Spatial and vertical extent of nucleation events in the Midwestern USA: insights from the Nucleation In ForesTs (NIFTy) experiment

Measurements of aerosol particle physical and chemical properties, gas phase concentrations and meteorological parameters were made along a transect in southern Indiana during the Nucleation In ForesTs (NIFTy) experiment conducted in May 2008. These measurements indicate nucleation was observed at all three measurement sites on almost half of all sampling days. The intensity of the nucleation events, as measured by the increase in ≥10 nm aerosol particle number concentrations of approximately 2 ×104 cm−3 over a layer of at least 300 m depth, is in good agreement with recent model results for the Midwestern USA derived using PMCAMx-UF. During the hour after termination of nucleation approximately half of the number concentration reduction is due to coagulation, while the remainder is due in equal parts to dry deposition and entrainment of relatively ultra-fine aerosol particle free troposphere air. Clear nucleation with continuous subsequent growth is only observed on days when the morning fractional cloud cover was less than 30%. It is associated with a clear transition from a strongly stratified atmosphere with low turbulence intensity and weak vertical velocities, to much a weaker vertical gradient of wind speed, increased turbulence intensity and stronger downwards vertical velocities, consistent with growth of the mixed layer and entrainment of air from the residual layer. Nucleation intensity is not very strongly determined by the prevailing condensational sink. However, there is a strong correlation between both a modified version of the Nucleation Parameter from Boy and Kulmala (2002) and Correspondence to: S. C. Pryor (spryor@ indiana.edu) ultrafine aerosol particle number concentrations, and mean morning H2SO4 concentrations and ultrafine aerosol particle number concentrations. Five A-class event days during NIFTy were characterized by values of the dimensionless nucleation parameter of Kuang et al. (2010) that are below 0.3, further indicating the applicability of their postulate that nucleation is favored byL0 values below 0.7. Based on aerosol particle composition measurements it appears that aerosol particle formation and initial growth to approximately 30 nm diameter is dominated by ammonium and sulfate. Conservative estimates of the percent contribution of H 2SO4 to aerosol particle growth (for sub-30 nm aerosol particles) on five Aclass event days ranged from 23 to 85%.


Introduction and objectives
Atmospheric aerosol particles affect climate both directly by scattering incoming solar radiation back to space and indirectly by acting as cloud condensation nuclei.Direct and indirect climate forcing by aerosol particles are two of the most uncertain parameters in past and future global change (Alley et al., 2007;Schwartz et al., 2010).Model simulations indicate nucleation i.e. gas-to-particle conversion is a significant source of aerosol particles in the global atmosphere (Spracklen et al., 2006;Yu and Luo, 2009), and there is experimental evidence that new particle formation is a worldwide phenomenon (Kulmala and Kerminen, 2008;Kulmala et al., 2004;Manninen et al., 2010).However, to assess model uncertainties and constrain climate projections, additional distributed measurements of aerosol particle concentrations and properties and related modeling are required.
The Nucleation In ForesTs (NIFTy) experiment was conducted in southern Indiana in the eastern USA during 5-31 May 2008.Simulations of the eastern USA during 12-28 July 2001 indicate ≥3 nm aerosol particle concentrations in the vicinity of the NIFTy sampling sites were 4×10 4 cm −3 (for ≥10 nm aerosol particles the concentration was approximately 2×10 4 cm −3 ) and exhibited a three-fold increase in near surface ≥10 nm aerosol particle concentrations when a scaled version of a ternary nucleation scheme was turned on relative to a no-nucleation simulation (Jung et al., 2010).Consistent with that model analysis, two years of continuous aerosol particle size distribution measurements at the central forest site used in the NIFTy experiment exhibit evidence of nucleation on 46% of classifiable days, and clear nucleation with consistent subsequent growth on 14% of days (Pryor et al., 2010).The context for and objectives of the NIFTy experiment were as follows: 1. To examine the spatial scales of nucleation and variations in ultra-fine aerosol particle characteristics in locations with differing land use and emission profiles.
Prior research has shown gas-to-particle conversion occurs on regional spatial scales as indicated by ground based measurements upto several hundred kilometers apart (Dal Maso et al., 2007;Hussein et al., 2009;Jeong et al., 2010;Komppula et al., 2006;Wehner et al., 2007).These events, while nearly coincident in time, have site-specific characteristics (Hussein et al., 2009).
Prior research has also shown qualitatively similar behavior in ultrafine size-resolved aerosol particle concentrations at upwind rural sites and downtown urban areas over scales of 10's of km (Stanier et al., 2004b, a;Yue et al., 2009).Accordingly, during NIFTy measurements were taken along an 80 km transect extending from the small town of Bloomington to the southwest, to the central long-term measurement forest site (in Morgan Monroe State Forest (MMSF)) and up to northeast Indianapolis to examine meso-scale variability in aerosol particle properties (Fig. 1).The vertical extent of new aerosol particle formation is uncertain, with evidence for both new aerosol particle formation at the surface layer mixing upwards (O'Dowd et al., 2009) and nucleation occurring aloft and subsequently blended downwards (Siebert et al., 2004).New aerosol particle formation frequently occurs concurrently with the destruction of the night-time stable layer and subsequent formation of the mixed boundary layer.Typically the time for mixing and for the growth to detectable size occurs in approximately same time scales (Nilsson et al., 2001), thus differentiating between the two theses is difficult.Further, increased turbulence could enhance the concentrations of nucleating species on a small scale and possibly facilitate the initial formation of new aerosol particles (Lauros et al., 2007;Lauros et al., 2006;Nilsson and Kulmala, 1998;Wehner et al., this inventory.The SO2 emissions derive principally from point sources with elevated 8 effective emission heights, while NH3 emissions are predominantly from ground-based 9 agricultural sources (Pryor et al., 2001).10 11 12 Fig. 1.Location of the sampling sites (ν); Bloomington (39.171 • N, 86.506 • W), MMSF (39.317 • N, 86.417 • W), and northeast Indianapolis (39.811 • N, 86.114 • W), and county-level aggregated emissions of sulfur dioxide (SO 2 ) and ammonia (NH 3 ) in 2002 in tonnes per square mile (data available at the EPA AirData portal; http://www.epa.gov/air/data/reports.html).Annual average NH 3 and SO 2 emissions in southern Indiana are estimated to be approximately 46 and 2.3 tons per square mile based on this inventory.The SO 2 emissions derive principally from point sources with elevated effective emission heights, while NH 3 emissions are predominantly from ground-based agricultural sources (Pryor et al., 2001).2010).Thus, to examine the vertical extent of nucleation and the fate of recently nucleated aerosol particles we flew instrumentation on an unmanned aerial vehicle (UAV) from a small private airfield on the southern portion of the experimental transect between Bloomington and MMSF (Table 1).

To examine the meteorological context for nucleation
events.Certain synoptic scale meteorological conditions, particularly cold front passages, appear to be associated with nucleation, in part because they are associated with a reduction in the sinks for freshly formed aerosol particles and condensable vapors.Further, since analyses of other long-term data sets have indicated preferential occurrence of nucleation with specific back-trajectories (Coe et al., 2000;Fiedler et al., 2005;Hussein et al., 2009;Komppula et al., 2006;Young et al., 2007), the synoptic context was examined and 24h back trajectories were computed for every classified day of data from MMSF using the HYSPLIT model.Analyses of local meteorological conditions, described using metrics such as the Nucleation Parameter (NP) of (Boy and Kulmala, 2002) and atmospheric stability indices (based on the Monin-Obukhov length), have indicated nucleation is frequently preceded by destabilization of the atmosphere possibly associated with entrainment of elevated pollution layers (Nilsson et al., 2001).Aerosol particle number size distributions (10-400 nm) @ 12 m lidar derived state parameters at Cabauw in the Netherlands also indicated that increased ultrafine aerosol particle number concentrations were observed in the residual layer between the growing boundary layer and freetroposphere (Stull, 1988).Further, these layers "were characterized by a sub-critical Richardson number and concomitant turbulence..." leading the authors to speculate that "turbulent mixing is likely to lead to local supersaturation of possible precursor gases. . .Observed peaks in the number concentrations of ultrafine aerosol particles at ground level are connected to the new aerosol particle formation in the residual layer by boundary layer development and vertical mixing."(Wehner et al., 2010).Thus during NIFTy we deployed a scanning doppler lidar system and tethersonde at MMSF to examine vertical profiles of meteorological conditions in addition to analyzing data from ongoing micrometeorological and ceilometer measurements (Table 1).

3.
To examine associations between organic and inorganic gas concentrations and ultra-fine aerosol particle composition and number concentration during nucleation events.The precise mechanism for nucleation remains uncertain, and the relative importance of binary, ternary and ion-induced nucleation likely vary in space and time (Gagné et al., 2010), but likely most nucleation events involve sulfuric acid (Nieminen et al., 2009;Sipila et al., 2010;Weber et al., 2001;Weber et al., 1997), ammonia (Korhonen et al., 1999;Napari et al., 2002) and/or organics (Metzger et al., 2010;Zhang et al., 2004b), or some combination thereof (Paasonen et al., 2010).Subsequent growth of the freshly formed aerosol particles occurs via multi-component condensation and to a smaller extent by coagulation (Boy et al., 2005;Stolzenburg et al., 2005;Wehner et al., 2005).Particle growth in locales with high sulfur dioxide emissions is dominated by sulfuric acid (Petäjä et al., 2007;Sakurai et al., 2005;Stolzenburg et al., 2005).Generally in less polluted regions particle growth is dominated by compounds other than sulfuric acid (Boy et al., 2005;Wehner et al., 2005), and there is evidence for a role of organics (particularly biogenics) in aerosol particle formation and initial growth in forested locations (Metzger et al., 2010;Place et al., 2010).Indiana, and indeed the whole Midwestern USA, is characterized by very high emissions of the precursors of ternary nucleation (SO 2 (and thus sulfuric acid (H 2 SO 4 )) and ammonia (NH 3 )).Accordingly, prior research using an aerosol mass spectrometer in Pittsburgh (east of the Ohio River Valley) showed that, during nucleation events, aerosol particles with physical diameters of 18 to 33 nm showed an initial increase in sulfate (SO 2− 4 ) concentrations followed by increased ammonium (NH + 4 ) concentrations (delayed by 10-40 min) and ultimately an increase in oxygenated organics (Zhang et al., 2004a), thus indicating a dominant role for the inorganic gases in the initial aerosol particle formation and growth at this location.Equally measurements during Border Air Quality and Meteorology Study (BAQS-Met 2007) conducted to the north of Indiana exhibited a high degree of association between sulfur dioxide (SO 2 ) concentrations and nucleation occurrence (Jeong et al., 2010).To examine the chemistry of ultra-fine aerosol particle concentrations and links to gas phase chemistry during NIFTy continuous measurements of SO 2 , NH 3 , H 2 SO 4 and timeaveraged VOC and aerosol particle composition were made at the MMSF site (Table 1).

Sampling sites
Measurements presented herein were conducted along an 80 km transect from the small town of Bloomington in the southwest, to northeastern Indianapolis to the northeast.The transect (Fig. 1) has at its approximate mid-point the long-term aerosol particle measurement site in the Morgan-Monroe State Forest (MMSF) (39 • 19 N, 86 • 25 W, Fig. 1).MMSF is an extensive broad-leafed forest with a total area of over 95.3 km 2 .The single-sided leaf area index (LAI) of the forest during the experiment was approximately 4 m 2 m −2 , and tree total basal area is approximately 26 m 2 ha −1 .The forest is dominated by five species: sugar maple (Acer saccharum, 27%), tulip poplar (Liriodendron tulipifera, 19%), sassafras (Sassafras albidum, 9.5%), white oak (Quercus alba, 9%), and black oak (Quercus nigra, 8.5%).Bloomington (on the southwest end of the transect) is a small college town (population approximately 80 000), while Indianapolis is the fifth most populous city in the USA (population of 6.5 million) and currently fails current ozone and particulate matter air quality standards (Pryor and Spaulding, 2009).The sampling site used herein is located approximately 5 km downwind of the urban core.

Aerosol physical characterization
Aerosol particle size distributions have been continuously measured at MMSF since December 2006 using two Scanning Mobility Particle Sizer (SMPS) systems from TSI Inc.
A Fast Mobility Particle Sizer (FMPS 3091) (TSI, Inc) (Tammet et al., 2002) was added in November 2007.During NIFTy the SMPS system comprising an Electrostatic Classifier (Model 3080), long-DMA (Model 3081), and a Condensation Particle Counter (CPC) (Model 3025A) was redeployed to the Bloomington site.The second SMPS comprising an Electrostatic Classifier (Model 3080), nano-DMA (Model 3085), and a Condensation Particle Counter (Model 3786) and a FMPS were operated at MMSF sampling from a height of 46 m (the forest canopy is at 26-28 m).A fourth particle sizing system -a FMPS was operated in northeast Indianapolis at an Indiana Department of Environmental Management air quality site (see Table 1).
For the flow rates and inlet nozzles used herein the aerosol particle size range reported by the SMPS operated in Bloomington is 9.8 to 414.2 nm, while for the SMPS operated at MMSF it is 3.22 to 105.5 nm.However, the long sampling tube used at MMSF functionally limits the sampling range to above 6 nm, since this is the minimum aerosol particle diameter (Dp) for which transmission efficiencies exceed 10%.In both cases the SMPS systems were deployed with the multiple charge and diffusion loss corrections in the AIM software turned on (Frank et al., 2008).The aerosol particle size resolved concentrations from the SMPS and FMPS deployed at MMSF were corrected for tubing losses computed using experimentally derived correction factors (Pryor et al., 2010).The FMPS systems used at MMSF and Indianapolis measure aerosol particle concentrations in 32 equally spaced (logarithmic scale) size channels from 6.04 to 523.3 nm but the sampling inlet used in Indianapolis limited the diameter range for which reliable data were collected to 10-200 nm.
Inter-comparison of aerosol particle sizing instruments used at the different sites during NIFTy indicates relatively good correspondence from three of the instruments -the SMPS system deployed in Bloomington and the FMPS and SMPS deployed at MMSF.As in prior research sub-30nm aerosol particle concentrations are slightly lower from the FMPS (Jeong and Evans, 2009), while concentrations of approximately 100 nm diameter particles are higher in both FMPS systems (Asbach et al., 2009).Size distributions from the FMPS deployed in Indianapolis differ substantially from those derived from the other instruments (Fig. 2).This discrepancy may derive from noise on the electrometers (specifically mis-counting aerosol particles at smaller sizes with the FMPS due to residue on the electrometer stages), or slight mis-alignment in the aerosol flow within the FMPS deployed in Indianapolis.In light of this discrepancy, the aerosol particle size distributions from Indianapolis must be viewed with caution, and growth rates computed there from are deemed too uncertain to present herein.
To examine the vertical extent of the layer through which enhanced ultra-fine aerosol particle concentrations are observed an instrumented UAV was flown from a private airstrip approximately 10 km SW of the MMSF tower.The UAV was flown in a pattern of displaced spirals to avoid  sample contamination from the engine.Vertical profiles up to a maximum height of 300 m a.g.l. of total ultrafine aerosol particle concentrations and size distributions were measured using a CPC (TSI 3007) and GRIMM Portable Aerosol Spectrometer (1.109, from Grimm Aerosol technik GmbH) (Table 1).

Aerosol chemical characterization
An MSP MOUDI-110 with nano-MOUDI attachment was installed at the MMSF tower at 28 m and operated to measure 24 h average inorganic ion concentrations.Nominal cut points for the MOUDI-110 (for 30 liters per minute (l.p.m.) flow rate) are; 18, 10, 5.6, 3.2, 1.8, 1.0, 0.56, 0.32, 0.18, 0.1, and 0.056 µm, while the nano-MOUDI nominal cut points are; 32, 18 and 10 nm.Aluminum foil substrates, coated using silicon spray to minimize aerosol particle bounce, were used for analysis of inorganic ions.After collection the foils were immediately placed in test-tubes and extracted for two hours in oxalic acid (10 −5 M) on a stirring table, and then analyzed on an ion chromatograph (IC, DIONEX DX-120).The ions for which standards were applied to the IC and for which observed concentrations in any sample exceeded the minimum detection limits (MDL) were; sodium (Na + ), ammonium (NH + 4 ), potassium (K + ), magnesium (Mg 2+ ), calcium (Ca 2+ ), chloride (Cl − ), nitrate (NO − 3 ), phosphate (PO 3− 4 ) and sulfate (SO 2− 4 ).Uncertainty bounds and MDL for the ion concentrations reported herein were computed based on replica analysis of standards (spanning the range of concentrations in the ambient samples) on the IC.Repeat extractions of ten randomly selected MOUDI foils did not show any ion concentrations above the MDL, indicating complete extraction of the sample by the aforementioned procedure.
A second MOUDI-110 was operated with quartz filters for capture and analysis of organics, but due to difficulties with the filter extractions no data are reported from that system.

Gas phase characterization
One minute average NH 3 concentrations were made at 32 m using a Wet Effluent Diffusion Denuder (WEDD) system (Pryor et al., 2001), while the 30-min average concentrations of SO 2 and H 2 SO 4 close to ground level reported herein were made with a TECO (Model 43S) monitor and a Chemical Ionization Mass Spectrometer (CIMS) (Eisele and Tanner, 1993;Mauldin III et al., 2003), respectively.A limited number of hydroxyl radical (OH) measurements were also made with the CIMS (Berresheim et al., 2000;Eisele and Tanner, 1993;Mauldin III et al., 2003).
VOC concentrations were measured in six approximately 2-h intervals starting 09:00 (LST) at a flow rate of 27 ml min −1 on multi-sorbent cartridges (MSC), following EPA Method TO-17.Samples were analyzed for the following indicators of biogenic VOCs; isoprene, α-pinene, limonene, cumene (isopropylbenzene), and for the following indicators of anthropogenic VOCs; benzene, toluene, ethyl benzene, o-, m-, and p-xylene.The resulting VOC concentrations are used herein to provide a first-order assessment of the production of low-volatility oxidation products using Fractional Aerosol Coefficients (FAC) (Grosjean, 1992) (see Supplement).

Meteorological instrumentation
To supplement the ongoing micro-meteorological and ceilometer (Vaisala CL31) measurements that are continuously made at the MMSF tower as part of the AmeriFlux network protocols, detailed meteorological profiles were obtained in a small clearing close to the MMSF tower using a scanning doppler lidar (Natural Power ZephIR lidar) and a Vaisala tethersonde system equipped with a temperature, humidity and wind package.The Natural Power ZephIR is a ground-based doppler lidar system designed to accurately measure horizontal and vertical wind speed, wind direction and turbulence up to 300 m height.The lidar measures in a cone of 30 • from the vertical enabling horizontal and vertical components of the wind speed to be determined from the radial velocity (Wagner et al., 2009).In effect the characteristics of a volume of air are measured from a conical scan pattern as the beam is moved and focused by a lens at different heights (Emeis et al., 2008).Each measurement takes about 20 ms enabling multiple sampling at one height and a full scan at each of the measurement heights in 3 s.The turbulence parameter is a slightly different metric of turbulence than is obtained using a cup or sonic anemometer in that it represents the turbulence intensity of the radial wind speed within the scanned circles, however it shows good correspondence with turbulence metrics from anemometers (Wagner et al., 2009).

Model description
Vertical profiling of the aerosol particle number concentrations with the UAV coupled with the number size distributions determined at the MMSF tower site enabled us to probe aerosol particle number concentration budgets in the boundary layer.The measured concentrations were used as input values for a box model applied assuming the layer is wellmixed and that the event was regionally coherent in character (allowing us to neglect the role of advection), to assess the relative importance of coagulation, dry deposition and entrainment/dilution due to growth of the boundary layer.
Aerosol particle coagulation within the model was computed using COAGSOLV, a semi-implicit aerosol particle coagulation solver (Jacobson et al., 1994), while dry deposition was calculated using the size-dependent deposition velocities for MMSF (Pryor et al., 2009).We further obtain an estimate of the degree to which the model estimates are reasonable by inverting this analysis.We apply a simple budget model to assess what the entrainment velocity would need to be if the source strength for ultrafine aerosol particles with diameters below 30 nm was zero, and the concentration of ultrafine aerosol particles with diameters below 30 nm was zero above the mixed layer.Thus the following budget equation was applied (Fairall and Larsen, 1984): Where C(Dp) is the concentration of aerosol particles in a size range (Dp = 10-30 nm) dC(Dp))/dt = the rate of change of concentration (corrected for coagulation losses computed as described above) h = mixed layer depth (estimated from the ceilometers backscatter data) v d (Dp) = the mean dry deposition velocity for aerosol particles of 10 to 30 nm (Pryor et al., 2009).
w e = entrainment velocity

Nucleation metrics
Data from all four aerosol particle sizing instruments operated during NIFTy were analyzed to determine event frequency and characteristics based on a subjective classification protocol (Boy and Kulmala, 2002;Dal Maso et al., 2005) (see Supplement).The condensational sink (CS) was computed from these data assuming the condensing vapors have a very low vapor pressure, an accommodation coefficient of 1, and exhibit properties similar to sulfuric acid (Kulmala et al., 2001).From the data collected at MMSF we additionally calculated a modified version of Boy and Kulmala's (2002) "nucleation parameter" (NP): Where: UV is derived from measurements 46 m of short-wave incoming solar radiation (as measured using a LI-COR LI-200SZ-50 pyranometer) [H 2 O] = water vapor concentration (in molecules m −3 ) (as measured using a Meteolabor AG, VTP37 thermohygrometer) T = air temperature (K) from the Meteolabor AG, VTP37 thermohygrometer.

Horizontal profiles along the NIFTy transect
In keeping with the prior climatology of ultrafine aerosol particle concentrations at MMSF (Pryor et al., 2010), high concentrations of ultrafine aerosol particles with a diameter below 20 nm and clear growth were observed on multiple days during the NIFTy experiment (Table 2 and Fig. 3).While division of the data between the three event classes and non-events is subjective, independent classifications for measurements at the three sites show a relatively high degree of coherence.Of the days when there is evidence for class A nucleation events at MMSF and for which measurements are also available from Indianapolis and Bloomington, measurements at all three sites indicated a significant increase in ultrafine aerosol particle concentrations, though on two days the event classification for the other two sites (Indianapolis and Bloomington) indicated C rather than A class events.
Growth rates (GR) computed from the number geometric mean diameter (nGMD) for Dp = 6-30 nm based on data from the SMPS at MMSF for the A-events had a mean value of 3.6 nm hr −1 , and ranged from a minimum of 2.0 nm hr −1 to a maximum of 5.4 nm hr −1 .These values are in agreement with typical values obtained elsewhere (Kulmala et al., 2004), and are very similar to GR computed from nGMD for Dp = 10-40 nm based on data from the SMPS in Bloomington for the A-events (mean value of 3 nm hr −1 , range of 2.3 to 4.9 nm hr −1 ).Given the discrepancy in the exact size range used to compute GR, these results indicate the two sites ex- hibited very similar growth rates.This may imply that the nucleation events are both regional in occurrence and regionally homogeneous in terms of the subsequent aerosol particle growth.

Case studies of double-peak event days
On two days (17 and 18 May), data from all three sites show evidence for a double peak in ultrafine aerosol particle concentrations, wherein an initial increase in aerosol particle number concentrations was observed followed by a decline in number concentrations, and then a second burst of ultrafine aerosol particles (see Fig. 4).In common with other days on which nucleation was observed, the initial aerosol particle number concentration increase was preceded by an increase in turbulence (as manifest by an increase in the momentum flux in see Fig. S1, and discussion of the turbulence data from the lidar given below).The increase in turbulence was associated with an increase in incoming solar radiation and an increase in both H 2 SO 4 and SO 2 concentrations, consistent with the postulate that entrainment is strongly coupled to aerosol particle nucleation.On 17 May the initial increase in aerosol particle number concentrations was associated with a small decrease in NH 3 concentrations possibly as a result of NH 3 consumption in aerosol particle nucleation and growth.On both days the second increase in ultrafine aerosol particle number concentrations was associated with a second peak in solar radiation after a period of increased cloud cover as recorded by the ceilometer.The observed increase in H 2 SO 4 and SO 2 followed an increase in turbulent activity and increased solar flux and thus may be due to renewed mixed layer growth and entrainment, and/or possibly enhanced photochemistry and production of H 2 SO 4 .Boy and Kulmala (2002) and the start time of the event as derived from data obtained using aerosol particle sizing instrumentation deployed at MMSF, Indianapolis and Bloomington during NIFTy.N/A denotes periods when the instrument was not operating at the specified site for that day.To allow comparison across the sites the start hour denotes the hour in which the maximum increase in the number concentration of 10 nm diameter aerosol particles was first noted.For event class B a threshold diameter of 20 nm is used.

Vertical profiles of ultrafine aerosol particle concentrations
One example of data collected on UAV flights prior to and following initiation of a nucleation event on 22 May is shown in Fig. 5. Vertical profiles of total aerosol particle concentrations for Dp≥10 nm prior to initiation of the event at MMSF (see Fig. 6), indicate a well-mixed layer with fairly homogeneous aerosol particle number concentrations of 1×10 4 cm −3 at approximately 10:00 LST.As the event was observed at MMSF, the vertical profile showed a substantial increase in total aerosol particle number concentrations and significant vertical structure.Within two hours of event initiation, the vertical profile of aerosol particle concentrations was well-mixed to a height of over 300 m.The homogenization of aerosol particle number concentrations over less than 30 min is consistent with time scale analysis for the mixed layer.Assuming the mixed layer depth (z) derived from the ceilometer backscatter density gradients for 10:30-11:00 (LST) of 800 m, and an average energy dissipation rate (ε) deriving from the synthesis of data from the tethersonde, lidar and sonic anemometer at 46 m, of 5×10 −3 m 2 s −3 , the mixing time scale (τ mix , τ mix = z 2 ε 1/3 (Kaimal and Finnigan, 1994;Wehner et al., 2010) is approximately 500 s (about 9 min).Little vertical structure is evident in the super-250 nm aerosol particle concentrations from the GRIMM deployed on the UAV during any of the flights on this day (Fig. 5), implying the structure in the ultrafine particle concentrations is due to in situ production rather than horizontal advection.During about a 90 min period the total aerosol particle number concentrations (Dp≥10 nm) throughout a layer of at least 300 m increased by a factor of over three (from approximately 1×10 4 cm −3 to 3.2 ×10 4 cm −3 , Fig. 5).This is consistent with SMPS data from the MMSF tower which showed an increase in total aerosol particle number concentrations of a factor of four for Dp≥10 nm.The magnitude of increases in ultrafine aerosol particle concentrations observed on 22 May provides some indirect validation of the modeling conducted using PMCAMx-UF based on ternary nucleation which indicated the increase in ≥10 nm aerosol particle number concentrations during nucleation events was approximately 2×10 4 cm −3 (Jung et al., 2010).

Atmospheric fate of recently produced ultrafine aerosol particles
The instrumented UAV was also flown subsequent to the initial phases of nucleation events to examine the subsequent fate of recently formed aerosol particles and the evolution of vertical concentration profiles.On 17 May 2008, the UAV was flown (take-off at 13:52 LST), about one hour after termination of the period with highest 6 nm aerosol particle concentrations at MMSF (Fig. 7a).Total aerosol particle number concentrations (Dp≥10 nm) over a 250 m layer were calculated for the first 15 min and last 15 min of the flight.The results indicate total aerosol particle number concentrations decreased from an average of 3.96×10 4 cm −3 to 3.28×10 4 cm −3 over the course of 30 min.We initialized the box model by applying the size distribution as measured using the SMPS at MMSF to the UAV total aerosol particle concentrations and then ran the model for 30 min.The results indicate coagulation accounts for approximately half of the number concentration reduction (i.e. a aerosol particle number concentration decrease of 3600 cm −3 ), dry deposition accounts for a number concentration reduction of approximately 1800 cm −3 , and thus the residual of approximately 1600 cm −3 is assumed to be due to entrainment/dilution. Based on (1), this aerosol budget analysis indicates an entrainment velocity of 7 cm s −1 (computed assuming zero ultrafine aerosol particle concentrations above the mixed layer) which shows good agreement with the lidar estimate of mean vertical velocities (in the period 14:00 to 14:30 at 300 m) of −8 cm s −1 (where the negative number indicates a downwards vertical velocity).While typical entrainment velocities over water are generally less than 1 cm s −1 (Lenschow et al., 1999), larger entrainment rates are likely over land (Batchvarova and Gryning, 1994), and thus the derived entrainment velocity is reasonable, and the box model analysis appears valid.
To test the budget estimates from 17 May, a second analysis was conducted for the flight on 19 May which again was initiated just shortly after termination of nucleation.Although the aerosol particle number concentration profile in the ascent portion of the flight exhibited more variability than on 17 May (notably associated with higher number concentration variability centered at about 100 m elevation, Fig. 7b), the results are similar to those for 17 May in terms of the relative importance of the aerosol particle loss processes.For 19 May the box model analysis indicates coagulation again accounts for approximately half of the number concentration reduction (i.e. a number concentration decrease of 3300 cm −3 ), dry deposition accounts for a number concentration reduction of approximately 1600 cm −3 , and thus entrainment/dilution (i.e. the residual) accounts for a aerosol particle number concentration reduction of approximately 1900 cm −3 .However, these estimates are subject to higher uncertainty than for the 17 May because scattered precipitation was observed in the region at the time of the UAV flight on 19 May.fronts associated with transitory mid-latitude cyclones interspersed with days on which anticyclonic conditions dominated.This variability is manifest in the back trajectories that terminated at MMSF at 50 and 500 m a.g.l.However, as in Pryor et al. (2010) there is something of a tendency towards a greater prevalence of northerly flow on the days on which evidence of nucleation was observed at MMSF (Fig. 8) consistent with cold front passages from the north and northwest.

Meteorology during nucleation events
A much clearer dependence on local meteorological parameters is apparent.The principal difference between nonevent days during NIFTy and days with A and B or C events is the presence of cloud cover (as diagnosed from the ceilometer back-scatter data) particularly during the hours after sunrise.A-events were only observed on days when fractional cloud cover between 06:00 and 12:00 (LST) was below 0.3.B and C events occurred on days with less than 60% cloud cover, while the non-event days exhibited almost total cloud cover during the morning hours (Fig. 9).
Based on data from the doppler lidar, there is also a clear link between initiation of nucleation events (i.e. the start of the period when high concentrations of 6 nm aerosol particles is observed at MMSF) and erosion of the stable nocturnal boundary layer.A composite of lidar derived wind speeds, turbulence intensity and vertical velocities for all event class A days indicates a clear transition from a strongly stratified atmosphere (with substantial increase in wind speed with height), low turbulence intensity and weak vertical velocities, to much a weaker vertical gradient of wind speed, increased turbulence intensity and stronger downwards vertical velocities, consistent with growth of the mixed layer and entrainment of air from the residual layer (Fig. 10).This transition occurs on average one hour prior to the appearance of significant quantities of ultrafine aerosol particles.The strong coupling to boundary layer dynamics and growth of the mixed layer may provide a partial explanation for the apparent delay in the onset of nucleation at Bloomington and Indianapolis relative to MMSF (Table 2, see also the example from 22 May given in Fig. 6).It may be that the higher surface roughness and different energy balance of the forest means that mixed layer growth and entrainment from the residual layer is initiated earlier over the forest.

Associations between organic and inorganic gas
concentrations and ultra-fine aerosol particle composition and number concentration during nucleation events

Chemical controls on nucleation
In keeping with prior research which has indicated that if the source strength for the nucleating vapors is strong enough to compete against the losses to the pre-existing particles (Kulmala et al., 2005;Pryor et al., 2010), gas-to-particle conversion can be observed in polluted air masses (e.g. in Atlanta,  GA (Stolzenburg et al., 2005), Pittsburgh, PA (Stanier et al., 2004a), Beijing, China (Wu et al., 2007) and even Mexico City (Iida et al., 2008)), analyses of data from NIFTy do not indicate a very strong influence of condensational sink (CS) on the occurrence or intensity of nucleation (as measured using the total number concentration of sub-30nm aerosol particles) (see Fig. S2).This is also consistent with relatively high aerosol particle loadings in the region and the high regional emissions of SO 2 and NH 3 that may mean nucleation can be initiated and sustained even when the condensational sink is comparatively strong.Average NP values at MMSF are demonstrably higher on event than non-event days (Fig. S2), although there is one event class B day that was characterized by relatively low NP values and one non-event day that exhibited high NP but no evidence for enhanced ultra-fine aerosol particle concentrations.Excluding those two outliers, NP values in excess of 3×10 −23 W m molecules −1 K −1 was observed on event days, while values below that threshold characterized non-event days.
Analysis of the dimensionless nucleation parameter (L from Eq. 3) of Kuang et al. (2010) on five A-class event days indicate that the threshold of 0.7 previously proposed based on data from five prior intensive field campaigns also appears robust for MMSF (Fig. 11).Indeed, all five events during NIFTy for which we have H 2 SO 4 measurements are characterized by L <0.3.The intensity of particle production on these five days is inversely correlated with L (correlation coefficient for L vs. total aerosol particle number concentration ∼ −0.8), though the small sample size precludes detailed interpretation.
A composite of H 2 SO 4 concentrations at MMSF during event days (Fig. 12) indicate that concentrations typically increased by almost an order of magnitude upto three hours prior to the number concentration maximum, reached a maximum of 2×10 7 molecules cm −3 during the aerosol particle number concentration maximum and then subsequently declined.A qualitatively similar time profile is observed on non-event days, though the maximum  on non-event days were almost an order of magnitude lower than on event days.The time-trace of average NH 3 concentrations exhibited a quite different temporal profile.Average NH 3 concentrations exhibited a small decline (of less than 5%) during the period of increasing H 2 SO 4 and ultra-fine aerosol particle number concentrations, and then increased (by approximately 10%) during the remainder of the daylight hours.The increase in H 2 SO 4 is likely due to entrainment of an elevated polluted layer during boundary layer growth and photochemical conversion of SO 2 to H 2 SO 4 .Conversely NH 3 is emitted principally from ground sources and the diurnal profile of NH 3 is largely controlled by the diurnal profile of temperature and resultant NH 3 emissions.The small reduction in NH 3 concentrations during the period of nucleation (prior to the appearance of aerosol particles with diameters in excess of 6 nm) may either be linked to dilution during boundary-layer growth or loss of NH 3 to aerosol particles either in the process of ternary nucleation or condensational growth.Composites of the micro-meteorological parameters are consistent with interpretation of data from the lidar (see Figs. 10 and 12).Prior to nucleation events at MMSF the data indicate a transition towards more unstable conditions (as manifest in the transition of the stability index (z/L) from positive to negative values) and increased momentum fluxes, along with increased sensible heat flux.All are consistent with growth of the boundary layer in the early morning hours.The accompanying increase in solar radiation receipt would also lead to increased efficiency of H 2 SO 4 production from SO 2 .
Measurements of OH were made sporadically from 23 May through 29 May 2008 to understand the behavior and role of OH prior to and during nucleation events.Composite diurnal profiles of 30 s OH measurements for three nucleation days and two non-event days show a substantial difference in OH distribution throughout the day.On the three nucleation days, OH concentrations peaked around noon with an averaged peak concentration of ∼4×10 6 molecules cm −3 .On non-event days OH concentrations were suppressed, reaching a maximum of approximately 8×10 5 molecules cm −3 by 09:00 LST with little fluctuation, before decreasing at approximately 17:00 LST.The absence of a noontime peak in OH concentrations on nonevents days is consistent with the observed much lower solar radiation flux (incoming shortwave radiation was ≤50% of the value on event days).Measured OH concentrations were suppressed further since measurements were performed at ground level, shaded by the forest canopy.While OH concentrations were nearly an order of magnitude greater on the three nucleation days, ground level OH was not likely to have contributed significantly to the increase in measured H 2 SO 4 concentration in the lead up to the nucleation events.Sulfur dioxide measurements on event days 24 May and 25 May showed a morning peak between 08:00-09:00 (LST), at which time the average OH concentration was <2×10 6 molecules cm −3 .Complete conversion of OH to H 2 SO 4 could not account for the measured H 2 SO 4 at these times.The lack of sufficient OH at ground level to produce the H 2 SO 4 concentrations measured in early morning and H 2 SO 4 peaking around the time SO 2 peaks on these days further supports the postulate of entrainment of both species from well above the forest canopy.

Chemical controls on growth
Using hourly GR derived for 6-30 nm diameter aerosol particles from the SMPS at MMSF and the approximation that the total condensing vapor concentration is ∼1.37×10 7 cm −3 ×GR (Fiedler et al., 2005), gives a condensing vapor concentration of approximately 5.2×10 7 molecules cm −3 on 21 May 2008, and 5.0×10 7 molecules cm −3 on 22 May 2008 (see below for a discussion of the aerosol particle composition on these days).The average measured H 2 SO 4 during these events were 1.23×10 7 molecules cm −3 and 1.24×10 7 molecules cm −3 , respectively which implies the percentage contribution of H 2 SO 4 to aerosol particle growth is approximately 24%.The contribution of H 2 SO 4 to growth on three other A-class event days ranged from 23 to 85%, although these must be considered conservative estimates of the actual contribution of H 2 SO 4 to initial aerosol particle growth since the H 2 SO 4 measurements were taken below the forest canopy.Comparable estimates for approximately 20 events at Heidelberg and Hyytiälä were 4.3% and 5.9%, respectively (Fiedler et al., 2005), while another study focused only on Hyytiälä gave an average of 8.9% and a range of 4.8 to 16.9% (Boy et al., 2005).Conversely, during some days H 2 SO 4 contributed 100 % to the growth in Atlanta (Stolzenburg et al., 2005).Thus southern Indiana appears to be a mid-point on the scale of relative importance of H 2 SO 4 , and the relatively high contribution of H 2 SO 4 to aerosol particle growth at MMSF may reflect the abundance of both H 2 SO 4 and NH 3 in southern Indiana.

Mass closure
A mass closure for physical aerosol particle measurements and the size-resolved inorganic concentrations from the MOUDI and nano-MOUDI instruments was conducted to examine whether the ultra-fine aerosol particles exhibit evidence of the presence of a substantial fraction of organics.The mass concentration was derived from the SMPS and FMPS using an assumed aerosol particle density of 1.6 g cm −3 .Of the inorganic ions measured, only NH + 4 and SO 2− 4 were routinely present in the sub-32 nm aerosol particles at MMSF.Comparisons of the sum of NH + 4 and SO 2− 4 concentrations from the nano-MOUDI and MOUDI with those from the physical size instruments indicates that within the uncertainty bounds, during nucleation event days the majority of the mass of sub-32 nm aerosol particles can be accounted for by NH + 4 and SO 2− 4 (plus associated water) (Fig. 13).While this mass closure is not definitive evidence that organics are not playing a role in nucleation and initial growth at MMSF, consistent with the preceding analysis of the contribution of H 2 SO 4 to aerosol particle growth, it appears to indicate a dominant role for the inorganic gases.Beyond approximately 32 nm diameter, there is evidence for a greater role for organics.For example, for the size stages at 56 and 100 nm, the mass of inorganics contributes only half of the total mass derived from the physical aerosol particle size distributions, which implies organics may dominate in this size range, and thus organic vapors may be significantly contributing to aerosol particle growth beyond approximately 30 nm diameter.The results are thus consistent with research from other forest environments that has resolved that neutralized sulfates are responsible for the initial formation and growth of aerosol particles, but they were not able to explain subsequent growth to observed sizes of over 50 nm (Boy et al., 2005;Ristovski et al., 2010), and with the study of aerosol particle growth in Leipzig Germany which found sulfuric acid explained the majority of growth (about 75%) to approximately 20 nm but after that secondary organic compounds significantly contributed to the continued growth (Wehner et al., 2005).

Concluding remarks
Analyses of data from the spatial sampling and vertical profiles of aerosol particle concentrations in the lowest 200-300 m of the atmosphere during the NIFTy experiment indicate: -High concentrations of ultrafine aerosol particles with a diameter below 20 nm and clear growth were observed on nearly half of sampling days, and were observed at all three sampling sites (in the small city of Bloomington, expansive forest of MMSF and downwind of the urban core of Indianapolis).The observed increase of ultrafine aerosol particles at the MMSF tower site during nucleation events (of approximately 2×10 4 cm −3 for Dp≥10 nm) is representative of increases observed through a layer of at least 300 m in depth.The intensity of the nucleation events is in good agreement with recent model results from PMCAMx-UF for the Midwestern USA (Jung et al., 2010).
-There is a relatively high degree of coherence in the occurrence and intensity of new aerosol particle formation over spatial scales of at least 80 km.The very similar GR from these sites implies the nucleation events are both regional in occurrence and regionally homogeneous in terms of the subsequent aerosol particle growth, since the specific local use appears to play little role in causing site-to-site variations in GR.For example, on 22 May the GR at MMSF computed from the SMPS was approximately 4 nm hr −1 averaged over the 6 h prior to the nGMD minimum, while the comparable www.atmos-chem-phys.net/11/1641/2011/value for Bloomington was approximately 5 nm hr −1 .The condensational sinks (CS) computed for 1 h prior to the event initiation at the two sites (as computed based on data from the SMPS at Bloomington and the SMPS and FMPS at MMSF) were also very similar; 1.36×10 −2 s −1 and 1.45×10 −2 s −1 , respectively.
-On the basis of box modeling, approximately half of the aerosol particle number concentration reduction in the hour after termination of nucleation is due to coagulation, about one-quarter is due to dry deposition to the forest and one quarter is due to dilution resulting from mixed layer growth and entrainment of relatively ultrafine aerosol particle free troposphere air.
Analysis of data from in situ meteorological measurements indicates: -Nucleation is strongly influenced by cloud cover particularly during the early morning hours.A-events were only observed on days when the fractional cloud cover was less than 30%.
-Nucleation was most frequently observed in the late morning (09:00 to 12:00 LST), and is coupled to a clear transition from a strongly stratified atmosphere with low turbulence intensity and weak vertical velocities, to much a weaker vertical gradient of wind speed, increased turbulence intensity and stronger downwards vertical velocities, consistent with growth of the mixed layer and entrainment of air from the residual layer.This transition occurs on average one hour prior to the appearance of significant quantities of ultrafine aerosol particles.The strong coupling to boundary layer dynamics and growth of the mixed layer may provide a partial explanation for the apparent delay in the onset of nucleation at Bloomington and Indianapolis relative to MMSF.It may be that the higher surface roughness and different energy balance of the forest means that mixed layer growth and entrainment from the residual layer is initiated earlier over the forest.
Analysis of the aerosol particle size distributions and gasphase concentrations indicates: -Nucleation intensity is not very strongly determined by the prevailing condensational sink (CS), but there is a strong correlation between a modified version of the Nucleation Parameter (NP) from Boy and Kulmala (2002) (see Supplement).Analysis of the dimensionless nucleation parameter (L ) of Kuang et al. (2010) on five A-class event days indicates the threshold of 0.7 for nucleation occurrence (i.e.nucleation is only predicted for L <0.7) as previously proposed, appears robust for MMSF.
-Nucleation is always preceded by an increase in H 2 SO 4 concentrations but appears to be independent of NH 3 concentrations, possibly due to the high observed NH 3 concentrations.Mean morning H 2 SO 4 concentrations have a relatively strong positive relationship with ultrafine aerosol particle number concentrations (r 2 =0.5 for a power law fit, Fig. S3).A conservative estimation of the percentage contribution of H 2 SO 4 to aerosol particle growth (for sub-30 nm aerosol particles) on five event class A days ranged from 23 to 85%.
-A mass closure for physical aerosol particle measurements and the size-resolved inorganic concentrations indicates the overwhelming majority of the mass of sub-32 nm aerosol particles can be accounted for by NH +

34Figure 2 .
Figure 2. Comparison of aerosol particle number size distributions over a four hour period from the four sizing instruments deployed during NIFTy.The data derive from an instrument inter-comparison conducted at MMSF.

Fig. 2 .
Fig. 2. Comparison of aerosol particle number size distributions over a four hour period from the four sizing instruments deployed during NIFTy.The data derive from an instrument inter-comparison conducted at MMSF.

Figure 3 . 35 Figure 3 .
Figure 3.Time series of size-resolved aerosol particle number concentrations during NIFTy 2 based on data from the (a) SMPS and (b) FMPS deployed at MMSF, (c) FMPS at 3 Indianapolis, and (d) SMPS in Bloomington.Note the time periods shown differ slightly by 4 site, and the aerosol particle diameter range shown varies according to the detection limits of 5 the instrument used.6 7

Fig. 3 .
Fig. 3. Time series of size-resolved aerosol particle number concentrations during NIFTy based on data from the (a) SMPS and (b) FMPS deployed at MMSF, (c) FMPS at Indianapolis, and (d) SMPS in Bloomington.Note the time periods shown differ slightly by site, and the aerosol particle diameter range shown varies according to the detection limits of the instrument used.

Fig. 4 .
Fig. 4. Time series of size-resolved aerosol particle number concentrations (dN /dlog 10 Dp) at MMSF based on data derived from the SMPS on (a) 17 May 2008 and (b) 18 May 2008.

Figure 5 .
Figure 5. (a) Profiles of total aerosol particle concentrations (Dp ≥ 10 nm) from the CPC on board the UAV on 22 May.The legend shows the time (in LST) at which the profile was taken.The lower frames (b) and (c) show vertically discretized size-resolved aerosol particle concentrations (dN/dlog10Dp) as measured with the GRIMM on the UAV on 22 May 2008 at (b) 10:00 and (c) 11:25 (LST).

Fig. 5 .
Fig. 5. (a) Profiles of total aerosol particle concentrations (Dp≥10 nm) from the CPC on board the UAV on 22 May.The legend shows the time (in LST) at which the profile was taken.The lower frames (b) and (c) show vertically discretized size-resolved aerosol particle concentrations (dN /dlog 10 Dp) as measured with the GRIMM on the UAV on 22 May 2008 at (b) 10:00 and (c) 11:25 (LST).

Figure 7 .Fig. 7 .
Figure 7. Vertical profiles of total aerosol particle number concentrations (Dp ≥ 10 nm) (cm - 2 3 ) from the UAV on (a) 17 May 2008 and (b) 19 May 2008.The profile from the ascent flight 3 on 17 May was derived as the average profiles from the first 15 minutes of the flight which 4 took off at 13:52, while the descent profile is the average of profiles obtained 30 minutes later 5 over a 15 minute period.The profile from the ascent flight on 19 May was derived as the 6 average profiles from the first 10 minutes of the flight which took off at 12:40, while the 7 descent profile is the average of profiles obtained approximately 20 minutes later over a 10 8 minute period.In each frame the horizontal bars show the standard deviation of the 1 second 9 measurements at each height.10 11 Fig. 7. Vertical profiles of total aerosol particle number concentrations (Dp≥10 nm) (cm −3 ) from the UAV on (a) 17 May 2008 and (b) 19 May 2008.The profile from the ascent flight on 17 May was derived as the average profiles from the first 15 min of the flight which took off at 13:52, while the descent profile is the average of profiles obtained 30 min later over a 15 min period.The profile from the ascent flight on 19 May was derived as the average profiles from the first 10 min of the flight which took off at 12:40, while the descent profile is the average of profiles obtained approximately 20 min later over a 10 min period.In each frame the horizontal bars show the standard deviation of the 1 s measurements at each height.
Figure 8.The height a.g.l. of back trajectories terminating at MMSF at a height of 50 m and 500 m on event and non-event days during the NIFTy field experiment (5-31 May, 2008).These 24-hour back trajectories were computed using the HYSPLIT trajectory model (run online at http://ready.arl.noaa.gov/HYSPLIT_traj.php) with the meteorological data provided by the EDAS (North American Model (Eta) Data Assimilation System) model output at 40 km.The back-trajectory was initialized at the receptor site (MMSF) at 12:00 (LST) from a receptor height of 50 and 500 m about the ground surface.

Fig. 8 .
Fig. 8.The height a.g.l. of back trajectories terminating at MMSF at a height of 50 m and 500 m on event and non-event days during the NIFTy field experiment (5-31 May, 2008).These 24-h back trajectories were computed using the HYSPLIT trajectory model (run online at http://ready.arl.noaa.gov/HYSPLITtraj.php) with the meteorological data provided by the EDAS (North American Model (Eta) Data Assimilation System) model output at 40 km.The backtrajectory was initialized at the receptor site (MMSF) at 12:00 (LST) from a receptor height of 50 and 500 m about the ground surface.

5 Fig. 9 .
Fig. 9.Diurnal profiles of mean (and ±1 standard error) cloud cover on event days (class A and B and C) and non-event days during NIFTy.The cloud cover estimates derive from analysis of backscatter signals from the ceilometer.

42Figure 10 .
Figure 10.Composites of data collected with the Natural Power ZephIR lidar deployed at MMSF during NIFTy event class A days.The data are 10-minute average values and in each frame the time co-ordinate is given in terms of the time prior to the 10 minute period when the maximum increase in 6-10 nm aerosol particle concentrations was observed at the 46-m measurement level.The individual frames show; the (a) profile of wind speeds, (b) profile of turbulence intensity (depicted as the log 10 of the turbulence intensity) and (c) and vertical wind speeds.

Fig. 10 .
Fig. 10.Composites of data collected with the Natural Power ZephIR lidar deployed at MMSF during NIFTy event class A days.The data are 10-min average values and in each frame the time coordinate is given in terms of the time prior to the 10 min period when the maximum increase in 6-10 nm aerosol particle concentrations was observed at the 46-m measurement level.The individual frames show; the (a) profile of wind speeds, (b) profile of turbulence intensity (depicted as the log 10 of the turbulence intensity) and (c) and vertical wind speeds.

1Figure 11 .Fig. 11 .
Figure 11.Scatterplot of the maximum [H2SO4] and the dimensionless nucleation parameter 2 of Kuang et al. (2010) (L) showing data for event and non-event days (i.e.days without 3 aerosol particle nucleation) from Kuang et al. (2010), and derived from measurements on five 4 event days at MMSF during NIFTy.Note the data from Kuang et al. (2010) are based on 5-5 minute averaged, while those from MMSF are based on half-hour averages.6 7 Fig.11.Scatterplot of the maximum [H 2 SO 4 ] and the dimensionless nucleation parameter of Kuang et al. (2010) (L ) showing data for event and non-event days (i.e.days without aerosol particle nucleation) from Kuang et al. (2010), and derived from measurements on five event days at MMSF during NIFTy.Note the data from Kuang et al. (2010) are based on 5-min averaged, while those from MMSF are based on half-hour averages.
Figure 12.(a) Average NH3, SO2 and H2SO4 concentrations at MMSF during NIFTy event class A, B and C days.The data are 30-minute average values and the time co-ordinate is given in terms of the time prior to the 30 minute period with maximum 6-30 nm aerosol particle number concentrations.Also shown is the average aerosol particle number concentration (6-30 nm) and the half-hour average change in ultrafine aerosol particle concentrations.(b) Average momentum flux, a stability index (measurement height divided by the Monin-Obukhov length, z/L), incoming shortwave radiation, and the sensible heat flux for the event days (time normalized as above for the gas concentrations).

Fig. 12 .
Fig. 12.(a) Average NH 3 , SO 2 and H 2 SO 4 concentrations at MMSF during NIFTy event class A, B and C days.The data are 30-min average values and the time co-ordinate is given in terms of the time prior to the 30 min period with maximum 6-30 nm aerosol particle number concentrations.Also shown is the average aerosol particle number concentration (6-30 nm) and the half-hour average change in ultrafine aerosol particle concentrations.(b) Average momentum flux, a stability index (measurement height divided by the Monin-Obukhov length, z/L), incoming shortwave radiation, and the sensible heat flux for the event days (time normalized as above for the gas concentrations).

Fig. 13 .
Fig. 13.Mass closure for physical aerosol particle measurement from the SMPS and FMPS versus inorganic concentrations from the nano-MOUDI and MOUDI operated at MMSF during the 21 and 22 May.The uncertainty bounds shown on the sum of the NH + 4 and SO 2− 4 concentration are based on replica analysis of standards on the ion chromatograph.

Table 1 .
Instruments deployed during the NIFTy experiment conducted during May 2008.

Table 2 .
Summary of the event days based on the classification of