝑀𝐹 𝑑𝑐 is sensitive to the fine-

mode aerosol dry geometric mean radius, as a result of

the better scattering efficiency small aerosol in the

wavelength of 550 𝑛𝑚 (Figure ‎3.8).

S4 𝐼𝑚 𝑅𝑒𝑓𝑓𝑖 sea salt

Near clouds, where the hygroscopic growth is dominant,

𝐹𝑀𝐹(𝑑𝑐) and 𝐴𝑂𝐷(𝑑𝑐) curve shapes are not sensitive

to the fine-mode aerosol absorption

efficiency 𝐼𝑚 𝑅𝑒𝑓𝑓𝑖 . Far from clouds, the contribution

of fine-mode aerosol to the total AOD increases with its

absorption efficiency (Figure ‎3.9).

D1 𝜅𝑓𝑖 desert dust

Only non-hygroscopic fine-mode aerosol shows a

decrease of the 𝐹𝑀𝐹(𝑑𝑐) near clouds, due to the low

hygroscopicity of desert dust.

D2 𝑚𝑓𝑖 desert dust

In agreement with the equivalent simulation set S2, no

curve shape sensitivity was found in response to

changes in the fine-mode aerosol mass content 𝑚𝑓𝑖 .

D3 𝑟𝑔 ,𝑓𝑖 desert dust

Unlike the equivalent simulation set S3, with coarse-

mode sea salt, no sensitivity was found to the fine-mode

aerosol geometric mean radius. The fine-mode aerosols

are more effectively scattering and more hygroscopic,

and therefore drive the curve shapes of 𝐴𝑂𝐷(𝑑𝑐) and

𝐹𝑀𝐹(𝑑𝑐).

D4 𝐼𝑚 𝑅𝑒𝑓𝑓𝑖 desert dust

The low hygroscopicity of desert dust keeps the fine-

mode aerosol absorption efficiency 𝐼𝑚 𝑅𝑒𝑓𝑓𝑖 as the

main contributor of the curve shapes of 𝐴𝑂𝐷 𝑑𝑐 and

𝐹𝑀𝐹(𝑑𝑐).

67

Table 4 The selected 14 atmospheric sounding station names, the WMO station

number, the station geo location (latitude, longitude), the station elevation above sea

level (m), and the total number of all obtained vertical measurement profiles between

1980 and 2011.

Station

WMO

Station

#

Location

(lat , lon)

Elev.

(m)

# of all

profiles

Lihue, Hawaii 91165 21.97 , -159.35 45 31364

Hilo, Hawaii 91285 19.70 , -155.05 11 23178

Marshall Islands 91376 7.07 , 171.38 3 18990

Darwin, Australia 94120 -12.40 , 130.88 30 19276

Le Raizet, Guadeloupe 78897 16.26 , -61.51 11 7799

Lord Howe Island, Australia 94995 28.37 , 129.55 295 27938

Naze-Funchatoge, Japan 47909 -31.53 , 159.05 7 11436

Budapest, Hungary 12843 40.52 , -80.23 357 22488

Munich, Germany 10868 48.25 , 11.55 489 22919

Nashville, Tennessee 72469 36.25 , -86.55 210 23278

Manaus, Brazil 82332 -3.15 , -59.97 84 11796

Blacksburg, Virginia 72318 37.20 , -80.41 654 11484

Pittsburgh, Pennsylvania 72520 40.52 , -80.23 357 22488

Nairobi, Kenya 63741 -1.29 , 36.75 1798 14359

68

Table 5 The selected 14 atmospheric sounding station names, the WMO station number, the station geo location (latitude, longitude), the station elevation above sea level

(m), the total number of all obtained vertical measurement profiles between 1980 and 2011, the selected measurement time (UTC), the number of profiles in the selected

season (June-July-August) and hour, the number of cloudy-atmosphere profiles in the selected season and hour, the calculated mean LCL height above sea level for the

selected season and hour (m), and the LCL standard deviation for the selected season and hour (m).

Station WMO

Station #

Location

(lat , lon)

Elev.

(m)

# of all

profiles

Time

UTC

# of

selected

profiles

# of cloudy

selected

profiles

Mean

LCL

(m)

LCL

std

(m)

Lihue, Hawaii 91165 21.97 , -159.35 45 31364 00Z 3954 3774 877 257

Hilo, Hawaii 91285 19.70 , -155.05 11 23178 00Z 2887 2807 912 316

Marshall Islands 91376 7.07 , 171.38 3 18990 00Z 2765 2659 679 283

Darwin, Australia 94120 -12.40 , 130.88 30 19276 00Z 2857 1495 1265 720

Le Raizet, Guadeloupe 78897 16.26 , -61.51 11 7799 12Z 1051 1039 791 269

Lord Howe Island, Australia 94995 28.37 , 129.55 295 27938 00Z 2896 2734 1042 564

Naze-Funchatoge, Japan 47909 -31.53 , 159.05 7 11436 00Z 2618 2458 1217 647

Budapest, Hungary 12843 40.52 , -80.23 357 22488 12Z 2768 2236 1313 528

Munich, Germany 10868 48.25 , 11.55 489 22919 12Z 2781 2402 1985 940

Nashville, Tennessee 72469 36.25 , -86.55 210 23278 00Z 2880 2548 1713 945

Manaus, Brazil 82332 -3.15 , -59.97 84 11796 00Z 908 882 1055 385

Blacksburg, Virginia 72318 37.20 , -80.41 654 11484 00Z 1410 1224 2040 913

Pittsburgh, Pennsylvania 72520 40.52 , -80.23 357 22488 00Z 2782 2425 1938 1179

Nairobi, Kenya 63741 -1.29 , 36.75 1798 14359 12Z 1310 1239 3184 1132

69

Table 6 The selected 14 atmospheric sounding station names, and the analyzed properties for the selected hour (see Table 1), during the months June-July-August: the mean

relative humidity value in the lower cloudy troposphere (%), the standard deviation of relative humidity value in the lower cloudy troposphere (%), the SALR cloudy layer

mean depth (m), depth standard deviation (m), and depth median (m), the MBP cloudy layer mean depth (m), depth standard deviation (m), and depth median (m), and the

difference of the mean relative humidity values in the lower cloudy troposphere, between developed SALR cloudy layer (whose SALR is deeper than the median), and

shallow SALR cloudy layer (%).

Station

mean

𝑹𝑯𝑳𝑪𝑻

(%)

𝑹𝑯𝑳𝑪𝑻

Std

(%)

SALR

mean

depth (m)

SALR

depth

std

(m)

SALR

depth

median

(m)

MBP

mean

depth

(m)

MBP

depth

std

(m)

MBP

depth

median

(m)

𝚫(𝑹𝑯𝑳𝑪𝑻)

(%)

Lihue, Hawaii 80.90 7.36 984 660 950 1564 856 1425 3.37

Hilo, Hawaii 79.41 9.29 1168 865 1100 1707 939 1525 6.53

Marshall Islands 86.42 6.53 2266 1805 1900 4694 1747 4875 2.04

Darwin, Australia 65.65 15.15 722 1003 481 1259 890 1101 6.80

Le Raizet, Guadeloupe 79.95 7.63 1663 1194 1375 2971 1941 2450 4.33

Lord Howe Island, Australia 85.02 8.75 1202 1270 775 3309 1997 3100 5.30

Naze-Funchatoge, Japan 70.29 11.49 859 974 650 1599 1167 1300 5.02

Budapest, Hungary 77.75 13.98 1153 1258 725 2515 1679 2243 2.11

Munich, Germany 69.35 16.85 973 975 700 2458 1473 2325 2.33

Nashville, Tennessee 68.92 14.56 1932 1579 1625 2970 1900 2875 8.04

Manaus, Brazil 75.34 11.52 2789 1620 2929 4255 1675 4157 7.50

Blacksburg, Virginia 71.11 14.69 1454 1333 1050 2694 1801 2425 4.77

Pittsburgh, Pennsylvania 69.92 16.01 1408 1322 1050 2370 1696 2075 7.67

Nairobi, Kenya 75.00 13.23 829 956 500 1752 1081 1625 5.77

70

5 References

Ackerman, A. S., O. B. Toon, D. E. Stevens, A. J. Heymsfield, V. Ramanathan, and E. J.

Welton (2000), Reduction of tropical cloudiness by soot, Science, 288(5468), 1042-1047.

Ackerman, S. A., K. I. Strabala, W. P. Menzel, R. A. Frey, C. C. Moeller, and L. E. Gumley

(1998), Discriminating clear sky from clouds with MODIS, J. Geophys. Res.-Atmos.,

103(D24), 32141-32157.

Adler, J., and D. Hancock (1994), Advantages of using a distance transform function in the

measurement of fractal dimensions by the dilation method, Powder Technol., 78(3), 191-196.

Albrecht, B. A. (1989), Aerosols, cloud microphysics, and fractional cloudiness, Science,

245(4923), 1227-1230.

Altaratz, O., I. Koren, Y. Yair, and C. Price (2010), Lightning response to smoke from

Amazonian fires, Geophys. Res. Lett., 37, L07801.

Andreae, M. O., and D. Rosenfeld (2008), Aerosol-cloud-precipitation interactions. Part 1.

The nature and sources of cloud-active aerosols, Earth-Science Reviews, 89(1-2), 13-41.

Andreae, M. O., D. Rosenfeld, P. Artaxo, A. A. Costa, G. P. Frank, K. M. Longo, and M. A.

F. Silva-Dias (2004), Smoking rain clouds over the Amazon, Science, 303(5662), 1337-1342.

Bar-Or, R. Z., I. Koren, and O. Altaratz (2010), Estimating cloud field coverage using

morphological analysis, Environmental Research Letters, 5(1), 014022.

Bar-Or, R. Z., I. Koren, and O. Altaratz (2011), Global analysis of cloud field coverage and

radiative properties, using morphological methods and MODIS observations, Atmos. Chem.

Phys., 1, 191-200.

Bar-Or, R. Z., I. Koren, O. Altaratz, and E. Fredj (2012), Radiative properties of humidified

aerosols in cloudy environment, Atmospheric Research, 118(0), 280-294.

Blanchard, D. C., and A. H. Woodcock (1980), The production, concentration, and vertical

distribution of the sea-salt aerosol, Ann. N. Y. Acad. Sci., 338(1), 330-347.

Bolton, D. (1980), The computation of equivalent potential temperature, Mon. Wea. Rev.,

108, 1046-1053.

71

Borde, R., and H. Isaka (1996), Radiative transfer in multifractal clouds, J. Geophys. Res.-

Atmos., 101(D23), 29461-29478.

Cahalan, R. F., and J. H. Joseph (1989), Fractal statistics of cloud fields, Monthly Weather

Review, 117(2), 261-272.

Cahalan, R. F., W. Ridgway, W. J. Wiscombe, T. L. Bell, and J. B. Snider (1994), The albedo

of fractal stratocumulus clouds, Journal of the Atmospheric Sciences, 51(16), 2434-2455.

Carrico, C. M., M. D. Petters, S. M. Kreidenweis, A. P. Sullivan, G. R. McMeeking, E. J. T.

Levin, G. Engling, W. C. Malm, and J. L. Collett (2010), Water uptake and chemical

composition of fresh aerosols generated in open burning of biomass, Atmos. Chem. Phys.,

10(11), 5165-5178.

Charlson, R. J., A. S. Ackerman, F. A. M. Bender, T. L. Anderson, and Z. Liu (2007), On the

climate forcing consequences of the albedo continuum between cloudy and clear air, Tellus

Series B-Chemical and Physical Meteorology, 59(4), 715-727.

Clarke, A. D., et al. (2002), Indoex aerosol: A comparison and summary of chemical,

microphysical, and optical properties observed from land, ship, and aircraft, J. Geophys. Res.-

Atmos., 107(D19), 8033.

Coakley, J. A., R. L. Bernstein, and P. A. Durkee (1987), Effect of ship-stack effluents on

cloud reflectivity, Science, 237(4818), 1020-1022.

Cotton, W. R., et al. (2003), RAMS 2001: Current status and future directions, Meteorology

and Atmospheric Physics, 82(1-4), 5-29.

Davidi, A., A. B. Kostinski, I. Koren, and Y. Lehahn (2012), Observational bounds on

atmospheric heating by aerosol absorption: Radiative signature of transatlantic dust, Geophys.

Res. Lett., 39, L04803.

Durre, I., R. S. Vose, and D. B. Wuertz (2006), Overview of the integrated global radiosonde

archive, Journal of Climate, 19(1), 53-68.

Dybbroe, A., K. G. Karlsson, and A. Thoss (2005), NWCSAF AVHRR cloud detection and

analysis using dynamic thresholds and radiative transfer modeling. Part i: Algorithm

description, J, Appl, Meteorol,, 44(1), 39-54.

Evans, K. F. (1998), The spherical harmonics discrete ordinate method for three-dimensional

atmospheric radiative transfer, Journal of the Atmospheric Sciences, 55(3), 429-446.

72

Feingold, G., and B. Morley (2003), Aerosol hygroscopic properties as measured by lidar and

comparison with in situ measurements, J. Geophys. Res.-Atmos., 108(D11), 4327.

Feingold, G., H. L. Jiang, and J. Y. Harrington (2005), On smoke suppression of clouds in

Amazonia, Geophys. Res. Lett., 32(2), L02804.

Forster, P., et al. (2007), Changes in atmospheric constituents and in radiative forcing., in

Climate change 2007: The physical science basis. Contribution of working group i to the

fourth assessment report of the intergovernmental panel on climate change, edited by S.

Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller

Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Gasparini, R., D. R. Collins, E. Andrews, P. J. Sheridan, J. A. Ogren, and J. G. Hudson

(2006), Coupling aerosol size distributions and size-resolved hygroscopicity to predict

humidity-dependent optical properties and cloud condensation nuclei spectra, J. Geophys.

Res.-Atmos., 111(D5), 16.

Hansen, J., M. Sato, and R. Ruedy (1997), Radiative forcing and climate response, J.

Geophys. Res.-Atmos., 102(D6), 6831-6864.

Heiblum, R. H., I. Koren, and O. Altaratz (2012), New evidence of cloud invigoration from

trmm measurements of rain center of gravity, Geophys. Res. Lett., 39, L08803.

Hinds, W. C. (Ed.) (1999), Aerosol technology: Properties, behavior, and measurement of

airborne particles, 281–283 pp., New York.

Jiang, H., G. Feingold, and I. Koren (2009), Effect of aerosol on trade cumulus cloud

morphology, J. Geophys. Res.-Atmos., 114, D11209.

Jiang, H. L., H. W. Xue, A. Teller, G. Feingold, and Z. Levin (2006), Aerosol effects on the

lifetime of shallow cumulus, Geophys. Res. Lett., 33(14).

Jouzel, J., et al. (2007), Orbital and millennial antarctic climate variability over the past

800,000 years, Science, 317(5839), 793-796.

Kaufman, Y. J., D. Tanre, and O. Boucher (2002), A satellite view of aerosols in the climate

system, Nature, 419(6903), 215-223.

Kaufman, Y. J., I. Koren, L. A. Remer, D. Rosenfeld, and Y. Rudich (2005a), The effect of

smoke, dust, and pollution aerosol on shallow cloud development over the Atlantic Ocean,

Proc. Natl. Acad. Sci. U. S. A., 102(32), 11207-11212.

73

Kaufman, Y. J., et al. (2005b), A critical examination of the residual cloud contamination and

diurnal sampling effects on modis estimates of aerosol over ocean, Ieee Transactions on

Geoscience and Remote Sensing, 43(12), 2886-2897.

King, M. D., W. P. Menzel, Y. J. Kaufman, D. Tanre, B. C. Gao, S. Platnick, S. A.

Ackerman, L. A. Remer, R. Pincus, and P. A. Hubanks (2003), Cloud and aerosol properties,

precipitable water, and profiles of temperature and water vapor from MODIS, Ieee

Transactions on Geoscience and Remote Sensing, 41(2), 442-458.

Koehler, K. A., S. M. Kreidenweis, P. J. DeMott, M. D. Petters, A. J. Prenni, and C. M.

Carrico (2009), Hygroscopicity and cloud droplet activation of mineral dust aerosol, Geophys.

Res. Lett., 36, L08805.

Koren, I., and G. Feingold (2011), Aerosol-cloud-precipitation system as a predator-prey

problem, Proc. Natl. Acad. Sci. U. S. A., 108(30), 12227-12232.

Koren, I., Y. J. Kaufman, L. A. Remer, and J. V. Martins (2004), Measurement of the effect

of Amazon smoke on inhibition of cloud formation, Science, 303(5662), 1342-1345.

Koren, I., J. V. Martins, L. A. Remer, and H. Afargan (2008a), Smoke invigoration versus

inhibition of clouds over the Amazon, Science, 321(5891), 946-949.

Koren, I., G. Feingold, H. L. Jiang, and O. Altaratz (2009), Aerosol effects on the inter-cloud

region of a small cumulus cloud field, Geophys. Res. Lett., 36, L14805.

Koren, I., Y. J. Kaufman, D. Rosenfeld, L. A. Remer, and Y. Rudich (2005), Aerosol

invigoration and restructuring of Atlantic convective clouds, Geophys. Res. Lett., 32(14),

L14828.

Koren, I., L. A. Remer, Y. J. Kaufman, Y. Rudich, and J. V. Martins (2007), On the twilight

zone between clouds and aerosols, Geophys. Res. Lett., 34(8), L08805.

Koren, I., L. Oreopoulos, G. Feingold, L. A. Remer, and O. Altaratz (2008b), How small is a

small cloud?, Atmos. Chem. Phys., 8(14), 3855-3864.

Koren, I., L. A. Remer, O. Altaratz, J. V. Martins, and A. Davidi (2010), Aerosol-induced

changes of convective cloud anvils produce strong climate warming, Atmos. Chem. Phys.,

10(10), 5001-5010.

Koren, I., O. Altaratz, L. A. Remer, G. Feingold, J. V. Martins, and R. H. Heiblum (2012),

Aerosol-induced intensification of rain from the tropics to the mid-latitudes, Nature

Geoscience, 5(2), 118-122.

74

Korolev, A., and G. A. Isaac (2006), Relative humidity in liquid, mixed-phase, and ice clouds,

Journal of the Atmospheric Sciences, 63(11), 2865-2880.

Kreidenweis, S. M., K. Koehler, P. J. DeMott, A. J. Prenni, C. Carrico, and B. Ervens (2005),

Water activity and activation diameters from hygroscopicity data - part i: Theory and

application to inorganic salts, Atmos. Chem. Phys., 5, 1357-1370.

Kuo, K. S., R. M. Welch, R. C. Weger, M. A. Engelstad, and S. K. Sengupta (1993), The 3-

dimensional structure of cumulus clouds over the ocean .1. Structural-analysis, J. Geophys.

Res.-Atmos., 98(D11), 20685-20711.

Lee, J., J. Chou, R. C. Weger, and R. M. Welch (1994), Clustering, randomness, and

regularity in-cloud fields .4. Stratocumulus cloud fields, J. Geophys. Res.-Atmos., 99(D7),

14461-14480.

Levoni, C., M. Cervino, R. Guzzi, and F. Torricella (1997), Atmospheric aerosol optical

properties: A database of radiative characteristics for different components and classes, Appl.

Opt., 36(30), 8031-8041.

Levy, R. C., L. A. Remer, S. Mattoo, E. F. Vermote, and Y. J. Kaufman (2007), Second-

generation operational algorithm: Retrieval of aerosol properties over land from inversion of

moderate resolution imaging spectroradiometer spectral reflectance, J. Geophys. Res.-Atmos.,

112(D13), D13211.

Lovejoy, S., and D. Schertzer (2006), Multifractals, cloud radiances and rain, Journal of

Hydrology, 322(1-4), 59-88.

Luo, Y., A. P. Trishchenko, and K. V. Khlopenkov (2008), Developing clear-sky, cloud and

cloud shadow mask for producing clear-sky composites at 250-meter spatial resolution for the

seven MODIS land bands over Aanada and North America, Remote Sensing of Environment,

112(12), 4167-4185.

Marshak, A., A. Davis, W. Wiscombe, and R. Cahalan (1995), Radiative smoothing in fractal

clouds, J. Geophys. Res.-Atmos., 100(D12), 26247-26261.

Marshak, A., Y. Knyazikhin, J. C. Chiu, and W. J. Wiscombe (2009), Spectral invariant

behavior of zenith radiance around cloud edges observed by ARM SWS, Geophys. Res. Lett.,

36, L16802.

Marshak, A., S. Platnick, T. Varnai, G. Y. Wen, and R. F. Cahalan (2006), Impact of three-

dimensional radiative effects on satellite retrievals of cloud droplet sizes, J. Geophys. Res.-

Atmos., 111(D9), D09207.

75

Marshak, A., G. Wen, J. A. Coakley, L. A. Remer, N. G. Loeb, and R. F. Cahalan (2008), A

simple model for the cloud adjacency effect and the apparent bluing of aerosols near clouds,

J. Geophys. Res.-Atmos., 113(D14), D14S17.

Martins, J. V., D. Tanre, L. Remer, Y. Kaufman, S. Mattoo, and R. Levy (2002), MODIS

cloud screening for remote sensing of aerosols over oceans using spatial variability, Geophys.

Res. Lett., 29(12), L013252

Meyers, M. P., R. L. Walko, J. Y. Harrington, and W. R. Cotton (1997), New RAMS cloud

microphysics parameterization .2. The two-moment scheme, Atmospheric Research, 45(1), 3-

39.

Michel Flores, J., R. Z. Bar-Or, N. Bluvshtein, A. Abo-Riziq, A. Kostinski, S. Borrmann, I.

Koren, and Y. Rudich (2012), Absorbing aerosols at high relative humidity: Linking

hygroscopic growth to optical properties, Atmos. Chem. Phys., 12(12), 5511-5521.

Nair, U. S., R. C. Weger, K. S. Kuo, and R. M. Welch (1998), Clustering, randomness, and

regularity in cloud fields - 5. The nature of regular cumulus cloud fields, J. Geophys. Res.-

Atmos., 103(D10), 11363-11380.

Pahlow, M., G. Feingold, A. Jefferson, E. Andrews, J. A. Ogren, J. Wang, Y. N. Lee, R. A.

Ferrare, and D. D. Turner (2006), Comparison between lidar and nephelometer measurements

of aerosol hygroscopicity at the southern great plains atmospheric radiation measurement site,

J. Geophys. Res.-Atmos., 111(D5), D05S15.

Petters, M. D., and S. M. Kreidenweis (2007), A single parameter representation of

hygroscopic growth and cloud condensation nucleus activity, Atmos. Chem. Phys., 7(8), 1961-

1971.

Petters, M. D., C. M. Carrico, S. M. Kreidenweis, A. J. Prenni, P. J. DeMott, J. L. Collett, and

H. Moosmuller (2009), Cloud condensation nucleation activity of biomass burning aerosol, J.

Geophys. Res.-Atmos., 114, D22205.

Platnick, S., M. D. King, S. A. Ackerman, W. P. Menzel, B. A. Baum, J. C. Riedi, and R. A.

Frey (2003), The MODIS cloud products: Algorithms and examples from Terra, Ieee

Transactions on Geoscience and Remote Sensing, 41(2), 459-473.

Quaas, J. (2012), Evaluating the "Critical relative humidity" As a measure of subgrid-scale

variability of humidity in general circulation model cloud cover parameterizations using

satellite data, J. Geophys. Res., 117(D9), D09208.

76

Quaas, J., O. Boucher, N. Bellouin, and S. Kinne (2008), Satellite-based estimate of the direct

and indirect aerosol climate forcing, J. Geophys. Res.-Atmos., 113(D5), D05204.

Quaas, J., B. Stevens, P. Stier, and U. Lohmann (2010), Interpreting the cloud cover - aerosol

optical depth relationship found in satellite data using a general circulation model, Atmos.

Chem. Phys., 10(13), 6129-6135.

Rauber, R. M., et al. (2007), Rain in shallow cumulus over the ocean - the RICO campaign,

Bulletin of the American Meteorological Society, 88(12), 1912-1928.

Remer, L. A., et al. (2008), Global aerosol climatology from the modis satellite sensors, J.

Geophys. Res.-Atmos., 113(D14), D14S07.

Remer, L. A., et al. (2005), The MODIS aerosol algorithm, products, and validation, Journal

of the Atmospheric Sciences, 62(4), 947-973.

Ripley, B. D. (Ed.) (1981), Spatial statistics, John Wiley Sons, New York.

Rissler, J., A. Vestin, E. Swietlicki, G. Fisch, J. Zhou, P. Artaxo, and M. O. Andreae (2006),

Size distribution and hygroscopic properties of aerosol particles from dry-season biomass

burning in Amazonia, Atmos. Chem. Phys., 6, 471-491.

Rogers, R. R. (1989), A short course in cloud physics / by r.R. Rogers and m.K. Yau,

Pergamon Press, Oxford ; New York.

Rosenfeld, D. (2000), Suppression of rain and snow by urban and industrial air pollution,

Science, 287(5459), 1793-1796.

Rosenfeld, D., U. Lohmann, G. B. Raga, C. D. O'Dowd, M. Kulmala, S. Fuzzi, A. Reissell,

and M. O. Andreae (2008), Flood or drought: How do aerosols affect precipitation?, Science,

321(5894), 1309-1313.

Rosin, P. L. (2009), A simple method for detecting salient regions, Pattern Recognition,

42(11), 2363-2371.

Sankaranarayanan, J., H. Samet, and A. Varshney (2007), A fast all nearest neighbor

algorithm for applications involving large point-clouds, Computers & Graphics-Uk, 31(2),

157-174.

Sengupta, S. K., R. M. Welch, M. S. Navar, T. A. Berendes, and D. W. Chen (1990),

Cumulus cloud field morphology and spatial patterns derived from high spatial-resolution

landsat imagery, J, Appl, Meteorol,, 29(12), 1245-1267.

77

Skamarock, W. C., J. B. Klemp, J. Dudhia, D. O. Gill, B. D. M, M. Duda, X.-Y. Huang, W.

E., and J. G. Powers (2008), NCAR Technical Note.

Sorooshian, A., S. Hersey, F. J. Brechtel, A. Corless, R. C. Flagan, and J. H. Seinfeld (2008),

Rapid, size-resolved aerosol hygroscopic growth measurements: Differential aerosol sizing

and hygroscopicity spectrometer probe (dash-sp), Aerosol Sci. Technol., 42(6), 445-464.

Tang, I. N. (1996), Chemical and size effects of hygroscopic aerosols on light scattering

coefficients, J. Geophys. Res.-Atmos., 101(D14), 19245-19250.

Tao, W.-K., X. Li, A. Khain, T. Matsui, S. Lang, and J. Simpson (2007), Role of atmospheric

aerosol concentration on deep convective precipitation: Cloud-resolving model simulations, J.

Geophys. Res.-Atmos., 112(D24), D24S18.

Trenberth, K. E., J. T. Fasullo, and J. Kiehl (2009), Earth's global energy budget, Bulletin of

the American Meteorological Society, 90(3), 311-323.

Twohy, C. H., J. A. Coakley, and W. R. Tahnk (2009a), Effect of changes in relative

humidity on aerosol scattering near clouds, J. Geophys. Res.-Atmos., 114, D05205.

Twohy, C. H., et al. (2009b), Saharan dust particles nucleate droplets in eastern Atlantic

clouds, Geophys. Res. Lett., 36, L01807.

Twomey, S. (1977), Influence of pollution on shortwave albedo of clouds, Journal of the

Atmospheric Sciences, 34(7), 1149-1152.

Varnai, T., and A. Marshak (2009), MODIS observations of enhanced clear sky reflectance

near clouds, Geophys. Res. Lett., 36, L06807.

Verver, G., M. Fujlwara, P. Dolmans, C. Becker, P. Fortuin, and L. Miloshevich (2006),

Performance of the Vaisala RS80A/H and RS90 humicap sensors and the meteolabor "Snow

white" Chilled-mirror hygrometer in paramaribo, suriname, Journal of Atmospheric and

Oceanic Technology, 23(11), 1506-1518.

Walko, R. L., W. R. Cotton, M. P. Meyers, and J. Y. Harrington (1995), New RAMS cloud

microphysics parameterization .1. The single-moment scheme, Atmospheric Research, 38(1-

4), 29-62.

Walko, R. L., W. R. Cotton, G. Feingold, and B. Stevens (2000), Efficient computation of

vapor and heat diffusion between hydrometeors in a numerical model, Atmospheric Research,

53(1-3), 171-183.

78

Wang, H. L., and G. Feingold (2009), Modeling mesoscale cellular structures and drizzle in

marine stratocumulus. Part i: Impact of drizzle on the formation and evolution of open cells,

Journal of the Atmospheric Sciences, 66(11), 3237-3256.

Wang, J. H., and W. B. Rossow (1995), Determination of cloud vertical structure from upper-

air observations, J, Appl, Meteorol,, 34(10), 2243-2258.

Wang, Y., and B. Geerts (2010), Humidity variations across the edge of trade wind cumuli:

Observations and dynamical implications, Atmospheric Research, 97(1-2), 144-156.

Weger, R. C., J. Lee, and R. M. Welch (1993), Clustering, randomness, and regularity in-

cloud fields .3. The nature and distribution of clusters, J. Geophys. Res.-Atmos., 98(D10),

18449-18463.

Weger, R. C., J. Lee, T. R. Zhu, and R. M. Welch (1992), Clustering, randomness and

regularity in cloud fields .1. Theoretical considerations, J. Geophys. Res.-Atmos., 97(D18),

20519-20536.

Wen, G. Y., A. Marshak, R. F. Cahalan, L. A. Remer, and R. G. Kleidman (2007), 3-d

aerosol-cloud radiative interaction observed in collocated modis and aster images of cumulus

cloud fields, J. Geophys. Res.-Atmos., 112(D13), D13204.

Xue, H. W., and G. Feingold (2006), Large-eddy simulations of trade wind cumuli:

Investigation of aerosol indirect effects, Journal of the Atmospheric Sciences, 63(6), 1605-

1622.

Yan, P., X. L. Pan, J. Tang, X. J. Zhou, R. J. Zhang, and L. M. Zeng (2009), Hygroscopic

growth of aerosol scattering coefficient: A comparative analysis between urban and suburban

sites at winter in Beijing, Particuology, 7(1), 52-60.

Zhu, T., J. Lee, R. C. Weger, and R. M. Welch (1992), Clustering, randomness, and regularity

in cloud fields .2. Cumulus cloud fields, J. Geophys. Res.-Atmos., 97(D18), 20537-20558.

Zinner, T., A. Marshak, S. Lang, J. V. Martins, and B. Mayer (2008), Remote sensing of

cloud sides of deep convection: Towards a three-dimensional retrieval of cloud particle size

profiles, Atmos. Chem. Phys., 8(16), 4741-4757.

79

6 List of publications

6.1 Bar-Or, R. Z., I. Koren, and O. Altaratz (2010) Estimating cloud field coverage

using morphological analysis, Environ. Res. Lett., 5, 014022, doi:10.1088/1748-

9326/5/1/014022.

6.2 Bar-Or, R. Z., O. Altaratz, and I. Koren (2011) Global analysis of cloud field

coverage and radiative properties, using morphological methods and MODIS

observations, Atmos. Chem. Phys., 11(1), 191, doi:10.5194/acp-11-191-2011.

6.3 Flores, J. M., R. Z. Bar-Or, N. Bluvshtein, A. Abu-Riziq, A. Kostinski, S.

Borrmann, I. Koren, and Y. Rudich (2012) Absorbing aerosols at high relative

humidity: closure between hygroscopic growth and optical properties, Atmos.

Chem. Phys., 12, 5511-5521, doi:10.5194/acp-12-5511-2012.

6.4 Bar-Or, R. Z., I. Koren, O. Altaratz, and E. Fredj: Humidified aerosol

properties in cloudy environment, Atmos. Res., 118, 280-294, doi:

10.1016/j.atmosres.2012.07.014.

6.5 Bar-Or, R. Z., I. Koren, O. Altaratz: Characterizing the relative humidity in the

lower cloudy troposphere, Geophys. Res. Lett., submitted.

80

7 Declaration

I declare that the thesis summarizes my independent research, under the

supervision of Prof. Ilan Koren and with the continuous consultancy of Dr. Orit

Altaratz. The following sections of the thesis have been conducted in collaboration

with additional researchers:

Sections ‎2.4.1 and ‎3.3.1 were supported by Prof. Erick Fredj, who provided

the WRF simulation results.

Section ‎3.3.3 was co-authored with Dr. Michel Flores, Mr. Nir Bluvshtein,

Prof. Alex Kostinski, Prof. Stephan Borrmann, Prof. Ilan Koren, and Prof.

Yinon Rudich (PI of this study).

81

Appendix A

Figure ‎0.1 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (dc),

for fine-mode hygroscopicity parameter κfi that varies between 0-1.1. The simulated

bimodal log-normal distribution contains coarse-mode desert dust, and fine-mode

biomass burning aerosols (set D1 in Table 2).

82

Figure ‎0.2 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (𝑑𝑐),

for fine-mode geometric mean radius 𝑟𝑔 ,𝑓𝑖 values that vary between 0.06 − 0.3 𝜇𝑚.

The simulated bimodal log-normal distribution contains coarse-mode desert dust, and

fine-mode biomass burning aerosols (set D3 in Table 2).

83

Figure ‎0.3 Simulated aerosol optical depth (AOD, upper panel) and fine-mode

fraction (FMF, lower panel) as a function of the distance from the nearest cloud (𝑑𝑐),

for the fine-mode imaginary part of the refractive index 𝐼𝑚 𝑅𝑒𝑓𝑓𝑖 values that vary

between 0 − 1. The simulated bimodal log-normal distribution contains coarse-mode

desert dust, and fine-mode biomass burning aerosols (set D4 in Table 2).

84

Appendix B

The results of all atmospheric sounding stations (listed in Table 4), as described in

Section ‎3.4.1, are presented below.

Figure ‎0.1 Vertical profiles of the sampled cloud layer fraction (left panel), the mean

RH values (right panel, black line), and the mean RH values that are in cloud layers,

but not inside clouds (right panel, blue line). The gap between each dashed line pair

represents two standard deviations (one for each direction). The analyzed data are all

00:00 UTC radiosonde observations of Lihue, Hawaii, between June and August,

from 1980 to 2011.

85

Figure ‎0.2 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 00:00 UTC radiosonde observations of

Marshall Islands, between June and August, from 1980 to 2011.

86

Figure ‎0.3 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 00:00 UTC radiosonde observations of

Darwin, Australia, between June and August, from 1980 to 2011.

87

Figure ‎0.4 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 00:00 UTC radiosonde observations of Le

Raizet, Guadeloupe, between June and August, from 1980 to 2011.

88

Figure ‎0.5 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 00:00 UTC radiosonde observations of Lord

Howe Island, Australia, between June and August, from 1980 to 2011.

89

Figure ‎0.6 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 00:00 UTC radiosonde observations of

Naze-Funchatoge, Japan, between June and August, from 1980 to 2011.

90

Figure ‎0.7 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 12:00 UTC radiosonde observations of

Budapest, Hungary, between June and August, from 1980 to 2011.

91

Figure ‎0.8 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 12:00 UTC radiosonde observations of

Munich, Germany, between June and August, from 1980 to 2011.

92

Figure ‎0.9 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption for

full description. The analyzed data are all 00:00 UTC radiosonde observations of

Nashville, Tennessee, between June and August, from 1980 to 2011.

93

Figure ‎0.10 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption

for full description. The analyzed data are all 00:00 UTC radiosonde observations of

Manaus, Brazil, between June and August, from 1980 to 2011.

94

Figure ‎0.11 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption

for full description. The analyzed data are all 00:00 UTC radiosonde observations of

Blacksburg, Virginia, between June and August, from 1980 to 2011.

95

Figure ‎0.12 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption

for full description. The analyzed data are all 00:00 UTC radiosonde observations of

Pittsburgh, Pennsylvania, between June and August, from 1980 to 2011.

96

Figure ‎0.13 Vertical profiles of cloudy layer and RH values, see Figure ‎0.1 caption

for full description. The analyzed data are all 12:00 UTC radiosonde observations of

Nairobi, Kenya, between June and August, from 1980 to 2011.