UC San Diego UC San Diego Electronic Theses and Dissertations

Title In-Situ Measurements of Dust, Soot, and Biological Species and their Effects on Mixed Phase Clouds /

Permalink https://escholarship.org/uc/item/1g6706dj

Author Suski, Kaitlyn Jo

Publication Date 2014

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library University of California

UNIVERSITY OF CALIFORNIA, SAN DIEGO

In-Situ Measurements of Dust, Soot, and Biological Species and their Effects on Mixed Phase Clouds

A dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy

Chemistry

Kaitlyn Jo Suski

Committee in charge:

Professor Kimberly A. Prather, Chair Professor Timothy H. Bertram Professor Seth M. Cohen Professor Mark H. Thiemens Professor Ray F. Weiss

2014

Kaitlyn Jo Suski, 2014

SIGNATURE PAGE

The dissertation of Kaitlyn Jo Suski is approved, and it is acceptable in quality and form for publication on microfilm and electronically:

______

Chair

University of California, San Diego

2014

iii

DEDICATION

To Katie, who brightened everyone’s day, and to David, who brightens every one of my days.

EPIGRAPH

“You must not blame me if I do talk to the clouds.”

-Henry David Thoreau

TABLE OF CONTENTS

SIGNATURE PAGE ...... iii DEDICATION ...... iv EPIGRAPH ...... v TABLE OF CONTENTS ...... vi LIST OF FIGURES ...... x LIST OF TABLES ...... xvi ACKNOWLEDGEMENTS ...... xviii VITA ...... xxv ABSTRACT OF THE DISSERTATION ...... xxvii Chapter 1. Introduction ...... 1 1.1 Aerosols ...... 1 1.1.1 Aerosol Sources ...... 1 1.1.2 Transport of Aerosols ...... 2 1.2 Aerosol Effects on Climate ...... 2 1.2.1 Cloud Condensation Nuclei ...... 3 1.2.2 Ice Nucleation ...... 4 1.3 Airborne Sampling ...... 5 1.4 A-ATOFMS ...... 6 1.5 Objectives of Thesis ...... 7 1.6 Synopsis of Thesis ...... 8 Chapter 2. Dust, Bioparticles, or Artifacts? Unraveling the Sources of Metals in Cloud Residues ...... 11 2.1 Abstract ...... 11 2.2 Introduction ...... 12 2.3 Methods ...... 14 2.3.1 Ice in Clouds Experiment – Tropical (ICE-T) ...... 14 2.3.2 CalWater ...... 16 2.3.3 A-ATOFMS ...... 17

2.3.4 Scratch Tests ...... 17 2.3.5 Laboratory Studies of Sea Spray Aerosol...... 18 2.4 Results ...... 19 2.4.1 Inlet Composition and Mass Spectra ...... 19 2.4.2 Isolating Artifact Periods ...... 20 2.4.3 In-Situ Cloud Residue Data ...... 21 2.4.4 Metals in Biological Particles ...... 24 2.5 Discussion/Summary ...... 25 2.6 Acknowledgments ...... 27 2.7 Figures ...... 29 2.8 Supplemental Figures: ...... 35 Chapter 3. Long Range Transport of Asian Soot Impacts Cloud Properties over California ...... 37 3.1 Abstract ...... 37 3.2 Introduction ...... 38 3.3 Measurement Details ...... 41 3.4 Results and Discussion ...... 43 3.5 Conclusions ...... 50 3.6 Acknowledgements ...... 51 3.7 Figures ...... 52 3.8 Supplementary Materials...... 57 Chapter 4. Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western US ...... 63 4.1 Abstract ...... 63 4.2 Introduction ...... 63 4.3 Precipitation ...... 66 4.4 In-Situ Cloud Analysis ...... 69 4.5 Back Trajectory Analysis ...... 75 4.6 Conclusions ...... 77 4.7 Acknowledgments ...... 79 4.8 Figures ...... 80

vii

4.9 Supplementary Material ...... 86 4.9.1 Materials and Methods: ...... 86 4.9.2 Supplemental Discussion ...... 100 4.10 Supplemental Figures ...... 105 4.11 Supplemental Tables ...... 124 Chapter 5. Biological Species Dominate Ice Nucleation in Mixed Phase Marine Clouds Warmer than -15 ºC ...... 127 5.1 Abstract ...... 127 5.2 Introduction ...... 128 5.3 Methods ...... 130 5.4 Results & Discussion ...... 133 5.5 Conclusions ...... 140 5.6 Acknowledgements ...... 141 5.7 Figures ...... 142 5.8 Tables ...... 147 5.9 Supplementary Figures ...... 148 5.10 Supplementary Tables ...... 149 Chapter 6. Principal Component Analysis on In-Situ Cloud Residual Data ...... 151 6.1 Abstract ...... 151 6.2 Introduction ...... 152 6.3 Methods ...... 153 6.4 Results & Discussion: ...... 155 6.4.1 CalWater ...... 155 6.4.2 ICE-T ...... 156 6.4.3 CalWater Sources ...... 157 6.5 Conclusions ...... 160 6.6 Acknowledgements ...... 161 6.7 Figures ...... 162 6.8 Tables ...... 163 6.9 Supplementary Materials...... 166

viii

Chapter 7. Conclusions ...... 167 7.1 Conclusions ...... 167 7.1.1 Dust, Bioparticles, or Artifacts? Unraveling the Sources of Metals in Cloud Residues ...... 167 7.1.2 Long Range Transport of Asian Soot Impacts Cloud Properties over California ...... 168 7.1.3 Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western U.S...... 169 7.1.4 Biological Species Dominate Ice Nucleation in Mixed Phase Marine Clouds Warmer than -15 ºC ...... 170 7.1.5 Principal Component Analysis on In-situ Cloud Residual Data ...... 171 7.2 Future Directions ...... 171 REFERENCES: ...... 175

LIST OF FIGURES

Figure 2.1. Comparison of spectra of titanium particles from (A) a scratch test sample from a sheet of grade 2 titanium and (B) a saturated CVI period during ICE-T. . 29

Figure 2.2. Comparison of spectra from a scratch test sample of 304 stainless steel (A), and aluminum and stainless steel artifact (B) and aluminum artifact (C) particles from cloud sampling during CalWater...... 30

Figure 2.3. Mass spectrum of dust sampled during CalWater...... 30

Figure 2.4. Mass spectra of titanium-containing particles sampled during ICE-T: silicate- rich dust (A), sea salt with small (B) and large titanium markers (C) and bio-sea salt with large titanium markers and copper chloride (D)...... 31

Figure 2.5. Mass spectra of copper-containing particles sampled during ICE-T: sea salt with copper (A), sea salt with bio and copper (B), and bio-sea salt with copper chloride (C)...... 32

Figure 2.7. Mass spectra of a sea spray aerosol sampled with the MART tank plotted with an iron-rich particle sampled through a CVI during CalWater...... 34

Figure 2.9. CVI inlet data for a representative flight showing possible artifact production conditions during CalWater. RH is shown in black and flow through the inlet tip is shown in purple in the bottom panel. Addflow (blue), sample (red), and ambient (green) temperatures are also shown...... 36

Figure 3.1. Chemical composition of aerosols and cloud residues plotted with altitude for the morning flight on 5 March 2011. Altitude is in black and absorption coefficient at 532 nm is in green. S1 and S2 denote passes through the soot layer and C1 indicates a cloud pass...... 52

Figure 3.2. Positive and negative mass spectra of key aerosol types observed on 5 March 2011...... 53

Figure 3.4. Modal diameter for the lower (grey) polluted and higher altitude pristine (blue) clouds during the afternoon flight on 5 March 2011...... 54

Figure 3.5. Picture of thin, grey clouds with soot in the cloud residues taken at 3562 m during the afternoon flight (22:02:06 5 March 2011 UTC)...... 55

Figure 3.6. CDP Re colored by CDP concentration (left), CDP liquid water content (LWC) (middle), and chemical composition of cloud residues (right) plotted against altitude for the afternoon flight on 5 March 2011. The right panel is colored by chemical composition...... 55

Figure 3.7. FLEXPART 10 day back trajectories initiated on 5 March 2011 over Sugar Pine Dam. Fires that occurred in Southeast Asia on February 27, 28 and 1 March 2011 are marked by black dots...... 56

Figure 3.8. K-soot observed in Shanghai in February of 2011 (black) and over California on March 5, 2011 (red). The March 5 data is offset by one mass/charge unit. .... 56

Figure 3.9. Flight tracks for the morning (pink) and afternoon (grey) flights on 5 March 2011. McClellan Airfield is marked by a black dot...... 57

Figure 3.10. Positive and negative mass spectra of aerosol types observed on 5 March 2011...... 58

Figure 3.11. Absorption coefficient (σabs), Absorption Ångström exponent (AAEBR) and altitude (Alt) measured during the mornning flight on 5 March 2011 using the PSAP. Passes through the soot layer are indicated by the S1 and S2 labels...... 59

Figure 3.12. Potential temperature (θ) plotted against altitude for the morning flight (left) and afternoon flight (right) on 5 March 2011. The data is colored by time in UTC. The temperature inversion is boxed in black...... 59

Figure 3.13. Modified CALIPSO image showing smoke at 5 km over the eastern Pacific Ocean on 4 March 2011 (top panel) and CALIPSO images showing pollution and smoke being transported across the Pacific Ocean (all other panels)...... 60

Figure 4.1. Precipitation and cloud characteristics during the 2011 CalWater study at SPD. Markers in (A) show the % of dust and bio precipitation residues sampled at the surface at SPD. Bars show the percentages of in-cloud residues. The % of cold rain, snow/graupel/hail, and warm rain per sample are shown in (B)...... 81

Figure 4.2. Vertical profiles of the number of cloud residues, liquid water content (LWC, g/m3), cloud ice fraction, and cloud temperature (°C) for the flight on (A) Feb 16th ascent (17:01-18:29 UTC) and (B) Feb 16th descent (18:29-20:19 UTC ). Also shown are IN concentrations (L-1)...... 82

Figure 4.4. Dust tracking for days during the (A) Feb 14-16 and (B) Feb 24-26 storms. The 10-day air mass back trajectories ended at cloud top heights over SPD on Feb 16th (7390 m, MSL) and Feb 25th (9340 m, MSL). The circled numbers mark the locations where dust was present along these trajectories...... 84

Figure 4.5. Model summarizing (A) prefrontal and (B) postfrontal conditions. Prefrontal conditions corresponded to a deep continuous cloud layer, AR, and terrain- parallel blocked flow (SBJ), which shifted to two cloud layers, and the disappearance of the AR and blocked flow during the postfrontal conditions. .... 85

Figure 4.7. Representative mass spectra for ATOFMS precipitation residues. Examples include dust mixed with biological material (A)-(C) and pure biological residues (D) and (E). Positive and negative ion intensities vary by type...... 107

Figure 4.8. Representative mass spectra for A-ATOFMS cloud residues. The silicon-rich dust (A), dust (B), dust mixed with biological material (dust/biological) (C), and

xii

dust mixed with salt and biological material (D). Also shown are two different biological residue types in (E) and (F), sea salt (G), and inlet artifact (H)...... 108

Figure 4.9. Percentages of dust, dust mixed with biological material, dust mixed with salt and biological material, and biological particles (A) as clean air aerosol, (B) in cloud residues, (C) as residues in precipitation samples, and (D) in ambient aerosol at the ground level during the CalWater field campaign in 2011...... 109

Figure 4.10. Map showing the flight tracks for all 25 CalWater flights. McClellan Airfield is labeled in red...... 111

Figure 4.11. Feb 16th flight track colored by time on the left and by ascent or descent on the right. The size of the markers in the left panel represents altitude (m, MSL)...... 111

Figure 4.12. The Feb 25th flight track colored by time with altitude (m, MSL) shown by the size of the markers...... 112

Figure 4.14. Representative 2D-S images including the classifications for 0-4. 0 = no images, 1 = liquid droplets with no ice crystals, 2 = isolated ice crystals, or liquid droplets with few ice crystals present, 3 = a mixture of ice crystals and liquid droplets, and 4 = all ice crystals (including rimed) and no liquid droplets...... 114

Figure 4.15. TEM analysis of collected ice nuclei categorized as dust and biological on Feb 16th. Includes: (A) image of dust IN, (B) spectrum for the dust IN shown in (A), (C) image of a biological IN, and (D) spectrum for the biological IN shown in (C). The overall TEM residue composition from Feb 16th is shown in (E). .. 115

Figure 4.16. Google Earth map showing locations of AERONET sites used for NAAPS time-height section plots...... 116

Figure 4.17. Example NAAPS time-height cross sections for (A) Sedeboker, (B) Solar Village, (C) Yinchuan, and (D) Beijing...... 117

Figure 4.18. NAAPS time-height cross section data used to determine the maximum height of dust plumes at sites in each of the dust regions. The times of the dust

xiii

plume heights correspond to times when trajectories passed over the dust regions. Also shown are the altitudes of these trajectories endpoints...... 118

Figure 4.19. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript. The CALIPSO longitude matches the longitude of the markers in Figure 4.4. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located...... 119

Figure 4.20. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript. The CALIPSO longitude matches the longitude of the markers in Figure 4.4. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located...... 120

Figure 4.21. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript. The CALIPSO longitude matches the longitude of the markers in Figure 4.4. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located...... 121

Figure 4.22. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript. The CALIPSO longitude matches the longitude of the markers in Figure 4.4. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located...... 122

Figure 4.23. (A) Wind profiler observations from the Concord, California atmospheric river observatory (ARO) showing wind speed and direction as a function of time. (B) Same as (A) but for a site farther east at Sloughhouse...... 123

Figure 5.1. Mass spectra from A) CalWater and B) ICE-T showing the salt+bio particle types observed during the studies...... 142

Figure 5.2. A) CalWater cloud residual chemical types plotted against temperature for different ice and liquid regions in mixed phase clouds. Markers are colored by ice type and sized by A-ATOFMS aerodynamic size. B) ICE-T cloud residual chemical types plotted against temperature...... 143

Figure 5.3. SEM images of salt+bio (A) and salt+organic carbon (B) particles sampled during ICE-T...... 144

Figure 5.4. Binary logistic regression fitted results plotted against mass spectral areas for different chemical species (aluminum, phosphate, and CNO). In the top panels, markers are colored by phosphate area...... 145

xiv

Figure 5.5. Example HYSPLIT air mass back trajectory initiated on July 7, 2011 at 01:00 UTC at St. Croix, USVI (17.73 N, 64.75 W) at 1000, 2500, and 5000 m, which shows the air masses typically came from the east during ICE-T...... 146

Figure 5.6. Binary logistic regression fitted results plotted against mass spectral areas (CNO, phosphate) and temperature for CalWater (left panels) and ICE-T (right panels)...... 148

Figure 6.1. Clustered HYSPLIT back-trajectory analysis from CalWater 2011 showing the four main back-trajectories: Long Asia (top left), Short Asia (top right), Long Pacific (bottom left), and Short Pacific (bottom right)...... 162

LIST OF TABLES

Table 3.1. FLEXPART back trajectory point source information...... 62

Table 4.1. Statistics for precipitation sample collection during winter storms in 2011 at SPD. Includes start and end dates and times for sample collection and number of residues that were chemically analyzed. Cloud temperature statistics are also provided. Storms are also labeled for corresponding samples...... 124

Table 4.2. Meteorological conditions during the 11 precipitation samples collected at SPD. Observations were used to determine if AR conditions and/or SBJ conditions were present, or if there was a cold frontal passage during each sampling period. Accumulated rainfall and average rain rate is also shown...... 125

Table 4.3. Aerosol concentrations from 0.6-10.0 μm, used as a proxy for dust concentrations at 3000-6000 m during selected CalWater flights. Also provided are the atmospheric pressures and total precipitation amounts at SPD...... 126

Table 5.1. Binary logistic regression results showing the chemical species that correlate to ice formation for CalWater and ICE-T. Negative correlations are italicized. 147

Table 5.2. Binary logistic regression loadings for the CalWater data set. Significance codes are Pr(>|z|) at values of 0-0.001(***), 0.001-0.01 (‘*), 0.01-0.05 (*), 0.05- 0.1 (.), and 0.1-1 ( )...... 149

Table 5.3. Binary logistic regression loadings for the ICE-T data set. Significance codes are Pr(>|z|) at values of 0-0.001(***), 0.001-0.01 (‘*), 0.01-0.05 (*), 0.05-0.1 (.), and 0.1-1 ( )...... 150

Table 6.1. CalWater chemical components derived from PCA loadings separated by liquid, warm ice, and cold ice...... 163

Table 6.2. ICE-T chemical components derived from PCA loadings separated by liquid, warm ice, and cold ice...... 163

Table 6.3. Back-trajectory cluster assignments by flight number for CalWater 2011. .. 163

Table 6.4. Chemical components derived from PCA loadings for the Long Asia trajectories...... 164

xvi

Table 6.5. Chemical components derived from PCA loadings for the Short Asia trajectories...... 164

Table 6.6. Chemical components derived from PCA loadings for the Long Pacific trajectories...... 164

Table 6.7. Chemical components derived from PCA loadings for the Short Pacific trajectories...... 165

Table 6.8. PCA chemical inputs, their abbreviations, and associated mass-to-charges for all of the PCA runs...... 166

xvii

ACKNOWLEDGEMENTS

First and foremost I would like to thank Professor Kimberly Prather, my Ph.D. advisor, for her support and guidance. I have learned so much from here especially how to be more assertive and confident. She has taught me to look at the big picture and her enthusiasm for her work is infectious. Most of all I’ve learned to never give up and always go for what you want. With enough hard work anything is possible!

Special thanks go out to Professor Daniel Rosenfeld. His guidance and kind words were priceless during the last year of my graduate work. I greatly appreciate his support and our weekly meetings were always enjoyable.

I would like to thank and acknowledge my committee members Professor

Timothy H. Bertram, Professor Seth M. Cohen, Professor Mark H. Thiemens, and

Professor Ray F. Weiss for their comments and thoughtful feedback during my exams.

I would also like to thank Professor Dina Merrer and Professor Linda Doerrer, who fostered a love of chemistry in me during my undergraduate studies and research. If

Professor Doerrer wouldn’t have suggested I major in chemistry, who knows where I would be now! Professor Merrer was a great mentor and I owe a lot to her support.

Prather Lab members past and present were invaluable during my time in graduate school. From preparing for field studies to editing papers or listening to me complain, their companionship and friendship means a lot. The list includes: Jack Cahill,

Doug Collins, Dr. Jessie Creamean, Dr. Alberto Cazorla, Dr. Jess Axson, Matt Ruppel,

Dr. Bob Moision, Dr. Melanie Zaucher, Liz Fitzgerald, Dr. Lindsay Hatch, Professor

xviii

Andy Ault, Professor Kerri Pratt, Professor Tim Guasco, Dr. Luis Cuadra-Rodriguez, Dr.

Andy Martin, Camille Sultana, Chris Lee, Mitchell Santander, Gavin Cornwell, Hash Al-

Mashat, Dr. Cassie Gaston, and Dr. Meagan Moore. I would also like to thank Joe Mayer and Paula Schachter for all of their help.

Field studies require many people to work together and I would like to thank all of my collaborators: Jason Tomlinson, Beat Schmidt, Jennifer Comstock, John Hubbe,

Celine Kluzek, Rahul Zaveri, Mike Hubble, Bill Svancara, Gene Dukes, Paul DeMott,

Ryan Sullivan, Fred Brechtel, Steven Day, Haf Jonsson, John Shilling, Andy Heymsfield, the entire NSF/NCAR C-130 crew, Cynthia Twohy, James Anderson, Scott Hersey, John

Seinfeld, Richard Flagan, Andrew Metcalf, Jill Craven, Jake Lipman, and many others who helped during field studies.

I would like to thank my parents, Sharon Wells, Ron Suski, Roy Wells, and Lori

Suski, and my brother, Shane Suski, for their encouragement and support. I would like to acknowledge my friends for all of their support, especially Viktoría Gisladottir, Silvana

Sartori, Katie Osterday, and Carrie James. My running group, In Motion Fit, helped me to shed stress and kick butt at the Carlsbad Marathon! My yoga teacher, Maria Sirriya, also helped me to de-stress and provided much emotional support during my journey.

And extra special thanks go out to David Barajas-Solano, without whom I would be lost.

I am extremely grateful for your support and encouragement in everything I do.

The work presented in this thesis is the result of several field campaigns I have participated in including: CalWater 2010, CalNex, the Carbonaceous Aerosol and

Radiative Effects Study (CARES), CalWater 2011, and the Ice in Clouds Experiment-

xix

Tropical (ICE-T). For all of these campaigns, I prepared the aircraft aerosol time-of-flight mass spectrometer (A-ATOFMS) by tuning, aligning, and calibrating the instrument. For some of the field studies this work was split between me and one other person. For

CalNex and CARES, the field study preparation and data acquisition was split equally between the dissertation author and John F. Cahill. For ICE-T, the field study preparation and data acquisition was split between the dissertation author and Alberto Cazorla.

During CalWater 2011, I did all of the field study preparation and all data acquisition except for two flights that were performed by John Cahill. I also led a group of three people with the goal of cutting 100 pounds of off the aircraft aerosol time-of-flight mass spectrometer and reduced its volume to fit in a smaller aircraft rack. The team lead by the dissertation author was successful in the mission and the instrument was able to fly on the

CIRPAS Twin Otter for CalNex. Once the data was acquired, I devised analysis techniques to use the data to answer my personal science questions. I used statistical analysis techniques as well as analysis of individual mass spectra and cloud imagery by hand to separate the data into different ice and chemical types. I did all of the by hand analysis by myself. Additionally, with the help of Daniel Rosenfeld, we designed laboratory experiments to probe the chemistry of possible inlet artifacts. I performed the experiments and analyzed the data. I wrote Chapters 1, 2, 3, 5, and 6 by myself and chapter 4 was written equally by Jessie M. Creamean and me. We contributed equally to the work in Chapter 6 both in the analysis and writing, as noted in the published version of the manuscript.

Chapter 2, in full, is currently being prepared for submission for publication of the material. Kaitlyn J. Suski, Cynthia H. Twohy, James R. Anderson, Daniel Rosenfeld,

Jennifer M. Comstock, Andrew J. Heymsfield, and Kimberly A. Prather (2014), “Dust,

Bioparticles, or Artifacts? Unraveling the Sources of Metals in Cloud Residues”, In preparation for submission to Aerosol Science and Technology. The dissertation author was the primary investigator and author of this material and collected A-ATOFMS data during CalWater and for the first half of ICE-T, analyzed A-ATOFMS and cloud probe data, and performed the experiments presented in this chapter. Cynthia Twohy ran the counterflow virtual impactor inlet during ICE-T, provided metal for the scratch tests, contributed intellectually to the analysis and to the manuscript. James Anderson provided the EDX data used in this chapter and contributed intellectually to the manuscript. Daniel

Rosenfeld contributed intellectually to the analysis, manuscript, and to the design of the laboratory experiments. Jennifer Comstock provided inlet data from the CalWater flights.

Alberto Cazorla is acknowledged for his part in obtaining A-ATOFMS data during the second portion of ICE-T. Doug Collins, Jess Axson, Chris Lee, Camille Sultana, Matt

Ruppel, and Mitch Santander are acknowledged for running the MART tank and collecting ATOFMS sea spray data. Darin Toohey is acknowledged for his role in operating the CVI and collection of UHSAS data during ICE-T. Andrew Heymsfield was the principal investigator for ICE-T and designed the ICE-T experiments. Kimberly

Prather was the principal investigator for CalWater, designed the CalWater flight experiments, and contributed intellectually to the analysis and the manuscript.

Chapter 3, in full, is currently being prepared for submission for publication of the material. Kaitlyn J. Suski, Alberto Cazorla, John F. Cahill, Jason Tomlinson, Duli Chand,

xxi

Beat Schmid, Daniel Rosenfeld, Kimberly A. Prather (2014), “Long Range Transport of

Asian Soot Impacts Cloud Properties over California”, In preparation for submission to

Atmospheric Environment. The dissertation author was the primary investigator and author of this material and collected and analyzed the A-ATOFMS data, completed back- trajectory analysis, and analyzed cloud probe and aerosol data. Alberto Cazorla and John

Cahill contributed intellectually to this work and helped collect A-ATOFMS data. Jason

Tomlinson contributed intellectually to this work and collected and analyzed cloud and aerosol probe data. Duli Chand and Beat Schmid analyzed and provided PSAP data. Xin

Yang provided the Shanghai ATOFMS data. Daniel Rosenfeld contributed intellectually to the analysis and the manuscript. Kimberly Prather was the principal investigator for

CalWater, designed the CalWater flight experiments, and contributed intellectually to the analysis and the manuscript.

Chapter 4, in full, is a reprint of the material as it appears in Science 2013. Jessie

M. Creamean, Kaitlyn J. Suski, Daniel Rosenfeld, Alberto Cazorla, Paul J. DeMott, Ryan

C. Sullivan, Allen B. White, F. Martin Ralph, Patrick Minnis, Jennifer M. Comstock,

Jason M. Tomlinson, and Kimberly A. Prather (2013), “Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western U.S.”, Science,

339(6127), 1572-1578. Reprinted with permission from AAAS. The dissertation author was a primary investigator of this material, made aircraft-based A-ATOFMS measurements in the field, analyzed aerosol data, analyzed cloud probe data, and wrote the manuscript. Jessie Creamean made ground-based ATOFMS precipitation measurements in the field and in the laboratory, analyzed aerosol data, analyzed GOES-

11 data, and wrote the manuscript. Daniel Rosenfeld provided intellectual contribution

xxii

and contributed to the writing of this manuscript. Alberto Cazorla assisted with aircraft- based A-ATOFMS measurements in the field and source analysis using HYSPLIT. Paul

DeMott and Ryan Sullivan provided ice nuclei concentration and composition measurements and intellectual contribution. Allen White and Martin Ralph analyzed the surface meteorological and S-PROF data and description of meteorological data. Patrick

Minnis directed the analysis of the GOES-11 data and contributed to the writing of the manuscript. Jennifer Comstock provided ice fraction calculations. Jason Tomlinson processed CDP data and provided PCASP and CAS data. Kimberly Prather was the principal investigator for this work, designed the CalWater experiments, and contributed intellectually to the analysis and the manuscript.

Chapter 5, in full, is currently being prepared for submission for publication of the material. Kaitlyn J. Suski, Daniel Rosenfeld, James R. Anderson, Cynthia H. Twohy,

Andrew J. Heymsfield, and Kimberly A. Prather (2014), “Biological Species Dominate

Ice Nucleation in Mixed Phase Marine Clouds Warmer than -15 ºC”, In preparation for submission to Nature Geosciences. The dissertation author was the primary investigator and author of this material and collected A-ATOFMS data during CalWater and for the first half of ICE-T, as well as carried out the statistical analysis and A-ATOFMS data analysis presented in this chapter. Daniel Rosenfeld contributed intellectually to the analysis and the manuscript. James Anderson provided SEM images for this work.

Cynthia Twohy operated the CVI inlet during ICE-T. Alberto Cazorla is acknowledged for collecting A-ATOFMS data during the second half of ICE-T. Andrew Heymsfield was the principal investigator for ICE-T and designed the ICE-T experiments. Kimberly

xxiii

Prather was the principal investigator for CalWater, designed the CalWater flight experiments, and contributed intellectually to the analysis and the manuscript.

Chapter 6, in full, is currently being prepared for submission for publication of the material. Kaitlyn J. Suski, Daniel Rosenfeld, Alberto Cazorla, and Kimberly A. Prather

(2014), “Principal Component Analysis on In-Situ Cloud Residual Data”, In preparation.

The dissertation author was the primary investigator and author of this material and collected A-ATOFMS data during CalWater and for the first half of ICE-T, as well as carried out the statistical analysis and A-ATOFMS data analysis presented in this chapter.

Daniel Rosenfeld contributed intellectually to the analysis and the manuscript. Alberto

Cazorla performed the HYSPLIT clustering analysis and collected A-ATOFMS data during the second half of ICE-T. Kimberly Prather was the principal investigator for

CalWater, designed the CalWater flight experiments, and contributed intellectually to the analysis and the manuscript.

xxiv

VITA

2008 Bachelor of Arts, Barnard College

2008-2009 Teaching Assistant, Department of Chemistry and Biochemistry University of California, San Diego

2011 Master of Science, University of California, San Diego

2008-2013 Graduate Student Researcher, University of California, San Diego

2014 Doctor of Philosophy, University of California, San Diego

PUBLICATIONS

Suski, K. J., D. Rosenfeld, J. R. Anderson, C. H. Twohy, A. J. Heymsfield, and K. A. Prather (2014), “Biological Species Dominate Ice Nucleation in Mixed Phase Marine Clouds Warmer than -15 ºC”, In preparation for submission to Nature Geosciences.

Suski, K. J., C. H. Twohy, J. R. Anderson, D. Rosenfeld, J. M. Comstock, A. J. Heymsfield, and K. A. Prather (2014), “Dust, Bioparticles, or Artifacts? Unraveling the Sources of Metals in Cloud Residues”, In preparation for submission to Aerosol Science and Technology.

Suski, K. J., A. Cazorla, J.F. Cahill, J. Tomlinson, D. Chand, B. Schmid, D. Rosenfeld, K.A. Prather (2014), “Long Range Transport of Asian Soot Impacts Cloud Properties over California”, In preparation for submission to Atmospheric Environment.

Creamean, J. M. and Suski, K. J., D. Rosenfeld, A. Cazorla, P. J. DeMott, R. C. Sullivan, A. B. White, F. M. Ralph, P. Minnis, and K. A. Prather (2013), “Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western US.”, Science, 339(6127), 1572-1578.

Rosenfeld, D., R. Chemke, K. A. Prather, K. J. Suski, J. M. Comstock, B. Schmid, J. Tomlinson, H. Jonsson (2014), “Polluting of Winter Convective Clouds upon transition from ocean inland Over Central California: Contrasting Case Studies”, Atmospheric Research, 135–136(0), 112-127. Cazorla, A., R. Bahadur, K. J. Suski, J. F. Cahill, D. Chand, B. Schmid, V. Ramanathan, and K. A. Prather (2013), “Relating aerosol absorption due to soot, organic carbon, and dust to emission sources determined from in-situ chemical

xxv

measurements”, Atmospheric Chemistry and Physics, 13, 9337-9350, doi:10.5194/acp-13-9337-2013.

Rosenfeld, D., R. Chemke , P. DeMott , R. Sullivan , R. Rasmussen , F. McDonough , J. Comstock , B. Schmid , J. Tomlinson , H. Jonsson , K. Suski , K. Prather (2013), “The Common Occurrence of Highly Supercooled Drizzle and Rain near the Coastal Regions of the Western United States”, Journal of Geophysical Research: Atmospheres, 118(17), 9819-9833.

Hersey, S. , J. Craven , A. Metcalf , J. Lin , T. Lathem , K. Suski , J. Cahill , H. Duong , A. Sorooshian , H. Jonsson , A. Nenes , K. Prather , R. Flagan, J.Seinfeld (2013), “Composition and hygroscopicity of the Los Angeles Aerosol: CalNex”, Journal of Geophysical Research: Atmospheres, 118, 3016–3036, doi:10.1002/jgrd.50307.

Cahill, J. F., K. Suski, J. H. Seinfeld, R. A. Zaveri, and K. A. Prather (2012), “The mixing state of carbonaceous aerosol particles in northern and southern California measured during CARES and CalNex 2010”, Atmospheric Chemistry and Physics, 12, 10989-11002, doi:10.5194/acp-12-10989-2012.

Zaveri R. A., W. J. Shaw, D. J. Cziczo, B. Schmid, R. Ferrare, M. L. Alexander, M. Alexandrov, R. J. Alvarez, W. P. Arnott, D. Atkinson, S. Baidar, R. M. Banta, J. C. Barnard, J. Beranek, L. K. Berg, F. J. Brechtel, W. A. Brewer, J. F. Cahill, B. Cairns, C. D. Cappa, D. Chand, S. China, J. M. Comstock, M. K. Dubey, R. C. Easter, Jr, M. H. Erickson, J. D. Fast, C. Floerchinger, B. A. Flowers, E. Fortner, J. S. Gaffney, M. K. Gilles, K. Gorkowski, W. I. Gustafson, Jr, M. S. Gyawali, J. Hair, M. Hardesty, J. W. Harworth, S. C. Herndon, N. Hiranuma, C. A. Hostetler, J. M. Hubbe, J. T. Jayne, H. Jeong, B. T. Jobson, E. I. Kassianov, L. I. Kleinman, C. D. Kluzek, B. Knighton, K. R. Kolesar, C. Kuang, A. Kubatova, A. O. Langford, A. Laskin, N. S. Laulainen, R. D. Marchbanks, C. Mazzoleni, F. Mei, R. C. Moffet, D. A. Nelson, M. Obland, H. Oetjen, T. B. Onasch, I. Ortega, M. Ottaviani, M. S. Pekour, K. A. Prather, J. G. Radney, R. Rogers, S. P. Sandberg, A. Sedlacek, C. Senff, G. Senum, A. Setyan, J. E. Shilling, M. K. B. Shrivastava, C. Song, S. R. Springston, R. Subramanian, K. Suski, J. M. Tomlinson, R. M. Volkamer, H. A. Wallace, J. Wang, A. M. Weickmann, D. R. Worsnop, X. Y. Yu, A. Zelenyuk, and Q. Zhang (2012), "Overview of the 2010 Carbonaceous Aerosols and Radiative Effects Study (CARES)", Atmospheric Chemistry and Physics 12(16):7647-7687. doi:10.5194/acp-12-7647-2012.

xxvi

ABSTRACT OF THE DISSERTATION

In-Situ Measurements of Dust, Soot, and Biological Species and their Effects on Mixed Phase Clouds

Kaitlyn Jo Suski

Doctor of Philosophy in Chemistry

University of California, San Diego, 2014

Professor Kimberly Prather, Chair

Aerosol effects on clouds contribute the largest uncertainty to predictions of anthropogenic induced climate change. Thus, chemical characterization of aerosols that nucleate cloud droplets and ice crystals is crucial to extending our knowledge of cloud formation and accurately predicting climate change. The studies presented in this dissertation utilize an aircraft aerosol time-of-flight mass spectrometer (A-ATOFMS) to study the chemical mixing state of individual aerosols, cloud droplets and ice crystal residues from two aircraft field campaigns: CalWater in the California Sierra Nevada and

xxvii

the Ice in Clouds Experiment-Tropical (ICE-T) in the Caribbean. Analysis of these datasets provided insights into aerosol effects on clouds and the chemical characteristics of ice nuclei (IN) as well as inlet artifacts.

During CalWater two trans-Pacific transported aerosol types, dust and soot, were observed to have vastly different effects on clouds. Soot, from Asian pollution, was present in clouds with more numerous, small droplets. Clouds with small droplets in the

Sierra Nevada have previously been linked to reduction of precipitation, thus increases in

Asian pollution could contribute to additional precipitation reduction in the region. In contrast, dust and biological aerosols are shown to have the opposite effect and enhance precipitation in the Sierra Nevada by serving as ice nuclei in glaciated clouds that coincided with increased ice-induced precipitation.

Chemically characterizing the initial ice nuclei is challenging, as numerous physical and chemical changes occur after nucleation. Using a combination of statistical analysis methods, chemical components key to ice formation were identified. Organic nitrogen and phosphate, indicators of biological species, were found to be important for ice formation in both study locations using binary logistic regression. Principal component analysis identified biological particles as key to ice formation and suggested they were of marine origin. Additionally, using in-situ cloud residual data, the ice nucleation temperature of a salt-biological particle type was identified between -10 and

-13 °C. These results provide insight into the chemical composition of IN active at warmer temperatures. Lastly, a combination of in-situ and laboratory measurements

xxviii

identified aircraft inlet artifacts produced during sampling in mixed phase clouds and showed the artifacts were distinct from other metal-containing residues.

xxix

Chapter 1. Introduction

1.1 Aerosols

Aerosols are solids or liquids suspended in a gas and in the case of atmospheric aerosols the suspension gas is air. Atmospheric aerosols can range in size from 10 nm to

100 µm and there are three main size modes: Aitken mode (10-100 nm), accumulation mode (100 nm- 1 µm), and coarse mode (> 1 µm) [Seinfeld and Pandis, 2006]. Aerosols ejected straight from the source are known as primary aerosols, and those formed indirectly through gas phase reactions are called secondary aerosols.

1.1.1 Aerosol Sources

Aerosols are produced from both natural and anthropogenic sources. Natural aerosols in part form by mechanical processes including wind erosion, bubble bursting, and wave crashing to produce airborne dust and sea spray aerosol, respectively [Alfaro and Gomes, 2001; Bigg and Leck, 2008]. Anthropogenic aerosols result from incomplete combustion of fossil fuels and biomass that produces black carbon or soot, as well as gas phase species. Aerosols can also be nucleated from gas phase precursors of both natural and anthropogenic origins. For instance, isoprene and other volatile organic carbon

(VOC) are emitted from the biosphere and become oxidized in the atmosphere to form secondary organic aerosol (SOA) [Carlton et al., 2009]. Particles of biological origin that are emitted directly into the atmosphere are termed primary biological aerosol particles

(PBAP) and can include whole or fragments of pollen, bacteria, spores, plant material, or viruses [Despres et al., 2012]. Additionally, biological species can be present on both sea

2 spray and dust aerosols [Blanchard and Syzdek, 1970; Guggenberger et al., 1994;

Kellogg and Griffin, 2006; Pratt et al., 2009a]. The term aerosol mixing state describes the distribution of chemical species among individual aerosol particles. Once emitted, aerosols can undergo various physical and chemical processes that change their mixing state including condensation and evaporation of semi-volatile gases, coagulation with other particles, heterogeneous reactions on the aerosol surface, and cloud processing

[Seinfeld and Pandis, 2006]. Thus, aerosols are complex mixtures of different chemical species that are continuously changing in time and are moving in space.

1.1.2 Transport of Aerosols

Aerosols have atmospheric lifetimes of hours to weeks [Seinfeld and Pandis,

2006]. Once aerosols are emitted, they can be transported far from their source regions.

Aerosols can be lofted to high altitudes, which allow them to travel far distances. Trans-

Pacific transport of dust and pollution from Asia across the Pacific Ocean to the western coast of North America is a frequent occurrence [Hadley et al., 2007; Husar et al., 2001;

Sun et al., 2001; Uno et al., 2011; Wuebbles et al., 2007; Yienger et al., 2000]. It has been shown that within 13 days an Asian dust plume made it the whole way around the world

[Uno et al., 2009]. Additionally, westward transport of dust from the Sahara toward the

Caribbean and South America is also well documented [Ben-Ami et al., 2010; Doherty et al., 2008; Petit et al., 2005; Prospero et al., 1981].

1.2 Aerosol Effects on Climate

Aerosols have both direct and indirect effects on climate. Directly, they scatter and absorb solar and terrestrial radiation. Additionally, aerosols can modify cloud

3 properties by serving as a site for water to condense forming cloud droplets and ice crystals, and the resulting cloud modification effects are collectively known as aerosol indirect effects [Lohmann and Feichter, 2005]. Clouds with pollution particles as their cloud seeds tend to have more numerous, smaller cloud droplets which results in more reflection of solar radiation, known as the Twomey effect [Twomey, 1974]. The Albrecht effect, or cloud lifetime effect, occurs when small cloud droplets are formed due to an increase in the number of cloud nuclei for a defined liquid water content [Albrecht,

1989]. The more numerous cloud droplets result in reduced or delayed precipitation thus extending the lifetime of the cloud [Albrecht, 1989]. The semi-direct effect occurs when absorbing aerosols, like black carbon, absorb radiation, either within cloud droplets causing them to evaporate, or around clouds causing adjacent water droplets to evaporate, resulting in reduced surface radiation [Hansen et al., 1997]. The glaciation effect is caused by aerosols serving as ice nuclei in mixed phase clouds [Lohmann, 2002]. An increase in ice increases the precipitation efficiency of the cloud [Lohmann, 2002]. Small cloud droplets lead to delayed ice formation which results in less efficient precipitation processes and clouds extend to higher altitudes and colder temperatures before dissipating (thermodynamic effect) [Khain et al., 2005]. The magnitudes of these effects are poorly understood, which result in aerosol-cloud interactions accounting for the largest uncertainty in predicting climate change [IPCC, 2007].

1.2.1 Cloud Condensation Nuclei

Aerosols can nucleate cloud droplets by condensing liquid onto the surface of the particles. Different supersaturations, or relative humidity’s above 100 %, are necessary

4 for aerosols to form cloud droplets based on their size and chemical composition

[Seinfeld and Pandis, 2006]. This nucleation process is described by Köhler theory, which combines Raoult’s law, which describes the vapor pressure of a solution based on its chemical composition, with the Kelvin effect, which explains how vapor pressure changes with size due to surface curvature of the droplets [ 1921]. Therefore, for a given supersaturation, aerosols with appreciable amounts of soluble material can serve as cloud condensation nuclei (CCN) at smaller sizes than particles with negligible soluble material [Seinfeld and Pandis, 2006].

1.2.2 Ice Nucleation

Aerosol particles upon which ice formation in clouds occurs are known as ice nuclei (IN). In the absence of aerosols, liquid droplets will form ice homogeneously at temperatures below -37 °C [Murray et al., 2010]; therefore, IN enable ice to form at warmer temperatures. There are several modes of ice nucleation: deposition, immersion, contact, and inside-out contact [Pruppacher and Klett, 2010]. Deposition nucleation occurs when vapor freezes on the surface of a particle. Immersion nucleation occurs when a nucleus is inside a liquid droplet, which then freezes. Contact nucleation occurs when a nucleus touches a liquid droplet and freezes on contact. Inside-out contact nucleation occurs via a mixture of immersion and contact freezing where the nucleus is inside a liquid droplet that touches the edge of the droplet from the inside, which causes freezing. Which nuclei form ice efficiently is poorly understood; however, what is known about ice nuclei is they tend to be insoluble, have large surface areas, and have active sites where ice germs can grow [Seinfeld and Pandis, 2006]. Asian dust has been shown

5 to glaciate clouds around -20 ºC [Rosenfeld et al., 2011] and Saharan dust has been shown to form ice in clouds with cloud top temperatures ~-20 ºC [Ansmann et al., 2008].

Certain biological particles have enhanced warm temperature IN activity with certain bacteria shown to form ice beginning around -2 °C [Maki et al., 1974; Yankofsky et al.,

1981] and pollen around -8 ºC [Morris et al., 2004a]. Biological species can be present on dust and it has been shown that the IN activity of dust and biological mixtures correlated with the presence of biological species [Conen et al., 2011; O'Sullivan et al.,

2013]. Ice formation can greatly impact precipitation and it has been suggested that more than half of precipitation events are initiated by ice [Lau and Wu, 2003]. Thus, to improve quantitative prediction of precipitation events, a better understanding of ice formation processes in clouds is crucial.

1.3 Airborne Sampling

To study cloud droplet and ice crystal residues in-situ, airborne measurements are frequently employed. There are certain considerations that must be taken into account when sampling in aircraft. Aerosols are sampled through an isokinetic aircraft inlet to reduce size biases [Pena et al., 1977; Rogers et al., 1989]. Cloud residues are sampled using a counter-flow virtual impactor (CVI) inlet, which uses a counter-flow gas to create a stagnation plane in the inlet through which only particles with a large enough inertia can pass [Noone et al., 1988]. Typical cut sizes for CVI inlets, the size where 50% of the particles make it through the stagnation plane, range from 9 to 30 µm [Noone et al.,

1988]. This effectively removes interstitial aerosols from the sample and allows for cloud droplets and ice crystals to pass through the inlet. Artifact production from sampling ice

6 crystals through CVIs is a concern. Ice crystals can scratch the inlet walls or dislodge particles that have coated the inlet [Murphy et al., 2004]; thus, proper artifact identification and inlet characterization is necessary.

1.4 A-ATOFMS

Individual aerosols, cloud droplets, and ice crystals were analyzed using aircraft aerosol time-of-flight mass spectrometry (A-ATOFMS) [Pratt et al., 2009b], which is a smaller, aircraft-compatible version of the field-ready ATOFMS [Gard et al., 1997].

Single particles are sized and chemically analyzed using laser desorption and ionization mass spectrometry. Particles enter the instrument through an aerodynamic lens that focuses the particles into a tight particle beam [Liu et al., 1995a; b]. Then, particles of a given size reach their terminal velocity, which is calculated by measuring the time it takes the particle to pass through two 532-nm laser beams (CDPS532M, JDSU, Eching,

Germany), which have a fixed distance of 6 cm between them. The speed is then converted to size based on calibration with a range of predetermined sizes of polystyrene latex spheres (PSLs). The speed is also used to trigger a 266-nm Neodymium:YAG laser

(Ultra 50, Quantel USA, Bozeman, MT) to fire, that desorbs and ionizes the particles in a

Z-TOF mass spectrometer (Tofwerk AG, Thun, Switzerland). The positive and negative ions are guided by ion extractors down flight tubes and enter a reflectron, which partially corrects for kinetic energy distributions thereby improving the mass resolution

[Bergmann et al., 1989; Wollnik and Przewloka, 1990]. Then, the ions are directed down another flight tube toward the detectors. Positive and negative mass spectra for each particle are generated using microchanel plate (MCP) detectors (25 mm bipolar time-of-

ﬂight detector, BURLE Industries Inc., Lancaster, PA), which are optically decoupled from the data acquisition board using a scintillator and photomultiplier tube (PMT). Thus, for each analyzed particle, size and positive and negative mass spectra are recorded. This

ATOFMS technique allows for the chemical composition of individual particles to be examined in real time.

1.5 Objectives of Thesis

The main goal of the work presented in this thesis involves determining how the chemistry of individual aerosol particles, with an emphasis on particle sources, affects cloud properties. Aerosol-cloud interactions contribute the largest uncertainty to climate change predictions [IPCC, 2007]; thus, a fundamental understanding of aerosol effects on clouds is key to accurately quantifying climate change due to anthropogenic activity. A major focus involves understanding which particle types serve as effective ice nuclei. The motivation for these studies stems from the fact that the characteristics of efficient IN are relatively unknown [Cantrell and Heymsfield, 2005] and there is a dearth of IN in the atmosphere [DeMott et al., 2010; Rogers et al., 2001], especially IN active at warm temperatures (> -15 °C). However, ice formation at warm temperatures occurs frequently and ice concentrations are frequently underestimated, especially in the tropics [Hobbs and Rangno, 1985; Mossop, 1968; Mossop et al., 1970]. Furthermore, the studies in this thesis aim to compare the effects of local pollution versus long-range transported natural aerosols on cloud properties in California. A deeper understanding of aerosol sources and their ice nucleating abilities will greatly improve our understanding of aerosol effects on mixed phase clouds.

The main questions that will be addressed herein include:

1. What are the sources of IN and CCN in the California Sierra Nevada and the

Caribbean?

2. How does aerosol chemistry affect IN activation?

3. What are the sources of biological particles sampled in California and over the

Caribbean and at which temperatures are they IN active? Are terrestrial or

marine biological species more prevalent in mixed phase clouds?

4. What are the effects of local pollution compared to long-range transported

aerosols, both natural and anthropogenic, on clouds in California?

1.6 Synopsis of Thesis

This thesis includes the results from two major aircraft campaigns: CalWater

(http://www.esrl.noaa.gov/psd/calwater/) and the Ice in Clouds Experiment- Tropical

(ICE-T) (https://www.eol.ucar.edu/field_projects/ice-t). Results from in-situ sampling of aerosol size and chemical mixing state, cloud droplet and ice crystal residues (size, chemical mixing state, phase), and cloud microphysics (droplet and ice crystal size, type, and concentration) were combined with statistical analysis methods, satellite retrievals, and ground-based precipitation measurements to provide insights into aerosol-cloud interactions. Using this combination of measurements and analysis, the work in this thesis shows that transported aerosols from Asia and Africa affected clouds in California and biological species play an important role in mixed phase marine clouds at temperatures above -15 °C. Chapter 2 examines the conditions under which inlet artifacts are produced

9 and details the chemical characterization of the artifacts in order to determine their contributions to the chemical composition of mixed phase cloud residues. Inlet artifact production limits the types of cloud conditions that can be sampled effectively; therefore, determining when artifacts are present and if certain markers can be used to identify them is essential to understanding cloud chemistry in mixed phase clouds. Chapter 3 documents a case study where long-range transported pollution impacted clouds over

California. This study shows via in-situ measurements that soot from Asia was present in clouds with small droplets. Previous studies have shown a reduction in precipitation in

California resulting from smaller cloud droplets [Rosenfeld et al., 2008] and Asian pollution transport to the western United States occurs frequently [Hadley et al., 2007;

Liang et al., 2004; Liu et al., 2003; Wuebbles et al., 2007; Yienger et al., 2000]. Together these results suggest that future increases in Asian pollution could have a large impact on cloud properties and possibly affect precipitation amounts in California. Chapter 4 uses a unique suite of measurements to investigate the impact of long-range transported dust and biological particles on cloud properties and precipitation in the California Sierra Nevada.

Dust and biological particles transported from over Asia and Africa were frequently observed in ice in clouds over California. Using aircraft observations this study identified dust and biological particles as the ice nuclei in upper level clouds and showed that these particle types were present in ice-induced precipitation in the region. Chapter 5 describes a novel approach for determining IN activation temperature ranges using in-situ cloud residual and microphysical measurements and identifies a biological particle type that is

IN active as warm as -10 to -13 °C. Additionally, binary logistic regression was used on cloud residual data to identify biological species as key chemical components in efficient

IN in both CalWater and ICE-T. These results increase the overall understanding of the key chemical species in aerosols that are IN active at warm temperatures. In particular, biological components are shown to be key to ice formation regardless of the composition of the rest of the particle. Principal component analysis is used on cloud residual data in Chapter 6 to investigate the sources of key chemical components present in ice in clouds both in the California Sierra Nevada and over the Caribbean Sea. Marine sources were shown to be important in both regions and during Pacific transport periods in CalWater, potassium phosphate is shown to be frequently present in ice. Overall, the work presented herein uses in-situ cloud residual and microphysical data to investigate the chemical composition and sources of ice nuclei and expands our understanding of how aerosols affect mixed phase cloud processes.

Chapter 2. Dust, Bioparticles, or

Artifacts? Unraveling the Sources of

Metals in Cloud Residues

2.1 Abstract

Aerosols indirectly affect climate by modifying cloud properties via nucleation of cloud droplets and ice crystals. Studies of heterogeneous ice nucleation processes are essential to understanding indirect climate effects; however, interpreting ice nuclei (IN) composition on the basis of analyzing the residual particles from collected and evaporated cloud droplets and ice crystals is made challenging by potential sampling artifacts and other factors. Sampling artifacts can include the generation of metallic particles when larger ice crystals or droplets collide with or break off of any part of a sampling inlet. Using results from a combination of in-situ and laboratory measurements, this study shows that the mass spectral signatures of metal artifact particles during two aircraft campaigns, CalWater and the Ice in Clouds Experiment – Tropical (ICE-T), in mixed phase clouds can be isolated from other metal-containing particles, such as mineral dust and sea spray. A relatively low percentage of particles, 4% during ICE-T and 2% during CalWater, were made up of inlet artifacts. It is shown herein that metals, such as iron and chromium, are often associated with biological markers in mixed phase cloud residues and, importantly, this same association between biological species and metals is

12 also observed in a significant fraction of laboratory generated sea spray aerosols during periods with bacterial enrichment. The observed correlation between metal and biological markers in both controlled sea spray studies and aircraft measurements suggests that metal-containing bioparticles, which are commonly excluded from analysis based on the presence of inlet metals, could play an important role in ice formation processes in marine mixed phase clouds.

2.2 Introduction

Aerosols can serve as nucleation sites for the formation of cloud droplets and ice crystals [Pruppacher and Klett, 2010]. Ice crystals grow to precipitable sizes faster than liquid droplets, which results in more efficient precipitation initiation when ice is present

[Pinsky et al., 1998]. Thus, precipitation predictions rely on an accurate understanding of ice formation. Unfortunately, the sources, chemical characteristics, and activation temperatures of ice nuclei (IN) are poorly understood [Cantrell and Heymsfield, 2005].

Laboratory experiments can provide insight into the ice nucleation temperatures of specific aerosol types [Hoose and Mohler, 2012]. Additionally, aircraft sampling of mixed phase clouds provides the opportunity to examine the ice nucleating ability of aerosols in-situ [Wendisch and Brenguier, 2013]. However, there are many complications to sampling clouds that contain ice [Baumgardner et al., 2012] including artifact production in inlets, which has been previously observed in cirrus [Cziczo et al., 2013;

Murphy et al., 2004] and mixed phase clouds [Stith et al., 2011].

Generation of metal particles from ice abrasion of the inside of inlets, which are typically composed of stainless steel or titanium, as well as dislodgement of particles

13 adhered to inlet walls have been suggested as artifact formation mechanisms [Murphy et al., 2004]. Additionally, large droplets and ice crystals (> 100 µm) can break up within inlets due to their substantial inertia, which contributes to artifact formation [Twohy et al.,

2003]. Some studies have taken a conservative approach and label all particles with metals, particularly iron, as artifacts and remove them from the analysis in addition to removing all particles from periods where 25% or more of the particles are identified as artifacts [Cziczo et al., 2013]. Unfortunately, cloud conditions that produce copious amounts of inlet artifacts are often the same conditions where little information is known about the cloud nuclei, including periods with large ice crystals (> 100 µm) or large supercooled liquid droplets. Understanding which compositions nucleate ice versus those which do not is essential to predicting cloud microphysical properties, including the probability of precipitation.

The presence of metals in ambient aerosols and cloud residues highlights the importance of being able to differentiate metals from inlet artifacts, ambient aerosols, and cloud residues. Industrial metal particles have been observed in cirrus clouds [Cziczo et al., 2013; Twohy and Poellot, 2005], cloud water samples from China [Li et al., 2013], and ambient aerosols [Ault et al., 2012; Lim et al., 2010; Moffet et al., 2008; Pacyna and

Pacyna, 2001; Park and Dam, 2010]. Furthermore, metals have been observed in sea spray bacteria-related particles [Guasco et al., 2013; Prather et al., 2013]. The presence of metals and biological species on the same particles could lead to enhanced IN ability as both biological particles [Conen et al., 2011; Maki et al., 1974; Morris et al., 2004b;

O'Sullivan et al., 2013; Yankofsky et al., 1981] and lead [Cziczo et al., 2009] have been

14 shown to nucleate ice efficiently. The collocation of metal with biological particles could be due in part to sea surface microlayers being enriched in transition metals [Barker and

Zeitlin, 1972], with the metals commonly associated with organic matter [Barker and

Zeitlin, 1972; Wallace, 1982]. Additionally, microorganisms can chelate iron and move it into cells using molecules called siderophores [Saha et al., 2013]. Biofilm production on many surfaces, including metal, and biofouling are also known processes that contribute to the collocation of biological residues with metals [Beech and Sunner, 2004; Costerton et al., 1995; Lappin-Scott and Bass, 2001; Lappin‐Scott and Costerton, 1989].

As dust, industrial, and biological particles can often contain some of the same metal constituents as those used in inlets, including iron, aluminum, titanium or chromium [Guieu et al., 2002], it is virtually impossible to make conclusions on any data based solely on the presence or absence of individual metal constituents. In order to better understand the major sources of IN in mixed phase clouds, this study strives to determine the conditions and sources leading to metals in ice residuals. Using the methodology developed herein, our results suggest that all metal particles should not be excluded from mixed phase cloud analysis and useful chemical insights can be gained on the unique particle types involved in ice nucleation processes.

2.3 Methods

2.3.1 Ice in Clouds Experiment – Tropical (ICE-T)

ICE-T flights were flown on the National Science Foundation/ National Center for Atmospheric Research C130, which was stationed in St.Croix, USVI. Thirteen flights were flown over the Caribbean Sea between July 1 and 30, 2011. The flights targeted

15 developing marine convective clouds at temperatures between 31 and -14 °C with the goal of observing first ice formation and development of the clouds. During ICE-T cloud particles were sampled using a titanium CVI (National Center for Atmospheric Research design). The CVI had a titanium porous tube, with the tip and downstream tube made of solid grade 2 titanium to limit artifact production. However, parts of the inlet were machined via electro-discharge machining (EDM) using a brass wire, which can impart a thin recast layer of copper and zinc [Choudhary et al., 2010]. Ice crystal habit was characterized using a two-dimensional cloud probe (Fast-2DC, Particle Measuring

Systems) during ICE-T.

Periods of possible artifact contamination during ICE-T were assessed by comparing Cloud Droplet Probe (CDP) cloud droplet number concentrations to Ultra

High Sensitivity Aerosol Spectrometer (UHSAS) number concentrations measured after the CVI. Regions where the UHSAS concentrations were greater than the CDP concentrations by 100 cm-3 or more were marked as shattering periods and are likely to be contaminated by inlet metals. Figure 2.8 shows example CDP and UHSAS data for research flight 9. Likely artifact periods are boxed in black. Artifacts could also be produced during periods with large condensed water contents, which indicate the presence of large liquid droplets or drizzle. These periods were marked by the CVI condensed water content monitor measuring liquid above its maximum threshold, approximately 2 g m-3 for most conditions.

Particles for microscopy studies during ICE-T were collected using a downstream impactor and chemically analyzed using Energy Dispersive Spectroscopy (EDS),

16 backscattered electron (BSE), and Scanning Electron Microscopy (SEM). EDS data were used to calculate mass percentages by using the EDS signal ratioed to the particle area calculated by using the 2D area of the images. Mass percentages were used to calculate the fraction of particles that contained copper or titanium. Standard error values for EDS data were calculated assuming simple random sampling with replacement and represent one standard deviation.

2.3.2 CalWater

The CalWater 2011 flight campaign took place out of the McClellan Airfield in

Sacramento, CA. Twenty eight flights were flown on the Department of Energy

Gulfstream-1 (G1) between February 5, 2011 and March 6, 2011. One goal of CalWater was to investigate the impacts of aerosols on precipitation in California; thus, orographic and convective clouds were targeted. During CalWater, cloud droplet and ice crystal residues were sampled through a Counterflow Virtual Impactor (CVI, Model 1204, BMI), which was made of 304 stainless steel. The inlet also has an expansion cone made of aluminum. Periods with ice crystals were identified using a two-dimensional stereo probe

(2D-S, SPEC Inc.). Relative humidity (RH) was measured with a closed-cell tunable diode laser (TDL) water vapor system.

During CalWater, UHSAS measurements behind the CVI were not made.

Therefore, periods of possible artifact contamination were evaluated using CVI inlet parameters including temperature and flow rates. The following criteria were used to flag periods of possible artifact contamination:

1. Sample temperatures below freezing, which could result in ice and liquid

droplets not melting or evaporating fully.

2. Decreased flow out of the tip of the inlet (addflow), indicating possible inlet

tip blockage from freezing of supercooled water onto the tip of the CVI inlet.

3. Large RH spikes during descents on multiple flights, which could indicate

possible melting or breakoff of ice frozen on the inlet walls.

Figure 2.9 shows an example of a sample temperature below zero, a decreased inlet tip flow and RH spikes. All of these periods were flagged as potential artifact periods, and the spectra of residuals were investigated.

2.3.3 A-ATOFMS

In both studies, an aircraft aerosol time-of-flight mass spectrometer (A-ATOFMS)

[Pratt et al., 2009b], an aircraft ready, compact version of the ATOFMS [Gard et al.,

1997], was used to measure the chemical composition and size of individual aerosol particles, cloud droplet, and ice crystal residues. An aerodynamic lens inlet [Liu et al.,

1995b] is used to focus aerosols (0.1-2.5 μm vacuum aerodynamic diameter) which are then sized based on the time it takes the particles to travel 6 cm between two 532 nm lasers. Then, a 266 nm laser is used to desorb, ionize, and chemically analyze individual particles. Standard error values for A-ATOFMS data were calculated assuming simple random sampling with replacement and represent one standard deviation.

2.3.4 Scratch Tests

The CVIs used in CalWater and ICE-T were composed of stainless steel and titanium, respectively. In order to determine the prevalent metals detected from each inlet

18 by the ATOFMS, laboratory scratch tests were used to dislodge metals that were sent directly into the ATOFMS for chemical analysis. To simulate the stainless steel CVI used during CalWater, 304 stainless steel was used. Grade 2 titanium, which is composed primarily of titanium with small amounts of aluminum, vanadium, and iron, was used to simulate the titanium CVI used during ICE-T. The stainless steel and titanium plates were scratched with a file, as well as ice cooled down to approximately -80 ºC using dry ice, to produce small metal pieces. The ice was cooled with dry ice to ensure the ice was as hard as possible to simulate the conditions encountered during flights. The scratched metals were placed in milli-q water (18.2 MΩ) and the ice samples were melted and both sample types were atomized into the ATOFMS to obtain mass spectra of pure inlet metals for comparison with the metals detected in each of the studies.

2.3.5 Laboratory Studies of Sea Spray Aerosol

To better understand the origin of a commonly detected salt-metal-bioparticle in cloud residuals [Creamean et al., 2013b], a comparison was made with sea spray aerosol generated from a Marine Aerosol Reference Tank (MART) [Stokes et al., 2013].

Seawater was taken from the Pacific Ocean and placed in the MART. Then, a mesocosm experiment was performed by adding inorganic nutrients to seawater to initiate a phytoplankton bloom [Roman et al., 1988]. Throughout the course of the bloom, sea spray aerosols were generated and sampled with an ATOFMS. It is important to note that any metals associated with the laboratory generated sea spray aerosol (SSA-lab) are specific to particles in nascent sea spray based on the fact that SSA-lab was sampled directly from the MART tank constructed from plexiglass and silicone. Stainless steel

19 ports were used on the tank; however, contact with the metal ports was limited. In addition, no large liquid droplets or ice crystals could impact or abrade the metal.

2.4 Results

2.4.1 Inlet Composition and Mass Spectra

As discussed in the Methods section, scratch tests involved scratching the inlet metals with a file and ice to generate small metal particles which were directly analyzed using ATOFMS. Representative spectra from the scratch tests are shown in Figures 2.1 and 2.2. Titanium (48Ti+) and titanium oxide (64TiO+) were the most prominent peaks from the titanium scratch sample with a small aluminum (27Al+) peak, as seen in Figure

2.1A. Figure 2.2A shows a spectrum from scratching 304 stainless steel. The most prominent ion peaks are chromium (52Cr+) and iron (54/56Fe+). The chromium peaks were generally larger than the iron peaks. Previous single particle mass spectral studies commonly detected molybdenum and nickel, in addition to chromium and iron, in their stainless steel artifact spectra using a different single particle mass spectrometer, the

Particle Analysis by Laser Mass Spectrometer (PALMS) [Murphy et al., 2004].

Noticeably absent in the ATOFMS scratch test spectra were molybdenum (96Mo+) and nickel (59Ni+) ion peaks, which could be due to the different wavelengths used for laser desorption and ionization in the ATOFMS (266 nm) versus the PALMS (193 nm)

[Murphy and Thomson, 1997]. Mass spectra of these scratch test metals indicate that while iron, chromium, and titanium are generated, the inlet metal-enriched particle mass spectra are quite distinct from dust and biological particles previously sampled with the

ATOFMS, as described in Creamean et al. [2013b]. Dust samples frequently contain

20 aluminum, iron, titanium, and silicate ion peaks; however, ATOFMS mass spectra of dust typically do not contain significant chromium peaks, likely a result of only trace chromium concentrations in dust [Engelbrecht and Jayanty, 2013]. Therefore, when large chromium peaks are observed along with iron in cloud residuals, it suggests that these metals are stainless steel artifacts. Additionally, when titanium and titanium oxide are present without other dust markers, including silicates [Silva et al., 2000], the titanium is likely a result of inlet abrasion based on typical known dust mass spectra and scratch test results. The distinct ion signature differences from inlet versus dust metal combination can be used to chemically identify artifact particles sampled in-situ. In addition, as described in the next section, metal information in combination with microphysics data can be used to more definitively isolate periods with artifact contamination.

2.4.2 Isolating Artifact Periods

Inlet contamination periods could be isolated using the criteria described in the methods section, which includes conditions such as below freezing sample temperatures, large droplets and ice crystals (> 100 µm), and conditions conducive to breakthrough of ice off the inlet walls. An example of a titanium-rich particle sampled during a CVI saturated period, defined by the condensed water content being above the maximum threshold (~2 g m-3), is shown in Figure 2.1B. The titanium and titanium oxide markers are intense and aluminum (27Al+), lithium (7Li+), and sodium (23Na+) are also present.

With the exception of calcium (40Ca+), the scratch test sample (Figure 2.1A) and the inlet artifact spectra (Figure 2.1B) look nearly identical. Additionally, under conditions with large ice crystals, the samples from CalWater (Figure 2.2C) had large aluminum (27Al+)

70 + 52 + 54/56 + and aluminum oxide ( Al2O ) markers as well as chromium ( Cr ) and iron ( Fe ) ion markers (Figure 2.2B). These mass spectra look very similar to scratch test samples with large chromium markers (52Cr+), shown in Figure 2.2A.

Comparisons between cloud residues and scratch tests demonstrate metals can be scraped from the inlet surface during flights and that the artifact metals have characteristic mass spectral signatures that can be utilized to distinguish between inlet artifact particles and metal-containing cloud residue data. Based on mass spectral signatures, inlet artifacts typically made up a small percentage of particles during

CalWater. Particles with large chromium peaks made up only 2.4 ± 0.2 % of all particles sampled through the CVI and 4.8 ± 0.5 % of particles sampled during ice periods in

CalWater, which were identified using 2D-S imagery. Aluminum-rich particles, shown in

Figure 2.2C, that lack other identifiable dust signatures, are also attributed to inlet artifacts based on the fact that this type is only seen in cloud residues and not in clear air aerosols during CalWater. These Aluminum-rich artifact particles were easily distinguishable from other aluminum-containing particles and their percentages were generally low (0.8 ± 0.1% in all CVI particles, 2.2 ± 0.3% in ice).

2.4.3 In-Situ Cloud Residue Data

Metal-containing cloud residues and clear-air aerosols were compared to inlet metal spectra to differentiate between inlet artifacts and naturally occurring metal particles. A variety of metals were present in cloud residues collected during both

CalWater and ICE-T, including titanium and copper during ICE-T and iron and aluminum during CalWater. Dust observed during CalWater had various metals

22 associated with it as seen in the dust spectrum in Figure 2.3. Iron (54/56Fe+) and aluminum

27 + 77 - 92 - 120 - ( Al ) were present along with silicates ( SiO3 , SiO4 , Si2O4 ). The observed dust lacked the large chromium (52Cr+) ion peak present in the inlet artifacts, thus it is unlikely that the commonly detected metals in the dust were the result of artifacts during

CalWater.

Figure 2.4 shows titanium-containing particle types observed during ICE-T.

Figure 2.4A shows a dust particle containing titanium, a naturally occurring metal in dust, especially Asian dust [Sullivan et al., 2007]. Titanium was also observed in other particle types that look similar to sea salt (B-C): some with small titanium markers (B) and some with large titanium markers (C-D). The mixed titanium and salt particles lack other dust signatures, including silicates, thus these particles are likely the result of sea salt mixed with metal from the inlet. A study using the same titanium CVI inlet in cold clouds showed titanium was more prevalent in large particles (> 1 um) [Stith et al., 2011] and the percentage of titanium particles they observed around 2 µm was between 7-8% of total particles. For submicron particles more consistent with the A-ATOFMS peak size mode, the percentage of titanium particles ranged from 1.5-4.5%. In ICE-T, the

A-ATOFMS data showed 3.9 ± 0.2% of CVI particles had titanium, in agreement with previous results.

Copper and zinc have been observed in other studies that used the same titanium

CVI inlet as in ICE-T [Twohy, 2013]. During ICE-T, A-ATOFMS data did not show significant amounts of zinc present on the particles that contained copper. Figure 2.5 shows copper-containing particles observed during ICE-T. These particles were sea salt

23 mixed with various components: copper (A), copper and biological markers (26CN-,

42 - 79 - 133, 135, 137 - CNO , PO3 ) (B), copper, biological markers, and copper chloride ( CuCl2 )

(C). Additionally, large CuCl2 markers were present on several metal particle types, which may be indicative of copper leaching from copper-containing metal (i.e. brass or brazing alloys) with a sea salt coating. EDS data of ICE-T cloud residues sampled with a downstream impactor also show evidence of copper and titanium in a CVI sample. Figure

2.6 shows pie charts of the fraction of metal-containing particles and residues using

A-ATOFMS data in the top row and EDS data on the bottom row. All particles, including inlet artifacts and suspected artifact-rich periods, are included in Figure 2.6 to give a high end estimate for the possible amount of artifact particles sampled. Copper was present on

37.4 ± 0.5% of A-ATOFMS particles sampled through the CVI and 3.9 ± 0.2% had titanium markers compared to 18 ± 1% of EDS particles from a CVI sample contained copper and 6.7 ± 0.9% had titanium. The EDS data also revealed the CVI ice particles frequently contained both copper and zinc, while the CVI liquid residues contained mostly copper. Copper could have been leached by sea salt and washed off in the large liquid droplets, as suggested by the extensive presence of copper chloride in the cloud residues. Additionally, ice could scrape off a chunk of metal containing both copper and zinc on the same particle. During ice periods, 71 ± 1% of copper particles had biological markers, which points to a possible biofilm on the inlet that could aid in leaching copper out of the inlet. Figure 2.8 shows periods of likely artifact contamination plotted with copper and biological particle counts. These results show that the amount of biological and copper particles is not entirely dictated by the artifact-prone conditions; however, the largest amounts of copper and biological particles do occur during the periods where the

CVI UHSAS concentrations are higher than the CDP concentrations. This further suggests that the copper particles are likely the result of inlet contamination.

IN activity during CalWater and ICE-T is investigated in Chapter 5. Those results reveal that the IN activity is similar between mixed salt and biological particle types observed during both studies. The main chemical difference between these types is the metals that are associated with them. During CalWater, aluminum and iron were more commonly associated with the salt and biological particles, while copper was dominant in the ICE-T particles. The copper is most likely an inlet artifact, but the fact that the rest of the cloud residue mass spectrum matches between both studies and the IN activity is similar suggests that the cloud residues are likely representative of biological particles found in the clouds. Excluding all metal particles would result in an underestimate of particle types that are commonly associated with metals, such as biological species. Thus, it is advised that other factors, such as those described in Sections 2.4.1 and 2.4.2, are used as well.

2.4.4 Metals in Biological Particles

Metals have been shown to be associated with biological activity in sea spray aerosol [Guasco et al., 2013; Prather et al., 2013]. Iron and chromium have been shown to be enhanced in sea spray aerosol generated using breaking waves in a wave flume after heterotrophic bacteria were added [Prather et al., 2013]. A related control study was conducted by sampling an inorganic salt solution using the ATOFMS before and after the addition of heterotrophic bacteria, which resulted in an increase in biological markers

26 - 42 - 79 - ( CN , CNO , PO3 ) after addition [Guasco et al., 2013]. These results together

25 suggest that the presence of iron and chromium in sea spray may be a good indicator of the presence of marine bacteria, which demonstrates the collocation of metals and biological species on particles and further adds importance to the proper identification of metals in aerosol and cloud residue sampling. Additionally, in similar sea spray laboratory studies conducted using the MART system, mass spectra of sea spray match signatures observed in cloud residues, as shown in Figure 2.7. In Figure 2.7, the chemical signature of sea spray aerosol (SSA-lab) is plotted on top of the chemical signature of a metal-containing biological cloud residue sampled during CalWater. The two sources show almost identical mass spectra, with both types containing large iron ion markers

54/56 + 26 - 42 - 79 - ( Fe ) and biological markers ( CN , CNO , PO3 ). This highlights that biological particles are commonly associated with metals and suggests that by removing all iron- containing particles one risks selectively removing the key particle types involved in ice formation. During CalWater, 83 ± 3% of iron particles had biological markers in ice, while in ICE-T 74 ± 7% of iron particles had biological markers in ice. Therefore, if all iron particles were removed a large amount of biological particles would be removed from the analysis.

2.5 Discussion/Summary

Here we have shown that there are many sources of metal-containing particles sampled through aircraft inlets including dust and biological particles, as well as inlet artifacts. Laboratory scratch tests were used to chemically identify and characterize the combinations of metals that can serve as indicators of inlet artifacts. Given that there are many other sources of iron and other inlet metals, critical information on the chemical

26 species in mixed phase cloud nuclei is lost when all inlet-metal-containing particles, such as iron, are omitted. This study demonstrates that one must use more than the presence of a single metal as a screening tool for mixed phase cloud residues:

1. Inlet metal chemical signatures identified using scratch tests are distinct from

other metal-containing particles including dust and bioparticles in sea spray.

2. It is important to use both the scratch test metal signatures coupled with

information on periods of likely artifact contamination. ICE-T showed a higher

percentage of particles containing copper (37%) due to material used to construct

one part of the inlet. During CalWater, with a stainless steel inlet only 2% of

cloud residues contained chromium, while during ICE-T the titanium CVI led to

titanium artifacts making up 4% of all cloud residues sampled.

3. We have developed a stronger understanding of a potential source of bioparticles

in sea spray. Specifically, the chemical signatures of laboratory generated sea

spray are nearly identical to iron-containing biological species sampled in cloud

residues. These lab and field results together verify that iron can be prevalent in

bacteria-enriched sea spray.

Taken together, these pieces of evidence suggest that the presence of specific metals alone, such as iron, is not enough to label a particle as an inlet artifact. The entire chemical signature of metal-containing particles must be examined. Rejection of all metal-containing particles will selectively eliminate specific particle types such as dust or metal-containing biological particles from the analysis. Such restrictive analysis will not

27 allow one to gain any insights into the sources of aerosol particles involved in ice nucleation.

Stainless steel inlets appear to limit the amount of artifacts produced; however, even small metal pieces that underwent electro-discharge machining appear to have large impacts on artifact production. To extend our understanding of how inlet abrasion can affect particle composition, detailed studies of aircraft inlets and artifact formation and drying processes are necessary. We suggest wind tunnel experiments that can investigate the morphology of residues as they progress through the inlet and water is evaporated to leave a residue behind and can probe the effects of different inlet metals on artifact production. Probing the factors leading to metal artifact production and the subsequent effects on morphology and chemical composition will provide valuable information about the relative contributions of artifacts to cloud residuals and how one can successfully distinguish between artifact and aerosol metals.

2.6 Acknowledgments

Chapter 2 contents are part of a manuscript in preparation, Kaitlyn J. Suski,

Cynthia H. Twohy, James R. Anderson, Daniel Rosenfeld, Jennifer M. Comstock,

Andrew J. Heymsfield, and Kimberly A. Prather (2014), “Dust, Bioparticles, or

Artifacts? Unraveling the Sources of Metals in Cloud Residues”, In preparation for submission to Aerosol Science and Technology. Funding was provided by the National

Science Foundation under award number 1118735 and the California Energy

Commission under contract CEC 500-09-043. Alberto Cazorla is acknowledged for his efforts in collecting A-ATOFMS data during ICE-T. Doug Collins, Jess Axson, Chris

Lee, Camille Sultana, Matt Ruppel, and Mitch Santander are acknowledged for running the MART tank and collecting ATOFMS sea spray data. Darin Toohey is acknowledged for his role in operating the CVI and collection of UHSAS data during ICE-T.

2.7 Figures

Figure 2.1. Comparison of spectra of titanium particles from (A) a scratch test sample from a sheet of grade 2 titanium and (B) a saturated CVI period during ICE-T.

Figure 2.3. Mass spectrum of dust sampled during CalWater.

Figure 2.5. Mass spectra of copper-containing particles sampled during ICE-T: sea salt with copper (A), sea salt with bio and copper (B), and bio-sea salt with copper chloride (C).

Figure 2.6. Pie charts showing the percentage of metal-containing particles, copper and titanium from ICE-T and chromium from CalWater, sampled through a CVI. A- ATOFMS data is shown in the top row and represents all CVI data during the studies. EDX data is shown on the bottom row and shows one representative CVI sample from research flight 9 on July 23, 2011.

Figure 2.7. Mass spectra of a sea spray aerosol sampled with the MART tank plotted with an iron-rich particle sampled through a CVI during CalWater.

2.8 Supplemental Figures:

Figure 2.8. The top panel shows CVI UHSAS versus CDP data measured during reasearch flight 9 of ICE-T (July 23, 2011). Periods with CVI UHSAS counts higher than CDP counts are indicative of shattering and are boxed in black. The number of particles that contain copper (orange) and biological (teal) markers are shown in the bottom panel.

Chapter 3. Long Range Transport of

Asian Soot Impacts Cloud Properties over

California

3.1 Abstract

Pollution plumes emanating from Asia can be transported intercontinentally resulting in regional surface dimming, atmospheric warming, and altered circulation patterns. This study is the first to show via coupled in-situ chemical and microphysical cloud measurements that soot and pollution transported from Asia strongly impacted cloud properties over the California Sierra Nevada. During the CalWater field campaign, the soot impacted an area of approximately 590 square mile region over California and had a chemical signature identical to pollution observed in Shanghai, China just 8 days before. This pollution lead to the formation of numerous extremely small cloud droplets

(mean radii of 5 µm). Local California pollution had been implicated in reducing orographic precipitation by reducing cloud droplet sizes, but here we show that when clouds were decoupled from the boundary layer Asian pollution reduced cloud droplet size below the warm rain threshold (Re = 14 µm) [Pinsky and Khain, 2002; Rosenfeld et al., 2002] in mid-level clouds. Thus, future increases in Asian pollution could also result in precipitation reduction in California. Importantly, long range transported pollution aerosols would also indirectly impact ground-level air quality, as less rain results in more

38 pollution build-up. The coincident timing of California’s peak precipitation season with

Asian pollution transport amplifies the importance of documenting and quantifying the impacts of such long range transport events.

3.2 Introduction

In California, significant efforts have been made to reduce air pollution by reducing emissions from local sources, thereby increasing the relative importance of long-range transport of pollutants at high altitudes. Trans-Pacific transport of Asian aerosols has traditionally been though to occur in spring [de Gouw et al., 2004; Jaffe et al., 1999; VanCuren and Cahill, 2002]. More recently, it has been shown that trans-

Pacific transport of pollutants, such as carbon monoxide (CO) and soot produced from the burning of fossil fuel and biomass, from Asia to the western United States occurs between February and April above 2 km [Hadley et al., 2007; Wuebbles et al., 2007;

Yienger et al., 2000]. This timeframe coincides with slash-and-burn agricultural fires in

Southeast Asia, which occur from October through June, peaking in March [Duncan et al., 2003; Streets et al., 2003]. Long-range transported fossil fuel and biomass burning aerosols from Asia have been observed at high altitudes over the western United States

[de Gouw et al., 2004]. Recent studies have shown that the majority of black carbon emitted in China is from fossil fuel combustion [Chen et al., 2013], while in South Asia both biomass burning and fossil fuels contribute [Gustafsson et al., 2009]. A study in

Shanghai showed that influences from elemental carbon and biomass burning emissions changed based on wind direction [Huang et al., 2013].

Soot specifically has been identified as a key player in climate change with only carbon dioxide having a larger effect [Ramanathan and Carmichael, 2008], thus the frequent trans-Pacific transport of Asian soot has significant climatic implications. Asian pollution can be lofted into the atmosphere via frontal, convective, or warm conveyor belt lifting, and is then transported across the Pacific by flow in the free troposphere [Liang et al., 2004; Liu et al., 2003]. More recently, Yu et al. [2012] showed via satellite observations that trans-Pacific transported aerosols can contribute up to 50% of the total

North America aerosol mass, with the majority of those aerosols at high altitudes. It was estimated that over 75% of soot transported into North America in April of 2004 originated in Asia [Hadley et al., 2007]. The processing of Asian aerosols during transport was studied during the Intercontinental Chemical Transport Experiment-Phase

B (INTEX-B) field campaign during which long-range transported Asian aerosol was identified to be enriched in sulfate [Dunlea, 2009]. Trans-Pacific transported carbonaceous aerosols have been shown to have heating effects across the United States

[Teng et al., 2012]; however, the impacts of those aerosols on North American clouds and precipitation have not been investigated.

Soot has many effects on clouds including absorbing solar radiation and heating the surrounding air (aerosol direct effect). This heat can evaporate proximate water vapor or cloud droplets (semi-direct effect) [Hansen et al., 1997] and increase local atmospheric stability [Ackerman et al., 2000]. Absorbing aerosols produced by combustion include soot and brown carbon and have been shown to heat the atmosphere more when located within cloud droplets as opposed to being located in between droplets [Jacobson, 2012].

Additionally, when absorbing aerosols are located above clouds their effect on radiative

40 forcing is increased [Podgorny and Ramanathan, 2001]. Pollution aerosols have been shown to form clouds with more numerous, smaller droplets, which reflect more solar radiation (Twomey effect) [Twomey, 1974] and have increased cloud lifetimes (Albrecht

Effect) [Albrecht, 1989]. The presence of increased pollution aerosol in clouds can increase or decrease the amount of precipitation depending on the environment [Khain et al., 2008]. In dry continental and orographic locations, smaller droplets as a result of pollution lead to suppression of precipitation [Khain et al., 2008].

Results from the Suppression of Precipitation (SUPRECIP) flight campaign in the

California Sierra Nevada Mountain Range demonstrated that higher aerosol concentrations caused more numerous smaller droplets in the area [Rosenfeld et al.,

2008], which increased the height to which the clouds had to develop above their bases to start precipitating. Measurements showed the higher aerosol concentrations were mixed into the clouds from the boundary layer of the California Central Valley, and thus attributed to local sources of air pollution. Trends of increasing aerosols were also found to be related to decreasing trends of orographic precipitation over the western USA [Givati and Rosenfeld, 2004; Rosenfeld and Givati, 2006] and China [Rosenfeld et al., 2007].

The CalWater field study was initiated to study aerosol and atmospheric river effects on winter precipitation in the California Sierra Nevada. Atmospheric rivers transport large amounts of water vapor in focused bands from the tropics [Zhu and

Newell, 1998] and have been shown to be associated with heavy rainfall and flooding in

California [Ralph et al., 2006]. A key hypothesis of CalWater was that local Central

Valley pollution was mixing into orographic clouds causing the formation of smaller droplets, resulting in suppression of precipitation [Rosenfeld et al., 2008]. CalWater

41 builds upon the SUPRECIP results, which included cloud microphysical measurements, by adding in-situ chemical composition measurements using single particle mass spectrometry (SPMS) with the main goal of identifying the sources of cloud seeds in the region. The aerosol time-of-flight mass spectrometer (ATOFMS) was used to differentiate aerosol pollution sources including fossil fuels [Shields et al., 2007;

Sodeman et al., 2005; Spencer et al., 2006; Toner et al., 2006], biomass burning [Qin and

Prather, 2006; Silva et al., 1999], ship emissions [Ault et al., 2010], and dust [Guazzotti et al., 2001a; Guazzotti et al., 2001b; Silva et al., 2000]. In this study, we show that in cases where clouds were decoupled from the boundary layer, long-range transported soot, as opposed to local pollution, was responsible for the reduction of cloud droplet radii over the California Sierra Nevada. We also show for the first time regional climate impacts in California of long-range transported pollution.

3.3 Measurement Details

The CalWater 2011 flight campaign (5 February 2011 – 6 March 2011) consisted of 28 flights conducted out of the McClellan Airfield in Sacramento, CA. The flights were flown on the Department of Energy Gulfstream-1 (G1) and multiple flight tracks spanned from the eastern Pacific Ocean to the Sierra Nevada Mountain Range. The G1 possessed a unique suite of instruments described in detail in Rosenfeld et al. [2013], which were chosen to probe both aerosol and cloud properties. Static temperature was measured using a Rosemount 102E temperature probe. Latitude, longitude, altitude, wind speed and wind direction were measured using an Aircraft-Integrated Meteorological

Measurement System (AIMMS-20, Aventech Research Inc.). Aerosols were sampled

42 through an isokinetic aerosol inlet (Model 1200, BMI), while cloud droplet and ice crystal residues were sampled through a Counterflow Virtual Impactor (Model 1204,

BMI). Cloud droplet diameters (2-50 µm), droplet concentrations (particles per cm3), and liquid water content (LWC, g/m3) were measured with a Cloud Droplet Probe (CDP,

Droplet Measurement Technologies). The chemical composition and size of individual aerosol particles and cloud droplet residues were measured with an aircraft aerosol time- of-flight mass spectrometer (A-ATOFMS) [Pratt et al., 2009b]. Particles (0.1-2.5 μm vacuum aerodynamic diameter) enter the A-ATOFMS through an aerodynamic lens inlet

[Liu et al., 1995a; b] and are sized using two 532 nm lasers at a fixed distance apart to get the speed of the particle which is converted to vacuum aerodynamic diameter based on a calibration with polystyrene latex spheres of known size. Particles are then individually desorbed, ionized and chemically analyzed using a 266 nm laser. ATOFMS (TSI model

3800) data were measured in the Baoshan district in Shanghai, China for 8 daylight hours per day from February 21-25, 2011. Both A-ATOFMS and ATOFMS single-particle mass spectra were imported into MATLAB (The MathWorks, Inc.) using the YAADA toolkit (www.yaada.org) and the mass spectra were clustered using ART-2a, an adaptive resonance theory clustering algorithm [Song et al., 1999]. The clusters were then classified by hand and grouped into particle types based on combinations of characteristic ion peaks. Standard error values for relative fractions of A-ATOFMS particle types were calculated assuming simple random sampling with replacement and represent one standard deviation. Aerosol concentrations were measured with a Passive Cavity Aerosol

Spectrometer Probe (PCASP, Droplet Measurement Technologies, 0.1-3 μm). The

PCASP data were filtered to remove periods affected by clouds. The aerosol absorption

43 coefficient was measured using a Particle Soot Absorption Photometer (PSAP, Radiance

Research); data were corrected based on the methods described by Bond et al. [1999] and

Ogren [2010].

Back trajectories were calculated using the Lagrangian particle dispersion model

FLEXPART [Stohl et al., 2005] version 8.1. The model was run with NOAA Global

Forecast System 0.5° x 0.5° global data at 26 levels using 38.52 N, 121.5 W (Sugar Pine

Dam) as the point source for 10-day backward air mass trajectory calculations. Times and altitudes for the various trajectories can be found in Table 3.1. The location of fires in

Asia was determined by the Moderate Resolution Imaging Spectroradiometer (MODIS) fire detection data, a compilation of fire and thermal anomaly data from the Terra and

Aqua Satellites [Kaufman et al., 2003]. The data were plotted using the Fire Information for Resource Management System (FIRMS) Web Fire Mapper

(http://firms.modaps.eosdis.nasa.gov/firemap/). Satellite imagery including the level 2.0 aerosol layer product from the Cloud-Aerosol LIDAR and Infrared Pathfinder Satellite

Observation (CALIPSO) satellite [Winker et al., 2010] was utilized to confirm the presence of smoke in the eastern Pacific Ocean as well as track the smoke as it traveled from the source location to California.

3.4 Results and Discussion

On March 5, two flights were flown out of McClellan Airfield northeast toward the Sierra Nevada mountain range: one flight was conducted in the morning (March 5a,

17:15 - 20:17 UTC) and the other in the afternoon (March 5b, 21:33 – 00:33 UTC)

(Figure 3.8). An elevated aerosol layer was sampled from 3500-5250 m over the Sierra

Nevada during the morning flight on March 5. Figure 3.1 shows the chemical composition of aerosols and cloud residues, altitude, absorption coefficient at 532 nm,

PCASP concentrations, and ozone concentrations for the morning flight on 5 March

2011. During the flight, the Gulfstream-1 intersected the pollution layer twice as highlighted by the black bars in Figure 3.1: once between 17:40 and 18:07 UTC (S1) and a second time between 18:50 and 19:16 UTC (S2). The layer had elevated aerosol concentrations of ~200 particles/cm3 with diameters between 70 - 1000 nm according to

PCASP (Figure 3.1). For comparison, PCASP particle concentrations in free troposphere regions not in the aerosol layer were typically less than 20 particles/cm3, except for the end of the flight when concentrations approached ~600 particles/cm3 due to enhanced pollution in the boundary layer.

ATOFMS data showed the elevated aerosol layer was composed of soot, potassium-containing soot (K-soot), and highly cloud processed aerosols. Soot was

+ 12 + 24 + 36 + characterized by large elemental carbon markers, Cn ( C , C2 , C3 , etc.) with

27 + 43 + smaller organic carbon markers ( C2H3 , C2H3O ) [Moffet and Prather, 2009] and was

-97 - internally mixed with significant amounts of sulfate ( HSO4 ), suggesting long-range transport from Asia and cloud processing. K-soot particles were identified by a large potassium marker (39K+), elemental and organic carbon markers [Pratt et al., 2011; Silva et al., 1999], and had large sulfate and sulfuric acid cluster ion markers. Air masses originating in Asia are frequently enriched in sulfur dioxide, which is converted to sulfate during transport [Dunlea, 2009]. In addition, sulfur dioxide can be oxidized in clouds to form sulfate [Hegg and Hobbs, 1981]. The most highly processed particles produced no

-97 - -195 - positive ions, but had large sulfate ( HSO4 ) and sulfuric acid cluster ( H2SO4HSO4 ) ion markers in the negative spectra. These particles likely contain soot at their core with a thick coating of sulfate, which can inhibit the production of positive ions [Pratt et al.,

2011]. The presence of these highly processed particles highlights the extensive atmospheric processing this air mass underwent during transport. Since soot was the main component of the elevated aerosol layer, it will herein be referred to as the “soot layer”.

Representative spectra of these aerosol types are shown in Figure 3.2; descriptions and spectra of other aerosol types observed in this study are given in the SOM (Figure 3.9).

To estimate the size of the area over California affected by this soot layer, aerosol concentrations from the PCASP were plotted over the flight track, as shown in Figure

3.3. The high altitude soot layer, as indicated by the increased PCASP concentrations, covered a wide area and is boxed in black. Interestingly, in the same area during the afternoon flight, PCASP aerosol concentrations were much lower (Figure 3.3 right panel). The box represents the area affected by the high altitude soot and is estimated to cover roughly 590 square miles. This estimate is a lower limit, as areas outside of this region were not sampled extensively and thus the soot layer could have covered an even broader area. Soot layers can have significant impacts on regional radiative forcing if they are present frequently and cover a large spatial area.

Figure 3.1 also shows the absorption coefficient, which increased in the soot layer, confirming the presence of absorbing aerosol. Absorption Ångström exponent

(AAE) is a measure of the dependence of absorption on wavelength. This dependence can elucidate the difference between elemental or black carbon (AAE of 1) [Bergstrom et al.,

2002] and brown carbon (AAE > 1) [Kirchstetter et al., 2004]. The AAE for the soot layer varied between 1 and 1.5 (Figure 3.10), which points to the presence of both black and brown carbon and is close to the reported AAE of 1.28 for biomass burning aerosol

[Bahadur et al., 2012; Cazorla et al., 2013]. It is important to note that the soot layer was decoupled from clouds at lower altitudes (3000 m, ~1000 m below the soot layer). While the soot layer was composed of soot, K-soot and highly processed particles, the cloud droplet residues in the lower altitude clouds were composed of mainly sea salt and was thus more pristine (Figure 3.1, blue bar (C1)). Ozone levels were up to 50 ppbv in the soot layer and fairly consistent throughout the flight. Ozone plumes in the western U.S. have been linked to long-range transport of Asian pollution [Berntsen et al., 1999;

Cooper et al., 2010; Jaffe et al., 2003], and thus, the higher concentrations of gaseous pollutants around the plume further support the hypothesis of trans-Pacific transport.

During the afternoon flight, two cloud layers existed with vastly different microphysical properties. As shown in Fig. 3.4, a shift in modal diameter from 9 µm for the lower altitude clouds (grey) to 25 µm in the higher altitude clouds (blue) was observed. The clouds with a smaller modal diameter of 9 µm were thin and grey as shown in Figure 3.5 and had an effective radius well below the warm rain threshold of

Re = 14 µm [Pinsky and Khain, 2002; Rosenfeld et al., 2002]. Clearly, the large difference between the modal diameters (9 vs 25 µm) in the different cloud layers directly corresponds to changes in the chemistry of the cloud seeds. Figure 3.6 shows the

CDP Re and CDP concentration (left), the CDP LWC (middle), and the vertical distribution of the chemical composition of cloud residues (right). The lower level clouds

(3680-4400 m) had higher concentrations (up to 130 cm-1) of smaller droplets

3 (Re ~5 µm), a low LWC (0.07 g/m ) and were composed of mostly soot and K-soot particles (37±6% soot, 32±6% K-soot, 22±5 % highly processed particles, 5±3% sea salt, and 3±3% black carbon). The higher altitude cloud layer (5540-6000 m) had lower

-1 3 concentrations (20-40 cm ) of larger droplets (Re = 5-13 µm), a high LWC (0.3 g/m ), and were composed entirely of sea salt. A mid-level cloud layer (4500-5000 m) had low

-1 3 concentrations (~20 cm ) of small droplets (Re ~5 µm) and a low LWC (0.03 g/m ). The cloud was composed of 50±20% sea salt, 33±19% K-soot, and 17±15% dust.

The presence of soot and K-soot particles in the lower level clouds suggests that the soot layer, sampled during the morning flight, became incorporated into these clouds.

This assertion is supported by the presence of a temperature inversion (Figure 3.11) around 4000 m during the morning flight which dissipated by the afternoon flight. This resulted in atmospheric instability, which allowed the soot and cloud layers to become inter-mixed. The high altitude aerosol concentrations were substantially reduced between the morning and afternoon flights, as shown in Figure 3.3. This suggests the aerosols were scavenged by the clouds in the afternoon. The cloud that contained soot had a much lower modal diameter than the cloud that contained sea salt, which shows the chemistry of the cloud residues is playing a role in modifying the cloud properties. In particular, soot appears to be reducing cloud droplet size to below the warm rain threshold which could result in precipitation suppression.

To investigate the source of the soot, air mass back trajectory analysis was conducted. FLEXPART back trajectories (Figure 3.7) indicate that the high altitude air

48 mass containing soot originated over Asia 8-10 days prior to reaching California, signifying long-range transport of the high-altitude soot. Agricultural fires peak in March in Southeast Asia [Duncan et al., 2003; Streets et al., 2003] and FIRMS fire data (Figure

3.7) show several fires in Southeast Asia on February 27, 28 and March 1, where the sampled air mass originated. The mix of soot and K-soot in the soot layer and cloud residues suggests that Asian pollution from both fossil fuel combustion and biomass burning was present. CALIPSO satellite images (Figure 3.12A) confirm the presence of smoke around 5 km over the eastern Pacific Ocean one day before the high altitude soot was observed in flight. FLEXPART shows the soot layer was over the eastern Pacific

Ocean at 5 km the day before sampling. Additional CALIPSO images (Figure 3.12B) show smoke over California on March 5 and traveling east from Asia days before. All of the trajectories for March 5 originated in the Asian outflow region, which is well documented [Liu et al., 2003]. Therefore, the satellite and back trajectory data strongly suggest the elevated soot layer, which affected clouds in the Sierra Nevada region, is a result of long-range transport of soot from Asia.

To compare the chemical composition of the soot layer to Asian pollution,

ATOFMS data sampled in Shanghai, China from February 21-25, 2011 was investigated.

The Shanghai data show soot particles with similar chemical composition as the soot layer particles measured over California on March 5, 2011. Figure 3.8 shows one of the most prevalent particle types measured in Shanghai plotted with the soot type sampled during CalWater, which is offset by one mass-to-charge unit for visual clarity. The positive ion spectra look very similar with potassium (39K+) and carbon cluster (12C+,

24 + 36 + C2 , C3 ) ion markers and the negative ion spectra are also remarkably similar with

97 - 62 - large sulfate ( HSO4 ) and smaller nitrate ( NO3 ) ion markers. The similarities in these mass spectra further support the assertion that the soot observed over California originated in Asia.

The observation of smaller droplets due to soot is consistent with previous observations of higher local PM2.5 levels in the Sierra Nevada Region leading to a reduction in cloud droplet size [Rosenfeld et al., 2008], but here we show that the clouds in California that have large concentrations (130 cm-1) of small droplets (5 µm) contain long-range transported Asian soot as their cloud seeds. These results suggest that trans-

Pacific transport of Asian soot containing high concentrations of mixed soot and sulfate particles could ultimately be impacting regional precipitation over California, as smaller droplets often correlate with reductions in precipitation [Rosenfeld et al., 2008]. Even though the long-range transported soot is present at higher altitudes and thus does not directly impact air quality, the presence of these additional cloud seeds leads to a reduction in wet scavenging which will directly result in the buildup of pollution at ground levels. Thus, long-range transported Asian pollution can indirectly impact

California air quality by reducing precipitation events. A recent study showed that the warming effects of Asian ABCs could be measured as far as the eastern United States, suggesting that ABCs may contribute to atmospheric warming far from their sources

[Teng et al., 2012]. Additionally, it has been shown that compared to interstitial soot, soot found within cloud droplets increases heating rates by approximately 200% in low clouds and 235% in mid-level clouds [Jacobson, 2012]. Thus, the geographical extent of the

50 transport of Asian aerosols coupled with these results suggests that Asian soot and pollution can affect clouds over different continents, with the potential to impact both regional climate and the radiation budget.

3.5 Conclusions

Elevated concentrations of soot sampled over a broad region of California near the Sierra Nevada Mountains were measured on the morning of 5 March 2011. The soot pollution likely originated from Asia, as evidenced by identical soot mass spectra measured in pollution 8 days prior in Shanghai. The soot layer became incorporated into clouds in the afternoon flight, resulting in high concentrations of small droplets containing soot particles. Such high concentrations of local pollution aerosols from

Central Valley have been shown to cause the formation of more numerous, smaller cloud droplets in California. However, here we show that long-range transport of Asian pollution modified cloud properties over California. The transported soot is present at higher altitudes, thus the direct effect on ground-based pollution levels is low. However due to reduced precipitation and the resulting reduction in wet scavenging of air pollution, this soot will indirectly contribute to poorer California air quality. Due to the high frequency of trans-Pacific transport of pollutants at high altitudes, these results demonstrate the potential for long-range transported Asian soot affecting cloud microphysics, regional climate, precipitation levels, and warming patterns across the

United States.

3.6 Acknowledgements

Funding was provided by the California Energy Commission under contracts 500-

09-043 and 500-09-032. The deployment of the DOE Gulfstream-1 involved many PNNL field staff who we would like to thank for their efforts, particularly J. Hubbe, C. Kluzek,

M. Pekour, J. Comstock, pilots M. Hubble and B. Svancara, and mechanic G. Dukes. H.

Jonsson acquired and analyzed the CDP data. F. Brechtel and S. Day aided in characterizing the inlets. Xin Yang is acknowledged for providing the Shanghai

ATOFMS data. The authors acknowledge the use of FIRMS data and imagery from the

Land Atmosphere Near-real time Capability for EOS (LANCE) system operated by the

NASA/GSFC/Earth Science Data and Information System (ESDIS) with funding provided by NASA/HQ.

Chapter 3 contents are part of a manuscript that is in preparation for submission for publication, Kaitlyn J. Suski, Alberto Cazorla, John F. Cahill, Jason Tomlinson, Duli

Chand, Beat Schmid, Daniel Rosenfeld, Kimberly A. Prather (2014), “Long Range

Transport of Asian Soot Impacts Cloud Properties over California”, In preparation for submission to Atmospheric Environment.

3.7 Figures

Figure 3.2. Positive and negative mass spectra of key aerosol types observed on 5 March 2011.

Figure 3.3. Aerosol concentrations measured with the PCASP (top panels) and altitude (bottom panels) for the morning flight (March 5a, left panels) and the afternoon flight (March 5b, right panels). The area affected by Asian soot is boxed in black. The high concentrations of aerosols observed at high altitude during the morning flight are absent in the same region during the afternoon flight.

Figure 3.4. Modal diameter for the lower (grey) polluted and higher altitude pristine (blue) clouds during the afternoon flight on 5 March 2011.

Soot Clouds

Figure 3.5. Picture of thin, grey clouds with soot in the cloud residues taken at 3562 m during the afternoon flight (22:02:06 5 March 2011 UTC).

Figure 3.7. FLEXPART 10 day back trajectories initiated on 5 March 2011 over Sugar Pine Dam. Fires that occurred in Southeast Asia on February 27, 28 and 1 March 2011 are marked by black dots.

Figure 3.8. K-soot observed in Shanghai in February of 2011 (black) and over California on March 5, 2011 (red). The March 5 data is offset by one mass/charge unit.

3.8 Supplementary Materials

During the CalWater flight campaign, many aerosol types were observed. Soot, highly processed, and K-soot types were described in the main text. Sea salt was

23 + 46 + 62 + characterized by the presence of sodium ( Na ), sodium clusters ( Na2 , Na2O ,

63 + 81/83 + -35/-37 - Na2OH , Na2Cl ) and chloride ( Cl ) [Gard et al., 1998; Gaston et al., 2011].

-44 - -60 - -76 - 40 + 96 + Dust was characterized by silicate ( SiO , SiO2 , SiO3 ) and calcium ( Ca , Ca2O ,

113 + (CaO)2H ) markers [Guazzotti et al., 2001a; Guazzotti et al., 2001b; Silva et al., 2000].

Figure S2 shows mass spectra for the major aerosol types.

March 5a March 5b 39.5

39.0 McClellan

38.5 Sacramento Latitude (º) Latitude 38.0 San Francisco 37.5

-124 -123 -122 -121 -120 -119 Longitude (º) Figure 3.9. Flight tracks for the morning (pink) and afternoon (grey) flights on 5 March 2011. McClellan Airfield is marked by a black dot.

Figure 3.10. Positive and negative mass spectra of aerosol types observed on 5 March 2011.

Figure 3.12. CALIPSO images showing smoke transport across the Pacific Ocean (continued).

Table 3.1. FLEXPART back trajectory point source information.

Time (UTC) Altitude (m ASL) Flight Description 5 March 2011 17:45 5000 Morning Soot Layer 1 5 March 2011 19:00 5000 Morning Soot Layer 2 5 March 2011 18:20 3000 Morning Cloud 1 5 March 2011 22:10 4000 Afternoon Lower Level Clouds 5 March 2011 23:00 5500 Afternoon Upper Level Clouds

Chapter 4. Dust and Biological Aerosols from the Sahara and Asia Influence

Precipitation in the Western US

4.1 Abstract

Winter storms in California’s Sierra Nevada increase seasonal snowpack and provide critical water resources for the state. Thus, the mechanisms influencing precipitation in this region have been the subject of research for decades. Previous studies suggest Asian dust enhances cloud ice and precipitation [Ault et al., 2011], while few studies consider biological aerosols as an important global source of ice nuclei (IN).

Here, we show that dust and biological aerosols transported from as far as the Sahara were present in glaciated high-altitude clouds coincident with elevated IN concentrations and ice-induced precipitation. This study presents the first direct cloud and precipitation measurements showing that Saharan and Asian dust and biological aerosols likely serve as IN and play an important role in orographic precipitation processes over the western

United States.

4.2 Introduction

Aerosols can modify cloud microphysical properties including droplet size and water phase, and thus can alter precipitation efficiency [Andreae and Rosenfeld, 2008]. In particular, dust aerosols which originate from various deserts around the world [Prospero

64 et al., 2002], have been shown to serve as effective ice nuclei (IN) and potentially enhance precipitation in mixed-phase clouds [Rosenfeld et al., 2011]. IN are atmospheric particles that catalyze the freezing of supercooled cloud droplets, producing ice crystals that would not form otherwise at warmer, mixed-phase cloud temperatures [Isono et al.,

1959]. Some types of biological aerosols such as bacteria have also been shown to serve as effective IN at relatively warm temperatures [Morris et al., 2004a], however recent modeling studies have concluded they are of minor importance to global IN concentrations and precipitation processes [Hoose et al., 2010]. Biological aerosols, such as bacteria, can be co-lofted and transported with dust [Hara and Zhang, 2012]. When lofted to high altitudes (approximately ≥ 5000 m), dust and biological aerosols can travel long distances. For example, Uno et al. showed dust from the Taklimakan desert in China circled the globe within 13 days [Uno et al., 2009]. Intercontinental transport of dust from Asia is well documented [Ault et al., 2011; Husar et al., 2001; Sun et al., 2001; Uno et al., 2011], while few have reported trans-Pacific transport from North Africa [Hsu et al., 2012; McKendry et al., 2007]. Modeling studies have shown that both Asian and

African dust influence ice formation in mixed-phase clouds [DeMott et al., 2003a;

Wiacek et al., 2010]. Small amounts of dust can be lofted into the free troposphere from the Taklimakan as early as February, with maximum concentrations detected between

April and May. Much of the Gobi desert is frozen in the winter months but in Siberia it can represent a significant Asian dust source in response to strong winter storms

[Prospero et al., 2002]. Notably, dust is lofted from the Sahara year round so this source region has the potential to affect ice formation in clouds throughout the year [Prospero et al., 2002].

IN concentrations are generally low (approximately 1 in every 105 aerosol particles in the free troposphere at -20C) [Rogers et al., 2001], highly variable, and strongly dependent upon the chemical composition of the aerosols present within a particular cloud [DeMott et al., 2010]. At mixed-phase cloud temperatures (above roughly -36°C), IN are necessary for ice formation. Ice crystals grow diffusionally at the expense of liquid droplets [Korolev, 2007], and can gain mass through accretion of supercooled liquid droplets or aggregation with other crystals to form graupel or snow

[Hosler et al., 1957; Houze, 1993]. Freezing of cloud droplets induced by IN produces a mixed-phase cloud that initiates precipitation more rapidly than a supercooled, liquid- only cloud due to the faster growth rate of ice particles versus droplets [Pinsky et al.,

1998]. The presence, absence, and abundance of IN can thus affect the intensity and spatial distribution of precipitation events. The efficiency of ice and precipitation processes has serious ramifications for mountainous regions such as the California Sierra

Nevada, where snowpack supplies copious amounts of water to reservoirs [Guan et al.,

2010]. Hence, cloud seeding experiments have been conducted in the Sierra Nevada since

1948 as a possible means of increasing precipitation [Deshler and Reynolds, 1990;

Reynolds and Dennis, 1986]. It has been suggested that over 50% of precipitation globally is initiated in the ice phase (e.g., [Lau and Wu, 2003]). Therefore, identifying the sources of IN within clouds and the mechanisms by which they influence precipitation processes is critical for future water resources.

4.3 Precipitation

Analysis of precipitation samples in combination with storm meteorology can provide insight into IN effects on orographic precipitation. For example, Ault et al. observed what was hypothesized to be Asian dust as insoluble residues in precipitation collected in the Sierra Nevada [Ault et al., 2011]. Precipitation samples enriched in dust coincided with increased snowfall and total precipitation at the surface by 60% and 40%, respectively, relative to another intense storm with nearly identical meteorological conditions [Ralph et al., 2012], thus suggesting the dust was affecting the precipitation processes. Two key unresolved questions from the Ault et al. study were: 1) Do the insoluble precipitation residues reflect the composition of the IN that initially nucleated cloud ice? and 2) What role does dust play in affecting cloud microphysics and precipitation? Further evidence from Pratt et al. provided aircraft measurements of dust and biological residues collocated with ice in clouds over Wyoming, but this was a limited dataset and no attempts were made to link the observations with precipitation

[Pratt et al., 2009a]. Herein, we present results building upon those from Ault et al. and

Pratt et al. using a combination of both ground-based precipitation and aircraft in-situ cloud measurements in California throughout the 2011 winter. Our results indicate a relationship between dust and biological aerosols detected in in-situ cloud residues in distinct mid-level ice layers, residues in precipitation collected at the ground, and long- range transport of dust and biological aerosols from Asia and the Sahara. Combined, these results strongly suggest the dust and biological aerosols impacted ice formation in mid-level clouds where precipitation processes were initiated.

The CalWater field campaign was designed to directly address aerosol impacts on clouds and precipitation in the Sierra Nevada during three consecutive winter seasons

(2009-2011). Measurements from a remote ground site at California’s Sugar Pine Dam

(SPD, 39°07'42.80"N, 120°48'04.90"W; 1064 m, MSL) included aerosol and meteorological instrumentation from 2009-2011. The same technique used by Pratt et al.

[2009a] and Ault et al. [2011], aerosol time-of-flight mass spectrometry (ATOFMS)

[Gard et al., 1997], was used to determine the chemical composition of re-suspended insoluble precipitation residues collected at SPD from Jan 30–Mar 8, 2011. S-band profiling radar (S-PROF) provided bulk microphysical information using vertical profiles of hydrometeor fall velocity and radar reflectivity [White et al., 2000]. Precipitation processes included warm rain, which started as liquid; cold rain, which started as ice and melted during descent; and snow/graupel/hail when surface temperatures were low

[White et al., 2003]. Precipitation that starts in the ice phase, i.e., snow/graupel/hail and cold rain, is termed “ice-induced precipitation.” S-PROF time-height cross sections during the times for all 11 of the precipitation sampling periods are shown in the Figure

4.6.

During all of the 2011 storms, dust and biological residues were frequently present in the precipitation, comprising up to 99% of the total insoluble residues per sample. Figure 4.1 shows the percentage of dust plus biological residues (%Dust+Bio), median cloud top temperature (°C), relative amount of cloud ice (%Ice), and percentages of different precipitation processes during each sample collection date in 2011. Dust and biological residue percentages were combined for the precipitation analysis because

68 chemical markers for each type were often observed within the same individual residue due to the precipitation collection and aerosolization processes, as shown in Figure 4.7 and discussed in the Supporting Online Material (SOM). Ice-induced precipitation comprised 74% of the total precipitation that fell at SPD, whereas warm rain only comprised 10% (the remaining 16% was unclassified). Notably, the highest %Dust+Bio in precipitation samples occurred during storms that contained a higher percentage of ice- induced precipitation (e.g., Jan 30, Feb 16-19, and Feb 24-26). During these storms, surface temperatures were sufficiently low (as shown in Table 4.1), enabling snow/graupel/hail to reach the surface. During the storms from Feb 14-16, Mar 1-3, and

Mar 5-7, surface temperatures were higher resulting in more rain than ice at the surface.

Further, more warm rain coincided with lower %Dust+Bio during these time periods. For the five samples where >50% of precipitation occurred as snow/graupel/hail, the average

%Dust+Bio was 90%, whereas for the four cases where >30% of the precipitation occurred as warm rain, the average %Dust+Bio was 69%. One possible explanation could be that a limited amount of dust and biological residues were available to serve as IN, which could result in less precipitation initiated in the ice phase. Another potential explanation is that the thermodynamic and dynamic structure of the storm resulted in shallower clouds (Figure 4.6) that produced warm rain and were perhaps unable to reach vertically into the dust plume. Typically, the cloud top temperatures were lower when the

%Dust+Bio in precipitation and %Ice in cloud were higher, suggesting dust and biological aerosols are more efficiently activated and scavenged via ice nucleation at colder temperatures. Both the Mar 1-2 and Mar 2-3 time periods had similar cloud top temperatures (-22°C), however, the %Dust+Bio and %Ice were higher on Mar 1-2 and

69 lower on Mar 2-3. Thus, this result suggests that the formation of in-cloud ice is not entirely dependent on the cloud temperature alone, and that the presence and abundance of IN feeding the clouds is important. Overall, these results suggest that dust and biological aerosols likely served as IN and influenced the precipitation phase in the clouds within colder sectors of the storms, expanding upon the precipitation measurements made by Ault et al. in 2009. Precipitation residues can provide useful information on the nature of aerosols and cloud seeds, however due to the caveats involved with analyzing precipitation and correlating ground-based measurements with mid-level cloud processes, in-situ aircraft measurements of cloud residues made during

CalWater 2011 were critical for gaining further insight into the role of the dust and biological particles in cloud and precipitation processes.

4.4 In-Situ Cloud Analysis

A compact aircraft version of the ATOFMS [Pratt et al., 2009b] measured in-situ cloud droplet and ice crystal residues onboard the G-1 during CalWater. Figure 4.1 includes cloud residue data during each precipitation sampling time period. Discussion regarding the collection of cloud residues and the relative amount of dust and biological residues in the precipitation compared to the cloud residues is provided in the SOM.

Interestingly, high percentages of dust and dust mixed with biological cloud residues were observed during the flights on Feb 16th and Feb 25th, respectively. On Feb 16th, the cloud residues were rich in aluminosilicate dust (50%), with smaller contributions from dust mixed with biological material (“dust/biological,” 16%) and purely biological residues (3%). However, on Feb 25th, residues were predominantly composed of dust

70 mixed with salt and biological material (“dust/salt/biological,” 66%), while dust and biological residues represented 9% and 10%, respectively. The precipitation residues were also characterized as more “dust-like” on Feb 16th and were composed of biological material on Feb 25th as further discussed in the SOM. Although the precipitation residue compositions were not exactly the same as for the cloud residues, the similarities between the two suggest the majority of precipitation residues likely originated from the cloud particles as opposed to being scavenged by precipitation from air below cloud base. The meteorological conditions were similar during these two days and corresponded to the passing of a cold front (discussed further in the SOM). The passing of these cold fronts likely enabled the transport of dust and biological aerosols into the region as suggested by previous studies [Ault et al., 2011]. However, if IN were hypothetically absent in these cold air masses, supercooled cloud droplets could remain as liquid instead of forming ice crystals as shown by Rosenfeld et al. [2013] and described herein. On days without these meteorological conditions, we observed only minor fractions of dust and biological particles in cloud residues, as shown in Figure 4.9, showing that dust and biological particles are not always present. Additionally, estimated dust concentrations between

3000-6000 m are provided and discussed in the SOM to show the variability of dust concentrations can range over an order of magnitude during flights. This strongly supports the concept proposed here that ice formation depends not only on the meteorological conditions, but also on the availability of effective IN such as certain mineral dusts and biological particles.

In order to investigate the impact of dust and biological particles on cloud properties during the storms on Feb 16-19 and Feb 24-26, we probed data from the flights

71 corresponding to these storms in more detail. Figure 4.2 shows the number of individual cloud residues measured by A-ATOFMS for the Feb 16th ascent (A) and Feb 16th descent

(B) flight segments (flight tracks shown in SOM). Feb 25th ascent and descent are shown in Figure 4.13. Flights on Feb 16th and Feb 25th are separated into the ascent and descent due to changing dynamics as discussed in the SOM. The early mornings on Feb 16th and

Feb 25th included prefrontal periods with an atmospheric river (narrow band of water vapor transported from sub-tropical oceanic regions) and terrain-parallel blocked flows

(further discussed in SOM), which introduced moist marine air masses into deep clouds over the Sierra Nevada (see Figure 4.6). In the afternoon during the flights on Feb 16th and Feb 25th, cold postfrontal conditions with lower-level convective clouds and mid- level orographic clouds were present. Liquid water content (LWC), the mass concentration of liquid in a cloud, was measured with a Cloud Droplet Probe (CDP). IN concentrations (IN per liter of air) were measured with the Continuous Flow Diffusion

Chamber (CFDC) [Rogers et al., 2001]. Cloud ice fraction was calculated using liquid and total water content measured with the WCM-2000. Details of the calculation are discussed in the SOM.

During the ascent to high altitudes on Feb 16th, the lower-level convective clouds

(≤ 3500 m) were pristine marine clouds that contained predominately sea salt residues and few dust particles (Figure 4.2A). Correspondingly, the IN concentrations were low

(0 to a maximum of 0.15±0.2 L-1 in a temperature range from -18.4 to -18.9°C). These measurements were consistent with little to no ice in-cloud as evidenced by a low cloud ice fraction (0-0.4) and high LWC (up to ~1 g/m3). The nearly ice-free, supercooled cloud reached temperatures down to -21°C. The presence of liquid water and lack of significant

72 amounts of ice was consistent with very few IN active in this region of the cloud. The IN concentrations in the mid-level orographic clouds (4000-6000 m) could be determined with greater certainty at 0.16±0.06 L-1 measured at -18.8°C, while the cloud top temperature reached -35°C, and contained ice along with supercooled water of <0.1 g/m3.

In addition, a spike in dust and dust/biological residues was observed between 4400-5500 m in these mid-level clouds. Transmission Electron Microscope (TEM) analysis showed that 88% of IN-activated aerosol in the CFDC was of dust and biological origin (53% dust, 26% dust/salt/biological, and 9% biological), thus, confirming the dominance of dust and biological cloud residues observed by the A-ATOFMS, as discussed above.

However, above 5500 m, we observed significantly less dust and biological residues coincident with supercooled liquid water at temperatures as low as -35°C, thus showing a distinct layer of dust and biological aerosols played a role in forming the ice-enriched layer at lower altitudes and warmer temperatures. During the descent back down to lower altitudes (Figure 4.2B), dust and dust/biological cloud residues were detected at several altitudes, but were most abundant between 3500-4500 m where the highest ice fractions were observed (up to 1). IN concentrations measured at a lower CFDC processing temperature of -32±0.5°C trended with the abundance of the dust and biological residues, reaching 2.2±0.3 L-1 between 4000-4500 m. The large increase in IN concentrations measured in dust layers at -32°C on the descent compared to -18.4 to -18.8°C on the ascent suggests that the dust was IN-active within the intermediate temperature range.

The collocation of these specific residues with ice crystals and higher IN concentrations suggest dust and biological aerosols served as IN in these mid-level clouds.

On Feb 25th, the meteorological and cloud conditions were similar to Feb 16th.

However, dust/salt/biological aerosols, as opposed to pure dust as on Feb 16th, were collocated with ice, particularly around 2700 and 3900 m. The residues were separated into two distinct layers as shown in Figure 4.13A; a lower layer from 1400-3100 m and an upper layer from 3800-4000 m. IN concentrations reached 0.39±0.2 L-1 measured at

-25.0ºC in the upper layer, while in the lower layer maximum IN concentrations of

0.51±0.24 L-1 were measured at -25.3 ºC; the CFDC temperatures were appropriate for the upper layer (-26 to -20 ºC) but colder than cloud temperatures in the lower layer (-17 to -10ºC). The mixed nature of these particles makes it difficult to extract which component (i.e., dust or biological) enabled the ice formation. The observations on Feb

16th and Feb 25th bracket the activation temperatures of the dust and biological particles between -18 and -32°C, with most of the activity occurring below -25°C. This is in agreement with previous laboratory studies, showing that many mineral dusts are more

IN-active below ~-20˚C, whereas biological particles such as spores become active at

~-28° C [Iannone et al., 2011], as well as with satellite observations of Asian desert dust glaciating cloud tops at -22°C on average [Rosenfeld et al., 2011].

On Feb 25th, the mid-level cloud contained supercooled cloud water of up to

0.3 g/m3, with some ice crystals that rimed into graupel at lower levels (see SOM), suggesting the ice was formed in the mid-level cloud and fell into the lower-level cloud while riming and depleting the cloud water. These results, along with the results on Feb

16th, are indicative of the seeder-feeder mechanism [Choularton and Perry, 1986], which has been documented previously in the Sierra Nevada [Meyers et al., 1992] as well as in

74 other mountain regions including those in Colorado, Arizona, and Central Europe [Dore et al., 1999; Fowler et al., 1995; Reinking et al., 2000; Saleeby et al., 2009]. These data suggest that particles with biological components, in addition to dust, play a role in ice formation in the orographic clouds in the Sierra Nevada and contribute to the efficiency of the seeder-feeder mechanism. Previous studies have shown that a graupel particle grows much faster than a supercooled raindrop of the same mass in the same convective cloud [Pinsky et al., 1998]. Therefore, seeder-feeder is an efficient precipitation mechanism, especially when the accreted drops are large [Borys et al., 2000]. Satellite observations have suggested the possibility of dust glaciating high altitude clouds [Choi et al., 2010], however there have not been any in-situ measurements of dust in clouds over the Sierra Nevada prior to this study. When urban pollution aerosols are incorporated into clouds, precipitation processes are often less efficient and suppression of both warm and mixed phase precipitation has been shown via satellite [Givati and

Rosenfeld, 2004; Rosenfeld, 2000]. Our results show the presence of dust and biological particles feeding deeper and colder orographic cloud levels appears to have the opposite effect, i.e., accelerating the efficiency of the precipitation forming processes, especially in situations where cloud layers are decoupled from the boundary layer likely due to the presence of terrain-parallel blocked flow (discussed in SOM). This implies that future studies must observe both influences, i.e., boundary layer and upper-level aerosols, as well as associated meteorological conditions.

4.5 Back Trajectory Analysis

The co-location of dust and biological aerosols with IN and ice in clouds, in addition to the abundance of dust and biological residues observed on days with more ice-induced precipitation during CalWater, leads to a question of the origin of the dust and biological aerosols. Ault et al. suggested Asian dust influenced snowfall during only one storm in 2009 [Ault et al., 2011]; however, herein we present six storms with influences from Asia and the Sahara. Local sources of dust and biological aerosols were likely minor contributors to the precipitation residues, as discussed in the SOM. Figure

4.3A shows 10-day air mass backward trajectories calculated using HYSPLIT [Draxler and Rolph, 2013] ending at SPD during the storms shown in Figure 4.1. Meyers et al. suggest that ice nucleation occurs at the tops of orographic clouds [Meyers et al., 1992], and thus trajectories were calculated to end at cloud top heights between 2000-10000 m,

MSL over SPD using data from the 11th Geostationary Operational Environmental

Satellite (GOES-11) averaged every 3 hours during storms. The boxes highlight major dust regions [Prospero et al., 2002]. Based on previous studies (e.g., [Hallar et al.,

2011]), the biological aerosols were likely co-lofted from arid regions and transported with the dust, although other sources of biological material such as the ocean and local sources could also contribute. Figure 4.3B shows the frequency of trajectories that traveled over each dust region during each storm. Trajectories that originated from over

North Africa frequently traveled over the other dust regions on the way to the United

States, suggesting these air masses contained a mixture of dust from the various source regions. Further, wet and dry deposition of the dust and biological aerosols could contribute to loss during transport from the more distant sources. The dust source

76 frequencies show that the storms were predominantly affected by Asia; however, the end of the study was additionally influenced by North Africa and the Middle East (15% and

21% of the time, respectively). Not only did the air masses travel over the various dust regions, the air intersected dust layers as shown in Figure 4.3C. Using time-height cross sections of dust concentrations from the Navy Aerosol Analysis and Prediction System

(NAAPS, http://www.nrlmry.navy.mil/aerosol/), we estimated the maximum height of dust layers at sites within/near each dust region on days where the trajectories traveled over the boxes. The height of trajectories that passed through each region typically fell within the height of the dust layers as shown in Figure 4.3C. Although transport within the dust layers was infrequent over North Africa, Saharan dust was likely picked up over the Middle East where trajectories, on average, skimmed the top of the layer. Overall, trajectories traveled through dust layers 6%, 23%, 73%, and 45% of the time in North

Africa, the Middle East, the Taklimakan, and Northeast China/Mongolia, respectively, as shown in Figure 4.18.

Further evidence that dust was present along the complete trajectory was obtained using Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) satellite data of aerosol subtypes. In combination with NAAPS, CALIPSO provides a method to track dust as it is transported from the source across the Pacific Ocean to SPD, as shown in Figure 4.19. Figure 4.4 shows NAAPS modeled aerosol optical depth (AOD) of dust in addition to locations where CALIPSO imagery transected the trajectories for

Feb 16th and Feb 25th. These two days represent the storms where the highest %Dust+Bio was present in the precipitation (93% and 95% on average during Feb 16-19 and Feb 24-

26). The trajectory ending on Feb 16th started 10 days earlier over the Atlantic Ocean and proceeded to travel north of the Saharan dust layer. Before reaching China, the air mass contained little dust based on CALIPSO imagery (the “N/A” circle). Once the air mass reached Northeast China/Mongolia, it picked up fresh dust and traveled across the Pacific

Ocean to SPD. In contrast, the trajectory ending on Feb 25th started 10 days back over

Oman where an abundance of fresh dust was present. The air mass continued to travel over Southern China where it likely picked up more dust from the Asian sources. The

CALIPSO imagery following the trajectory over the Pacific Ocean shows both fresh and polluted dust. Combined analysis of air mass back trajectories, NAAPS, and CALIPSO imagery provide evidence of lofted dust being transported across the Pacific Ocean and show the probable sources of IN that influence precipitation processes in the Sierra

Nevada. Conventionally, dust from the Sahara has been shown to travel west over the

Atlantic Ocean from Africa to South America, however, herein we show African dust traveling east across the Pacific and into the United States at higher altitudes.

4.6 Conclusions

Herein we present a direct link between long-range transported dust and biological aerosols affecting cloud ice formation and precipitation processes in the Sierra

Nevada. As summarized in Figure 4.5, long-range transport of the dust and biological aerosols is shown to be an important mechanism for introducing IN into mid-level clouds and enhancing precipitation formation via seeder-feeder in the Sierra Nevada. It is important to note that developing this whole conceptual picture requires combining evidence from multiple measurements which include: 1) ground-based precipitation

78 chemistry, 2) satellite data showing dust activity and transport, 3) aircraft in-situ cloud microphysics and residual aerosol composition, 4) detailed meteorology at the ground- level as well as within the clouds, and 5) out-of-cloud aerosol measurements (>0.6 m) particles. This study is the first to show dust originating from further west than Asia plays a role in clouds and precipitation processes over the Western U.S. Dust and biological aerosols measured as insoluble residues in precipitation samples corresponded to periods with more ice-induced precipitation. In-situ aircraft measurements of the same storm systems confirmed the dust and biological precipitation residues were collocated in a distinct cloud layer with higher IN concentrations and larger fractions of ice relative to supercooled liquid. Many factors contribute to the type and quantity of precipitation including dynamics, meteorological forcings, and transport conditions; however, our results suggest dust and biological particles may play a key role in ice formation and potentially precipitation initiation, by causing glaciation of mid-level clouds. Further, our results also suggest biological particles may play a larger role than modeling studies suggest, which indicates further studies are needed on the surface emission rates and IN efficiency of biological aerosols, especially when mixed with other species. In-situ cloud measurements such as these provide critical insights into the key chemical species involved in the formation of ice in clouds, however, controlled laboratory studies are needed to better understand the mechanisms involved as well as sort out critical chemical details such as whether metals, minerals, biological material, or some combination control the initial ice formation process.

Prior to this study, most work has focused on modeling the impacts of regional aerosol pollution sources on precipitation. Our results demonstrate the potentially critical impact of intercontinental transport of natural aerosols on precipitation formation processes. Further, these findings motivate the challenging study of quantifying aerosol effects on not only the initial cloud phase and formation of precipitation, but also the intensity and location of precipitation. Such studies must also quantify the role of non- aerosol-related meteorological forcings of orographic precipitation. Due to the ubiquity of dust and possibly co-lofted biological particles such as bacteria in the atmosphere, these findings have global significance. Furthermore, the implications for future water resources become even more substantial when considering the possible increase in

Aeolian dust as a result of a warming climate and land use changes.

4.7 Acknowledgments

Funding was provided by the California Energy Commission under contract CEC

500-09-043. D. Rosenfeld was funded under the ASR/DOE program. P. Minnis was supported by the ASR/DOE program under DE-SC0000991/003 and the NASA MAPS

Program. Mayer, E. Fitzgerald, D. Collins, and J. Cahill provided assistance with

UCSD/SIO equipment set-up. The authors gratefully acknowledge the NOAA Air

Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www.arl.noaa.gov/ready.php) used in this publication. J. Ayers and R. Palikonda (Science Systems and Applications) provided GOES-11 cloud top heights used for HYSPLIT back trajectory analysis and

%Ice in cloud. The deployment of the NOAA and UCSD/SIO equipment at the Sugar

Pine site involved many field staff, particularly C. King (NOAA/ESRL/PSD). The deployment of the DOE Gulfstream-1 involved many PNNL/ARM field staff, particularly J. Hubbe and C. Kluzek. H. Jonsson contributing cloud probe data. M.

Hubbell and B. Svancwara flew the G-1 for the CalWater flight campaign. D. Collins and

R. Spackman provided insightful discussions during the editing stages of this manuscript. Traci Lersch of R.J. Lee Inc. provided TEM analyses of collected IN. Data available in this paper are available in the SOM.

Chapter 4, in full, is a reprint of the material as it appears in Science 2013. Jessie

M. Creamean, Kaitlyn J. Suski, Daniel Rosenfeld, Alberto Cazorla, Paul J. DeMott, Ryan

C. Sullivan, Allen B. White, F. Martin Ralph, Patrick Minnis, Jennifer M. Comstock,

Jason M. Tomlinson, and Kimberly A. Prather (2013), “Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western U.S.”, Science,

339(6127), 1572-1578. Reprinted with permission from AAAS.

4.8 Figures

Figure 4.1. Precipitation and cloud characteristics during each sample collection date during the 2011 CalWater study at SPD. Markers in (A) show the percentages of dust and biological precipitation residues combined (Dust+Bio) sampled at the surface at SPD and bars show the percentages of in-cloud residues sampled on the G-1 aircraft. Sample dates with more than one column of bars for the aircraft data signify more than one flight took place; for instance, 3 flights were taken Feb 24-26 and 2 flights were taken Mar 2-3. The percentages of cold rain, snow/graupel/hail, and warm rain per sample (taken from S- PROF radar profiles) are shown in (B). Red markers represent the median cloud top temperature (°C) per sample collection date. Error bars are shown for temperature and represent the minimum and maximum of the range. Violet markers represent the average relative amount of the cloud consisting of ice (%Ice) within the 10-km radius over SPD derived from satellite measurements. Storms 1-6 are highlighted in order to identify the evolution of cloud and precipitation residues, cloud top temperature, %Ice, and precipitation processes over the course of the storms.

Figure 4.2. Vertical profiles of the number of cloud residues (dust, dust mixed with biological material called “dust/biological,” dust mixed with salt and biological material called “dust/salt/biological,” biological, sea salt, and residues from pollution such as soot and biomass burning), liquid water drops or liquid water content (LWC, g/m3), cloud ice fraction, and cloud temperature (°C) for the flight on (A) Feb 16th ascent (17:01-18:29 UTC) and (B) Feb 16th descent (18:29-20:19 UTC ). Figure 4.13 shows the ascent and descent for Feb 25th. Also shown are cloud particle residual IN concentrations (L-1) with the color representing the temperature at which the CFDC measurement was taken. Error bars represent twice the Poisson sampling errors in IN concentrations calculated for 3-10 minute time periods.

Figure 4.3. Dust source analysis including (A) 10-day back trajectories ending at cloud top heights over SPD during storms in 2011, with the boxes highlighting the four dust source regions. Trajectories are colored based on which dust region they traveled over, including the Sahara (separated into North Africa and the Middle East) and East Asia (separated into the Taklimakan desert and Northeast China/Mongolia). Storms were investigated in order to achieve statistical significance (i.e., more trajectories per time period). The percentages under each box represent the frequency that the trajectories passed through each arid region. (B) The frequency with which 10-day back trajectories passed over the dust regions defined in (A) in addition to other regions (non-dust origin). The four shades of green represent the four different dust regions. The total number of trajectories analyzed per storm is also provided. (C) The average altitudes of trajectories that passed through each region and dust layer heights determined using NAAPS analysis time-height sections at Sedeboker (North Africa), Solar Village (Middle East), Yinchuan (site nearest to the Taklimakan), and Beijing (Northeast China/Mongolia) for the time periods studied. A map of the sites and examples of the time-height sections are shown in Figure 4.16. Data originate from times periods that include back trajectory endpoints shown in Figure 4.18. The middle line of the shaded regions represents the average maximum height of the dust layers in each region from NAAPS analysis. The error bars show the minimum and maximum altitudes for the trajectory endpoints and dust layers.

Figure 4.4. Dust tracking for days during the (A) Feb 14-16 and (B) Feb 24-26 storms. The 10-day air mass back trajectories ended at cloud top heights over SPD (shown by the star) on Feb 16th (7390 m, MSL) and Feb 25th (9340 m, MSL). Within each trajectory, the color scale represents the altitude of each hourly endpoint while the circles show each endpoint at 00:00, i.e., each marker represents 1 day back. The last two days from the trajectory ending on Feb 16th are not shown. The circled numbers mark the locations where dust was present along these trajectories determined using CALIPSO imagery (imagery shown in Figure 4.19). The green overlays represents dust aerosol optical depth (AOD) from 0.1-0.8 modeled using NAAPS. The dust AOD overlays are composites of the 10 days during each trajectory: (A) Feb 6-16 and (B) Feb 15-25.

Figure 4.5. Conceptual model summarizing results from (A) prefrontal and (B) postfrontal conditions presented herein. Prefrontal conditions corresponded to a deep continuous cloud layer, atmospheric river (AR), and terrain-parallel blocked flow (Sierra Barrier Jet, SBJ), which shifted to two cloud layers, and the disappearance of the AR and blocked flow during the postfrontal conditions. Dust and biological aerosols were long- range transported and form ice in mid-level orographic clouds. The ice becomes rimed by supercooled drops and falls into lower-level convective clouds formed in pristine, marine air masses. Thus, precipitation is formed via the seeder-feeder mechanism.

4.9 Supplementary Material

4.9.1 Materials and Methods:

Cloud top temperatures were acquired from the University of Wyoming air soundings at Oakland, CA (http://weather.uwyo.edu/upperair/sounding.html). Surface temperature measurements (2-minute resolution) at Sugar Pine Dam (SPD) were acquired from the NOAA Hydrometeorological Testbed Network (HMT-West)

(http://hmt.noaa.gov/field_programs/hmt-west/2011/). The S-band profiling radar

(S-PROF) provided vertical profiles of radar reflectivity and Doppler vertical velocity, with 60-m vertical resolution and 40 s temporal resolution, that were used to monitor the radar brightband melting level [White et al., 2000; White et al., 2003]. The brightband results from the difference in the fall speed and dielectric factor for ice and water and the aggregation of ice particles as they descend and melt [White et al., 2002]. The total accumulation and percentage of snow/graupel/hail, brightband (BB) or cold rain, and non-brightband (NBB) or warm rain were estimated using the rainfall process- partitioning algorithm developed by White et al. [2003], which was applied to the

S-PROF profiles. Analysis was performed on all half-hour periods when the rain rate exceeded ~1 mm/h. This is illustrated in Figure 4.6, which shows time-height cross sections of signal-to-noise-ratio (SNR) measured by the S-PROF radar at SPD. These sections include the times for all 11 of the aerosol sampling periods (Table 4.1). The results from White et al. [2003] precipitation process partitioning algorithm are also shown in Figure 4.6.

Eleven precipitation samples were collected during six storms from Jan 30─Mar

8, 2011 at SPD. Collection times are provided in Table 4.1. The size-resolved chemical composition of individual ambient aerosols (0.2-3.0 μm) was measured during sample collection time periods using the aerosol time-of-flight mass spectrometer (ATOFMS), which is described in full detail elsewhere [Gard et al., 1997]. Precipitation samples were manually collected using beakers cleaned with ultrapure Milli Q water (18 MΩ/cm) and methanol. Once collected, precipitation samples were transferred to 500 mL glass bottles, frozen, and stored for 6-10 days until analysis. Insoluble residues in the precipitation samples were resuspended and dried using a Collison nebulizer and two silica gel diffusion driers, then sampled using the ATOFMS. This aerosolization method can produce agglomerates of different particle types as well as soluble coatings on insoluble residues, therefore do not fully represent the particles that nucleated the ice in-cloud

[Holecek et al., 2007]. However, this method can still provide useful information of the nature of the aerosols in-cloud, as verified in this study using in-situ cloud data. Residues between 0.2-3.0 μm were individually sized and chemically analyzed by the ATOFMS, providing dual polarity mass spectra for each particle. Particles were visually inspected and classified into different types based on combinations of characteristic ion peaks in both negative and positive spectra; however, only dust and biological particle types are presented. In order to be classified as a particular dust or biological residue type, the mass spectrum of the residue must contain a specific combination of ion peaks; the combinations of ion peaks used for classification are based on previous studies [Ault et al., 2011; Holecek et al., 2007; Pratt et al., 2009a; Silva et al., 2000] and are shown in

Figure 4.7. Dust mixed with biological (Dust+Bio) residues varied in mineralogy as

88 shown in (A-C), but were classified as containing a combination of lithium (6Li+), sodium

23 + 24 + 27 + 39,41 + 40 + ( Na ), magnesium ( Mg ), aluminum ( Al ), potassium ( K ), calcium ( Ca ,

56CaO+/57CaOH+), titanium (48Ti+, 64TiO+), iron (54,56Fe+), and/or aluminum oxide

70 + ( Al2O ) in the positive ion mass spectra. The negative ion mass spectra for the dust residues varied, but commonly contained ion peaks for aluminum and aluminosilicates

43 - 59 60 - 76 - 77 - 26 - 42 - ( AlO , AlO2-, SiO2 , SiO3 , HSiO3 ) and biological markers ( CN , CNO , and

79 - PO3 ). Residues containing combinations of these peaks are representative of soil dust and have been previously observed in ambient conditions using ATOFMS [Silva et al.,

2000]. A unique type of calcium-rich dust particularly present in samples at SPD in 2011

23 + 24 + 39,41 + 40 + 56 + 57 + 84 + was classified by ion peaks at Na , Mg , K , Ca , CaO , CaOH , Ca2 ,

96 + 113 + Ca2O , and (CaO)2H as shown in (C). These residues commly contained sulfate

48 - 64 - 80 - peaks in the negative mass spectra ( SO , SO2 , and SO3 ), in addition to what are

121 - 137 - likely higher mass aluminosilicate peaks ( AlSiO4 and HSi2O5 ; not shown).

Residues containing combinations of these peaks have been previously observed in ambient conditions using ATOFMS [Silva et al., 2000]. Not all ion peaks discussed are

39,41 + 26 - labeled in Figure 4.7. The first biological residue type contained markers at K , CN ,

42 - 63 - 79 - CNO , PO2 , and PO3 shown in (D). The second type shown in (E) contained ion

12 + 23 + 27 + + + 29 + + 37 + peaks of weaker intensity at C , Na , Al /C2H3 /NCH , C2H4 /N2H , C3H ,

38 + 39 + + 40 + + 41 + + 43 + 55 + 56 + + C3H2 , C3H3 /K , C3H4 /Ca , C3H5 /K , AlO , C2HNO , Fe /CaO , and

57CaOH+ in the positive ion mass spectra. Positive ion mass spectra from biological particles containing peak combinations similar to these were previously observed in cloud ice crystals [Pratt et al., 2009a]. The negative ion markers were similar to those of

26 - 42 - 79 - the first biological residue type mentioned above, notably CN , CNO , and PO3 , with

35,37 - 45 - 59 - additional contribution from chloride ( Cl ) and carbohydrates ( CHOO , CH3COO ,

71 - 73 - C3H3OO , CH3CH2CHOO ). In order to be classified as a biological residue, the

26 - 42 - 79 - negative mass spectrum must contain CN , CNO , and PO3 ; if any of these peaks were absent the residue was not classified as biological. The percentage of each type of precipitation residue was calculated based on the number of that type of residue compared to the total number of residues analyzed per sample.

In-situ aircraft measurements included chemical composition of aerosols as cloud residues, temperature, ice nuclei (IN) concentrations, and cloud droplet and ice crystal images. During the CalWater 2011 flight campaign, 25 flights (68.5 flight hours) were flown on the Department of Energy Gulfstream-1 (G-1) aircraft based out of the

McClellan Airfield in Sacramento, CA. Multiple flight tracks provided a large sampling area that spanned from the eastern Pacific Ocean into the Sierra Nevada. See Figure 4.10 for flight tracks of all 25 flights. The flight tracks for the Feb 16th and Feb 25th flights are shown in Figures 4.11 and 4.12, respectively. The color of the flight tracks shows time, with the size of the marker representing the altitude. On the right panel of Figure 4.11, the flight track shows the ascent in black and the descent in white. The ascent and descent were made in similar geographical areas by backtracking over the initial flight track.

Therefore, comparing the ascent and descent is valid because they are sampling the same cloud system in the same geographical area over time. Data collected east of the Sierra

Nevada ridgeline on Feb 25th were excluded because this area includes cloud outflow, which is not representative of the rest of the developing cloud system when ice and droplets were initially formed. Additionally, CFDC and A-ATOFMS data were excluded

90 between on Feb 25th for the period between 23:25-23:40 due to possible ice forming on/in the inlet, which caused flow disturbances and sampling artifacts as evidenced by distinct ion markers. It is quite challenging to fly through supercooled liquid and avoid ice buildup. However, we were able to identify and isolate these periods which occurred on

Feb 25th from the rest of the analysis by using a combination of inlet flows, sample and inlet temperatures, and the presence of a distinct particle type, termed “inlet artifact” containing Al and Al2O (see Figure 4.8H). Notably, one distinguishing feature of CVI inlet artifacts is the marker ions only detected on the CVI. Previous research has also shown metals such as chromium (Cr) and iron (Fe) can also be present due to abrasion of inlet materials by ice in cirrus clouds [Murphy et al., 2004]. On Feb 16th, we detected Cr and Fe on many of the CVI particles (Figure 4.8A), but also detected Fe on clear air dust particles. Thus, it is difficult to discern whether Fe was mostly from the inlet or dust because it is also a common metal species detected previously by our group in Asian dust

[Sullivan et al., 2007]. Additionally, metals have been shown to serve as ice nuclei in cases where a CVI was not used [DeMott et al., 2003b]. Due to the fact these same metals can also be present in dust and potentially other aerosols, this is an ongoing critically important area of research. Bacteria which can serve as an effective ice nucleus at relatively warm temperatures has been shown to be co-lofted with Asian dust and provides one possible explanation for the presence of biological markers and metals in the same residuals [Hara and Zhang, 2012].

The aircraft data used in Figures 4.1 and 4.2 of the manuscript were limited to times when particles were sampled using the Counterflow Virtual Impactor inlet (CVI

Inlet Model 1204, BMI) to include only cloud particles with a variable cut-off size of

6–18 µm. The CVI works by using a counterflow of air to allow the passage of only cloud droplets and ice crystals that are of sufficient size to have enough inertia to overcome this counterflow and pass into a minor flow beyond the inlet plane [Ogren et al., 1985]. The cloud particles thus captured melt/evaporate, leaving residual particles that are sampled by the ATOFMS and the continuous flow diffusion chamber (CFDC).

The Feb 16th flight was 3 hours 17 minutes and 30 seconds long with 1 hour 57 minutes and 36 seconds of the flight on the CVI inlet. The Feb 25th flight was 3 hours 34 minutes and 52 seconds long with 1 hour 56 minutes and 54 seconds on the CVI inlet. Sampling on the CVI comprised roughly 2/3 of the sampling time for each flight. Additionally,

Figure 4.9 includes clear air data sampled from an isokinetic aerosol inlet (Isokinetic Inlet

Model 1200, BMI).

The aircraft (A)-ATOFMS is a compact version of the ATOFMS, packaged to fit on a double wide aircraft rack [Pratt et al., 2009b]. It uses a compact Z-shaped dual polarity mass spectrometer and optically decoupled detectors. An aerodynamic lens inlet focuses particles from 0.1-2.5 μm. These particles are then sized, chemically analyzed and the spectra are grouped, as described above. Representative spectra from the A-

ATOFMS are shown in Figure 4.8. There were several distinct dust types detected and some dust had biological ion markers on the same particle, while others did not. It is not possible to know whether these mixed dust/salt/biological residues were initially present together before ice formation occurred or whether they formed subsequently by riming or during sampling. Research is ongoing on how to improve the sampling of ice crystals,

92 particularly in regions of large crystal (>100 µm) sizes. Nevertheless, Feb 25th and Feb

16th showed significant differences in the composition of ice cloud residuals. Specifically, a higher fraction of biological material relative to dust, and different transition metal ion combinations (more aluminum than iron) were present on Feb 25th. The fact that these unique signatures appeared in distinct ice layers in mid-level clouds when combined with precipitation measurements at ground level, satellite data showing dust transport, and complementary TEM data suggest that both dust and biological components played a role in ice formation. This is consistent with a recent study showing the combination of biological material mixed with dust enhances ice formation [Conen et al., 2011]. Future research is needed on the chemical composition of ice nuclei measured at the start of ice nucleation. In-situ cloud measurements such as those from this study provide critical insight into the chemical species most abundant in the ice phase of clouds, but lab studies are required to sort out whether metals, minerals, or biological material or some combination control ice formation at the microscopic level. An aluminosilicate dust type was identified by the presence of aluminum (27Al+) in the positive spectra and silicates

60 - 88 - ( SiO2 , Si2O2 ) in the negative spectra as shown in (A). (B) shows a different dust type

23 + 39,41 + 35 - 46 - with sodium ( Na ), potassium ( K ), chlorine ( Cl ), nitrite ( NO2 ), and silicates

60 - 76 - ( SiO2 , SiO3 ). The dust mixed with salt and biological material (dust/salt/biological) is shown in (D) and was classified by sodium (23Na+), aluminum (27Al+), potassium

(39,41K+), and occasionally iron (54,56Fe+) in the positive spectra with chlorine (35Cl-),

62 - 46 - 79 - 60 - nitrate ( NO3 ), nitrite ( NO2 ) phosphate ( PO3 ), and occasionally silicates ( SiO2 ,

76 - SiO3 ) in the negative spectra. Residues containing dust mixed with biological material

(dust/biological) were similar to the dust/salt/biological residues as shown in (C), but did

93 not contain makers for salt (23Na+, 35Cl-). Biological particles were characterized by the

26 - 42 - 63 - 79 - presence of biological ion markers ( CN , CNO , PO2 , PO3 ) and the absence of silicates or lithium as shown in (F). Biological particles generally had organic carbon and

12 + 27 + + + 29 + 77 + 91 + organic nitrogen ( C , Al /C2H3 /NCH , N2H , C6H5 , C7H7 ) in the positive

26 - 42 - 63 - 79 - spectra with biological markers ( CN , CNO , PO2 , PO3 ) and sometimes nitrate

62 - 46 - ( NO3 ) and nitrite ( NO2 ) in the negative spectra [Fergenson et al., 2003; Pratt et al.,

2009a; Russell, 2009]. Sea salt residues shown in (G) contained sodium (23Na+), sodium

81/83 + 93/95 - -35/-37 - 46 - 62 - clusters ( Na2Cl , NaCl2 ), chloride ( Cl ), nitrite ( NO2 ), and nitrate ( NO3 )

[Gard et al., 1998; Gaston et al., 2011]. The aluminum-rich type shown in (H) was observed when the CVI was used in mixed-phase clouds and was deemed an artifact caused by ice scraping metal off the inlet. For the results presented herein, this aluminum-rich type was detected during several isolated portions of the Feb 25th flight.

The percentage of each type of cloud residue was calculated based on the number of that type of residue compared to the total number of residues analyzed for each respective flight. In addition, on Feb 25th, data from the isokinetic inlet were compromised due to icing of the inlet; therefore these data were not included in Figure 4.9.

The static air temperature was measured onboard the Gulfstream-1 using a

Rosemount 102E. Cloud ice fraction was calculated using total water content (TWC) and liquid water content (LWC) measured with a Multi-Element Water Content System

(WCM-2000, SEA Inc.). The WCM uses heating elements to measure TWC collecting into a trap, and LWC striking wires. Ice water content (IWC) is determined as

IWC=TWC-LWC. Ice fraction (fIce) is calculated by dividing IWC by the sum of the

LWC and the IWC (fIce = IWC/(IWC+LWC)). The WCM had operating issues during parts of the ascent on Feb. 25th; therefore, 2D-S imagery was used to verify the presence of ice crystals or liquid droplets in cloud for this portion of the flight. Cloud droplet and ice crystal images were collected with a 2D-S (2D-S Stereo Probe, SPEC, Inc.). 2D-S images were classified into 4 categories based on the amount and type of ice present.

Representative images for each type are shown in Figure 4.14. Liquid water content was also measured using a Cloud Droplet Probe (CDP, Droplet Measurement Technologies).

Concentrations of aerosols between 1.0 and 10.0 µm were measured using a Cloud and

Aerosol Spectrometer (CAS, DMT). Ice nuclei measurements in the condensation/immersion-freezing regime above water saturation were made from separate isokinetic and CVI inlets on the G-1 using the Colorado State University CFDC

[DeMott et al., 2010; Rogers et al., 2001]. The sample aerosol (1.5 vlpm) is focused by particle-free sheath air (8.5 vlpm) in the annular space between two cylindrical, ice- coated walls in the processing section of the CFDC. Processing temperature and humidity at the aerosol lamina are defined by the inner and outer wall temperature difference.

Processing temperature was set to be representative of conditions in sampled clouds.

Processing relative humidity was typically set to 104-105% with respect to water to favor the contributions of ice nucleation mechanisms as occur in mixed phase cloud conditions

(condensation and immersion freezing) [DeMott et al., 2010]. After ~5s in the growth section, the aerosols enter ice saturated conditions for ~3s to evaporate activated cloud droplets and allow detection of ice nuclei as particles larger than ~3 μm exiting the CFDC into an optical particle counter (OPC). An inertial impactor was used upstream of the

CFDC to restrict assessment of ice nuclei to aerosols smaller than ∼2.5 μm (aerodynamic

95 diameter), and to ensure that large aerosol particles are not falsely counted as ice crystals.

To improve sampling statistics for the low sample flow rates used, ice nuclei concentrations were calculated for integrated time periods for which uncertainties could be well defined based on Poisson sampling errors. Error bars are present on all CFDC data. The typical sample period was 3 to 10 minutes long. Sample periods were alternated with periods sampling particle free air in order to correct for any background frost production [DeMott et al., 2010].

An inertial impactor with a 50% cut-point aerodynamic diameter of 2.9 μm located immediately downstream of the CFDC OPC was used to capture ice crystals on

Transmission Electron Microscope (TEM) grids, allowing for subsequent identification of the elemental composition of activated ice nuclei. IN were collected onto carbon- coated Formvar films supported by 200 mesh Cu TEM grids (Ted Pella Inc.). TEM analyses was performed by the RJLee Group, Inc. (Monroeville, PA) using a Hitachi HD-

2300 dedicated 200 kV scanning transmission electron microscope (STEM). The analysis was conducted primarily in the bright field transmission electron mode at magnifications up to 450,000x. Compositional information was obtained through collection of characteristic X-rays using a Thermo Scientific Si(Li) energy dispersive

X-ray spectroscopy (EDS) system. On the order of 50 particles were examined per sample and micrographs of each particle were recorded to illustrate morphological characteristics. An energy dispersive spectrum was also acquired for each particle and in some cases for separate features within particles. Background spectra acquired from particle-free areas indicate the presence of carbon and oxygen (from the support film),

96 copper (from the TEM grid), and silicon (from an unknown source within the TEM).

Therefore identification of these elements in particles can be less certain in some cases; however, Si (e.g., from minerals) and C were distinguished on the basis of significantly higher EDS counts compared to background spectra. Particle categories were defined from the elemental spectra. For this study, categories were selected for alignment with those defined for single particle mass spectral categorization and include mineral dusts, dust/salt/biological, biological, and carbonaceous particles with likely sources from biomass burning and other organic materials. Dusts were defined as particles of irregular morphology containing oxidized Ca-Mg-Fe-Al silicates, sometimes with S and Cl. No attempt was made to distinguish industrial from natural dusts, so metal oxides also went into this category. Nevertheless, desert dust-like particles predominated in the sample.

The dust/salt/biological particles contained more carbonaceous material and salts than the pure minerals. Biological particles were typed based on a smooth morphology, excess of

C, the presence of non-oxidized Si, and a relative deficiency of Al or other metals, when present, compared to dust particles. Carbonaceous particles showed few spectral signatures aside from C, but appeared as solid carbon components within an apparent organic C matrix.

Data from the geostationary GOES-11 were used to define cloud top phase and cloud top heights over the SPD. The GOES-11 satellite is centered at 135°W over the eastern Pacific Ocean. Cloud properties from Jan 30─Mar 8, 2011 were retrieved for the

CalWater field campaign. The ﬁve channels on the GOES-11 imager include a visible channel (0.65 µm), which was calibrated to the Aqua MODIS 0.64-µm channel, as well

97 as four infrared channels, including a central wavelength at 3.9 µm used to discriminate water from ice clouds. The 4-km pixel GOES-11 data were analyzed each hour for a domain bounded by 30°N-42.5N° latitude and 112.5°W-130°W longitude using the methods described by Minnis et al. [Minnis et al., 2011]. They are available at website, http://cloudsgate2.larc.nasa.gov/cgi-bin/site/showdoc?docid=4&cmd=field-experiment- homepage&exp =calwater. Data from all parallax-corrected pixels within a 10-km radius of the SPD were averaged to produce a variety of parameters including percent ice clouds

(%ice) and cloud top heights (Ztops). Air mass back trajectories (10-day) were calculated using HYSPLIT 4 [Draxler and Rolph, 2013] ending at the average Ztops during precipitation sample collection time periods above SPD. To attain the most accuracy in estimating the transport paths during each sample, Ztops ranging from 1-10 km ASL were averaged every 3 hours and used as the ending altitudes for the back trajectory analysis, resulting in 122 total trajectories for the six storms. Back trajectories were calculated ending at 0000, 0300, 0600, 0900, 1200, 1500, 1800, and 2100 each day a precipitation sample was collected and ended at the averaged Ztop. The averaged Ztops also typically corresponded to precipitation time periods during the storms.

Data from the Navy Aerosol Analysis and Prediction System (NAAPS) provided by the Naval Research Laboratory (NRL) were acquired from the NRL website

(http://www.nrlmry.navy.mil/aerosol/). NAAPS is a transport model driven by wind fields and described in detail on the NRL website. The NAAPS global aerosol model uses meteorological fields from the Navy Operational Global Atmospheric Prediction System

(NOGAPS) [Hogan and Rosmond, 1991; Hogan and Brody, 1993] and land use types

98 from the United States Geological Survey (USGS) Land Cover Characteristics Database to predict the spatial distribution and relative amount of dust in the atmosphere. Dust emission occurs whenever the friction velocity exceeds a threshold value, snow depth is less than a critical value, and the surface moisture is less than a critical value. The model is validated daily with satellite data from MODIS and the National Environmental

Satellite, Data, and Information Service (NESDIS) of NOAA. Archived global data were used in Figure 4.4 of the manuscript from the global NAAPS plot generator and included simulations from Feb 6-16 (A) and Feb 15-25 (B). Time-height sections used to calculate dust plume altitudes were also acquired from the NRL website using NAAPS. These plots provide unitless dust concentrations at specific AERONET (Aerosol Robotic

Network) sites. Site used for this study are location within the dust regions shown in

Figure 4.3C of the manuscript and include Sedeboker (30°N, 34°E), Solar Village (24°N,

46°E), Yinchuan (38°N, 106°E), and Beijing (39°N, 116°E) (see Figure 4.16). There were no sites directly in the Taklimakan dust region; therefore the closest site to the east was chosen (Yinchuan). A trajectory that passed over more than one region suggests the air mass contained dust from more than one region, i.e., the air mass on Feb 25th likely contained dust from both the Sahara and Asian deserts. Figure 4.17 shows examples of images used to determine the dust plume heights. These data were used during each time a trajectory passed through each of the dust regions. Figure 4.18 shows each point in time when a trajectory passed through a dust region in addition to mean altitude of that trajectory point and maximum heights of the dust plumes acquired from the NAAPS time-height sections. Each data point in Figure 4.18 was used for the average, minimum, and maximum values in Figure 4.3C of the manuscript.

Imagery from the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite

Observation (CALIPSO) satellite was used to determine the presence of dust over arid regions where air mass back trajectories travelled, including North Africa, the Middle

East (specifically Saudi Arabia, Oman, Iran, Iraq, and Afghanistan), the Taklimakan

(West China), and East Asia (specifically Northeast China and Mongolia). CALIPSO flies at 705 km in the sun-synchronous, A-train satellite constellation and measures total attenuated backscatter at 532 nm and 1064 nm, and the depolarization at 532 nm through its Lidar, CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization), with a 60-m vertical resolution [Winker et al., 2010]. CALIPSO determines the coverage, water/ice content, and altitude of clouds in addition to the type of aerosol, including clean marine, dust, polluted continental, clean continental, polluted dust, and smoke. Images shown in

Figure 4.19 were constructed using the CALIPSO Lidar Level 2 Vertical Feature Mask

(VFM) data available from the CALIPSO data archive

(http://eosweb.larc.nasa.gov/HORDERBIN/HTML_Start.cgi). The time periods were chosen based on when the CALIPSO path transected the back trajectories; i.e., CALIPSO flies in the A-train configuration, thus traveling roughly southeast to northwest during the day and southwest to northeast night, while the back trajectories generally traveled from west to east. The “N/A” and 1-6 markers in Figure 4.4 show points where these transects occurred. The longitudes of the CALIPSO data correspond to the longitudes of each marker in Figure 4.4. The latitudes of the markers along the back trajectories are centered in each panel of Figure 4.19. The panels show an additional area of ±10 degrees of latitude around the latitude of the markers in Figure 4.4. Dust and polluted dust are shown, with other CALIPSO types combined and shown in grey.

100

4.9.2 Supplemental Discussion

Figure 4.9 shows the relative amount of dust and biological particles (A) as aerosols in clear air (i.e., not in cloud residues or interstitial aerosol), (B) in-cloud residues during each of the flights, (C) in each of the precipitation samples, and (D) as ambient aerosols measured at the ground at SPD. Cloud residues, clear air aerosol, precipitation residues, and ambient aerosols at the ground were separated into purely dust, dust mixed with salt and biological material, dust mixed with biological material, and purely biological particles. Sea salt is included as well, however, was not observed in the precipitation samples due to its soluble nature. Generally the percentages of dust and biological residues were higher in the precipitation compared to the cloud residues, which is likely a result of sampling mostly insoluble residues in the precipitation. For instance, sea salt comprised high percentages of cloud residues on several flights as shown in Figure 4.7B. Because sea salt is soluble, it would not be observed in the precipitation samples, thus increasing the percentage of dust and biological residues. Due to the nature of the sampling and aerosolization processes described above, purely dust residues were not observed in the precipitation; the dust in precipitation could have originally been purely dust and developed a biological coating during aerosolization or could have originally been dust mixed with biological material. Dust and biological particles in cloud residues were generally observed in high relative amounts when a substantial portion of clear air aerosols were dust and biological particles; however dust and biological aerosols are not always present at high altitudes or in clouds. On Feb 16th, the A-ATOFMS sampled only a few dust and biological particles in clear air, which could be explained by either selective removal from the air mass by ice forming

101 processes at mid-level altitudes and the fact that most sampling on the isokinetic inlet was done at lower altitudes in marine air masses. The relative amount of dust and biological residues in the precipitation is higher compared to the cloud residues because only insoluble residues are analyzed in the precipitation, thus the soluble material that originates from dust, biological, and other aerosol types are not observed. Further, dust and biological aerosols from local sources are potentially scavenged by falling precipitation, adding to the percentage of dust and biological residues seen in the precipitation samples. Ambient aerosol data measured at SPD during the storms was compared with the composition of precipitation residues; dust composed 15% and dust mixed with biological material composed 2% of the ambient aerosol (0.2-3.0 μm) on average, while no purely biological aerosols were observed. Further, the time periods with the highest relative amount of dust residues in the precipitation samples did not correspond to the highest relative amount of ambient dust aerosol. Thus, long-range transported dust and biological aerosols likely contributed more to the precipitation residue composition than local dust aerosols by serving as IN versus scavenging by falling precipitation. Additionally, many of the dust residue types in clouds and precipitation samples contained biological markers (i.e., soil dust), which makes disentangling the ice nucleating ability of dust and biological residues difficult.

However, both types of aerosols appeared to contribute to heterogeneous ice freezing in clouds.

The 11 samples included in Figure 4.1 and Table 4.2 covered roughly 12 days during 6 storms. The events include represented both the postfrontal and prefrontal conditions. The prefrontal environment included atmospheric river (AR) conditions and

102 a Sierra barrier jet (SBJ). Atmospheric rivers are relatively narrow regions in the lower troposphere where strong horizontal winds and large water vapor contents are focused.

They transport large amounts of water vapor (mostly below 2.5 km altitude) and are ideal for generating strong orographic precipitation [Ralph et al., 2005]. Sierra barrier jets occur on the windward side of the Sierra Nevada and are generally found below 1.5 km altitude parallel to the mountains [Neiman et al., 2010]. Recent observations have shown that when an AR strikes central/northern California the AR can rise up and over the SBJ if one is present [Kingsmill et al., 2012]. While ARs can be identified effectively offshore due to their signature in SSM/I satellite-observed vertically integrated water vapor (IWV;

[Ralph et al., 2004]), land-based observations were collected onshore during the

CalWater experiment by NOAA’s Physical Sciences Division to monitor AR and SBJ conditions at landfall and inland to the Sierra Nevada. These included several wind profilers that were able to identify both the SBJ and AR conditions, including GPS-met observations. An example is shown in Figure 4.20, where the wind profiler at Concord,

CA observed an AR passage, but no SBJ because the site was west of the western edge of the SBJ. In contrast, the Sloughhouse wind profiler (SHS) clearly documented the SBJ as well since it was closer to the base of the Sierra Nevada (SBJ events were identified per criteria established by Neiman et al. [Neiman et al., 2010]). SHS observations also showed evidence of AR conditions above the SBJ. Additionally, wind profiler and GPS- met observations at Bodega Bay on the coast (BBY) showed passage of the AR. The propagation across this network from west to east was also apparent. These signatures of

AR and SBJ conditions at SHS and BBY were used to determine if an AR and/or SBJ were present during the aerosol precipitation sampling at SPD in the Sierra Nevada

103 shown in Figure 4.1. These results are shown in Table 4.2, which provides some of the foundation for the summary schematic in prefrontal conditions (Figure 4.5A).

The flights on Feb 16th and 25th both corresponded to conditions a few hours after the passage of a cold front. Just 4-8 hours prior to each of these flights, the experimental meteorological observations showed that a SBJ was present in the lowest 2000 m, MSL, and a weak AR was present above that. In both cases the AR and SBJ had ended several hours prior to the flight, and winds were westerly. In short, conditions were postfrontal with weak upslope forcing. This weak upslope forcing could help explain why there was an apparent lack of local aerosol incorporated into clouds on these days. On the Feb 16th ascent, there were distinct cloud layers, including lower-level convective clouds (≤ 3500 m), which were decoupled from the boundary layer containing local pollution aerosol, and mid-level orographic clouds (~4000-6000 m, MSL). During the descent, the boundary layer rose higher in altitude, coupling with the convective clouds. In addition, the orographic clouds began to precipitate into the convective clouds.

The chemical markers of the dust and biological cloud residues on Feb 16th and

Feb 25th were distinct. On Feb 16th, a high fraction of the residues contained ion markers

23 + 27 + 60 - at Na , Al , and/or SiO2 and strongly absorbed the UV wavelengths at 266 nm (e.g.,

Figure 4.8A and B). However, on Feb 25th, a high fraction of the cloud residues were dust mixed with salt and biological markers as seen in Figure 4.8D. The biological markers

26 - 42 - 79 - th ( CN , CNO , and PO3 ) were prominent. These results suggest the Feb 16 residues were more “dust-like” whereas on Feb 25th, relevant air masses for ice formation were also influenced by biological species. This distinction is consistent with the ATOFMS

104 analysis of precipitation samples collected, which were more often categorized as “dust- like” on Feb 16th and biological on Feb 25th. Feb 16th precipitation residues were similar to those shown in Figure 4.7A-D, while on Feb 25th precipitation residues resembled those shown in Figure 4.7E and F. The precipitation residues on Feb 25th did not contain the same salt markers as the cloud residues because the salts dissolved in solution (as explained above). This evidence supports the trajectory analysis that shows that different source regions affected the clouds. Figure 4.15 shows images and spectra from TEM analysis for a dust IN residue (A and B) and a biological residue (C and D). Overall, 43 residues were analyzed from the IN TEM grids from the flight on Feb 16th. Dust residues represented 53% of the total, while a mix of dust, biological, and salt represented 26%.

What were thought to be purely biological residues represented 9% and the remaining

12% was carbonaceous, likely from biomass burning or other organic sources. The cloud residues measured by the A-ATOFMS on Feb 16th were 50% dust, 16% dust mixed with biological material, 3% biological, 19% sea salt, and 12% were other residues such as organic carbon or soot (Figure 4.9B). Thus, the TEM and A-ATOFMS residue composition analysis were in close agreement on Feb 16th. Unfortunately, due to contamination issues, no TEM data were available for Feb 25th.

Out-of-cloud aerosol concentrations in the size range of 0.6-10 μm may be considered as a proxy for dust particle concentrations, although it is not possible to exclude contributions of sea salt particles at times. To limit this interference, aerosol concentrations measured only between 3000-6000 m during flights are averaged and shown in Table 4.3. These results highlight a range of at least 20 times in the average

105 abundance of larger particles and dust at higher altitudes, which has been directly associated with IN concentrations in previous studies [DeMott et al., 2010]. Also included in Table 4.2 are the atmospheric pressures and precipitation totals measured at

SPD during each flight day. Interestingly, as the pressure decreased and precipitation increased, the aerosol concentrations decreased. This is likely due to the additional role of cloud and precipitation scavenging on aerosol concentrations observed at times in this altitude range that typically encompasses clouds over the Sierra Nevada. For example, on

Feb 16th and Feb 25th, precipitation totals were highest and out-of-cloud aerosol concentrations were relatively low (98 and 49 L-1, respectively). These are the flights that corresponded to high percentages of dust and biological cloud residues. These are the flights that corresponded to high percentages of dust and biological cloud residues. Table

4.3 shows the average values, however maximum values on Feb 16th and Feb 25th were

100-200 L-1 and ~50 L-1, respectively. If we assume an IN-active fraction of 0.005

[Niemand et al., 2012] at the lowest temperature at the top of the layer (-35C), that gives estimated IN concentrations of 0.5-1 L-1 on Feb 16th and 0.25 L-1 on Feb 25th, which is consistent with the CFDC data. These results suggest that the out-of-cloud aerosols were scavenged via serving as cloud seeds.

4.10 Supplemental Figures

106

Figure 4.6. Time-height cross sections (with time reversed per meteorological convention) of S-PROF radar-observed vertical profiles of precipitation above Sugar Pine Dam (SPD) in early 2011 (start times are shown on right side of each panel). The parameter shown is the signal-to-noise ratio (SNR), essentially the strength of the radar pulse scattered back to the radar as it propagated up through the atmosphere. Most of these radar “echoes” come from precipitation particles, but some are from cloud particles. Blue represents background noise where no echo was detected. The time period for each of the co-located precipitation samples used to diagnose aerosol residues using ATOFMS (Table 4.1) is shown, along with the rain gauge-observed precipitation accumulation during each sample period. Also shown are the time periods corresponding to the Feb 16th and Feb 25th flights. Color-filled squares are shown when the precipitation process- partitioning algorithm identified a type of precipitation. A key feature that distinguishes the precipitation type is the presence of a radar bright band, which occurs at an altitude range where snow falls and melts. It is evidenced by a band of bright (strong) echo that is nearly horizontal in the type of plot shown in this figure. Horizontal dashed lines mark the upper and lower edges of the lowest radar beams from the two nearest NEXRAD scanning radars.

107

108

Figure 4.8. Representative mass spectra for A-ATOFMS cloud residues. The silicon-rich dust (A), dust (B), dust mixed with biological material (dust/biological) (C), and dust mixed with salt and biological material (D) represent different dust types. Also shown are two different biological residue types in (E) and (F), sea salt (G), and inlet artifact (H). Positive and negative ion intensities vary by type.

109

(A) 100 Dust Dust/biological

80 Dust/salt/biological r

i Biological

o 60

r Sea salt

l e

C 40

a % 20

3 4 5 6 8 3 4 8 a a a

a a 3 4

b b b

b b

1 1 1 1 1 2 2 2 2 5 6

1 5

2 5 6

1 5

r r

2 2

r r r

a a

r r r

b b b b b b b b

a a a

e e e e e e e e

b b

b b M M

e e

F F F F F F F F M M M

e e

M M M

F F

F F Flight date

(B) 100

y d

t 60

C d

i 40

% s

e r 20

3 4 5 6 8 3 4 8 a a a

a a 3 4

b b b

b b

1 1 1 1 1 2 2 2 2 5 6

1 5

2 5 6

1 5

r r

2 2

r r r

a a

r r r

b b b b b b b b

a a a

e e e e e e e e

b b

b b M M

e e

F F F F F F F F M M M

e e

M M M

F F

F F Flight date

y a

t 60

d i

e 40

P r

% 20

0 5 6 7 8 9 6 2 3 6 7

- - - -

3 1 1 1 1 1 2

1 2 5 6

------

n 4 5 6 7 8 4

r r r r

a 1 1 1 1 1 2

a a a a

b b b b b b

M M M M

e e e e e e

F F F F F F Precipitation sample date Figure 4.9. Percentages of dust, dust mixed with biological material, dust mixed with salt and biological material, and biological particles (A) as clean air aerosol measured onboard the G-1, (B) in cloud residues measured onboard the G-1, (C) as residues in precipitation samples, and (D) in ambient aerosol at the ground level during the CalWater field campaign in 2011.

110

(D) 100

n l

e o

i 60

m e

A 40

a % 20

0 5 6 7 8 9 6 2 3 6 7

- - - -

3 1 1 1 1 1 2

1 2 5 6

------

n 4 5 6 7 8 4

r r r r

a 1 1 1 1 1 2

a a a a

b b b b b b

M M M M

e e e e e e

F F F F F F

Precipitation sample date

Figure 4.9. Percentages of dust, dust mixed with biological material, dust mixed with salt and biological material, and biological particles (A) as clean air aerosol measured onboard the G-1, (B) in cloud residues measured onboard the G-1, (C) as residues in precipitation samples, and (D) in ambient aerosol at the ground level during the CalWater field campaign in 2011 (continued).

111

Figure 4.10. Map showing the flight tracks for all 25 CalWater flights. McClellan Airfield is labeled in red.

Figure 4.11. Feb 16th flight track colored by time on the left and by ascent or descent on the right. The size of the markers in the left panel represents altitude (m, MSL).

112

Figure 4.12. The Feb 25th flight track colored by time with altitude (m, MSL) shown by the size of the markers.

113

Figure 4.13. Similar to Figure 4.2 but for the (A) ascent on Feb 25th (20:57-22:41 UTC) and (B) descent on Feb 25th (22:42-00:36 UTC). Vertical profiles of the number of cloud residues (dust, dust mixed with salt and biological material called “dust/salt/biological,” biological, and residues from pollution such as soot and biomass burning), liquid water drops or liquid water content (LWC, g/m3), cloud ice fraction, and cloud temperature (°C). Also shown are cloud particle residual IN concentrations (L-1) with the color representing the temperature at which the CFDC measurement was taken. Error bars represent twice the Poisson sampling errors in IN concentrations calculated for 3-10 minute time periods. The 2D-S cloud particle classifications are shown, with 0 = no images/out-of-cloud, 1 = droplets with no ice, 2 = isolated ice crystals among droplets, 3 = mixed droplets and ice crystals, and 4 = all ice (shown in Figure 4.14).

114

Figure 4.14. Representative 2D-S images including the classifications for 0-4. 0 = no images due to the aircraft flying out of cloud, 1 = liquid droplets with no ice crystals, 2 = isolated ice crystals, or liquid droplets with few ice crystals present, 3 = a mixture of ice crystals and liquid droplets, and 4 = all ice crystals (including rimed) and no liquid droplets. Ice crystals that have flat sides were out of the image field of view.

115

116

Figure 4.16. Google Earth map showing locations of AERONET sites used for NAAPS time-height section plots.

117

Figure 4.17. Example NAAPS time-height cross sections for (A) Sedeboker, (B) Solar Village, (C) Yinchuan, and (D) Beijing.

118

Figure 4.18. NAAPS time-height cross section data used to determine the maximum height of dust plumes at sites in each of the dust regions. The times of the dust plume heights correspond to times when trajectories passed over the dust regions. Also shown are the altitudes of these trajectories endpoints.

119

Figure 4.19. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript (markers labeled “N/A” and 1-6). The CALIPSO longitude matches the longitude of the markers in Figure 4.4, i.e., where the CALIPSO orbital path transected the back trajectories. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located. The center of each CALIPSO panel represents the latitude of the markers in Figure 4.4, with an additional 10 degrees latitude on either side. N/A and panels 1-3 correspond to the trajectory ending on Feb 16th while panels 4-6 correspond to the trajectory ending on Feb 25th above SPD.

120

Figure 4.20. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript (markers labeled “N/A” and 1-6). The CALIPSO longitude matches the longitude of the markers in Figure 4.4, i.e., where the CALIPSO orbital path transected the back trajectories. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located. The center of each CALIPSO panel represents the latitude of the markers in Figure 4.4, with an additional 10 degrees latitude on either side. N/A and panels 1-3 correspond to the trajectory ending on Feb 16th while panels 4-6 correspond to the trajectory ending on Feb 25th above SPD (continued).

121

Figure 4.21. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript (markers labeled “N/A” and 1-6). The CALIPSO longitude matches the longitude of the markers in Figure 4.4, i.e., where the CALIPSO orbital path transected the back trajectories. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located. The center of each CALIPSO panel represents the latitude of the markers in Figure 4.4, with an additional 10 degrees latitude on either side. N/A and panels 1-3 correspond to the trajectory ending on Feb 16th while panels 4-6 correspond to the trajectory ending on Feb 25th above SPD (continued).

122

Figure 4.22. CALIPSO images that correspond to the numbered markers in Figure 4.4 of the manuscript (markers labeled “N/A” and 1-6). The CALIPSO longitude matches the longitude of the markers in Figure 4.4, i.e., where the CALIPSO orbital path transected the back trajectories. Times of CALIPSO transect also correspond to times in the back trajectories where the markers are located. The center of each CALIPSO panel represents the latitude of the markers in Figure 4.4, with an additional 10 degrees latitude on either side. N/A and panels 1-3 correspond to the trajectory ending on Feb 16th while panels 4-6 correspond to the trajectory ending on Feb 25th above SPD (continued).

123

Figure 4.23. (A) Wind profiler observations from the Concord, California atmospheric river observatory (ARO) showing wind speed (color coded, knots) and direction as a function of time. Observations are hourly with 100-m vertical resolution and are measured using clear-air radar methods. (B) Same as (A) but for a site farther east at Sloughhouse. Comparison of (A) and (B) in the region highlighted by the lower rectangle in (B) reveals that the Sierra barrier jet (SBJ; southerly winds in the lowest 2 km MSL containing a low-latitude wind speed maximum) was present at Sloughhouse, but not at Concord. Conversely, the signature of AR conditions are seen in the period of strong southwesterly winds aloft (highlighted in the upper rectangles of (A) and (B)). Note that the AR conditions at Sloughhouse were found above the SBJ. Time is reversed.

124

4.11 Supplemental Tables

Storm Sample Start End Number Surface Median Minimum Maximum Date Date of Temp. Cloud Top Cloud Top Cloud Top (UTC) (UTC) Residues (°C) Temp. Temp. Temp. Analyzed (°C) (°C) (°C) 1 S1 1/30/11 1/30/11 130 2.24 -29 -24 -34 02:53 20:00 2 S2 2/14/11 2/15/11 360 5.54 -7 -6 -8 18:40 17:00 2 S3 2/15/11 2/16/11 266 3.22 -31.5 -31 -32 17:05 18:00 3 S4 2/16/11 2/17/11 233 -0.67 -29 -28 -30 19:45 17:30 3 S5 2/17/11 2/18/11 208 -0.11 -36 -32 -40 17:30 18:40 3 S6 2/18/11 2/19/11 163 0.37 -35 -35 -35 19:15 18:40 4 S7 2/24/11 2/26/11 94 -2.07 -26 -22 -30 20:30 21:00 5 S8 3/1/11 3/2/11 26 4.55 -22 -13 -31 23:00 23:00 5 S9 3/2/11 3/3/11 398 3.71 -22 -14 -30 23:00 19:00 6 S10 3/5/11 3/6/11 351 3.76 -25 -25 -25 21:00 18:15 6 S11 3/6/11 3/7/11 204 4.57 -18.5 -16 -21 18:15 18:00

125

n n the

fall and

own own in Figure 4.1).

), ), and at the coast using an

]

Total Total accumulated rain

. .

]

2004

2012

Ralph Ralph et al.,

[

were were diagnosed at SHS using criteria established

et et al.

Kingsmill et al.,

[

et et al.

, , and cold frontal passages were identified from SHS wind profiler data i

during during the 11 precipitation samples collected at SPD (sh

]

2010

Neiman Neiman et al.,

[

et et al.

Meteorological Meteorological conditions

in rate during each sampling period is also is shown. period sampling during each rate in

2 . 4

ing ing SSM/I satellite images (per criteria developed by Ralph

average ra average Table Observations were used to determine if atmospheric river conditions and/or SBJ conditions were present, was or a distinct cold if frontal passage there during each sampling period. Atmospheric river conditions were assessed offshore us atmospheric river observatory at Bodega Bay. maximum water hourly vapor transport (“bulk flux”). SBJ conditions AR amplitudes were provided by Neiman by IWV values and estimates lowest 1 of km AGL. The signature of AR conditions profiler observations overriding and patterns an shown SBJ by Kingsmill at SHS was identified using the SHS wind

126

Flight Average N Standard N Atmos. Precipitation (#/L) Deviation Pressure at Total at SPD SPD (mb) (mm) Feb 13 30 41 597 900.65 0 Feb 14 440 270 982 896.34 8.128 Feb 15 142 134 3361 897.2 12.7 Feb 16 98 131 1593 887.17 71.628 Feb 18 65 59 11 889.29 19.05 Feb 21a 127 352 437 894.5 0 Feb 21b 321 471 1144 Feb 23 0 0 0 895.68 0 Feb 24 209 141 450 896.91 1.27 Feb 25a 0 0 0 Feb 25b 49 57 1236 899.4 57.658 Feb 28 0 0 0 899.24 0 Mar 2a 477 213 1807 895.41 14.986 Mar 2b 97 233 699 Mar 3 0 0 0 899.41 15.24 Mar 4 0 0 0 902.94 0 Mar 5a 415 250 5608 Mar 5b 526 435 1770 903.04 0 Mar 6a 138 130 1845 Mar 6b 350 319 2260 895.82 26.67

Chapter 5. Biological Species Dominate

Ice Nucleation in Mixed Phase Marine

Clouds Warmer than -15 ºC

5.1 Abstract

Worldwide more than 50% of precipitation is initiated in the ice phase [Lau and

Wu, 2003] thus, understanding ice formation processes in clouds is an essential step towards quantitatively predicting precipitation events. Unfortunately, heterogeneous ice formation is poorly understood, partly because sampling ice in clouds presents many logistical and physical problems [Cantrell and Heymsfield, 2005]. Ice can shatter, complicating size and concentration measurements [Field et al., 2003; Gardiner and

Hallett, 1985; Korolev and Isaac, 2005]. Ice residuals do not always represent the ice nuclei (IN) that initially formed the ice crystals. Secondary ice multiplication is a physical process that does not relate directly to the presence of IN [Hallett and Mossop,

1974]. Also, IN are scarce [DeMott et al., 2010; Rogers et al., 2001] and their chemical signals can easily be overwhelmed by aerosols in liquid droplets. Due to these sampling obstacles, statistical analysis methods were employed in this study to chemically identify the key species involved in ice formation. Using in-situ cloud residual data from two different aircraft studies in regions impacted by marine sources, the ice nucleation temperature range of particle types containing salts and biological markers (salt+bio type)

127

128 and dust and biological markers (dust+bio type) were identified. These particle types were found to be present in pristine ice between -10 and -13 °C in both the Calwater and

ICE-T studies over the California Sierra Nevada, Pacific Ocean, and the Caribbean Sea.

Microscopy results reveal the salt+bio types have similar morphology to sea spray aerosol detected during periods of high biological activity, which suggests a marine origin. Statistical analysis of the mass spectral data shows that organic nitrogen (26CN-,

42 - 79 - CNO ) and phosphate ( PO3 ) are indicators of biological IN in both study locations; however, in the Caribbean the phosphate signal is overwhelmed by the presence of inorganic phosphate in Saharan desert dust particles that is present but not IN active at these warmer temperatures. These results provide insight into the importance and chemical composition of marine IN active at warmer temperatures than desert dust activates (< -15 °C [Ansmann et al., 2008; Connolly et al., 2009; DeMott and Prenni,

2010]). Additionally, these results suggest that the presence of biological species is critical to ice nucleation and the remainder of the particle, salt, pollution, or dust, is less important.

5.2 Introduction

Extensive supercooled liquid has been observed down to -21 °C in orographically formed convective clouds [Rosenfeld et al., 2013], -37.5 °C in convective clouds

[Rosenfeld and Woodley, 2000], -40.7 °C in orographic wave clouds [Heymsfield and

Miloshevich, 1993], and –40°C in cirrus clouds [Sassen et al., 1985]. The abundance of supercooled liquid at cold temperatures in a variety of cloud types suggests the limiting factor in ice formation is the availability of particle types that can initiate ice formation.

129

Therefore, identifying the aerosols that form ice is essential to accurately predicting ice nuclei and ice concentrations. Dust has been shown to be an efficient IN at colder cloud temperatures with ice formation in clouds containing dust occurring around -20 ºC

[Ansmann et al., 2008; Rosenfeld et al., 2011]. However, above -15 ºC mineral dust has limited IN activity [Connolly et al., 2009; DeMott and Prenni, 2010; O'Sullivan et al.,

2013]. Additionally, in dust scarce regions, such as over oceans, marine biological particles have been suggested to be a significant source of IN [Burrows et al., 2013]. A study in the Amazon showed biological particles to be a larger fraction of IN than desert dust above -25 ºC [Prenni et al., 2009]. It has been suggested for soil dust that the IN activity is mainly a result of the biological residues on the surface of the dust [Conen et al., 2011; O'Sullivan et al., 2013]. Additionally, biological particles and residues have been shown to be effective IN at warm temperatures, including pollen freezing around

-8 ºC [Morris et al., 2004a] and a specific bacterium, Pseudomonas-Syringae, freezing as warm as -2 °C [Maki et al., 1974; Yankofsky et al., 1981].

Ice concentrations at warm temperatures are difficult to predict, in part because most known IN sources become active at colder temperatures (~-20 °C, as discussed above). Creamean et al. [2013b] showed that dust and biological particles were important for ice formation in the California Sierra Nevada with a focus on ice present primarily below -20 ºC. Here we focus on residue compositions in ice at warmer temperatures, particularly above -15 ºC. Using in-situ cloud residue and microphysical data from two separate field studies, this work shows that particles with biological components were the most common type present in ice residuals and identifies the warmest temperature range where they were found in pristine ice.

130

5.3 Methods

The Ice in Clouds Experiment- Tropical (ICE-T) took place out of St. Croix,

USVI with flights over the Caribbean during July of 2011. The CalWater field campaign was between February and March of 2011 and flights occurred over the California Sierra

Nevada and the Pacific Ocean. Cloud microphysical properties and cloud residue chemistry were measured in-situ on board the National Science Foundation/ National

Center for Atmospheric Research C-130 during ICE-T and the Department of Energy G-1 during CalWater. A two-dimensional stereo probe (2D-S, SPEC Inc.) and a two- dimensional cloud probe (Fast-2DC, Particle Measuring Systems) were used to obtain images of cloud droplets and ice crystals. Counterflow virtual impactors (CVI) were used to sample cloud particles during the flights. During CalWater a stainless steel Brechtel

CVI (Model 1204, BMI) was used, while a titanium CVI (National Center for

Atmospheric Research design) was used during ICE-T. Cloud droplet and ice crystal residuals were chemically characterized using an aircraft- aerosol time-of-flight mass spectrometer (A-ATOFMS) [Pratt et al., 2009b]. Aerosols (0.1-2.5 μm vacuum aerodynamic diameter) are focused into the A-ATOFMS using an aerodynamic lens inlet

[Liu et al., 1995b]. The speeds of the particles are then measured using two 532 nm lasers spaced 6 cm apart, which are converted into vacuum aerodynamic diameters based on a calibration with polystyrene latex spheres of varying sizes. Finally, species are desorbed, ionized, and chemically analyzed using a pulsed 266 nm laser and the composition is measured using microchanel plate (MCP) detectors. Particles were classified by hand and grouped into similar particle types based on combinations of characteristic marker ions.

131

Scanning Electron Microscopy (SEM) analysis was performed on aerosol and cloud residue samples collected by impaction during ICE-T.

Ice periods were identified by viewing 2D-S and 2D-C imagery. Secondary ice crystals are formed by physical processes, and thus residuals from secondary ice formation periods do not always chemically represent aerosols involved in primary ice nucleation. Additionally, many ice crystals were rimed, indicating that the chemical signatures of the liquid droplets that froze to the ice are mixed with the chemical signature of the initial ice nucleus. Therefore, pristine ice residues provide the best insight into the original ice nucleus composition with the least amount of liquid residue interference. Unfortunately, in typical sampling conditions during both studies, liquid droplets were frequently present with ice in pristine ice periods in mixed phase clouds.

Therefore, chemical components from the liquid were present in most ice periods due to either liquid droplets or rime. Choosing the periods with the least liquid present allows for the most clear chemical signals of IN. Liquid amounts and ice types were noted for each ice period. Ice periods were defined as in Creamean et al. [2013b]. Needles found between -3 and -8 °C were labeled separately from other pristine ice to distinguish periods with possible secondary ice formation despite the fact that needles are the pristine habit for the temperature range -3 to -8 °C. The label “pristine” was used to describe other pristine ice habits (i.e. non-needle periods) characteristic of growth in their temperature range. Supercooled liquid periods were defined as periods with all liquid and no ice with temperatures below 0 °C. Chemical composition of the residuals was determined and the presence of salt, dust, biological, or soot components was noted.

132

Example sea salt and dust spectra are shown in Chapter 3, Figure 3.8 and Chapter 4,

Figure 4.8. Example biological spectra are shown in Figure 1 and Chapter 4, Figure 4.8.

Example soot spectra are in Chapter 3, Figure 3.2. A-ATOFMS data were corrected for a sampling time delay caused by aerosol and cloud residual transit time through the aircraft inlet by subtracting 30 seconds from the A-ATOFMS data so it lined up with the cloud probe data.

HYSPLIT was used to calculate back trajectories for the ICE-T flights [Draxler and Rolph, 2013; Rolph, 2013]. Back-trajectories were initiated at 1000, 2500, and 5000 m at St. Croix, USVI (17.73 N, 64.75 W). Sea surface temperature (SST) was retrieved from 100 km global composite images derived from 8 km global SST measurements from the Advanced Very High Resolution Radiometer (AVHRR) and High Resolution

Infrared Radiation Sounder (HIRS) satellites. Images were produced using NOAA’s

Comprehensive Large Array-Data Stewardship System (CLASS, http://www.nsof.class.noaa.gov). Statistical analysis was carried out using R

(http://www.r-project.org/). A binary logistic regression was performed using mass spectral areas, aerodynamic size, and temperature as inputs. Ice was treated as the binary outcome: 1= ice, 0= no ice, all liquid. The regression treated ice as a function of temperature, size, and chemical components. The full list of the chemical components used is included in the Supplementary Material (Tables 5.2 and 5.3). Each chemical component consisted of a series of absolute areas from the A-ATOFMS mass spectra of individual particles. The regression outputs are presented in the Supplementary Material

(Tables 5.2 and 5.3).

133

5.4 Results & Discussion

To investigate the effect of chemical composition on ice formation, periods of pristine ice in clouds at a variety of temperatures were targeted. Several metal-containing biological particles were observed in the pristine ice. Figure 5.1 shows the mass spectra of the most common biological particle types observed during CalWater (A) and ICE-T

(B), which had mixtures of sea salt, metals, and biological components (salt+bio).

Aluminum (27Al+) and iron (54/56Fe+) were the most commonly observed metals during

CalWater. While it has been reported in prior studies in cirrus clouds that ice cloud sampling can result in metal artifact production [Cziczo et al., 2013; Murphy et al., 2004], the artifacts from these studies have been characterized to minimize the impacts from inlet contamination [Suski et al., 2014a]. Common sea spray elements were also present in the biological particles including sodium (23Na+), magnesium (24Mg+), and potassium

(39K+). Due to similarities between dust and biological chemical types, strict criteria were followed to distinguish between marine biological and dust particles. Dust was

60 - 76 - 27 + characterized by having silicate markers ( SiO2 , SiO3 ) as well as other metals ( Al and/or 54/56Fe+) and biological components were identified by the presence of three

79 - 26 - 42 - marker ions, phosphate ( PO3 ) and organic nitrogen ( CN , CNO ) [Creamean et al.,

2013b; Pratt et al., 2009a]. The majority of dust particles also contained biological components. Therefore, the IN activity of biological particles can only be compared to particles with both dust and biological components (dust+bio), as opposed to pure dust.

To investigate the IN activity of the different chemical types, the in-situ cloud residual data for CalWater was plotted by chemical type on the x-axis and temperature on

134

the y-axis, shown in Figure 5.2A. The markers are colored by liquid or ice type. The results show that several chemical types were present in both liquid and ice over a range of temperatures. Salt and pollution residues were present in the coldest supercooled liquid periods (-20 to -28 °C), shown below the dashed line in Figure 5.2A. The composition of the residuals in cold supercooled liquid periods yields insight into less IN active particle types within the temperature range of -20 to -28 °C. Salt+bio and dust+bio types were the most prominent types in pristine ice. These types were not present in supercooled liquid at cold temperatures. This suggests that the salt+bio and dust+bio types were active as ice nuclei in the temperature range they were observed in pristine ice: -10 to -13 ºC and points to the biological component being key to ice formation. The particle types that were present in both supercooled liquid and ice, salt and pollution, were unlikely to be IN active in this temperature range and were most likely present in the liquid droplets around the ice. Dust was typically mixed with biological species; therefore, there was limited dust-only residual data in ice which limits our ability to determine the IN activation temperature of dust without biological species. However, dust residues were present in ice only below -20 °C, which is consistent with previous observations [Ansmann et al.,

2008; Rosenfeld et al., 2011]. These results suggest that the biological components played a critical role in IN formation in the mixed salt+bio and dust+bio residues detected in cloud residues between -10 and -13 °C.

The salt+bio type most frequently observed during ICE-T, shown in Figure 5.1B, was very similar to the chemical type observed during CalWater except that it had large amounts of copper (63/65Cu+). Copper was present on 80.2±0.6% of biological particles

135 sampled during ICE-T. This may suggest the presence of a biofilm on the inlet that leaches copper resulting in a copper inlet artifact. However, the fact that the rest of the mass spectrum matches cloud residues from CalWater, and the IN activity is similar suggests there is important information to be gleaned in the rest of the cloud residue as discussed in Suski et al. [2014a]. During CalWater, 30±1% of biological particles had iron, while in ICE-T 4.6±0.3% of biological particles had iron. If all particles with inlet metals were excluded from analysis, a substantial fraction of the salt+bio particles would be missed. Additionally, lead has been suggested to increase IN activity of mixed particle types by creating active sites for ice germs to grow [Cziczo et al., 2009]. Therefore, the metals, including iron, present on these salt+bio particles could potentially serve as an important activation site. This study cannot address whether metals are enhancing the IN activity of these salt+bio particles, but future studies will explore this possibility. There was not a chemical difference observed between the liquid, pristine ice, and needles during ICE-T, shown in Figure 5.2B. This could be due to the large amount of rime present on the ice and the large amount of liquid water present in the clouds.

Additionally, ice formed at warm temperatures, which could point to an abundance of IN coupled with a large amount of secondary ice formation. For addition cloud microphysical analysis, see Suski et al. [2014b].

To determine the origin of the biological particles, microscopy studies were performed. Figure 5.3A shows a scanning electron microscope (SEM) image of a salt+bio particle sampled during ICE-T. These particles are of marine origin, as indicated by the cubic structure of the salt, and were present in both clear air and cloud residue samples.

136

These images resemble sea spray particles sampled during a wave flume study. Ault et al.

[2013] showed that submicron sea spray particles composed of organic carbon mixed with sea salt lost their cubic structure and appeared more spherical in bacteria and phytoplankton enriched seawater. Figure 5.3B shows a salt+organic carbon particle that is semi-spherical and resembles those shown in Ault et al. [2013]. The ICE-T salt+bio type had similar IN activity to the CalWater salt+bio type as seen in Figure 5.2B. During

ICE-T, these residues were detected in pristine ice between -10 and -13 ºC. Cold supercooled liquid was not observed during ICE-T; therefore, we cannot deduce which residues were likely in liquid versus ice during mixed phase periods. However, the similarity in the mass spectra, the temperature range where these salt+bio types were found in ice in both studies, and their morphological similarity to sea spray during periods of biological activity suggests that these salt+bio particles were of marine origin and were possibly IN active between -10 to -13 °C.

These results suggest that marine biological species are important for ice formation at relatively warm temperatures, -10 to -13 °C. Unfortunately, conclusive information regarding aerosol types forming ice at temperatures below -13 ºC was not obtained in these studies. During ICE-T, sampling at temperatures colder than -14 ºC did not occur as most measurements were made between 0 and -10 ºC. This temperature range was chosen as the goal of the study was to sample the first ice to form in developing tropical cumulus clouds. Thus, typical sampling started at 0 ºC and followed clouds up as they developed. Ice was present at very warm temperature (~-5 ºC). In

CalWater, extensive sampling was done at colder temperatures (down to -28 °C);

137 however, the aerosol types that were present in ice regions at colder temperatures were mostly mixed dust and biological residues. As described above, there was not sufficient dust data without biological components to determine the IN activation temperature range of dust with this method.

To investigate which chemical components showed the strongest correlation with ice formation during these studies, statistical analysis was performed on the data sets.

Due to complications with quantifying ice amounts (ice volume does not correlate to the number of IN present due to the mechanisms of ice crystal growth and ice multiplication processes), a binary logistic regression (BLR) was used with the condition that ice was treated as a binary outcome: ice was either present or absent. Chemical contributions from liquid are usually present in most ice conditions in mixed phase clouds due to riming and the presence of liquid droplets alongside ice; therefore, the aim of the statistical analysis was to isolate chemical components that were found in ice regions more frequently than in liquid regions. The results from the BLR are presented in Table

5.1. For each study, as expected, temperature has the best correlation with the presence of ice, as shown in Figure 5.6. For the CalWater data set, aluminum (27Al+), phosphate

79 - 26 - 23 + 39 + ( PO3 ), organic nitrogen ( CN ), and salt components ( Na , K ) also correlate well with ice formation. Aluminum is present in dust as well as biological particles while phosphate and organic nitrogen are markers indicative of biological species in ATOFMS data. These binary logistic regression results suggest that biological components are particularly important for ice formation during CalWater. In the ICE-T analysis, an organic nitrogen marker (42CNO-) correlated best with ice formation. Again, this suggests

138 the importance of biological components for ice formation. Additionally, ammonium

18 + ( NH4 ) and soot were anticorrelated to ice formation during ICE-T, which suggests the soot was not IN active in the sampled temperature range. Figure 5.4 shows the BLR fitted

42 - 79 - results against aluminum area and biological markers ( CNO and PO3 ). For both studies, aluminum appears to be a strong indicator of ice. Interestingly, in the CalWater data there is a cluster of large phosphate areas when the probability of ice is high, shown in the top left panel of Figure 5.4. This suggests that there are two types of aluminum- containing particles: one with phosphate, which is possibly of biological origin, and one without phosphate. This result also shows that the phosphate-containing particles are more likely to be found in ice. This further verifies that the phosphate or biological component of the particles is driving the IN activity.

The ICE-T data do not show the same relationship between phosphate area and probability of ice. However, the probability of ice does increase with increasing CNO areas, as shown as Figure 5.4. CNO, an organic nitrogen marker, is one of three common markers in biological particles, along with phosphate and CN. The fact that the biological markers that correlate with ice formation differ between the two studies is intriguing and may provide insight into the sources or abundance of phosphate and organic nitrogen in the regions. For instance, different biological particles could be present in the two studies.

In CalWater, sampling occurred around the Pacific Ocean in winter with daily sea surface temperatures averaging between 10 and 14 ºC, as measured with satellite data, while in

ICE-T sampling occurred over the Caribbean Sea in summer with sea surface temperatures between 28 and 30 ºC. These differences could correspond to diverse

139 marine biological activity, which could account for the different biological signals

42 - 79 - ( CNO vs PO3 ) correlating with ice. An alternative explanation is the abundance of inorganic phosphate present in Africa dust. Recent observations have shown inorganic phosphorous in elevated levels in Saharan dust due to magmatic hot spots [Gross et al.,

2013] and an abundance of apatite: (Ca10(PO4)6(OH)2, Ca10(PO4)6F2 and/or

Ca10(PO4)6Cl2) [Dall'Osto et al., 2010; Singer et al., 2004], with decreased phosphate after transport, which may be due to matrix effects of the changing aerosol [Dall'Osto et al., 2010]. The abundance of phosphate provides a unique tracer for Saharan dust; however, biological components, including phosphate, are commonly associated with dust. Due to the fact that A-ATOFMS cannot distinguish between organic and inorganic phosphate, if both types are present on the same particle, the abundance of inorganic phosphate in dust cannot be deconvoluted from bioorganic phosphate.

Back trajectory analysis for the CalWater study revealed that dust plumes originated over Asia and Africa [Creamean et al., 2013b]. Similar back trajectory analysis for ICE-T shows that all of the sampled air masses trace back to Africa. Figure

5.5 shows a representative back trajectory at 1000, 2500, and 5000 m. These trajectories suggest that even on days when a clear dust layer was not present; background Saharan dust could still be present. This dust, while not likely active as IN at temperatures above

-10 ºC, could contribute to cloud residuals by acting as cloud condensation nuclei (CCN).

The chemical contribution of inorganic phosphate to cloud residuals could overwhelm the biological phosphate signal in particles where both inorganic and biological phosphate is present causing phosphate and ice formation to have no correlation. Therefore, the ice

140 correlation to 42CNO- is indicative of the presence of biological species acting as IN, as these particles do contain all three required biological markers and the SEM images are suggestive of marine biological particles.

5.5 Conclusions

This study demonstrates that biological components are key to ice formation at

< -15 °C in marine impacted regions. The IN activation temperature of salt+bio and dust+bio particle types were bracketed using in-situ cloud residual data. The biological types were present in pristine ice between -10 and -13 ºC, suggesting they are active as

IN as warm as that temperature range. These results reveal additional IN sources that are active at temperatures warmer than desert dust IN activation. Using binary logistic regression, markers associated with biological activity, including organic nitrogen in

ICE-T and phosphate in CalWater, are shown to correlate to the presence of ice in the two studies. SEM images from ICE-T reveal organic carbon and sea salt mixed particles that are suggestive of marine biological particles. These results together suggest that biological components play a key role in the IN activity of mixed particle types and that the dust and salt components are of secondary importance. Further study of marine biological particles and their impact on ice formation is warranted, particularly over the

Caribbean where ice forms at warm temperatures. Ongoing wave flume studies investigating bioparticles as IN are being conducted to explore the range of behavior for different biological species including phytoplankton, bacteria, and viruses. Speciation of the biological IN is a necessary next step to understanding and predicting ice formation in clouds. It is anticipated that wave flume studies will guide future field investigations

141 probing the varying IN ability of sea spray as biological activity changes with location, season, and sea surface temperature.

5.6 Acknowledgements

Chapter 5 contents are part of a manuscript in preparation, Kaitlyn J. Suski,

Daniel Rosenfeld, James R. Anderson, Cynthia H. Twohy, Andrew J. Heymsfield, and

Kimberly A. Prather (2014), “Biological Species Dominate Ice Nucleation in Mixed

Phase Marine Clouds Warmer than -15 ºC”, In preparation for submission to Nature

Geosciences. Funding was provided by the California Energy Commission under contract

CEC 500-09-043 and the National Science Foundation under award number 1118735.

Alberto Cazorla is acknowledged for helping collect the A-ATOFMS data during ICE-T.

Darin Toohey is acknowledged for his role in operating the CVI during ICE-T. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website

(http://www.ready.noaa.gov) used in this publication. The authors also acknowledge

NOAA’s Comprehensive Large Array-Data Stewardship System (CLASS, http://www.nsof.class.noaa.gov) for the provision of sea surface temperature from the

Advanced Very High Resolution Radiometer (AVHRR) and High Resolution Infrared

Radiation Sounder (HIRS) satellites used in this manuscript.

142

5.7 Figures

Figure 5.1. Mass spectra from A) CalWater and B) ICE-T showing the salt+bio particle types observed during the studies.

143

Figure 5.2. A) CalWater cloud residual chemical types plotted against temperature for different ice and liquid regions in mixed phase clouds. Markers are colored by ice type and sized by A-ATOFMS aerodynamic size. Cold supercooled liquid residuals are the purple markers below the black dashed line. B) ICE-T cloud residual chemical types plotted against temperature. Markers are colored by ice or liquid type.

144

Figure 5.3. SEM images of salt+bio (A) and salt+organic carbon (B) particles sampled during ICE-T.

145

146

Hours from Site 0 -50 -100 -150 -200

5000 5000 4000 4000 3000 3000 2000 2000 1000 1000

Altitude (mASL) Altitude 60 60

40 40

20 20 Latitude

0 0

-20 -20 -80 -60 -40 -20 0 20

Longitude

147

5.8 Tables

Table 5.1. Binary logistic regression results showing the chemical species that correlate to ice formation for CalWater and ICE-T. Negative correlations are italicized.

Significance CalWater ICE-T Temperature, CNO, Mg, CN, PO , Al, Na, *** 3 Temperature, NH K, Sulfate 4 ** Silicates, soot CNO * Ca, Cr Soot . --- Silicates

148

5.9 Supplementary Figures

Figure 5.6. Binary logistic regression fitted results plotted against mass spectral areas (CNO, phosphate) and temperature for CalWater (left panels) and ICE-T (right panels).

149

5.10 Supplementary Tables

Table 5.2. Binary logistic regression loadings for the CalWater data set. Significance codes are Pr(>|z|) at values of 0-0.001(***), 0.001-0.01 (‘*), 0.01-0.05 (*), 0.05-0.1 (.), and 0.1-1 ( ).

Chemical Mass-to- Estimate Standard z value Pr(>|z|) Significance Input charge Error (Intercept) -1.80E+00 6.00E-02 -29.957 2.00E-16 *** Size 1.76E-05 7.66E-05 0.229 0.818687 Temperature -4.16E-02 3.30E-03 -12.619 2.00E-16 *** Iron 54/56 -6.22E-06 3.99E-06 -1.556 0.119726 CN -26 1.55E-05 3.72E-06 4.16 3.19E-05 *** CNO -42 -3.05E-05 7.06E-06 -4.322 1.54E-05 *** Phosphate -79 3.38E-05 6.57E-06 5.148 2.63E-07 *** Aluminum 27 9.71E-06 1.11E-06 8.721 2.00E-16 *** Sodium 23 8.75E-06 2.53E-06 3.457 0.000545 *** Magnesium 24 -2.43E-05 4.25E-06 -5.718 1.08E-08 *** Copper 63/65 -2.04E-05 4.72E-05 -0.431 0.666431 Zinc 64/66/68 1.16E-04 1.79E-04 0.647 0.517326 Calcium 40 5.85E-06 2.28E-06 2.57 0.010175 * Titanium 48 8.73E-07 2.43E-05 0.036 0.97128 Titanium 64 -4.00E-06 4.26E-05 -0.094 0.925312 Oxide Potassium 39/41 2.65E-05 3.65E-06 7.275 3.46E-13 *** Silicates 1 -76 -9.70E-06 1.22E-05 -0.793 0.427622 Silicates 2 -77 -3.53E-05 1.25E-05 -2.817 0.004848 ** Silicates 3 -60 3.32E-06 6.44E-06 0.515 0.606405 Soot 36 1.56E-04 5.34E-05 2.92 0.003503 ** Sulfate -97 1.93E-04 5.63E-05 3.426 0.000612 *** Nitrate -46 -4.68E-06 3.56E-06 -1.315 0.188427 Nitrate 2 -62 -6.89E-06 4.79E-06 -1.438 0.150323 Ammonium 18 -1.05E-05 8.73E-05 -0.12 0.904773 Chromium 52 1.11E-05 4.98E-06 2.226 0.026038 *

150

Table 5.3. Binary logistic regression loadings for the ICE-T data set. Significance codes are Pr(>|z|) at values of 0-0.001(***), 0.001-0.01 (‘*), 0.01-0.05 (*), 0.05-0.1 (.), and 0.1-1 ( ).

Chemical Mass-to- Estimate Standard z value Pr(>|z|) Significance Input charge Error (Intercept) -6.14E-01 4.90E-02 -12.521 2.00E-16 *** Size -3.80E-05 1.03E-04 -0.368 0.712686 Temperature -1.05E-01 3.65E-03 -28.883 2.00E-16 *** Iron 54/56 -4.76E-06 4.20E-06 -1.133 0.257233 CN -26 3.99E-06 2.76E-06 1.447 0.147863 CNO -42 9.47E-06 3.51E-06 2.702 0.006901 ** Phosphate -79 8.54E-07 3.90E-06 0.219 0.826736 Aluminum 27 3.88E-06 2.39E-06 1.624 0.104426 Sodium 23 6.71E-08 1.19E-06 0.056 0.955084 Magnesium 24 -2.13E-07 1.76E-06 -0.121 0.903577 Copper 63/65 2.68E-06 1.94E-06 1.383 0.166635 Zinc 64/66/68 -6.11E-06 1.29E-05 -0.473 0.635948 Calcium 40 -1.04E-06 1.24E-06 -0.841 0.400305 Titanium 48 -4.42E-07 3.73E-06 -0.118 0.905801 Titanium 64 4.14E-06 5.34E-06 0.775 0.438621 Oxide Potassium 39/41 -5.21E-07 1.36E-06 -0.383 0.70184 Silicates 1 -76 -8.39E-06 6.16E-06 -1.363 0.172884 Silicates 2 -77 1.22E-05 7.35E-06 1.663 0.096298 . Silicates 3 -60 -7.46E-06 6.39E-06 -1.169 0.242541 Soot 36 -8.85E-04 4.41E-04 -2.007 0.044723 * Sulfate -97 2.19E-06 3.28E-06 0.669 0.503726 Nitrate -46 -4.43E-06 4.37E-06 -1.013 0.311028 Nitrate 2 -62 -9.48E-06 9.71E-06 -0.976 0.328991 Ammonium 18 -7.31E-03 1.90E-03 -3.855 0.000116 ***

Chapter 6. Principal Component Analysis on In-Situ Cloud Residual Data

6.1 Abstract

The methods used to sample ice in-situ in mixed phase clouds are complicated by many physical processes, resulting in a poor understanding of heterogeneous ice formation. Statistical analysis can be used to extract relevant information about what makes certain aerosols more efficient ice nuclei (IN), as demonstrated in the previous chapter. Here we present an alternative statistical analysis method used to elucidate the sources of cloud condensation nuclei and ice nuclei in clouds. Principal component analysis (PCA) can identify how mixtures of particle types can influence ice formation, while binary logistic regression can only assess chemical types separately. PCA was performed using single particle mass spectrometer data of cloud droplet and ice crystal residuals from the CalWater and Ice in Clouds Experiment – Tropical (ICE-T) field campaigns. A mixed salt-dust-biological chemical type was ubiquitous in CalWater and

ICE-T and back-trajectory analysis suggests this type was of marine origin. The presence of a potassium phosphate chemical type in ice suggests that biological particles play an important role in ice formation during ICE-T and during periods of transport over the

Pacific Ocean during CalWater. These results illustrate the unique ability of statistical analysis methods to identify the sources of dominant chemical components in ice nucleation processes.

151

152

6.2 Introduction

Ice formation in clouds affects both precipitation and climate [Lohmann, 2002;

Lohmann and Feichter, 2005]; however, there are many unanswered questions regarding heterogeneous ice nucleation [Cantrell and Heymsfield, 2005]. One major complication to in-situ ice nuclei analysis is the fact that ice undergoes many physical and chemical changes in clouds and thus the ice residuals do not always chemically represent the original IN that nucleated the ice. For instance, liquid droplets can collide with ice and freeze on contact to the surface of the ice to form rime, which occurs frequently in mixed phase clouds [Houze, 1993], resulting in a mixed chemical signal made up of both liquid and ice residues. In addition, aerosols are complex mixtures of various chemical species

[Seinfeld and Pandis, 2006], which complicates the attribution of ice nucleation to one specific chemical type. Previous studies have shown, however, that complex mixtures are efficient IN [Ebert et al., 2011]; therefore, understanding the individual and mixed abilities of different chemical species in aerosols is extremely important to predicting IN and ice concentrations in clouds.

Various online methods have been used to attempt to uncover the identity of IN in clouds. Most notably, using a continuous flow diffusion chamber (CFDC) [Rogers et al.,

2001] in-situ to activate aerosols as ice nuclei and derive the fraction of aerosols that activate. Comparisons of CFDC concentrations to aerosol chemistry have revealed much about which chemical species are efficient IN, including dust and biological particles

[Creamean et al., 2013b; DeMott et al., 2003a; Pratt et al., 2009a]. However, where chemistry, morphology, or biology dictates ice formation is not well understood and

153 requires further study. Statistical analysis methods can be employed to uncover which chemical species in a mixture may be important for ice nucleation. Binary logistic regression was used in Chapter 5 to show biological components correlated to the presence of ice in both CalWater and ICE-T. Here we show an alternative statistical analysis method that may be used to learn about the sources of aerosols that are frequently observed in ice. Principal component analysis (PCA), which uses empirical orthogonal functions, has frequently been used to identify patterns in meteorological datasets and aims to simplify datasets by using a subset of the available variables to explain the majority of the variance in the data [Hannachi et al., 2007]. This technique is distinct from the binary logistic regression in that it provides information about mixtures of chemical species that describe the behavior of ice formation in clouds, while binary logistic regression examines chemical constituents individually. In this chapter, PCA is used to identify the sources of chemical components that describe the presence of ice and reveals marine source to be dominate during both CalWater and ICE-T.

6.3 Methods

During the CalWater and Ice in Clouds Experiment – Tropical (ICE-T) field campaigns, cloud residues were sampled through counterflow virtual impactor (CVI) inlets as described in the previous chapter. Cloud residues were chemically characterized using aircraft-aerosol time-of-slight mass spectrometry (A-ATOFMS) [Pratt et al.,

2009b], which is described in detail in the previous chapters. Principle Component

Analysis (PCA) was performed on the CalWater and ICE-T data sets. R (http://www.r- project.org/) was used to perform the analysis using the prcomp command. Individual

154 particle information was input into the analysis in a matrix with one mass spectral area for each key mass-to-charge for each particle. Key mass spectral mass-to-charge points are found in table 6.8. Cloud imagery from a two-dimensional stereo probe (2D-S, SPEC

Inc.) and a two-dimensional cloud probe (Fast-2DC, Particle Measuring Systems) was used to classify cloud data into ice and liquid time periods. The classification method used in Creamean et. al. [2013b] was used to define ice periods. Then, temperature was matched to each time period and periods containing ice were separated into ice > -12 and ice < -12 °C. The threshold for warm and cold ice (-12 °C) was chosen based on dust’s limited IN activity below -15 °C and the warm cloud temperatures observed during

ICE-T. Sampling did not occur below -14 °C during ICE-T, thus -12 °C was chosen to allow for cold ice periods (-12 to -14 °C) during ICE-T. The mass spectral area data were separated based on these phase and temperature designations and PCA was run for a variety of different subgroups including: ice > -12 °C (warm ice), ice < -12 °C (cold ice), all liquid, and all residues together (All). From the output loadings chemical compositions were constructed by combining all chemical markers that had a loading above 0.1. A component that had all three silicate markers was labeled “Sil”, all three biological markers was labeled “Bio”, and salt markers was labeled “Salt”. Then, the top five components for each ice and liquid region were compared to deduce which chemical types were most important to ice formation. Next, the data were further separated based on back-trajectory analysis in order to trace distinct chemical types to their source locations. Back-trajectories were obtained using HYSPLIT clustering [Draxler and

Rolph, 2013]. Ten day back-trajectories were initiated twice daily at 00:00 UTC and

12:00 UTC from Sugar Pine Dam (39.13 N, 120.80 W) at 3 km AGL. Then, the

155 trajectories were clustered, which resulted in four main trajectory types: Long Asia, Short

Asia, Long Pacific, and Short Pacific, shown in Figure 6.1.

6.4 Results & Discussion:

6.4.1 CalWater

Principal component analysis was used to identify chemical components key to ice formation. In-situ cloud residual data were separated based on the presence of ice or liquid and then further separated at the temperature the ice was observed. There were two ice categories: ice warmer than -12 °C and ice colder than -12 °C. PCA was used on these separate data sets and the results were compared to try to determine whether any chemical components stood out as being important to ice formation. In all cases (All, liquid, warm ice and cold ice), the primary component was a salt-dust-bio chemical type, as seen in table 6.1. This indicates that salt components (sodium, 23Na+), dust components

60 - 76 - 26 - 42 - 79 - (silicates, SiO2 and SiO3 ) and biological markers ( CN , CNO , and PO3 ) were contributing to this one component. The presence of the salt-dust-bio component in all ice and liquid periods suggests these types may frequently be present as background aerosols. The second component for the All category was a mixed pollution and dust type that contained both dust markers, including metals (48Ti+, 54/56Fe+) and silicates, as well as

+ pollution markers like carbon clusters from soot and ammonium (NH4 ). This mixed pollution-dust type was also the third component in the liquid data. The presence of this type in liquid and All and absence in the ice periods suggests that while pollution was present during the flights, it was likely present in liquid droplets and did not nucleate ice at the temperatures sampled.

156

Dust signatures were present in all five of the top components of the cold ice periods. This indicates that dust plays a significant role in ice formation at temperature below -12 °C during CalWater, which is consistent with previous studies [O'Sullivan et al., 2013]. The difference between the warm ice and cold ice components further highlights the role of dust at cold temperatures. Warm ice had components mostly composed of soot, biological, and salt types. This mixture of pollution and natural chemical types suggests that pollution liquid residues could be mixed with the warm ice to a large extent and the biological components could be responsible for ice formation at these warmer temperatures.

6.4.2 ICE-T

As in the CalWater study, salt-dust-bio components were the top component for all of the different liquid and ice categories for ICE-T (table 6.2). In the cold ice, two different salt-biological types were the top two components, which could indicate that biological species, possibly of marine origin, were responsible for ice formation at

< -12 ºC. These results verify the findings from Chapter 5, which showed the importance of biological components to ice formation. Dust was the second component for All, liquid, and warm ice indicating that background dust was likely present at all temperatures and altitudes during ICE-T. As in CalWater, soot types were present in the liquid, warm, and cold ice.

Dust was present in the top components for < -12 ºC ice during CalWater, while biological components were more prominent in cold ice during ICE-T. This could be a result of colder sampling temperatures during CalWater, which allow for dust to become

157

IN active below -20 °C [Ansmann et al., 2008; Connolly et al., 2009], while the coldest sampled temperatures during ICE-T were down to -14 °C. Interestingly, in both CalWater and ICE-T, the first component is the salt-dust-bio type for all categories. This result points to the frequent presence of these chemical types at various temperatures in the cloud residuals.

Copper and zinc were present in several components in the CalWater and ICE-T

PCA results. These metals are commonly found in brazing alloys used to join metal pieces. Pieces of the stainless steel CVI used during CalWater were joined with brazing alloy containing nickel, silver, and copper. The copper and zinc were not noticeably large peaks in the CalWater spectra, but the PCA pulled them out as important to explain the variance of the dataset. Additionally, copper was present on a large portion of particles sampled through a titanium CVI during ICE-T, as discussed in Chapter 2. The presence of metals used to join pieces of the CVI in both inlets, including brazing alloy on the

CalWater CVI and electro-discharge machining with brass wire on the ICE-T CVI, may be responsible for the common occurrence of these metals in the PCA components.

6.4.3 CalWater Sources

To differentiate between aerosol sources in the components, back-trajectory analysis was conducted. Trajectory analysis for the ICE-T data set revealed that all of the trajectories came from the east and most originated over Africa, as shown in Chapter 5.

This suggests that Saharan dust and sea spray, along with local pollution, were the primary aerosol sources encountered during ICE-T. CalWater back-trajectories were clustered using the HYSPLIT clustering algorithm [Draxler and Rolph, 2013]. The

158 clustering resulted in four different back-trajectory types: Long Asia, Short Asia, Long

Pacific, and Short Pacific. The trajectory clusters are shown in Figure 6.1. The Long Asia trajectories originate either over the Asian continent or as far back as Africa. The Short

Asia trajectories also originate in Asia, but closer to the eastern coast of Asia than the

Long Asia trajectories. The Long Pacific and Short Pacific trajectories both originate over the Pacific Ocean, with the Long Pacific trajectories originating further west than the Short Pacific trajectories. These trajectories were used to separate the cloud residual data then, principal component analysis was run on each of these trajectory groups for each ice and liquid category (liquid, warm ice, cold ice, and All).

There was a large contribution of metals to the components from the Long Asia trajectories. The fifth component from the All category and the third component from warm ice were made up of several metals found in stainless steel including iron (54/56Fe+), chromium (52Cr+), and nickel (59Ni+), which suggests the presence of stainless steel inlet artifacts during these sampling times. Warm ice from Long Asia trajectories was sampled during the February 18, 2011 and February 25b, 2011 flights. Extensive regions of supercooled liquid were observed during the February 18th flight, as described in

Rosenfeld et al. [2013]. These supercooled conditions can result in freezing of supercooled liquid to the inlet walls, which can subsequently break or melt off forming artifacts, as described in Chapter 2. During warm ice periods on the February 25b flight, the sample temperature was below zero and there were large relative humidity (RH) spikes, indicating possible artifact production, detailed in Chapter 2. These conditions

159 suggest that the metals present during warm ice from Long Asia trajectories were the result of stainless steel artifact contamination.

Soot was present in components in the Short Pacific (All, liquid, and warm ice ) data and in all categories for the Long Pacific data, which suggests that soot was transported across the Pacific Ocean and further verifies long-range transport of Asian pollution, which was described in Chapter 3. Salt-dust-bio components appear in the

Short Asia, Long Pacific, and Short Pacific trajectories suggesting that Asian dust could be mixing with sea spray during transport or in clouds. Additionally, both the warm and cold ice periods for the Long Asia trajectories have top components with all negative loading values, which could indicate that there was not a dominant chemical type that explained the variance of the data and might suggest that the absence of certain chemical types explained the variance.

A potassium phosphate (K-PO3) component appears in the Long Pacific data

(cold ice, All and liquid), Short Pacific cold ice, and the warm ice in Short Asia data.

This same chemical type was present in precipitation samples collected during CalWater and aerosolized into the ATOFMS [Creamean et al., 2013b]. Potassium phosphate residues, suspected to be vegetative detritus, were also observed in precipitation samples collected in both the northeast Pacific and the Indian Ocean [Holecek et al., 2007].

Additionally, in laboratory control studies, leaf litter samples produced similar potassium phosphate signatures as both the precipitation samples and the PCA components

[Creamean et al., 2013a]. Thus, the potassium phosphate chemical type is frequently present in cloud and precipitation samples from trajectories that spent a significant

160 amount of time over the Pacific Ocean, which could point to this type being of marine origin or terrestrial origin, which was suggested by previous work [Creamean et al.,

2013a; Holecek et al., 2007]. Future studies will compare potassium phosphate signatures from both marine and terrestrial sources to differentiate between the two and investigate their ice nucleating abilities.

6.5 Conclusions

Principal component analysis was utilized to probe the chemical composition of aerosols key to ice formation in mixed phase clouds. Overall, a mixed salt-dust-bio type was omnipresent during CalWater and ICE-T and was likely the result of transported dust mixing with sea spray during transport and in cloud processing. PCA results suggest potassium phosphate particles, most likely of biological origin, play an important role in ice formation during Pacific transport in CalWater. Further, the presence of potassium phosphate residues observed in precipitation samples from CalWater [Creamean et al.,

2013b] verifies the presence of potassium phosphate in rain from precipitating clouds.

Additionally, the presence of dust in several cold ice components during CalWater suggests dust is important to formation below -12 °C, analogous to previous work by

O'Sullivan et al. [2013]. Further exploration of statistical analysis techniques applied to single particle mass spectrometer and cloud microphysical data could provide invaluable insights into ice formation in mixed phase clouds, which presents many in-situ sampling complications.

161

6.6 Acknowledgements

Chapter 6 contents are part of a manuscript in preparation, Kaitlyn J. Suski,

Daniel Rosenfeld, Alberto Cazorla, and Kimberly A. Prather. Funding was provided by the California Energy Commission under contract CEC 500-09-043 and the National

Science Foundation under award number 1118735. Cynthia Twohy and Darin Toohey are acknowledged for operating the CVI during ICE-T. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www.ready.noaa.gov) used in this publication.

162

6.7 Figures

163

6.8 Tables

Table 6.1. CalWater chemical components derived from PCA loadings separated by liquid, warm ice, and cold ice.

Component All Liquid Ice > -12 C Ice < -12 C 1 Salt-Dust-Bio Salt-Dust-Bio Salt-Dust-Bio Salt-Dust-Bio 2 Ti-Fe-Al-Sil-Soot- Al-Sil-Fe-Ti-PO3-CN Na-Zn Fe-Cu-Ti NH4 3 Salt-Cu-Zn-Ca-Ti- Ti-Cu-Na-Fe-Soot-Nit- Na-Mg-Ca-Nit- Al-Sil Dust Soot-Nit-NH4 NH4 NH4 4 Fe-Ca-Nit-NH4 Na-Mg-Cu-Zn-Ca Soot-Sulf-Zn Al-Sil Dust 5 Salt-Ti-Ca-Zn-Cu Fe-Al-Cu-Zn-Soot- Zn-Cu-Al-Ti-PO3- Na-Mg-Ca-Ti NH4 CN

Table 6.2. ICE-T chemical components derived from PCA loadings separated by liquid, warm ice, and cold ice.

Component All Liquid Ice > -12 C Ice < -12 C 1 Salt-Dust-Bio Salt-Dust-Bio Salt-Dust-Bio Salt-Dust-Bio 2 Fe-Al-Ti-Sil Dust Fe-Al-Ti-Sil Dust Fe-Al-Ti-Sil Dust Na-Ca-K-Bio 3 Na-Ca-Mg-Al-Ti Ca-Mg-Na-Ti-Al- Soot-Nit-NH4 Al-Fe-Ti-Cu- CN Ca-Soot-Nit 4 Fe-Sil-Nit Silicates Zn-Ca-Al-Mg-Ti-CN- Zn-Ca-Mg-Al- Soot Fe-Soot-Sulf 5 Cu-Na-CNO-PO3- PO3-CNO-Cu- Fe-Al-Cu-Zn Na-Mg-K-Ti- Nit Soot-Nit CN

Table 6.3. Back-trajectory cluster assignments by flight number for CalWater 2011.

Long Asia Short Asia Long Pacific Short Pacific Feb 18 Feb 15 Feb 21 a Feb 14 Feb 23 Feb 16 Feb 21 b Mar 2 b Feb 25 b Feb 24 Mar 2 a Mar 4 Feb 25 a Mar 6 a Feb 28 Mar 6 b Mar 3 Mar 5 a Mar 5 b

164

Table 6.4. Chemical components derived from PCA loadings for the Long Asia trajectories.

Component All Liquid Ice>-12 Ice<-12 1 Salt-Dust-Bio Salt-Dust-Bio All negative All negative 2 Na-Mg-Ca-Mo Na-Mg-Zn-Ca-Mo Na-Ca-Mo-Zn- Na-Mg-Ca-Mo- Nit Nit 3 Al-Fe-Sil Dust Al-Fe-CN-Mn-Sulf-Sil Fe-Al-Sil-Cr-Ni- Fe-Ag-Ti-TiO NH4 4 Ca-Na-Mg-Mo-Ti- Na-Mg-Cu-Sil-Sulf- Zn-CN-CNO- Na-Mg-Ca-Cu-Ti- TiO Mo-NH4 Sil-Sulf TiO-Ag-Mo-Soot 5 Fe-PO3-Al-Cr-Ag- Fe-Mg-Zn-Ti-Ni-Cr- Fe-Mn Fe-Al-PO3-Soot- Ni-Mo Soot-Sulf Cr-Ag-Mo-Ni-Mn

Table 6.5. Chemical components derived from PCA loadings for the Short Asia trajectories.

Component All Liquid Ice>-12 Ice<-12 1 All negative Salt-Dust-Bio All negative Salt-Dust-Bio 2 Al-Cr-Ag-Mo- Na-Mg-Ca-K-Nit- Fe-Cr-Ag-Ti-TiO- Al-Cr-Mo-Ni-Mn- Ni-Mn-Sil-Nit- NH4 Soot-NH4 Nit-CN-CNO CN-CNO 3 Na-Mg-Ca-K- Ag-Mo-Ni-Mn-Nit- Al-Mn-Bio-Sil-Sulf Fe-Al-Cu-Cr-Ti- Ag-Mo-Ni-Mn- CN TiO-CNO Nit 4 Ni-Mn-Sil-Nit- Ca-Mg-Ag-Mo-Sil- Fe-Ca-Cr-Ag-Mo- Na-Mg-Zn-Cr-Ag- NH4-PO3 CNO PO3-Nit Al-Sil-CNO 5 Ag-Mo-Sil-PO3 Fe-Cr-Ca-Cu-Mg-Na- K-PO3-Sil-Nit Fe-Al-Cr-PO3-NH4 Al-Ti-TiO

Table 6.6. Chemical components derived from PCA loadings for the Long Pacific trajectories.

Component All Liquid Ice>-12 Ice<-12 1 All negative All negative Salt-Dust-Bio All Negative 2 Fe-Cr-Ag-Ti- Fe-Ti-TiO-Soot-NH4- Na-Mg-Ca-K-Mo-Ni Fe-Cr-Ag-Ti-TiO- TiO-Soot- Cr-Ag-Ni Soot-NH4 NH4 3 Al-Mn-Sil- Na-Mg-Cu-Zn-Ca-Ni- Cu-Zn-Mn-PO3-Sil Al-Mn-Bio-Sil- Sulf-Bio nit Sulf 4 Ca-Fe-Cr- Cr-Ag-Mo-Ni-PO3- Mn-Cr-Al-Sulf-Soot- Fe-Ca-Cr-Ag-Mo- Ag-Mo-PO3- Sil-Nit-NH4 Sil-CN-CNO PO3-Nit Nit 5 K-PO3-Sil- K-PO3-Ti-TiO-Sil K-Mn-Mg-Sil-NH4 K-PO3-Sil-Nit Nit

165

Table 6.7. Chemical components derived from PCA loadings for the Short Pacific trajectories.

Component All Liquid Ice>-12 Ice<-12 1 Salt-Dust-Bio Salt-Dust-Bio All negative --- 2 Fe-Al-Cr-Ni-CN-PO3-Ti- Fe-Al-Ti-TiO-Cr-Ni- Fe-Cr-Ag-Ti-TiO- --- TiO-Sil-Soot-Sulf-NH4 PO3-Soot-Sil-Sulf-NH4 Soot-NH4 3 Ca-Mo-Mn-CN-Nit Ca-Mn-Mo-CN-Nit Al-Sil-Bio-Mn-Sulf --- 4 Fe-Cr-Mn-Ti-Sil Al-Fe-Cr-Ti-Sil Fe-Ca-Cr-Ag-Mo- --- Nit-PO3 5 Al-Cr-Sil-CN-CNO Al-Cr-Sil-CN-CNO K-PO3-Sil-Nit ---

166

6.9 Supplementary Materials

Table 6.8. PCA chemical inputs, their abbreviations, and associated mass-to-charges for all of the PCA runs.

Chemical Input Abbreviation Mass-to-charge Iron Fe 54/56 Organic Nitrogen 1 (CN) CN -26 Organic Nitrogen 2 (CNO) CNO -42 Phosphate PO3 -79 Aluminum Al 27 Sodium Na 23 Magnesium Mg 24 Copper Cu 63/65 Zinc Zn 64/66/68 Calcium Ca 40 Titanium Ti 48 Titanium Oxide TiO 64 Potassium K 39/41 Silicates 1 Sil1 -76 Silicates 2 Sil2 -77 Silicates 3 Sil3 -60 Soot Soot 36 Sulfate Sulf -97 Nitrate Nit1 -46 Nitrate 2 Nit2 -62 Ammonium NH4 18 Chromium Cr 52 Nickel Ni 59 Manganese Mn 55 Molybdenum Mo 96 Silver Ag 107

Chapter 7. Conclusions

7.1 Conclusions

This dissertation investigates the effects of aerosols on clouds with a focus on ice formation during two aircraft field studies: CalWater and the Ice in Clouds Experiment-

Tropical (ICE-T). During CalWater, trans-Pacific transported natural and anthropogenic aerosols showed opposite effects on cloud properties and precipitation over the California

Sierra Nevada. Dust was shown serve as ice nuclei in clouds leading to enhanced ice- induced precipitation, while pollution and soot from Asia was shown to decrease cloud droplet sizes which results in a decrease in precipitation. Additionally, biological particles were found to be important to warm ice formation (> -15 °C) in both studies and thus likely play an important role in precipitation processes.

7.1.1 Dust, Bioparticles, or Artifacts? Unraveling the Sources of Metals in Cloud

Residues

Laboratory scratch tests were performed to better understand the metal combinations that could be detected by ATOFMS in aircraft inlet artifacts. Particles comprised of metal flakes were generated by scratching inlet materials used in ICE-T and

CalWater, titanium and stainless steel. These metal particles were deposited in in milli-q water, aerosolized, and sent directly into the ATOFMS for chemical analysis. The mass spectra of the particles produced by the scratch tests showed distinct chemical signatures with unique metal combinations. Once the unique inlet metal spectra were determined, known artifact periods in cloud flight data were isolated using specific criteria. Residues

167

168 from periods with ice shattering, high condensed water content, and sample temperatures below 0 °C were analyzed and residues sampled during these periods were compared to the scratch test mass spectra. A relatively low percentage of particles (4% had titanium during ICE-T and 2% had chromium during CalWater) sampled through the CVIs during periods of likely artifact contamination showed similar chemical signatures as the scratch test spectra, verifying that artifact particles can be chemically identified and separated from other metal-containing particles.

The combination of iron and chromium with biological signatures in mixed phase cloud residues were similar to spectra obtained in bacteria-enriched sea spray aerosol

[Guasco et al., 2013]. While previous studies have removed all inlet metal-containing particles observed during cold cloud sampling [Cziczo et al., 2013], these results show that metal-containing particles should not be excluded from in-situ data solely based on the presence of certain metals.

7.1.2 Long Range Transport of Asian Soot Impacts Cloud Properties over

California

A high altitude soot layer was observed during a flight on March 5, 2011. The layer had high concentrations of soot and biomass burning particles and was highly absorbing. The soot was transported across the Pacific Ocean and originated over Asia, likely the result of regional Asian pollution and was chemically very similar to soot observed in Shanghai, China 8 days prior. The long-range transported soot was later incorporated into clouds over the California Sierra Nevada resulting in high concentrations of small droplets that contained soot and biomass burning particles. This

169 study is the first to show using coupled in-situ chemical and microphysical measurements that clouds in California were modified by Asian pollution. Previous studies have shown that high concentrations of small droplets lead to reduced regional precipitation

[Rosenfeld et al., 2008] and thus, future increases in Asian pollution could contribute to further precipitation reduction in the California.

7.1.3 Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western U.S.

Long-range transported dust and biological particles from Asia and the Sahara were shown to be related to cloud ice formation and enhanced ice-induced precipitation in the California Sierra Nevada using a combination of in-situ cloud microphysical and chemical residual data, ground-based precipitation chemistry, clear air aerosol measurements, dust transport patterns and frequency from satellite data, and meteorology. Precipitation samples from periods of enhanced ice-induced precipitation contained dust and biological residues. Furthermore, in-situ cloud residual data from aircraft sampling showed dust and biological particles were also present in clouds with high IN concentrations and co-located with ice.

The dust and biological particles were shown to participate in the seeder-feeder mechanism, where they were entrained into “seeder” clouds and served as ice nuclei.

Then, the ice grew and fell into “feeder” clouds, where large droplets with sea salt rimed the ice, which grew to large, precipitable sizes. These results suggest that dust and biological particles play a key role in ice formation in mid-level clouds that result in enhanced ice-induced precipitation. Thus, trans-Pacific transport of dust from Asia and

170 the Sahara can have significant impacts on precipitation amounts and the water budget in the Western United States.

7.1.4 Biological Species Dominate Ice Nucleation in Mixed Phase Marine Clouds

Warmer than -15 ºC

Biological components have been implicated as key players in warm ice formation across the globe [Burrows et al., 2013; Prenni et al., 2009]. In-situ cloud residual chemistry and microphysical data were used to identify the IN activation temperature of salt-biological and dust-biological particle types between -10 and -13 ºC.

These particle types were present in pristine ice in clouds in both California and the

Caribbean. Using binary logistic regression, it is shown that in California, phosphate is a key marker that tracks ice formation, while organic nitrogen tracked ice formation in the tropics. Microscopy images of particles sampled during ICE-T show the collocation of sea salt and biological components on the same particles and the morphology resembles sea spray during periods of biological activity. These results together suggest that marine biological particles play a key role in ice formation near oceans.

Despite phosphate being a previously identified constituent of biological aerosols, it did not track ice formation during ICE-T. It is hypothesized that the presence of

Saharan dust with large amounts of inorganic phosphorous overwhelmed the phosphate signal in the statistical analysis resulting in a lack of a correlation with ice. Desert dusts are known to be efficient IN at colder temperatures [Ansmann et al., 2008; Connolly et al., 2009; DeMott and Prenni, 2010]. These results provide insights about IN active at warmer temperatures than dust activation. Additionally, these results suggest the presence

171 of biological components has the largest impact on the ice nucleating ability of aerosols, and the rest of the particle, whether it is salt, dust, or pollution has less of an effect.

7.1.5 Principal Component Analysis on In-situ Cloud Residual Data

Principal component analysis was used to investigate the sources and chemical components key to ice formation in clouds. A salt-dust-biological type was present as the primary component in all liquid and ice types. This suggests that salt, biological, and dust components were ubiquitous in the clouds sampled in both studies. Additionally, biological components were present in the top components in ICE-T and in CalWater during transport across the Pacific Ocean.

7.2 Future Directions

This dissertation investigated aerosol effects on cloud properties using in-situ cloud residual and microphysical data from two aircraft field campaigns. The studies in this thesis involved the chemical characterization of ice residues over the Sierra Nevada and the Caribbean, identified their IN activation temperatures, and linked dust and biological aerosols with increased ice-induced precipitation processes. This serves as an initial glimpse into the importance of chemical composition on cloud properties and precipitation processes; however, there are still many outstanding questions regarding aerosol effects remaining.

In order to fully understand the impact of inlet artifacts, we recommend further wind tunnel studies using various CVI inlet materials. These studies should employ cloud probes at various positions in the inlet to pinpoint locations where artifact formation most frequently occurs. Wind speed and particle size should be varied to fully characterize the

172 range of conditions where artifact production occurs. These tests could result in quantitative prediction of artifact frequency, amount, and size. Additionally, development and testing of a new CVI inlet to reduce sampling artifacts is crucial to future sampling of mixed phase clouds. The use of a gold-plated inlet may help with artifact reduction and identification [Froyd et al., 2010; Murphy et al., 2004].

The next phase of the CalWater field study, CalWater 2, is planned for 2015

(http://www.esrl.noaa.gov/psd/calwater/). When future flights are conducted as part of

CalWater 2, the flights will extend later into March to overlap more with the burning season in South East Asia. Additional observations of high altitude soot could provide further insights into the roles transported soot and Asian pollution play in clouds in

California. Detailed studies of liquid water content, cloud droplet size, and concentrations related to high-altitude soot concentrations will help define the frequency and extent to which soot aerosols affect clouds in California. Dust has been shown to be transported regularly across the Pacific, and two studies have shown dust in precipitation samples collected in the region [Ault et al., 2011; Creamean et al., 2013b]. As well as flights, additional precipitation sampling should be performed to cover a broader spatial range and identify the frequency and patterns of dust transport and its effects on precipitation.

We also suggest detailed laboratory studies of marine biological IN to identify which species of biological particles affect ice nucleation at various locations, sea surface temperatures, and seasons.

To identify the sources of ice nuclei at all temperatures present in mixed phase clouds, additional in-situ cloud sampling should extend measurements from temperatures

173 around zero down to -20 ºC, with substantial amount of time spent sampling between -10 and -15 °C. It is essential to study the biologically active IN in this temperature range and to include the in-situ study of dust IN activity around -20 ºC. An understanding of IN activity and location of IN with respect to temperature in clouds will enable better prediction of aerosol effects on mixed phase clouds. While ice-induced precipitation was shown to be enhanced when dust and biological particles were incorporated into clouds, quantification of these effects through laboratory studies of ice nucleation and modeling of cloud modification could greatly improve quantitative precipitation predictions.

Additionally, this dissertation showed that in cases where the boundary layer was decoupled from clouds long-range transported aerosols played a key role in modifying cloud properties in the Sierra Nevada region. Studies of conditions where the boundary layer is coupled to the clouds would provide additional insights into which aerosol sources are most important for cloud modification in the region and will allow for better prediction of precipitation amounts.

Finally, this dissertation utilized two statistical analysis methods, binary logistic regression and principal component analysis, to probe the roles of various chemical components in ice formation. Principal component analysis results shed light on sources of ice nuclei, but quantification of ice nucleation temperature was not possible. Binary logistic regression revealed trends between chemical types and ice formation in clouds and could be used in other studies of cold clouds to help identify important sources of IN.

However, the binary logistic regression analysis is limited by the fact that it treats ice as a binary condition. Therefore, if ice could be quantified in meaningful ways, such as

174 comparing ice crystal concentrations to the concentrations of aerosols in regions where secondary ice formation is not a concern, then other statistical analysis techniques could be employed that may produce quantitative results regarding aerosol effects on ice in clouds.

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.

Albrecht, B. A. (1989), Aerosols, Cloud Microphysics, and Fractional Cloudiness, Science, 245(4923), 1227-1230.

Alfaro, S. C., and L. Gomes (2001), Modeling mineral aerosol production by wind erosion: Emission intensities and aerosol size distributions in source areas, J Geophys Res-Atmos, 106(D16), 18075-18084.

Andreae, M. O., and D. Rosenfeld (2008), Aerosol-cloud-precipitation interactions. Part 1. The nature and sources of cloud-active aerosols, Earth Sci. Rev., 89(1-2), 13- 41.

Ansmann, A., M. Tesche, D. Althausen, D. Muller, P. Seifert, V. Freudenthaler, B. Heese, M. Wiegner, G. Pisani, P. Knippertz, and O. Dubovik (2008), Influence of Saharan dust on cloud glaciation in southern Morocco during the Saharan Mineral Dust Experiment, J Geophys Res-Atmos, 113(D4).

Ault, A. P., C. J. Gaston, Y. Wang, G. Dominguez, M. H. Thiemens, and K. A. Prather (2010), Characterization of the Single Particle Mixing State of Individual Ship Plume Events Measured at the Port of Los Angeles, Environ Sci Technol, 44(6), 1954-1961.

Ault, A. P., T. M. Peters, E. J. Sawvel, G. S. Casuccio, R. D. Willis, G. A. Norris, and V. H. Grassian (2012), Single-Particle SEM-EDX Analysis of Iron-Containing Coarse Particulate Matter in an Urban Environment: Sources and Distribution of Iron within Cleveland, Ohio, Environ Sci Technol, 46(8), 4331-4339.

Ault, A. P., C. R. Williams, A. B. White, P. J. Neiman, J. M. Creamean, C. J. Gaston, F. M. Ralph, and K. A. Prather (2011), Detection of Asian dust in California orographic precipitation, J. Geophys. Res., 116.

Ault, A. P., R. C. Moffet, J. Baltrusaitis, D. B. Collins, M. J. Ruppel, L. A. Cuadra- Rodriguez, D. Zhao, T. L. Guasco, C. J. Ebben, F. M. Geiger, T. H. Bertram, K. A. Prather, and V. H. Grassian (2013), Size-Dependent Changes in Sea Spray Aerosol Composition and Properties with Different Seawater Conditions, Environ Sci Technol, 47(11), 5603-5612.

Bahadur, R., P. S. Praveen, Y. Y. Xu, and V. Ramanathan (2012), Solar absorption by elemental and brown carbon determined from spectral observations, P Natl Acad Sci USA, 109(43), 17366-17371.

175

176

Barker, D. R., and H. Zeitlin (1972), Metal-Ion Concentrations in Sea-Surface Microlayer and Size-Separated Atmospheric Aerosol Samples in Hawaii, J Geophys Res, 77(27), 5076-&.

Baumgardner, D., L. Avallone, A. Bansemer, S. Borrmann, P. Brown, U. Bundke, P. Y. Chuang, D. Cziczo, P. Field, M. Gallagher, J. F. Gayet, A. Heymsfield, A. Korolev, M. Kramer, G. McFarquhar, S. Mertes, O. Mohler, S. Lance, P. Lawson, M. D. Petters, K. Pratt, G. Roberts, D. Rogers, O. Stetzer, J. Stith, W. Strapp, C. Twohy, and M. Wendisch (2012), IN SITU, AIRBORNE INSTRUMENTATION Addressing and Solving Measurement Problems in Ice Clouds, Bulletin of the American Meteorological Society, 93(2), E529-E534.

Beech, W. B., and J. Sunner (2004), Biocorrosion: towards understanding interactions between biofilms and metals, Curr Opin Biotech, 15(3), 181-186.

Ben-Ami, Y., I. Koren, Y. Rudich, P. Artaxo, S. T. Martin, and M. O. Andreae (2010), Transport of North African dust from the Bodele depression to the Amazon Basin: a case study, Atmos Chem Phys, 10(16), 7533-7544.

Bergmann, T., T. P. Martin, and H. Schaber (1989), High-Resolution Time-of-Flight Mass Spectrometers .1. Effects of Field Distortions in the Vicinity of Wire Meshes, Rev Sci Instrum, 60(3), 347-349.

Bergstrom, R. W., P. B. Russell, and P. Hignett (2002), Wavelength dependence of the absorption of black carbon particles: Predictions and results from the TARFOX experiment and implications for the aerosol single scattering albedo, J Atmos Sci, 59(3), 567-577.

Berntsen, T. K., S. Karlsdottir, and D. A. Jaffe (1999), Influence of Asian emissions on the composition of air reaching the North Western United States, Geophys Res Lett, 26(14), 2171-2174.

Bigg, E. K., and C. Leck (2008), The composition of fragments of bubbles bursting at the ocean surface, J Geophys Res-Atmos, 113(D11).

Blanchard, C., and L. Syzdek (1970), Mechanism for Water-to-Air Transfer and Concentration of Bacteria, Science, 170(3958), 626-&.

Bond, T. C., T. L. Anderson, and D. Campbell (1999), Calibration and Intercomparison of Filter-Based Measurements of Visible Light Absorption by Aerosols, Aerosol Sci Tech, 30(6), 582-600.

Borys, R. D., D. H. Lowenthal, and D. L. Mitchell (2000), The relationships among cloud microphysics, chemistry, and precipitation rate in cold mountain clouds, Atmos. Environ., 34(16), 2593-2602.

177

Burrows, S. M., C. Hoose, U. Poschl, and M. G. Lawrence (2013), Ice nuclei in marine air: biogenic particles or dust?, Atmos Chem Phys, 13(1), 245-267.

Cantrell, W., and A. Heymsfield (2005), Production of ice in tropospheric clouds - A review, Bulletin of the American Meteorological Society, 86(6), 795-807.

Carlton, A. G., C. Wiedinmyer, and J. H. Kroll (2009), A review of Secondary Organic Aerosol (SOA) formation from isoprene, Atmos Chem Phys, 9(14), 4987-5005.

Cazorla, A., R. Bahadur, K. J. Suski, J. F. Cahill, D. Chand, B. Schmid, V. Ramanathan, and K. A. Prather (2013), Relating aerosol absorption due to soot, organic carbon, and dust to emission sources determined from in-situ chemical measurements, Atmos. Chem. Phys., 13, 9337-9350.

Chen, B., A. Andersson, M. Lee, E. N. Kirillova, Q. F. Xiao, M. Krusa, M. N. Shi, K. Hu, Z. F. Lu, D. G. Streets, K. Du, and O. Gustafsson (2013), Source Forensics of Black Carbon Aerosols from China, Environ Sci Technol, 47(16), 9102-9108.

Choi, Y. S., R. S. Lindzen, C. H. Ho, and J. Kim (2010), Space observations of cold- cloud phase change, Proc. Natl. Acad. Sci. U.S.A., 107(25), 11211-11216.

Choudhary, R., H. Kumar, and R. K. Garg (2010), Analysis and evaluation of heat affected zones in electric discharge machining of EN-31 die steel, Indian J Eng Mater S, 17(2), 91-98.

Choularton, T. W., and S. J. Perry (1986), A Model of the Orographic Enhancement of Snowfall by the Seeder-Feeder Mechanism, Q. J. R. Meteorol. Soc., 112(472), 335-345.

Conen, F., C. E. Morris, J. Leifeld, M. V. Yakutin, and C. Alewell (2011), Biological residues define the ice nucleation properties of soil dust, Atmos Chem Phys, 11(18), 9643-9648.

Connolly, P. J., O. Mohler, P. R. Field, H. Saathoff, R. Burgess, T. Choularton, and M. Gallagher (2009), Studies of heterogeneous freezing by three different desert dust samples, Atmos Chem Phys, 9(8), 2805-2824.

Cooper, O. R., D. D. Parrish, A. Stohl, M. Trainer, P. Nedelec, V. Thouret, J. P. Cammas, S. J. Oltmans, B. J. Johnson, D. Tarasick, T. Leblanc, I. S. McDermid, D. Jaffe, R. Gao, J. Stith, T. Ryerson, K. Aikin, T. Campos, A. Weinheimer, and M. A. Avery (2010), Increasing springtime ozone mixing ratios in the free troposphere over western North America, Nature, 463(7279), 344-348.

Costerton, J. W., Z. Lewandowski, D. E. Caldwell, D. R. Korber, and H. M. Lappinscott (1995), Microbial Biofilms, Annu Rev Microbiol, 49, 711-745.

178

Creamean, J. M., T. C. Hill, C. Lee, A. P. Ault, C. J. Gaston, P. J. DeMott, K. A. Prather, A. B. White, and F. M. Ralph (2013a), Understanding the Chemical Properties of Insoluble Precipitation Residues, J Aerosol Sci, In Preparation.

Creamean, J. M., K. J. Suski, D. Rosenfeld, A. Cazorla, P. J. DeMott, R. C. Sullivan, A. B. White, F. M. Ralph, P. Minnis, J. M. Comstock, J. M. Tomlinson, and K. A. Prather (2013b), Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western U.S., Science, 339(6127), 1572-1578.

Cziczo, D. J., K. D. Froyd, C. Hoose, E. J. Jensen, M. H. Diao, M. A. Zondlo, J. B. Smith, C. H. Twohy, and D. M. Murphy (2013), Clarifying the Dominant Sources and Mechanisms of Cirrus Cloud Formation, Science, 340(6138), 1320-1324.

Cziczo, D. J., O. Stetzer, A. Worringen, M. Ebert, S. Weinbruch, M. Kamphus, S. J. Gallavardin, J. Curtius, S. Borrmann, K. D. Froyd, S. Mertes, O. Mohler, and U. Lohmann (2009), Inadvertent climate modification due to anthropogenic lead, Nat Geosci, 2(5), 333-336.

Dall'Osto, M., R. M. Harrison, E. J. Highwood, C. O'Dowd, D. Ceburnis, X. Querol, and E. P. Achterberg (2010), Variation of the mixing state of Saharan dust particles with atmospheric transport, Atmospheric Environment, 44(26), 3135-3146. de Gouw, J. A., O. R. Cooper, C. Warneke, P. K. Hudson, F. C. Fehsenfeld, J. S. Holloway, G. Hubler, D. K. Nicks, J. B. Nowak, D. D. Parrish, T. B. Ryerson, E. L. Atlas, S. G. Donnelly, S. M. Schauffler, V. Stroud, K. Johnson, G. R. Carmichael, and D. G. Streets (2004), Chemical composition of air masses transported from Asia to the U. S. West Coast during ITCT 2K2: Fossil fuel combustion versus biomass-burning signatures, J Geophys Res-Atmos, 109(D23).

DeMott, P. J., and A. J. Prenni (2010), New Directions: Need for defining the numbers and sources of biological aerosols acting as ice nuclei, Atmospheric Environment, 44(15), 1944-1945.

DeMott, P. J., K. Sassen, M. R. Poellot, D. Baumgardner, D. C. Rogers, S. D. Brooks, A. J. Prenni, and S. M. Kreidenweis (2003a), African dust aerosols as atmospheric ice nuclei, Geophys Res Lett, 30(14).

DeMott, P. J., D. J. Cziczo, A. J. Prenni, D. M. Murphy, S. M. Kreidenweis, D. S. Thomson, R. Borys, and D. C. Rogers (2003b), Measurements of the concentration and composition of nuclei for cirrus formation, Proc. Natl. Acad. Sci. USA, 100(25), 14655-14660.

DeMott, P. J., A. J. Prenni, X. Liu, S. M. Kreidenweis, M. D. Petters, C. H. Twohy, M. S. Richardson, T. Eidhammer, and D. C. Rogers (2010), Predicting global atmospheric ice nuclei distributions and their impacts on climate, Proc. Natl. Acad. Sci. U.S.A., 107(25), 11217-11222.

179

Deshler, T., and D. W. Reynolds (1990), Physical Response of Winter Orographic Clouds over the Sierra-Nevada to Airborne Seeding Using Dry Ice or Silver-Iodide, J. Appl. Meteorol., 29(4), 288-330.

Despres, V. R., J. A. Huffman, S. M. Burrows, C. Hoose, A. S. Safatov, G. Buryak, J. Frohlich-Nowoisky, W. Elbert, M. O. Andreae, U. Poschl, and R. Jaenicke (2012), Primary biological aerosol particles in the atmosphere: a review, Tellus B, 64.

Doherty, O. M., N. Riemer, and S. Hameed (2008), Saharan mineral dust transport into the Caribbean: Observed atmospheric controls and trends, J Geophys Res-Atmos, 113(D7).

Dore, A. J., M. Sobik, and K. Migala (1999), Patterns of precipitation and pollutant deposition in the western Sudete mountains, Poland, Atmos. Environ., 33(20), 3301-3312.

Draxler, R. R., and G. D. Rolph (2013), HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website (http://www.arl.noaa.gov/HYSPLIT.php). NOAA Air Resources Laboratory, College Park, MD.

Duncan, B. N., R. V. Martin, A. C. Staudt, R. Yevich, and J. A. Logan (2003), Interannual and seasonal variability of biomass burning emissions constrained by satellite observations, J. Geophys. Res., 108(D2), 4100.

Dunlea, E. J., DeCarlo, P. F., Aiken, A. C., Kimmel, J. R., Peltier, R. E., Weber, R. J., Tomlinson, J., Collins, D. R., Shinozuka, Y., McNaughton, C. S., Howell, S. G., Clarke, A. D., Emmons, L. K., Apel, E. C., Pfister, G. G., van Donkelaar, A., Martin, R. V., Millet, D. B., Heald, C. L., and Jimenez, J. L. (2009), Evolution of Asian aerosols during transpacific transport in INTEX-B, Atmos. Chem. Phys., 9, 7257-7287.

Ebert, M., A. Worringen, N. Benker, S. Mertes, E. Weingartner, and S. Weinbruch (2011), Chemical composition and mixing-state of ice residuals sampled within mixed phase clouds, Atmos Chem Phys, 11(6), 2805-2816.

Engelbrecht, J. P., and R. K. M. Jayanty (2013), Assessing sources of airborne mineral dust and other aerosols, in Iraq, Aeolian Research, 9(0), 153-160.

Fergenson, D. P., M. E. Pitesky, H. J. Tobias, P. T. Steele, G. A. Czerwieniec, S. C. Russell, C. B. Lebrilla, J. M. Horn, K. R. Coffee, A. Srivastava, S. P. Pillai, M.-T. P. Shih, H. L. Hall, A. J. Ramponi, J. T. Chang, R. G. Langlois, P. L. Estacio, R. T. Hadley, M. Frank, and E. E. Gard (2003), Reagentless Detection and Classification of Individual Bioaerosol Particles in Seconds, Anal. Chem., 76(2), 373-378.

180

Field, P. R., R. Wood, P. R. A. Brown, P. H. Kaye, E. Hirst, R. Greenaway, and J. A. Smith (2003), Ice Particle Interarrival Times Measured with a Fast FSSP, J Atmos Ocean Tech, 20(2), 249-261.

Fowler, D., I. D. Leith, J. Binnie, A. Crossley, D. W. F. Inglis, T. W. Choularton, M. Gay, J. W. S. Longhurst, and D. E. Conland (1995), Orographic enhancement of wet deposition in the United Kingdom: Continuous monitoring, Water Air. Soil Poll., 85(4), 2107-2112.

Froyd, K. D., D. M. Murphy, P. Lawson, D. Baumgardner, and R. L. Herman (2010), Aerosols that form subvisible cirrus at the tropical tropopause, Atmos Chem Phys, 10(1), 209-218.

Gard, E., J. E. Mayer, B. D. Morrical, T. Dienes, D. P. Fergenson, and K. A. Prather (1997), Real-Time Analysis of Individual Atmospheric Aerosol Particles: Design and Performance of a Portable ATOFMS, Anal Chem, 69(20), 4083-4091.

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

Gardiner, B. A., and J. Hallett (1985), Degradation of In-Cloud Forward Scattering Spectrometer Probe Measurements in the Presence of Ice Particles, J Atmos Ocean Tech, 2(2), 171-180.

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

Givati, A., and D. Rosenfeld (2004), Quantifying precipitation suppression due to air pollution, J Appl Meteorol, 43(7), 1038-1056.

Gross, A., T. Goren, J. Cardoso, C. Pio, O. Tirosh, Y. Erel, M. C. Todd, D. Rosenfeld, T. Weiner, D. Custódio, and A. Angert (2013), Magmatic hotspots drive up Phosphorus in Saharan dust that nourishes the biosphere, Nat Geosci, In Preparation.

Guan, B., N. P. Molotch, D. E. Waliser, E. J. Fetzer, and P. J. Neiman (2010), Extreme snowfall events linked to atmospheric rivers and surface air temperature via satellite measurements, Geophys. Res. Lett., 37.

Guasco, T. L., L. A. Cuadra-Rodriguez, B. E. Pedler, A. P. Ault, D. B. Collins, D. Zhao, M. J. Kim, M. J. Ruppel, V. H. Grassian, F. Azam, T. H. Bertram, and K. A. Prather (2013), Identification of transition metals as unique tracers for single-

181

particle identification of heterotrophic bacterial-associated particles in sea spray aerosol, Environ Sci Technol, Submitted.

Guazzotti, S. A., K. R. Coffee, and K. A. Prather (2001a), Continuous measurements of size-resolved particle chemistry during INDOEX-Intensive Field Phase 99, J. Geophys. Res., 106(D22), 28607-28627.

Guazzotti, S. A., J. R. Whiteaker, D. Suess, K. R. Coffee, and K. A. Prather (2001b), Real-time measurements of the chemical composition of size-resolved particles during a Santa Ana wind episode, California USA, Atmos Environ, 35(19), 3229- 3240.

Guggenberger, G., B. T. Christensen, and W. Zech (1994), Land-Use Effects on the Composition of Organic-Matter in Particle-Size Separates of Soil .1. Lignin and Carbohydrate Signature, Eur J Soil Sci, 45(4), 449-458.

Guieu, C., M. D. Loye-Pilot, C. Ridame, and C. Thomas (2002), Chemical characterization of the Saharan dust end-member: Some biogeochemical implications for the western Mediterranean Sea, J Geophys Res-Atmos, 107(D15).

Gustafsson, O., M. Krusa, Z. Zencak, R. J. Sheesley, L. Granat, E. Engstrom, P. S. Praveen, P. S. P. Rao, C. Leck, and H. Rodhe (2009), Brown Clouds over South Asia: Biomass or Fossil Fuel Combustion?, Science, 323(5913), 495-498.

Hadley, O. L., V. Ramanathan, G. R. Carmichael, Y. Tang, C. E. Corrigan, G. C. Roberts, and G. S. Mauger (2007), Trans-Pacific transport of black carbon and fine aerosols (D < 2.5 μm) into North America, J. Geophys. Res., 112(D5), D05309.

Hallar, A. G., G. Chirokova, I. McCubbin, T. H. Painter, C. Wiedinmyer, and C. Dodson (2011), Atmospheric bioaerosols transported via dust storms in the western United States, Geophys. Res. Lett., 38.

Hallett, J., and S. C. Mossop (1974), Production of Secondary Ice Particles during Riming Process, Nature, 249(5452), 26-28.

Hannachi, A., I. T. Jolliffe, and D. B. Stephenson (2007), Empirical orthogonal functions and related techniques in atmospheric science: A review, Int J Climatol, 27(9), 1119-1152.

Hansen, J., M. Sato, and R. Ruedy (1997), Radiative forcing and climate response, J. Geophys. Res., 102(D6), 6831-6864.

Hara, K., and D. Z. Zhang (2012), Bacterial abundance and viability in long-range transported dust, Atmos. Environ., 47, 20-25.

182

Hegg, D. A., and P. V. Hobbs (1981), Cloud Water Chemistry and the Production of Sulfates in Clouds, Atmos Environ, 15(9), 1597-1604.

Heymsfield, A. J., and L. M. Miloshevich (1993), Homogeneous Ice Nucleation and Supercooled Liquid Water in Orographic Wave Clouds, J Atmos Sci, 50(15), 2335-2353.

Hobbs, P. V., and A. L. Rangno (1985), Ice Particle Concentrations in Clouds, J Atmos Sci, 42(23), 2523-2549.

Hogan, T. F., and T. E. Rosmond (1991), The Description of the Navy Operational Global Atmospheric Prediction Systems Spectral Forecast Model, Mon. Weather Rev., 119(8), 1786-1815.

Hogan, T. F., and L. R. Brody (1993), Sensitivity Studies of the Navy Global Forecast Model Parameterizations and Evaluation of Improvements to Nogaps, Mon. Weather Rev., 121(8), 2373-2395.

Holecek, J. C., M. T. Spencer, and K. A. Prather (2007), Analysis of rainwater samples: Comparison of single particle residues with ambient particle chemistry from the northeast Pacific and Indian oceans, J. Geophys. Res., 112(D22).

Hoose, C., and O. Mohler (2012), Heterogeneous ice nucleation on atmospheric aerosols: a review of results from laboratory experiments, Atmos Chem Phys, 12(20), 9817- 9854.

Hoose, C., J. E. Kristjansson, and S. M. Burrows (2010), How important is biological ice nucleation in clouds on a global scale?, Environmental Research Letters, 5(2).

Hosler, C. L., D. C. Jensen, and L. Goldshlak (1957), On the Aggregation of Ice Crystals to Form Snow, J Meteorol, 14(5), 415-420.

Houze, R. A. (1993), Cloud dynamics, Academic Press, San Diego, USA.

Hsu, S.-C., C.-A. Huh, C.-Y. Lin, W.-N. Chen, N. M. Mahowald, S.-C. Liu, C. C. K. Chou, M.-C. Liang, C.-J. Tsai, F.-J. Lin, J.-P. Chen, and Y.-T. Huang (2012), Dust transport from non-East Asian sources to the North Pacific, Geophys. Res. Lett., 39(12), L12804.

Huang, Y., L. Li, J. Li, X. Wang, H. Chen, J. Chen, X. Yang, D. S. Gross, H. Wang, L. Qiao, and C. Chen (2013), A case study of the highly time-resolved evolution of aerosol chemical and optical properties in urban Shanghai, China, Atmos Chem Phys, 13(8), 3931-3944.

Husar, R. B., D. M. Tratt, B. A. Schichtel, S. R. Falke, F. Li, D. Jaffe, S. Gasso, T. Gill, N. S. Laulainen, F. Lu, M. C. Reheis, Y. Chun, D. Westphal, B. N. Holben, C. Gueymard, I. McKendry, N. Kuring, G. C. Feldman, C. McClain, R. J. Frouin, J.

183

Merrill, D. DuBois, F. Vignola, T. Murayama, S. Nickovic, W. E. Wilson, K. Sassen, N. Sugimoto, and W. C. Malm (2001), Asian dust events of April 1998, J. Geophys. Res., 106(D16), 18317-18330.

Iannone, R., D. I. Chernoff, A. Pringle, S. T. Martin, and A. K. Bertram (2011), The ice nucleation ability of one of the most abundant types of fungal spores found in the atmosphere, Atmos. Chem. Phys., 11(3), 1191-1201.

IPCC (2007), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. , Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996 pp.

Isono, K., M. Komabayasi, and A. Ono (1959), The nature and origin of ice nuclei in the atmosphere, J. Meteorol. Soc. Japan, 37, 211-233.

Jacobson, M. Z. (2012), Investigating cloud absorption effects: Global absorption properties of black carbon, tar balls, and soil dust in clouds and aerosols, J Geophys Res-Atmos, 117.

Jaffe, D., H. Price, D. Parrish, A. Goldstein, and J. Harris (2003), Increasing background ozone during spring on the west coast of North America, Geophys Res Lett, 30(12).

Jaffe, D., T. Anderson, D. Covert, R. Kotchenruther, B. Trost, J. Danielson, W. Simpson, T. Berntsen, S. Karlsdottir, D. Blake, J. Harris, G. Carmichael, and I. Uno (1999), Transport of Asian air pollution to North America, Geophys Res Lett, 26(6), 711- 714.

Kaufman, Y. J., C. Ichoku, L. Giglio, S. Korontzi, D. A. Chu, W. M. Hao, R. R. Li, and C. O. Justice (2003), Fire and smoke observed from the Earth Observing System MODIS instrument - products, validation, and operational use, Int J Remote Sens, 24(8), 1765-1781.

Kellogg, C. A., and D. W. Griffin (2006), Aerobiology and the global transport of desert dust, Trends Ecol Evol, 21(11), 638-644.

Khain, A., D. Rosenfeld, and A. Pokrovsky (2005), Aerosol impact on the dynamics and microphysics of deep convective clouds, Q J Roy Meteor Soc, 131(611), 2639- 2663.

Khain, A. P., N. BenMoshe, and A. Pokrovsky (2008), Factors Determining the Impact of Aerosols on Surface Precipitation from Clouds: An Attempt at Classification, J Atmos Sci, 65(6), 1721-1748.

184

Kingsmill, D. E., P. J. Neiman, B. J. Moore, M. Hughes, S. E. Yuter, and F. M. Ralph (2012), Kinematic and thermodynamic structures of Sierra barrier jets and overrunning atmospheric rivers during a land-falling winter storm in Northern California., Mon. Weather Rev., In Press.

Kirchstetter, T. W., T. Novakov, and P. V. Hobbs (2004), Evidence that the spectral dependence of light absorption by aerosols is affected by organic carbon, J Geophys Res-Atmos, 109(D21).

hler, . (1921), ur ondensation des Wasserdampfes in der Atmosph re, Geofysiske publikationer, 2, 3-15.

Korolev, A. (2007), Limitations of the wegener-bergeron-findeisen mechanism in the evolution of mixed-phase clouds, J. Atmos. Sci., 64(9), 3372-3375.

Korolev, A., and G. A. Isaac (2005), Shattering during Sampling by OAPs and HVPS. Part I: Snow Particles, J Atmos Ocean Tech, 22(5), 528-542.

Lappin-Scott, H. M., and C. Bass (2001), Biofilm formation: Attachment, growth, and detachment of microbes from surfaces, Am J Infect Control, 29(4), 250-251.

Lappin‐Scott, H. M., and J. W. Costerton (1989), Bacterial biofilms and surface fouling, Biofouling, 1(4), 323-342.

Lau, K. M., and H. T. Wu (2003), Warm rain processes over tropical oceans and climate implications, Geophys Res Lett, 30(24).

Li, W. J., Y. Wang, J. L. Collett, J. M. Chen, X. Y. Zhang, Z. F. Wang, and W. X. Wang (2013), Microscopic Evaluation of Trace Metals in Cloud Droplets in an Acid Precipitation Region (vol 47, pg 4172, 2013), Environ Sci Technol, 47(11), 6067- 6067.

Liang, Q., L. Jaeglé, D. A. Jaffe, P. Weiss-Penzias, A. Heckman, and J. A. Snow (2004), Long-range transport of Asian pollution to the northeast Pacific: Seasonal variations and transport pathways of carbon monoxide, J. Geophys. Res., 109(D23), D23S07.

Lim, J. M., J. H. Lee, J. H. Moon, Y. S. Chung, and K. H. Kim (2010), Airborne PM10 and metals from multifarious sources in an industrial complex area, Atmospheric Research, 96(1), 53-64.

Liu, H., D. J. Jacob, I. Bey, R. M. Yantosca, B. N. Duncan, and G. W. Sachse (2003), Transport pathways for Asian pollution outflow over the Pacific: Interannual and seasonal variations, J. Geophys. Res., 108(D20), 8786.

185

Liu, P., P. J. Ziemann, D. B. Kittelson, and P. H. Mcmurry (1995a), Generating Particle Beams of Controlled Dimensions and Divergence .1. Theory of Particle Motion in Aerodynamic Lenses and Nozzle Expansions, Aerosol Sci Tech, 22(3), 293-313.

Liu, P., P. J. Ziemann, D. B. Kittelson, and P. H. Mcmurry (1995b), Generating Particle Beams of Controlled Dimensions and Divergence .2. Experimental Evaluation of Particle Motion in Aerodynamic Lenses and Nozzle Expansions, Aerosol Sci Tech, 22(3), 314-324.

Lohmann, U. (2002), A glaciation indirect aerosol effect caused by soot aerosols, Geophys Res Lett, 29(4).

Lohmann, U., and J. Feichter (2005), Global indirect aerosol effects: a review, Atmos Chem Phys, 5, 715-737.

Maki, L. R., E. L. Galyan, Changchi.Mm, and D. R. Caldwell (1974), Ice Nucleation Induced by Pseudomonas-Syringae, Appl Microbiol, 28(3), 456-459.

McKendry, I. G., K. B. Strawbridge, N. T. O'Neill, A. M. Macdonald, P. S. K. Liu, W. R. Leaitch, K. G. Anlauf, L. Jaegle, T. D. Fairlie, and D. L. Westphal (2007), Trans- Pacific transport of Saharan dust to western North America: A case study, J. Geophys. Res., 112(D1).

Meyers, M. P., P. J. Demott, and W. R. Cotton (1992), New Primary Ice-Nucleation Parameterizations in an Explicit Cloud Model, J. Appl. Meteorol., 31(7), 708-721.

Minnis, P., S. Sun-Mack, D. F. Young, P. W. Heck, D. P. Garber, Y. Chen, D. A. Spangenberg, R. F. Arduini, Q. Z. Trepte, W. L. S. Jr., J. K. Ayers, S. C. Gibson, W. F. Miller, V. Chakrapani, Y. Takano, K.-N. Liou, Y. Xie, and P. Yang (2011), CERES Edition-2 cloud property retrievals using TRMM VIRS and Terra and Aqua MODIS data, Part I: Algorithms, IEEE Trans. Geosci. Remote Sens, 49(11), doi: 10.1109/TGRS.2011.2144601.

Moffet, R. C., and K. A. Prather (2009), In-situ measurements of the mixing state and optical properties of soot with implications for radiative forcing estimates, Proceedings of the National Academy of Sciences, 106(29), 11872-11877.

Moffet, R. C., Y. Desyaterik, R. J. Hopkins, A. V. Tivanski, M. K. Gilles, Y. Wang, V. Shutthanandan, L. T. Molina, R. G. Abraham, K. S. Johnson, V. Mugica, M. J. Molina, A. Laskin, and K. A. Prather (2008), Characterization of aerosols containing Zn, Pb, and Cl from an industrial region of Mexico City, Environ Sci Technol, 42(19), 7091-7097.

Morris, C. E., D. G. Georgakopoulos, and D. C. Sands (2004a), Ice nucleation active bacteria and their potential role in precipitation, J Phys Iv, 121, 87-103.

186

Morris, C. E., D. G. Georgakopoulos, and D. C. Sands (2004b), Ice nucleation active bacteria and their potential role in precipitation, Journal De Physique Iv, 121, 87- 103.

Mossop, S. C. (1968), Comparison between concentrations of ice crystals in cloud and concentrations of ice nuclei, J. Rech. Atmos., 3, 119–124.

Mossop, S. C., A. Ono, and E. R. Wishart (1970), Ice particles in maritime clouds near Tasmania, Q J Roy Meteor Soc, 96(409), 487-508.

Murphy, D. M., and D. S. Thomson (1997), Chemical composition of single aerosol particles at Idaho Hill: Positive ion measurements, J Geophys Res-Atmos, 102(D5), 6341-6352.

Murphy, D. M., D. J. Cziczo, P. K. Hudson, D. S. Thomson, J. C. Wilson, T. Kojima, and P. R. Buseck (2004), Particle generation and resuspension in aircraft inlets when flying in clouds, Aerosol Sci Tech, 38(4), 400-408.

Murray, B. J., S. L. Broadley, T. W. Wilson, S. J. Bull, R. H. Wills, H. K. Christenson, and E. J. Murray (2010), Kinetics of the homogeneous freezing of water, Phys Chem Chem Phys, 12(35), 10380-10387.

Neiman, P. J., E. M. Sukovich, F. M. Ralph, and M. Hughes (2010), A Seven-Year Wind Profiler-Based Climatology of the Windward Barrier Jet along California's Northern Sierra Nevada, Mon. Weather Rev., 138(4), 1206-1233.

Niemand, M., O. Mohler, B. Vogel, H. Vogel, C. Hoose, P. Connolly, H. Klein, H. Bingemer, P. DeMott, J. Skrotzki, and T. Leisner (2012), A Particle-Surface- Area-Based Parameterization of Immersion Freezing on Desert Dust Particles, J. Atmos. Sci., 69(10), 3077-3092.

Noone, K. J., J. A. Ogren, J. Heintzenberg, R. J. Charlson, and D. S. Covert (1988), Design and Calibration of a Counterflow Virtual Impactor for Sampling of Atmospheric Fog and Cloud Droplets, Aerosol Sci Tech, 8(3), 235-244.

O'Sullivan, D., B. J. Murray, T. L. Malkin, T. Whale, N. S. Umo, J. D. Atkinson, H. C. Price, K. J. Baustian, J. Browse, and M. E. Webb (2013), Ice nucleation by soil dusts: relative importance of mineral dust and biogenic components, Atmospheric Chemistry and Physics Discussion, 13(8), 20275-20317.

Ogren, J. A. (2010), Comment on “Calibration and Intercomparison of Filter-Based Measurements of Visible Light Absorption by Aerosols”, Aerosol Sci Tech, 44(8), 589-591.

Ogren, J. A., J. Heintzenberg, and R. J. Charlson (1985), Insitu Sampling of Clouds with a Droplet to Aerosol Converter, Geophys. Res. Lett., 12(3), 121-124.

187

Pacyna, J. M., and E. G. Pacyna (2001), An assessment of global and regional emissions of trace metals to the atmosphere from anthropogenic sources worldwide, Environmental Reviews, 9(4), 269-298.

Park, K., and H. D. Dam (2010), Characterization of metal aerosols in PM10 from urban, industrial, and Asian Dust sources, Environ Monit Assess, 160(1-4), 289-300.

Pena, J. A., J. M. Norman, and D. W. Thomson (1977), Isokinetic Sampler for Continuous Airborne Aerosol Measurements, Japca J Air Waste Ma, 27(4), 337- 341.

Petit, R. H., M. Legrand, I. Jankowiak, J. Molinie, C. A. de Beauville, G. Marion, and J. L. Mansot (2005), Transport of Saharan dust over the Caribbean islands: study of an event, J Geophys Res-Atmos, 110(D18).

Pinsky, M., A. Khain, D. Rosenfeld, and A. Pokrovsky (1998), Comparison of collision velocity differences of drops and graupel particles in a very turbulent cloud, Atmospheric Research, 49(2), 99-113.

Pinsky, M. B., and A. P. Khain (2002), Effects of in-cloud nucleation and turbulence on droplet spectrum formation in cumulus clouds, Q J Roy Meteor Soc, 128(580), 501-533.

Podgorny, I. A., and V. Ramanathan (2001), A modeling study of the direct effect of aerosols over the tropical Indian Ocean, J Geophys Res-Atmos, 106(D20), 24097- 24105.

Prather, K. A., T. H. Bertram, V. H. Grassian, G. B. Deane, M. D. Stokes, P. J. DeMott, L. I. Aluwihare, B. P. Palenik, F. Azam, J. H. Seinfeld, R. C. Moffet, M. J. Molina, C. D. Cappa, F. M. Geiger, G. C. Roberts, L. M. Russell, A. P. Ault, J. Baltrusaitis, D. B. Collins, C. E. Corrigan, L. A. Cuadra-Rodriguez, C. J. Ebben, S. D. Forestieri, T. L. Guasco, S. P. Hersey, M. J. Kim, W. F. Lambert, R. L. Modini, W. Mui, B. E. Pedler, M. J. Ruppel, O. S. Ryder, N. G. Schoepp, R. C. Sullivan, and D. F. Zhao (2013), Bringing the ocean into the laboratory to probe the chemical complexity of sea spray aerosol, P Natl Acad Sci USA, 110(19), 7550-7555.

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

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

188

Pratt, K. A., J. E. Mayer, J. C. Holecek, R. C. Moffet, R. O. Sanchez, T. P. Rebotier, H. Furutani, M. Gonin, K. Fuhrer, Y. X. Su, S. Guazzotti, and K. A. Prather (2009b), Development and Characterization of an Aircraft Aerosol Time-of-Flight Mass Spectrometer, Anal Chem, 81(5), 1792-1800.

Prenni, A. J., M. D. Petters, S. M. Kreidenweis, C. L. Heald, S. T. Martin, P. Artaxo, R. M. Garland, A. G. Wollny, and U. Poschl (2009), Relative roles of biogenic emissions and Saharan dust as ice nuclei in the Amazon basin, Nature Geoscience, 2(6), 401-404.

Prospero, J. M., R. A. Glaccum, and R. T. Nees (1981), Atmospheric Transport of Soil Dust from Africa to South-America, Nature, 289(5798), 570-572.

Prospero, J. M., P. Ginoux, O. Torres, S. E. Nicholson, and T. E. Gill (2002), Environmental characterization of global sources of atmospheric soil dust identified with the Nimbus 7 Total Ozone Mapping Spectrometer (TOMS) absorbing aerosol product, Rev Geophys, 40(1).

Pruppacher, H. R., and J. D. Klett (2010), Microphysics of clouds and precipitation, 954 p. pp., Springer, Dordrecht ; New York.

Qin, X., and K. A. Prather (2006), Impact of biomass emissions on particle chemistry during the California Regional Particulate Air Quality Study, International Journal of Mass Spectrometry, 258(1–3), 142-150.

Ralph, F. M., P. J. Neiman, and G. A. Wick (2004), Satellite and CALJET aircraft observations of atmospheric rivers over the eastern north pacific ocean during the winter of 1997/98, Mon. Weather Rev., 132(7), 1721-1745.

Ralph, F. M., P. J. Neiman, and R. Rotunno (2005), Dropsonde observations in low-level jets over the northeastern Pacific Ocean from CALJET-1998 and PACJET-2001: Mean vertical-profile and atmospheric-river characteristics, Mon. Weather Rev., 133(4), 889-910.

Ralph, F. M., T. Colman, P. J. Neiman, R. Zamora, and M. D. Dettinger (2012), Observed impacts of duration and seasonality of atmospheric-river landfalls on soil moisture and runoff in coastal northern California, J. Hydrometeor.(doi: http://dx.doi.org/10.1175/JHM-D-12-076.1), in press.

Ralph, F. M., P. J. Neiman, G. A. Wick, S. I. Gutman, M. D. Dettinger, D. R. Cayan, and A. B. White (2006), Flooding on California's Russian River: Role of atmospheric rivers, Geophys Res Lett, 33(13).

Ramanathan, V., and G. Carmichael (2008), Global and regional climate changes due to black carbon, Nature Geosci, 1(4), 221-227.

189

Reinking, R. F., J. B. Snider, and J. L. Coen (2000), Influences of Storm-Embedded Orographic Gravity Waves on Cloud Liquid Water and Precipitation, J. Appl. Meteorol., 39(6), 733-759.

Reynolds, D. W., and A. S. Dennis (1986), A Review of the Sierra Cooperative Pilot Project, Bull. Am. Meteorol. Soc., 67(5), 513-523.

Rogers, C. F., J. G. Watson, and C. V. Mathai (1989), Design and Testing of a New Size Classifying Isokinetic Sequential Aerosol Sampler, Japca J Air Waste Ma, 39(12), 1569-1576.

Rogers, D. C., P. J. DeMott, S. M. Kreidenweis, and Y. L. Chen (2001), A continuous- flow diffusion chamber for airborne measurements of ice nuclei, Journal of Atmospheric and Ocean Technololgy, 18(5), 725-741.

Rolph, G. D. (2013), Real-time Environmental Applications and Display sYstem (READY) Website (http://www.ready.noaa.gov). NOAA Air Resources Laboratory, College Park, MD.

Roman, M. R., H. W. Ducklow, J. A. Fuhrman, C. Garside, P. M. Glibert, T. C. Malone, and G. B. Mcmanus (1988), Production, Consumption and Nutrient Cycling in a Laboratory Mesocosm, Mar Ecol Prog Ser, 42(1), 39-52.

Rosenfeld, D. (2000), Suppression of rain and snow by urban and industrial air pollution, Science, 287(5459), 1793-1796.

Rosenfeld, D., and W. L. Woodley (2000), Deep convective clouds with sustained supercooled liquid water down to-37.5 degrees C, Nature, 405(6785), 440-442.

Rosenfeld, D., and A. Givati (2006), Evidence of Orographic Precipitation Suppression by Air Pollution–Induced Aerosols in the Western United States, Journal of Applied Meteorology and Climatology, 45(7), 893-911.

Rosenfeld, D., R. Lahav, A. Khain, and M. Pinsky (2002), The role of sea spray in cleansing air pollution over ocean via cloud processes, Science, 297(5587), 1667- 1670.

Rosenfeld, D., W. L. Woodley, D. Axisa, E. Freud, J. G. Hudson, and A. Givati (2008), Aircraft measurements of the impacts of pollution aerosols on clouds and precipitation over the Sierra Nevada, J. Geophys. Res., 113(D15), D15203.

Rosenfeld, D., J. Dai, X. Yu, Z. Y. Yao, X. H. Xu, X. Yang, and C. L. Du (2007), Inverse relations between amounts of air pollution and orographic precipitation, Science, 315(5817), 1396-1398.

Rosenfeld, D., X. Yu, G. H. Liu, X. H. Xu, Y. N. Zhu, Z. G. Yue, J. Dai, Z. P. Dong, Y. Dong, and Y. Peng (2011), Glaciation temperatures of convective clouds

190

ingesting desert dust, air pollution and smoke from forest fires, Geophys. Res. Lett., 38.

Rosenfeld, D., R. Chemke, P. DeMott, R. C. Sullivan, R. Rasmussen, F. McDonough, J. Comstock, B. Schmid, J. Tomlinson, H. Jonsson, K. Suski, A. Cazorla, and K. Prather (2013), The common occurrence of highly supercooled drizzle and rain near the coastal regions of the western United States, Journal of Geophysical Research: Atmospheres, 118(17), 9819-9833.

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

Saha, R., N. Saha, R. S. Donofrio, and L. L. Bestervelt (2013), Microbial siderophores: a mini review, Journal of Basic Microbiology, 53(4), 303-317.

Saleeby, S. M., W. R. Cotton, D. Lowenthal, R. D. Borys, and M. A. Wetzel (2009), Influence of Cloud Condensation Nuclei on Orographic Snowfall, J Appl Meteorol Clim, 48(5), 903-922.

Sassen, K., K. N. Liou, S. Kinne, and M. Griffin (1985), Highly Supercooled Cirrus Cloud Water - Confirmation and Climatic Implications, Science, 227(4685), 411- 413.

Seinfeld, J. H., and S. N. Pandis (2006), Atmospheric chemistry and physics : from air pollution to climate change, 2nd ed., xxviii, 1203 p. pp., J. Wiley, Hoboken, N.J.

Shields, L. G., D. T. Suess, and K. A. Prather (2007), Determination of single particle mass spectral signatures from heavy-duty diesel vehicle emissions for PM2.5 source apportionment, Atmos Environ, 41(18), 3841-3852.

Silva, P. J., R. A. Carlin, and K. A. Prather (2000), Single particle analysis of suspended soil dust from Southern California, Atmos Environ, 34(11), 1811-1820.

Silva, P. J., D. Y. Liu, C. A. Noble, and K. A. Prather (1999), Size and chemical characterization of individual particles resulting from biomass burning of local Southern California species, Environ Sci Technol, 33(18), 3068-3076.

Singer, A., S. Dultz, and E. Argaman (2004), Properties of the non-soluble fractions of suspended dust over the Dead Sea, Atmos Environ, 38(12), 1745-1753.

Sodeman, D. A., S. M. Toner, and K. A. Prather (2005), Determination of Single Particle Mass Spectral Signatures from Light-Duty Vehicle Emissions, Environ Sci Technol, 39(12), 4569-4580.

Song, X. H., P. K. Hopke, D. P. Fergenson, and K. A. Prather (1999), Classification of single particles analyzed by ATOFMS using an artificial neural network, ART- 2A, Anal Chem, 71(4), 860-865.

191

Spencer, M. T., L. G. Shields, D. A. Sodeman, S. M. Toner, and K. A. Prather (2006), Comparison of oil and fuel particle chemical signatures with particle emissions from heavy and light duty vehicles, Atmos Environ, 40(27), 5224-5235.

Stith, J. L., C. H. Twohy, P. J. DeMott, D. Baumgardner, T. Campos, R. Gao, and J. Anderson (2011), Observations of ice nuclei and heterogeneous freezing in a Western Pacific extratropical storm, Atmos Chem Phys, 11(13), 6229-6243.

Stohl, A., C. Forster, A. Frank, P. Seibert, and G. Wotawa (2005), Technical note: The Lagrangian particle dispersion model FLEXPART version 6.2, Atmos Chem Phys, 5, 2461-2474.

Stokes, M. D., G. B. Deane, K. Prather, T. H. Bertram, M. J. Ruppel, O. S. Ryder, J. M. Brady, and D. Zhao (2013), A Marine Aerosol Reference Tank system as a breaking wave analogue for the production of foam and sea-spray aerosols, Atmos. Meas. Tech., 6(4), 1085-1094.

Streets, D. G., K. F. Yarber, J. H. Woo, and G. R. Carmichael (2003), Biomass burning in Asia: Annual and seasonal estimates and atmospheric emissions, Global Biogeochem. Cycles, 17(4), 1099.

Sullivan, R. C., S. A. Guazzotti, D. A. Sodeman, and K. A. Prather (2007), Direct observations of the atmospheric processing of Asian mineral dust, Atmos Chem Phys, 7, 1213-1236.

Sun, J. M., M. Y. Zhang, and T. S. Liu (2001), Spatial and temporal characteristics of dust storms in China and its surrounding regions, 1960-1999: Relations to source area and climate, J. Geophys. Res., 106(D10), 10325-10333.

Suski, K. J., C. H. Twohy, J. Anderson, D. Rosenfeld, A. J. Heymsfield, and K. A. Prather (2014a), Dust, Bioparticles, or Artifacts?: Unraveling the Sources of Metals in Cloud Residues, In preparation for submission to Aerosol Science and Technology.

Suski, K. J., D. Rosenfeld, J. R. Anderson, C. H. Twohy, A. J. Heymsfield, and K. A. Prather (2014b), Biological Species Dominate Ice Nucleation in Mixed Phase Marine Clouds Warmer than -15 ºC, In preparation for submission to Nature Geosciences.

Teng, H. Y., W. M. Washington, G. Branstator, G. A. Meehl, and J. F. Lamarque (2012), Potential impacts of Asian carbon aerosols on future US warming, Geophys Res Lett, 39.

Toner, S. M., D. A. Sodeman, and K. A. Prather (2006), Single Particle Characterization of Ultrafine and Accumulation Mode Particles from Heavy Duty Diesel Vehicles

192

Using Aerosol Time-of-Flight Mass Spectrometry, Environ Sci Technol, 40(12), 3912-3921.

Twohy, C. H. (2013), Personal Communication.

Twohy, C. H., and M. R. Poellot (2005), Chemical characteristics of ice residual nuclei in anvil cirrus clouds: evidence for homogeneous and heterogeneous ice formation, Atmos. Chem. Phys., 5(8), 2289-2297.

Twohy, C. H., J. W. Strapp, and M. Wendisch (2003), Performance of a counterflow virtual impactor in the NASA Icing Research Tunnel, J Atmos Ocean Tech, 20(6), 781-790.

Twomey, S. (1974), Pollution and the planetary albedo, Atmospheric Environment (1967), 8(12), 1251-1256.

Uno, I., K. Eguchi, K. Yumimoto, Z. Liu, Y. Hara, N. Sugimoto, A. Shimizu, and T. Takemura (2011), Large Asian dust layers continuously reached North America in April 2010, Atmos. Chem. Phys., 11(14), 7333-7341.

Uno, I., K. Eguchi, K. Yumimoto, T. Takemura, A. Shimizu, M. Uematsu, Z. Y. Liu, Z. F. Wang, Y. Hara, and N. Sugimoto (2009), Asian dust transported one full circuit around the globe, Nat. Geosci., 2(8), 557-560.

VanCuren, R. A., and T. A. Cahill (2002), Asian aerosols in North America: Frequency and concentration of fine dust, Journal of Geophysical Research: Atmospheres, 107(D24), 4804.

Wallace, G. T. (1982), The Association of Copper, Mercury and Lead with Surface- Active Organic-Matter in Coastal Sea-Water, Mar Chem, 11(4), 379-394.

Wendisch, M., and J.-L. Brenguier (2013), Airborne measurements for environmental research : methods and instruments, xxxii, 655 pages pp., Wiley-VCH, Weinheim.

White, A. B., J. R. Jordan, B. E. Martner, F. M. Ralph, and B. W. Bartram (2000), Extending the dynamic range of an S-band radar for cloud and precipitation studies, J. Atmos. Ocean Tech., 17(9), 1226-1234.

White, A. B., D. J. Gottas, E. T. Strem, F. M. Ralph, and P. J. Neiman (2002), An automated brightband height detection algorithm for use with Doppler radar spectral moments, J. Atmos. Ocean. Technol., 19(5), 687-697.

White, A. B., P. J. Neiman, F. M. Ralph, D. E. Kingsmill, and P. O. G. Persson (2003), Coastal orographic rainfall processes observed by radar during the California land-falling jets experiment, J. Hydrometeorol., 4(2), 264-282.

193

Wiacek, A., T. Peter, and U. Lohmann (2010), The potential influence of Asian and African mineral dust on ice, mixed-phase and liquid water clouds, Atmos. Chem. Phys., 10(18), 8649-8667.

Winker, D. M., J. Pelon, J. A. Coakley, S. A. Ackerman, R. J. Charlson, P. R. Colarco, P. Flamant, Q. Fu, R. M. Hoff, C. Kittaka, T. L. Kubar, H. Le Treut, M. P. McCormick, G. Megie, L. Poole, K. Powell, C. Trepte, M. A. Vaughan, and B. A. Wielicki (2010), THE CALIPSO MISSION A Global 3D View of Aerosols and Clouds, Bulletin of the American Meteorological Society, 91(9), 1211-1229.

Wollnik, H., and M. Przewloka (1990), Time-of-Flight Mass Spectrometers with Multiply Reflected Ion Trajectories, Int J Mass Spectrom, 96(3), 267-274.

Wuebbles, D. J., H. Lei, and J. Lin (2007), Intercontinental transport of aerosols and photochemical oxidants from Asia and its consequences, Environ Pollut, 150(1), 65-84.

Yankofsky, S. A., Z. Levin, T. Bertold, and N. Sandlerman (1981), Some Basic Characteristics of Bacterial Freezing Nuclei, J Appl Meteorol, 20(9), 1013-1019.

Yienger, J. J., M. Galanter, T. A. Holloway, M. J. Phadnis, S. K. Guttikunda, G. R. Carmichael, W. J. Moxim, and H. Levy, II (2000), The episodic nature of air pollution transport from Asia to North America, J. Geophys. Res., 105(D22), 26931-26945.

Yu, H. B., L. A. Remer, M. Chin, H. S. Bian, Q. Tan, T. L. Yuan, and Y. Zhang (2012), Aerosols from Overseas Rival Domestic Emissions over North America, Science, 337(6094), 566-569.

Zhu, Y., and R. E. Newell (1998), A proposed algorithm for moisture fluxes from atmospheric rivers, Mon Weather Rev, 126(3), 725-735.