«REMOTE CONTINENTAL AEROSOL CHARACTERISTICS IN THE ROCKY MOUNTAINS OF COLORADO AND WYOMING Submitted by Ezra JT Levin Department of Atmospheric ...»
REMOTE CONTINENTAL AEROSOL CHARACTERISTICS IN THE
ROCKY MOUNTAINS OF COLORADO AND WYOMING
Ezra JT Levin
Department of Atmospheric Science
In partial fulfillment of the requirements
For the Degree of Doctor of Philosophy
Colorado State University
Fort Collins, Colorado
Advisor: Sonia M. Kreidenweis Jeffrey L. Collett Jr.
Susan C. van den Heever Jay Ham
REMOTE CONTINENTAL AEROSOL CHARACTERISTICS IN THE
ROCKY MOUNTAINS OF COLORADO AND WYOMING
Relevant aerosol observations were obtained in several long-term field studies: the Rocky Mountain Atmospheric Nitrogen and Sulfur study (RoMANS, Colorado), the Grand Tetons ii Reactive Nitrogen Deposition Study (GrandTReNDS, Wyoming) and as part of the Bio-hydro- atmosphere interactions of Energy, Aerosols, Carbon, H2O, Organics & Nitrogen project (BEACHON, Colorado). Average number concentrations (0.04 Dp 20 μm) measured during the field studies ranged between 1000 – 2000 cm-3 during the summer months and decreased to 200 – 500 cm-3 during the winter. These seasonal changes in aerosol number concentrations were correlated with the frequency of events typical of new particle formation. Measured submicron organic mass fractions were between 70 – 90% during the summer months, when new particle formation events were most frequent, suggesting the importance of organic species in the nucleation or growth process, or both. Aerosol composition derived from hygroscopicity measurements indicate organic mass fractions of 50 – 60% for particles with diameters larger than 0.15 μm during the winter. The composition of smaller diameter particles appeared to be organic dominated year-round.
High organic mass fractions led to low values of aerosol hygroscopicity, described using the κ parameter. Over the entire year-long BEACHON study, κ had an average value of 0.16 ± 0.08, similar to values determined during biologically active periods in tropical and boreal forests, and lower than the commonly assumed value of κcontinental = 0.3. There was also an observed increase in κ with size, due to external mixing of the fine mode aerosol. Incorrect representations of κ or its size dependence led to erroneous values of calculated CCN concentrations, especially for supersaturation values less than 0.3%. At higher supersaturations, most of the measured variability in CCN concentrations was captured by changes in total measured aerosol number concentrations.
While data from the three measurement sites were generally well correlated, indicating similarities in seasonal cycles and in total number concentrations, there were some variations
to the effects of local emissions. The averaged data provide reasonable, observationally-based parameters for modeling of aerosol number size distributions and corresponding CCN concentrations.
Field observations clearly indicated the episodic influence of wildfire smoke on particle number concentrations and compositions. However, the semi-volatile nature of the organic carbon species emitted makes it difficult to predict how much of the emitted organic mass will remain in the condensed phase downwind. To better constrain the volatility of organic species in smoke, emissions from laboratory biomass combustion experiments were subjected to quantified dilution, resulting in reduction of aerosol mass concentrations over several orders of magnitude and a corresponding volatilization response of the organic particles that was fit to the commonlyapplied Volatility Basis Set. Organic emissions from all burns with initial organic aerosol concentrations greater than 1000 μg m-3 contained material with saturation concentration values ranging between 1 and 10,000 μg m-3, with most of the organic mass falling at the two extremes of this range. For most burns, a single distribution was able to capture the volatility behavior of the organic material, within experimental uncertainty, despite the considerable variability in fuel and fire characteristics, suggesting that a simplified two-product model of gas-aerosol partitioning may be adequate to describe the evolution of biomass burning organic aerosol in models.
Sonia Kreidenweis, my advisor, has been a constant source of education, motivation and inspiration throughout my time at CSU. I appreciate her willingness and availability to answer questions and give advice, her enthusiasm for her students and their work and her scientific insights. I have learned a great deal from working with Sonia, and am extremely thankful that she offered me the opportunity to do my graduate research with her.
I would also like to thank my committee members, Jeff Collett, Sue van den Heever and Jay Ham who all provided great advice and feedback as I completed this work.
As is the nature of all field measurement campaigns, many people were involved in the RoMANS, GrandTReNDS, BEACHON and FLAME studies. Thus, while there can only be one name listed as author for this dissertation, the data and results presented here represent a collaborative effort from a great number of participants, and many thanks and acknowledgments are due. During RoMANS and GrandTReNDS, Amy Sullivan, Derek Day, Florian Schwandner, Katie Benedict, Kip Carrico, Misha Schurman, Taehyoung Lee, Tony Prenni, Yi Li and Yury Desyaterik all operated instruments and performed data analysis. I would specifically like to thank Misha and Taehyoung for supplying the AMS data used in this work and Sam Atwood for running the HYSPLIT model for me. Sonia, Jeff, Kip, Tony, Bill Malm and Bret Schichtel all helped with planning and logistics for these studies.
During BEACHON, Christina McCluskey, Elvin Garcia, John Ortega, Tony and Yutaka Tobo all assisted with the CCN measurements. Brett Palm operated the AMS and performed initial quality control and data analysis. Paul DeMott, Jim Smith, Tony and Sonia were all involved in planning and organization for BEACHON.
would also like to thank Taehyoung for his willingness to answer all my AMS questions over the last few years, although that instrument remains a mystery to me. Kip Carrico built and operated the dilution system used during FLAME 3, Andy May ran the CMU thermodenuder, Dan WelshBon provided VOC data, Gavin McMeeking performed SP2 measurements and the AERODYNE gas instruments were operated by Tim Onasch. Kyle Wold assisted with the burns and ran the FSL gas instrumentation and Jose Jiminez, Wei Min Hoa, Allen Robinson, Jeff and Sonia were all involved in the planning and successful completion of this study.
Parts of this work were funded by the National Park Service, the National Science Foundation the Environmental Protection Agency and the Joint Fire Sciences Program.
Finally, I must thank Jenny for putting up with being married to a student for so long. I am truly fortunate to be married to such a patient and wonderful woman. Thank you!
For my parents, who taught me to read and write and, through great patience, perseverance and personal pain, instilled in their delinquent son a lifelong love of learning.
TABLE OF CONTENTS
LIST OF TABLES
LIST OF FIGURES
1.2. Aerosols, models and the need for measurements
1.3. Why the mountains of Colorado and Wyoming?
1.4. IMPROVE measurements
1.5. Dissertation layout
2. Aerosol size distributions and concentration
2.1. Introduction and background
2.2. Measurement site locations and wind characteristics
2.3.1. Aerosol concentration
126.96.36.199. Spatial variability
188.8.131.52. Inter-annual variability
184.108.40.206. Annual variability
2.3.2. Aerosol size distribution
2.3.3. Factors influencing aerosol size and concentration
220.127.116.11. Wind speed and direction
18.104.22.168. Meteorological events
22.214.171.124. High concentration events
2.3.4. Comparison between concentration and size
2.4. Summary and Conclusions
3. Aerosol hygroscopicity in the Rocky Mountains
3.1. Importance of aerosol hygroscopicity
3.2.1. Biogenic aerosol at the BEACHON site
3.2.2. Site location and characteristics
3.2.3. Instrumentation and measurement technique
Hygroscopicity parameter: κ
3.2.5. BEACHON Results
126.96.36.199. Seasonal cycle of kappa and CCN
188.8.131.52. Aerosol composition and hygroscopicity
184.108.40.206. Hygroscopicity and chemistry closure during BEACHON-RoMBAS............... 91 220.127.116.11. Seasonal cycle of aerosol composition from CCN measurements
18.104.22.168. Small Particle Events
3.2.6. Aerosol hygroscopicity during RoMANS 2010
3.2.7. Aerosol hygroscopicity during GrandTReNDS
3.2.8. Relevance of ground based measurements
3.3. Summary and conclusions
4. Volatility of primary organic emissions from biomass burning
4.2.1. Volatility basis set
4.2.2. Semi-volatile species and gas/particle partitioning
4.2.3. Determining organic aerosol volatility
4.3.1. Description of burns, sampling and dilution setup
4.3.2. Calculating emission factors
4.3.3. Dilution in the combustion lab
4.3.4. Dilution in the dilution barrels
4.3.5. Calculating partitioning
4.4. Results and discussion
4.4.1. Volatility of organic material
4.4.2. Sources of uncertainty in estimated OA volatility
4.4.3. Organic aerosol composition
ix 4.5. Summary and conclusions
5. Summary, conclusions and future work
5.1. Summary and conclusions
5.2. Future work
Appendix 1. Instrumentation
Appendix 2. AMS charge balance
Appendix 3. What effect does condensing organic material have on CCN concentrations?..... 220 Appendix 4. Laminar flow element calibrations
Appendix 5. Infiltration rates from gas measurements
x LIST OF TABLES
Table 2.1 Median (± 1 standard deviation) number and volume concentrations and fine mode volume fraction during RoMANS 2, RoMANS 2010 and GrandTReNDS as well as published data from RoMANS 1 [Levin et al.
, 2009]. Values are also shown for RoMANS 2 and 2010 during the time period corresponding to GrandTReNDS.
Table 2.2 Median (± 1 standard deviation) geometric means for the number and volume volume distributions during the three studies as well as published data from the RoMANS 1 [Levin et al.
, 2009], YACS [McMeeking et al., 2005b] and BRAVO [Hand et al., 2002] campaigns. Volume distribution parameters are divided into fine and coarse modes.
Table 2.3 Median (± 1 standard deviaition) geometric standard deviations for the number and volume volume distributions during the three studies as well as published data from the RoMANS 1 [Levin et al.
, 2009], YACS [McMeeking et al., 2005b] and BRAVO [Hand et al., 2002] campaigns. Volume distribution parameters are divided into fine and coarse modes...... 42 Table 3.1 Study average (± 1 standard deviation) supersaturation (s) values determined from ammonium sulfate calibrations at each CCNC ΔT setting and corresponding critical activation diameters (Dc) for a particle with κ = 0.2 as well as the range of κ, from the range in s at the Dc.
Table 3.2 Average (± 1 standard deviation) kappa values measured from the ground (3 m) and tower (25 m) inlets.
Ground based measurements are for May – June 2010 while tower measurements were made May – June 2011
Table 4.1 List of fuels burned as well as initial fuel weight, remaining ash, fuel moisture content, combustion efficiency (CE) and modified combustion efficiency (MCE).
.................. 142 Table 4.2 Fraction of total organic mass in each C* bin for all burns with initial OA concentration greater than 1000 µg m-3
Table 4.3 The best fit volatility distribution determined from thermal denuder (TD) data as well as an alternate fit to these data constrained by the dilution fit.
Table 4.4 Fraction of total organic mass and organic family mass in each C* bin.
xi LIST OF FIGURES
Figure 1.1 Maps of PM2.