Open Access Climate of the Past

Abstract. Aerosol hygroscopicity describes the ability of a particle to take up water and form a cloud droplet. Modeling studies have shown sensitivity of precipitation-producing cloud systems to the availability of aerosol particles capable of serving as cloud condensation nuclei (CCN), and hygroscopicity is a key parameter controlling the number of available CCN. Continental aerosol is typically assumed to have a representative hygroscopicity parameter, κ, of 0.3; however, in remote locations this value can be lower due to relatively large mass fractions of organic components. To further our understanding of aerosol properties in remote areas, we measured size-resolved aerosol chemical composition and hygroscopicity in a forested, mountainous site in Colorado during the six-week BEACHON-RoMBAS (Bio-hydro-atmosphere interactions of Energy, Aerosols, Carbon, H2O, Organics and Nitrogen–Rocky Mountain Biogenic Aerosol Study) campaign. This campaign followed a year-long measurement period at this site, and results from the intensive campaign shed light on the previously reported seasonal cycle in aerosol hygroscopicity. New particle formation events were observed routinely at this site and nucleation mode composition measurements indicated that the newly formed particles were predominantly organic. These events likely contribute to the dominance of organic species at smaller sizes, where aerosol organic mass fractions were between 70 and 90%. Corresponding aerosol hygroscopicity was observed to be in the range κ = 0.15–0.22, with hygroscopicity increasing with particle size. Aerosol chemical composition measured by an aerosol mass spectrometer and calculated from hygroscopicity measurements agreed very well during the intensive study, with an assumed value of κorg = 0.13 resulting in the best agreement.


Introduction
Whether or not a particle will act as a cloud condensation nucleus (CCN) at a given supersaturation depends on the particle's size and hygroscopicity, the latter referring to how readily the particle takes up water and grows to droplet size.For particles in the accumulation mode, where the majority of atmospheric particle number concentrations are typically found, aerosol hygroscopicity can have large impacts on the number of available CCN (Petters and Kreidenweis, 2007), which in turn can impact cloud droplet number and size.Higher droplet number concentrations are expected to cause changes in both cloud structure and lifetime, leading to enhancements in cloud reflectivity (Twomey, 1974), suppression of drizzle (Albrecht, 1989), changes in intensity and distribution of precipitation (Rosenfeld et al., 2008;Khain et al., 2005;Khain and Lynn, 2009;Storer et al., 2010;van den Heever et al., 2006;van den Heever and Cotton, 2007;van den Heever et al., 2011) and even changes in updraft and storm intensity (Lerach and Cotton, 2012).It is thus essential to accurately represent the number and characteristics of CCN in order to model these important processes.
Published by Copernicus Publications on behalf of the European Geosciences Union.

E. J. T. Levin et al.: Size-resolved aerosol composition and its link to hygroscopicity
Hygroscopicity can be described by the hygroscopicity parameter, κ (Petters and Kreidenweis, 2007).Typical atmospheric aerosol components can be divided into three broad categories based on hygroscopicity.Inorganic salts, such as ammonium sulfate, ammonium nitrate, and sodium chloride, are highly hygroscopic, with κ values ranging from ∼0.6 to 1.2 (Petters and Kreidenweis, 2007).Laboratory and field studies of secondary organic aerosol (SOA) species have found κ values around ∼0.1 (Prenni et al., 2007;Engelhart et al., 2008;Engelhart et al., 2011;Wex et al., 2009;Pierce et al., 2012) although individual species, when separated by polarity, show a larger range (Suda et al., 2012).Studies have shown that κ of more oxidized secondary organic species can be as high as 0.22 (Chang et al., 2010), although Frosch et al. (2011) found no relationship between κ and O / C ratio for α-pinene SOA aged in a smog chamber.Black carbon and dust are typically measured or assumed to have κ values of ∼0-0.05 (Koehler et al., 2009;Yamashita et al., 2011).As a representative average value for the fine mode aerosol that dominates CCN number concentrations, continental aerosol particles are often assumed to be adequately represented by a hygroscopicity parameter of ∼0.3 (Andreae and Rosenfeld, 2008;Pringle et al., 2010;Pöschl et al., 2009).However, in remote locations, far from sources of anthropogenic sulfate and nitrate, organic species may represent the dominant fraction of the aerosol, leading to lower κ values (Gunthe et al., 2009;Juranyi et al., 2011).Biogenic secondary organic aerosol (BSOA) is of particular interest in remote regions where oxidized volatile organic compounds, emitted from biogenic sources, play a key role in aerosol formation.These oxidized biogenic compounds can condense onto existing particles or contribute to new particle formation and growth (Kulmala et al., 2004a, b;Dusek et al., 2010), and have been shown to be a major source of organic aerosol mass (Kanakidou et al., 2005;Robinson et al., 2011).
Given the global significance of organic aerosol, particularly of biogenic origin, many laboratory studies have examined the hygroscopicity of various model BSOA species (Prenni et al., 2007;Petters et al., 2009b;King et al., 2010;Asa-Awuku et al., 2010;Massoli et al., 2010;Engelhart et al., 2011) while field studies have measured aerosol hygroscopicity in areas dominated by BSOA (Cerully et al., 2011;Gunthe et al., 2009;Paramonov et al., 2013;Levin et al., 2012;Sihto et al., 2011;Pierce et al., 2012).These studies have consistently found average κ values below 0.3, as well as variability in aerosol hygroscopicity with size, the latter feature likely due to the differing sources and gas-to-particle conversion processes of the organic and inorganic aerosol components.For example, Levin et al. (2012) measured aerosol hygroscopicity (14 nm < D p < 350 nm) from March 2010 to May 2011 at a forested, mountainous site in Colorado as part of the ongoing Bio-hydro-atmosphere interactions of Energy, Aerosols, Carbon, H 2 O, Organics and Nitrogen (BEA-CHON) project.Although the site is representative of a semiarid, mountain ecosystem, the average κ value determined for the measured aerosol was 0.16 ± 0.08, similar to values measured in highly biologically active areas such as Amazonian (Gunthe et al., 2009;Pöschl et al., 2010) and boreal forests (Sihto et al., 2011).Levin et al. (2012) hypothesized that the low hygroscopicity measured during the BEA-CHON annual cycle period was due to a large mass fraction of secondary organic species, likely of biogenic origin, in the aerosol.However, chemical composition measurements were not available during the year-long BEACHON study.In this work, we report on size-resolved CCN concentrations measured during the subsequent Rocky Mountain Biogenic Aerosol Study (BEACHON-RoMBAS), and we compare our results to concurrent measurements of size-resolved aerosol chemical composition.

Measurement site and instrumentation
Measurements were made during two phases of the BEA-CHON project: from March, 2010 to May, 2011 to study a full seasonal cycle, as previously reported (Levin et al., 2012), and during the BEACHON-RoMBAS intensive study which took place during July and August, 2011.The measurement site for both periods was located in Manitou Experimental Forest (lat.39.10 • N, long.105.10 • W, elevation 2370 m).Kim et al. (2010), Levin et al. (2012) and Ortega et al. (2014) describe the site as well as typical transport conditions for air masses arriving at the site.
During both the year-long and the BEACHON-RoMBAS campaigns, we made measurements of size-resolved CCN concentrations at five different supersaturation (s = RH -100 %) conditions and used these data to calculate aerosol hygroscopicity following Petters et al. (2009a).The specific details of the instrument setup and data collection and processing were reported in Levin et al. (2012).Briefly, we used a differential mobility analyzer (DMA; TSI 3081), operating at a 10:1.5 sheath to sample flow ratio, to select particles in 20 different size bins between ∼14 and 350 nm in diameter.The quasi-monodisperse sample was then split and sent to a condensation particle counter (CPC; TSI 3010), 1 LPM, and a cloud condensation nucleus counter (CCNC; Droplet Measurement Technologies), 0.5 LPM.This technique provides number size distributions of total aerosol and CCN concentrations, at a specified s, with a ∼15 min temporal resolution.We stepped s inside the CCNC through five different set points between 0.15 and 0.95 %.Supersaturation at each CCNC instrument setting was determined via ammonium sulfate calibrations using the thermodynamic properties of ammonium sulfate aqueous solutions as prescribed by the Aerosol Inorganics Model (Clegg et al., 1998).Calibrations at each s setting were performed about once a day.Activated fractions were calculated from the size-resolved total aerosol and CCN data.The critical activation diameter (d c ), the diameter at which the aerosol was able to take up water and grow to cloud droplet size at a given supersaturation, was defined as the diameter where the activated fraction = 0.5.Finally, we used the CCNC s setting and the d c to determine the best-fit aerosol hygroscopicity parameter, κ.
During BEACHON-RoMBAS, submicron aerosol composition was measured with a high-resolution time-of-flight aerosol mass spectrometer (AMS, Aerodyne Research Inc.) described by (Canagaratna et al., 2007;DeCarlo et al., 2006;Drewnick et al., 2005).The AMS measures nonrefractory material, which includes most organic species and inorganic salts other than NaCl, but not crustal material or black carbon.While the instrument nominally measures PM 1 it is most sensitive to particles with diameters between 35 and 600 nm (Zhang et al., 2004).The AMS was calibrated with size-selected ammonium nitrate every four days during the study.Limits of detection (LOD) for each of the AMS size bins were calculated as three times the standard error of the mean during filter measurement periods.For both organic and sulfate components, local LODs, for daily averaged data, increased roughly linearly with size across the measurement range from 0.17-0.43µg m −3 for organics and 0.015-0.06µg m −3 for sulfate.
Nanoparticle composition was also measured with a thermal desorption chemical ionization mass spectrometer (TD-CIMS) (Smith et al., 2004;Voisin et al., 2003).The TD-CIMS first charges the aerosol and then uses a Nano DMA (TSI 3085) to size-select particles.The size-selected particles are then collected onto a charged wire via electrostatic precipitation.Once a suitable mass of aerosol has been collected, the wire is inserted into the chemical ionization mass spectrometer and gradually heated from room temperature to ∼600 ºC, thus desorbing the sample.For the current study, the volatilized sample was then chemically ionized and detected with a high resolution time-of-flight mass spectrometer.Due to the low mass concentrations at smaller sizes, during BEACHON-RoMBAS size selected TDCIMS measurements were only made at 30 nm.The data presented here are averaged over two measurement periods: 3-8 August and 15-18 August 2013.

Aerosol composition and hygroscopicity
The κ value of a mixed-composition particle depends on its chemical constituents and can be calculated from volumeweighted aerosol composition (Petters and Kreidenweis, 2007): where κ i and ε i are the hygroscopicity parameter and volume fraction of the individual components.For the Rocky Mountain region during the summer months, SOA and inorganic salts have been shown to be the main components of PM 2.5 (Levin et al., 2009).While black carbon can be a signifi-cant contributor to aerosol mass during the wildfire season in the western US (Park et al., 2007), during BEACHON-RoMBAS, black carbon measured with a single particle soot photometer (SP2; Droplet Measurement Technologies) accounted for only 2 % of the aerosol mass averaged across all diameters below 350 nm (Ortega et al., 2014), similar to values reported by Levin et al. (2009) during a nonsmoke impacted time period in Rocky Mountain National Park.Dust particles may also be present in the aerosol, particularly during spring.However, dust particles typically have diameters larger than the 350 nm upper limit used here (Malm et al., 2009).Thus, for this study we assume that the aerosol below 350 nm diameter was composed mainly of SOA and hygroscopic inorganic salts, predominantly ammonium sulfate.Gunthe et al. (2009) employed similar assumptions for field measurements made in the Amazon Basin during the rainy season and found good agreement between κ determined from size-resolved CCN measurements and calculated from AMS chemical composition measurements.In this work we follow a similar approach to Gunthe et al. (2009) to compare two-component κ values calculated from chemical composition to κ determined from measured CCN activity.We also expand this analysis to show that a representative twocomponent aerosol composition can be derived from κ, and use the year-long BEACHON κ measurements to deduce seasonal changes in aerosol composition.

Aerosol hygroscopicity
Figure 1 shows daily average κ values (colored circles) and standard deviations during BEACHON-RoMBAS at the five different CCNC s settings.Study averaged values (± 1 standard deviation) are also reported in Fig. 1.Overall, aerosol hygroscopicity was low during this study, more similar to values reported for SOA than ammonium sulfate, and, on average, below the typically assumed value of κ continental = 0.3.Only at the lowest s value did κ values reach 0.3 during a few time periods.There was also little variation in κ during the study period, especially at higher s conditions.The values determined during the BEACHON-RoMBAS summer intensive study were slightly higher than those measured during the summer months of the year-long BEACHON study (± 1 standard deviation of July/August means shown by gray shading in Fig. 1).However, averaged over the BEACHON-RoMBAS study, κ values at every supersaturation setting fell within 1 standard deviation of the values from the previous year.
The right axis of Fig. 1 shows activation diameters corresponding to the left axis κ values for each s setting.At lower s, particles of similar hygroscopicity must be larger to activate as CCN.Thus, by operating at five different s settings, we are probing the hygroscopicity of particles at different diameters.While κ was low at all s settings, there was a small increase in κ with decreasing s, corresponding to increasing activation diameter.That is, there was an apparent change in aerosol composition with size, with particles larger than ∼100 nm containing a larger fraction of more hygroscopic material, similar to what was observed during the year-long study (Levin et al., 2012).We must also note that the κ values presented here only represent the aerosol near the activation diameters.Particles larger or smaller than this range could have different chemical composition, and thus different κ.Furthermore, the size-resolved CCNC measurements provide a single κ value for each critical activation diameter.If the aerosol at that size is internally mixed, this κ value will be representative of any particle at that size.However, if there are externally mixed aerosol populations at the d c , the calculated κ value will represent an average κ for particles at that size.Both Su et al. (2010) and Petters et al. (2009a) show that externally mixed aerosol with differing κ values will result in a bimodal distribution of activated fraction or distributions that never reach a value of 1.During BEACHON-RoMBAS we did not observe either of these phenomena, and thus conclude that at a given activation diameter the aerosol was internally mixed or at least contained aerosol with similar hygroscopicity.Levin et al. (2012) suggested that the aerosol measured during the year-long BEACHON study was predominantly organic, based on the low κ values, but there were no direct chemical measurements available to corroborate this claim.The similarly low κ values determined during BEACHON-RoMBAS also suggest a highly organic aerosol.During BEACHON-RoMBAS, submicron aerosol chemical composition was measured with the AMS and TDCIMS.Contour plots of daily averaged organic and inorganic aerosol mass distributions determined from AMS measurements are shown in Fig. 2, and the study-averaged organic mass fraction (X org ) as a function of particle diameter is shown in Fig. 3. To compare AMS results with those from the CCNC system, we converted vacuum aerodynamic diameter, measured by the AMS, to mobility diameter by dividing by the study averaged aerosol density, using values of 1.4 (Bahreini et al., 2005;Kroll and Seinfeld, 2005;Poulain et al., 2010) and 1.8 g cm −3 (Tang, 1996) for the organic and inorganic fractions, respectively.Orange shading in Fig. 3 indicates the range in AMS X org due to instrument uncertainty (Bahreini et al., 2009).Although the AMS can resolve a number of organic and inorganic species, we have binned the data into these two components to compare them with the twocomponent composition which can be estimated from sizeresolved CCNC measurements (discussed below).The AMS data indicate that the inorganic mass concentration consisted primarily of ammonium sulfate, calculated as the measured SO 2− 4 mass multiplied by 1.37, assuming that all sulfate was in the form of (NH 4 ) 2 SO 4 .This method was used rather than using the AMS ammonium measurements since the preci-sion of sulfate measurements was much higher than for ammonium in the particle time-of-flight mode.While acidic forms of sulfate could exist in the aerosol, AMS measurements, using the higher precision "mass spec" mode, indicated sufficient NH + 4 mass to fully neutralize the aerosol during BEACHON-RoMBAS.Nitrate mass was extremely low, often below the detection limit, and, when present, was associated with organic nitrate aerosol (Fry et al., 2013).Other inorganic species were negligible in the AMS measurements during BEACHON-RoMBAS.As expected, the organic fraction was the dominant component of the aerosol throughout the study, with a few periods of increased inorganic mass concentration.Averaged over the whole study (Fig. 3), organics were the main aerosol component at all sizes, with their relative contribution increasing at smaller sizes.

Aerosol composition
Extending composition measurements to smaller sizes, the TDCIMS provided composition data for particles 30-50 nm in diameter.These data are also shown in Fig. 3 (blue bar) with the width of the bar representing the range in X org for a range of assumed organic aerosol molecular weights (100-500 g mol −1 ).These data are consistent with the observed trends from the AMS, showing that these small particles were predominantly organic.Organic compounds appear to play a critical role in the growth of nucleation mode particles to CCN sizes (Dusek et al., 2010).The role of organic compounds in growth of small particles was supported by observations in the year-long BEACHON study, in which aerosol hygroscopicity appeared to be affected by shifts in size distributions typical of new particle formation (NPF) events, with κ over most of the measured size range decreasing as the newly formed particles grew to sizes large enough to activate as CCN (Levin et al., 2012).While newly formed, organic particles dominated at small sizes, AMS measurements showed an increasing inorganic fraction at larger aerosol diameters.Given this enhanced inorganic fraction and the lack of local sulfate emissions, it is most likely that the inorganic aerosol was more aged and may have undergone some cloud processing, as cloud droplet residual particles have been shown to have increased sulfate concentrations (Hao et al., 2013).Similarly, enhanced hygroscopicity for particles larger than ∼100 nm was observed by Fors et al. (2011) during long-term measurements at a background site in Sweden.This increased hygroscopicity was also attributed to an enhanced inorganic component due to cloud processing.

Kappa composition closure
To compare the AMS measurements with those from the size-resolved CCNC, we calculated organic volume fractions (ε org ) from measured κ values by rearranging Eq. 1: We used a κ inorg value of 0.6, assuming the inorganic aerosol fraction was predominately ammonium sulfate, and allowed κ org to vary from 0 to 0.15.We chose 0.15 as the maximum allowed value for κ org as this was the lowest average κ value determined during BEACHON-RoMBAS; assuming a higher value of κ org would have led to a physically meaningless, negative inorganic volume fraction, given the twocomponent system.We also assumed that the range in κ org did not change over the course of the study or as a function of size.To convert from volume fractions calculated from Eq 2 to mass fractions, we multiplied the volume fractions by the assumed densities of 1.4 and 1.8 g cm −3 for the organic and inorganic fractions, respectively.Figure 3 shows the studyaveraged organic mass fraction (X org ) calculated from κ as a function of study-averaged critical activation diameter.The gray shaded region shows the range in X org resulting from the range in assumed κ org values (0-0.15).Regardless of the assumption made about the hygroscopicity of the organic material during the study, Figure 3 clearly shows that the organic fraction dominated aerosol mass at all measured sizes.There was also a decrease in X org with increasing activation diameter, as expected from the higher average κ values at these sizes.
Averaged over the whole study, the best agreement (minimum χ 2 ) between X org determined from κ values and AMS measurements was for a κ org value of 0.13 (black line in Fig. 3).In the diameter range where the CCNC and AMS measurements overlap, the agreement in organic mass fraction using this value of κ org is very good, indicating little size dependence in κ org or any of the other assumptions inherent in these calculations.Further, the trend of decreasing X org with increasing size is very similar for the two measurements.
We also calculated daily averaged κ distributions from the AMS organic and inorganic mass measurements (κ AMS ).To do this, we first converted AMS measured mass to volume fractions using the assumed densities mentioned above and then used Eq.(1).and the best fit value of size-independent κ org , 0.13, to calculate κ AMS .These values are shown in Fig. 4. We then selected the κ AMS values at the diameters corresponding to the d c for each CCNC s setting and plotted these in Fig. 1 to compare with κ derived from CCNC measurements (κ CCNC ).The error bars for κ AMS in Fig. 1 represent AMS uncertainty.
Overall, there is good agreement between κ AMS and κ CCNC .For all but the lowest s setting, the largest discrepancies between κ AMS and κ CCNC occurred when total AMS mass concentration at the d c was near the detection limit , with better agreement at higher mass concentration.At the lowest s, largest d c , there is some difference in κ at the beginning of the study, during a period of relatively high AMS mass concentration, but then there is good agreement after this time period.During the early time period (7/18-7/22), AMS measurements indicated an increased inorganic mass concentration, and thus higher κ, at sizes slightly larger than the largest CCNC d c (∼150 nm).There was also a sharp gradient in inorganic mass concentrations between 150 and 250 nm.The higher κ CCNC values during this time period could be affected by larger, multiply charged particles appearing in smaller DMA size bins.Although we correct total aerosol and CCN number distributions for multiply charged particles (Petters et al., 2009a), size-dependent chemical differences in the aerosol are not accounted for in the inversion and could impact measured κ values.
From Fig. 2 it can be seen that the κ AMS values shown in Fig. 1 were calculated at diameters where mass loadings were very low, especially at the smaller d c values.At smaller sizes, this resulted in more variable κ AMS values, and in many cases we were unable to calculate κ AMS because mass was below the AMS detection limit.At larger diameters, above the measurable range for the CCNC system, the AMS values are likely more certain, due to the higher mass concentrations at these sizes.By contrast, the CCNC is a number-based measurement, and is thus most reliable at diameters with the highest number concentrations, so that comparing results from these two systems is challenging.Nevertheless, the general agreement between X org and κ determined from these two measurement systems in this case supports the use of the AMS data to calculate hygroscopicity, and the size-resolved CCNC data to calculate two-component aerosol composition.Furthermore, while the AMS can only detect nonrefractory particles, it appears to capture the important species for determining aerosol hygroscopicity in this size range at this site.

Seasonal aerosol composition
Based on the generally good agreement between aerosol composition measured by the AMS and calculated from CCNC data during BEACHON-RoMBAS, we used the yearlong CCNC data reported by Levin et al. (2012), and the same methods discussed above, to calculate monthly averaged, two-component aerosol composition distributions for a full year.These distributions are shown in Fig. 5 with the green and blue bars showing the organic and inorganic components, respectively.As before, the gray shaded region shows the range in calculated organic mass fraction for assumed κ org values.However, the range in assumed κ org had to be limited to 0 < κ org < 0.12 for the year-long data set in order to avoid a negative calculated X inorg value during some months.Since a best-fit value of κ org = 0.13 was observed during BEACHON-RoMBAS, this could indicate some seasonal changes in the composition of the organic aerosol at this site or the influence of other low-κ aerosol components, such as dust or black carbon.
We observed a similar trend throughout the year-long study period as that seen during the intensive BEACHON-RoMBAS campaign, with aerosol organic fraction decreasing with increasing diameter.There was, however, some seasonality in this pattern.Aerosol composition was least sizedependent during the late summer and early fall, and then the variability with particle diameter steadily increased throughout the winter and into the spring.The largest change in organic fraction with size was observed during April, when X org decreased to ∼0.5 at the largest measured size.The seasonal change in X org for particles at the largest measured size is consistent with the seasonal variability of volatile organic compound (VOC) emissions (Saarikoski et al., 2008), which are precursors for SOA production.Likewise, while NPF events occur year round, they were most common in the summer months at Manitou Experimental Forest (Levin et al., 2012), and could be a contributing factor to the increased X org values during this time period.Less seasonality was observed in the inferred composition of the smallest measured particles, which remained almost entirely organic throughout the year-long study period.

Summary and conclusions
Atmospheric particles that act as CCN are important in controlling cloud and precipitation formation and lifetime, and can have considerable effects on local to global scales.Highquality measurements of CCN concentration and aerosol hygroscopicity are needed to constrain modeled cloud and precipitation processes, and explore their subsequent impacts on water cycles and radiative transfer.In this paper we present aerosol composition and hygroscopicity data from a forested site in the eastern Rocky Mountains of Colorado made as part of BEACHON-RoMBAS.We used size-resolved CCN measurements to determine aerosol hygroscopicity and measured aerosol chemical composition with an AMS and TDCIMS, and found consistency between measured hygroscopicity and aerosol composition from the AMS and TDCIMS.Likewise, we show that CCN-derived κ values can be used to calculate aerosol composition, assuming conditions are met for a simple two-component system composed of organic aerosol of size-independent κ and an inorganic aerosol dominated by sulfate.Over the intensive summer study, the best agreement between measured and calculated aerosol composition was achieved for an assumed value of κ org = 0.13.
Aerosol chemical composition and hygroscopicity measurements showed that the aerosol measured at the BEA-CHON site was predominantly organic, and thus had low κ values.Averaged over the entire year-long BEACHON study, and all s settings, κ had an average value of 0.16 ± 0.08, similar to values determined in biogenic organic dominated locations such as a tropical forest in the Amazon during the wet season (Gunthe et al., 2009;Pöschl et al., 2010) and a boreal forest in Finland (Sihto et al., 2011), and lower than the commonly assumed value of κ continental = 0.3.The relatively low κ values measured at these locations dominated by biogenic emissions suggest a predominance and potential importance of organic species, which generally have low hygroscopicity (Prenni et al., 2007;Petters et al., 2009b;King et al., 2010).It is known that highly biologically active areas such as boreal and tropical forests will have large contributions from organic aerosol leading to low κ values.However, results from this study indicate that the aerosol in this Colorado, semi-arid, high altitude location is also dominated by organic species.
While organic species were dominant at all sizes, there were some changes in aerosol composition with size, with the organic fraction typically decreasing at larger diameters.The increased inorganic fraction of larger particles likely indicates that these particles were more aged and had likely undergone some cloud processing.The trend of decreasing organic fraction with larger diameters was observed year round, however, there was some seasonality observed during the year-long BEACHON study.During the winter months, there was a sharper decrease in calculated organic mass fraction with size, although even at the largest diameter (∼ 150 nm) the aerosol was still predominantly organic.During the summer months, there was less size dependence in composition.We postulate that the increased organic fraction at larger diameters during the summer is due to increased biological activity during this time period, resulting in increased emissions of biogenic volatile organic compounds, which are precursors of organic aerosol.

Fig. 1 .
Fig. 1.Daily averaged κ values at each supersaturation setting during BEACHON-RoMBAS (colored circles) as well as calculated κ from AMS measurements (black circles).Gray bars show ± one standard deviation for measured κ values from July to August, 2011, collected during the year-long BEACHON study.Right axis indicates activation diameter corresponding to the left axis κ values at the indicated supersaturation.

Fig. 5 .
Fig. 5. Monthly averaged aerosol size distributions (black) and two-component chemical composition (green = organic, blue = inorganic) calculated from hygroscopicity measurements during the year-long BEACHON study.Gray shading shows the range in organic mass fraction (X org ) due to uncertainty in κ org .