During Biogenic Aerosols—Effects on Clouds and Climate (BAECC), the U.S. Department of Energy’s Atmospheric Radiation Measurement (ARM) Program deployed the Second ARM Mobile Facility (AMF2) to Hyytiälä, Finland, for an 8-month intensive measurement campaign from February to September 2014. The primary research goal is to understand the role of biogenic aerosols in cloud formation. Hyytiälä is host to the Station for Measuring Ecosystem–Atmosphere Relations II (SMEAR II), one of the world’s most comprehensive surface in situ observation sites in a boreal forest environment. The station has been measuring atmospheric aerosols, biogenic emissions, and an extensive suite of parameters relevant to atmosphere–biosphere interactions continuously since 1996. Combining vertical profiles from AMF2 with surface-based in situ SMEAR II observations allows the processes at the surface to be directly related to processes occurring throughout the entire tropospheric column. Together with the inclusion of extensive surface precipitation measurements and intensive observation periods involving aircraft flights and novel radiosonde launches, the complementary observations provide a unique opportunity for investigating aerosol–cloud interactions and cloud-to-precipitation processes in a boreal environment. The BAECC dataset provides opportunities for evaluating and improving models of aerosol sources and transport, cloud microphysical processes, and boundary layer structures. In addition, numerical models are being used to bridge the gap between surface-based and tropospheric observations.
Observations obtained during an 8-month deployment of AMF2 in a boreal environment in Hyytiälä, Finland, and comprehensive in situ data from the SMEAR II station enable the characterization of biogenic aerosol, clouds, and precipitation and their interactions.
The boreal forest atmosphere provides an ideal locale to study aerosol and cloud microphysical processes. In this clean environment, the contribution of biogenic emissions and subsequent increases in secondary aerosol number (e.g., Kulmala et al. 2007, 2013) and mass (e.g., Tunved et al. 2006) enables assessment of the importance of natural aerosol in the aerosol–cloud interactions. Anthropogenic influences and signatures in aerosol particle size distributions (PSDs) are evident even at the cleanest sites (Carslaw et al. 2013). However, with a suitably long time series, analyses of aerosol–cloud interactions in such an environment can be used for assessing the role of both natural and anthropogenic emissions in global climate, especially over boreal and Arctic regions, which are experiencing drastic changes because of the ongoing climate change (Epstein et al. 2013; Koenigk et al. 2013).
The largest uncertainty in climate predictions comes from aerosol–cloud interactions (Boucher et al. 2013). A lack of comprehensive observations of cloud and precipitation systems in different climatic regimes is one of the key reasons for this uncertainty (Stephens and Kummerow 2007). Because of the complexity of different feedback mechanisms linking global climate with aerosols, trace gases, and precipitation, regional and global climate models are unable to reliably predict precipitation patterns and their changes (Rosenfeld et al. 2008; Ren et al. 2013). As a result, climatologically meaningful connections between aerosols, clouds, and precipitation have not yet been established (Stevens and Feingold 2009). An approach postulated to address this problem is the “deployment of arrays of ground-based remote sensors that can both vertically and temporally resolve the aerosol, clouds, precipitation and the meteorological state” and to document cloud and precipitation processes in different regimes (Stevens and Feingold 2009, p. 611). This holistic approach capturing the life cycles of aerosols, clouds, and precipitation was adopted for the campaign Biogenic Aerosols–Effects on Clouds and Climate (BAECC; Petäjä 2013).
During BAECC, the U.S. Department of Energy (DOE)’s Atmospheric Radiation Measurement (ARM) Program deployed their Second ARM Mobile Facility (AMF2) in Hyytiälä, Finland (61°51´N, 24°17´E), for an 8-month intensive measurement campaign from February to September 2014 (Fig. 1). Hyytiälä hosts the Station for Measuring Ecosystem–Atmosphere Relations II (SMEAR II; Hari and Kulmala 2005; see sidebar on “Hyytiälä: History and significant scientific advances”). The primary research goal of BAECC is to understand how biogenic aerosols affect cloud microphysical properties.
The history of the Hyytiälä for-estry field station goes back over 100 years. The University of Helsinki founded the station to enable forestry teaching and research in field conditions in 1910. Atmospheric research has been part of Hyytiälä station since the mid-1980s as a result of an interdisciplinary Finnish Acidification Research Programme (HAPRO; Kauppi et al. 1990) and because of the Chernobyl nuclear power plant accident in April 1986 as Hyytiälä was in the area of the radioactive fallout (Raunemaa et al. 1987; Nygren et al. 1994). These events paved the road to establishing a continuous, comprehensive measurement station dedicated to forest–atmosphere interactions. In 1995, the operational phase of SMEAR II (Hari and Kulmala 2005) was started. After 20 years of measurements and incessant development, SMEAR II features the most comprehensive atmospheric observations with the longest continuous time series, for example, on submicron aerosol number size distributions.
One of the key findings of SMEAR II has been that secondary aerosol formation from gas-phase precursors is a frequent phenomenon in the boreal environment (Mäkelä et al. 1997; see Fig. 2 for an example from the BAECC campaign), occurring on average every fourth day. This process is most active during spring, when on almost half of the days we observe the appearance of sub-3-nm aerosol particles and their subsequent growth (Dal Maso et al. 2005). A secondary maximum of the events takes place in autumn. A typical source strength associated with the events is 1 cm−3 s−1 with growth rates varying from 0.5 to 8 nm h−1 (Kulmala et al. 2004). As this process takes place in a clean environment, this relatively small source, occurring over a large area, is able to produce particles large enough to activate as cloud droplets (Kerminen et al. 2012) and contribute up to 60% of the CCN-sized particles in the region (Merikanto et al. 2009).
Once the events were discovered, the quest to find the precursor vapors, their concentrations, and detailed mechanism commenced. The pathway is dominated by neutral nucleation, and the role of ions is limited (Kulmala et al. 2007). Laboratory experiments have identified several key vapors contributing to the initial steps of aerosol formation, such as sulfuric acid (Sipilä et al. 2010), ammonia (e.g. Kirkby et al. 2011), amines (Petäjä et al. 2011; Almeida et al. 2013), and organics (Riccobono et al. 2014; Tröstl et al. 2016). All of these compounds can enhance cluster stability and initial growth and therefore contribute to the process (Kulmala et al. 2013). In continental locations, such as Hyytiälä, sulfuric acid has been identified as a crucial ingredient (Weber et al. 1995; Petäjä et al. 2009), but its concentration is not high enough to explain further growth to climatically relevant particle sizes (Kulmala et al. 1998). In the boreal environment, the role of organic vapors in the growth is important (Riipinen et al. 2012), supported by the dominance of organics in the bulk chemical composition (Jimenez et al. 2009). A very plausible candidate for the vapor responsible for the aerosol growth to CCN sizes is a group of compounds called ELVOCs that readily participates both in the early stages of clustering and the growth (Ehn et al. 2014).
The modern high-resolution mass spectrometry (e.g., Ehn et al. 2010; Smith et al. 2010; Jokinen et al. 2012; Lopez-Hilfiker et al. 2014) together with the capacity to perform physical characterization of nanoparticles below 3 nm in size (Kulmala et al. 2007, 2013b) has led to breakthroughs in understanding the initial steps of aerosol formation in the boreal environment as well as in their subsequent growth to CCN sizes (Kerminen et al. 2012). The next stage of research requires the combination of measurements (in situ and active remote sensing) and multiscale modeling tools (quantum chemical modeling, process-level modeling, and regional and global models) to resolve the vertical and spatial variability of the processes affecting aerosol–cloud–climate interactions.
To reach the holistic research goals, a comprehensive measurement approach is required. During BAECC this was provided through the complementary strengths of SMEAR II and AMF2 capabilities. SMEAR II ecophysiological measurements and in situ observations of the physical, chemical, and optical properties of aerosol particles and ions were complemented with the ground-based active remote sensing capacity of the AMF2 radar and lidar systems. Additional supplementary observations on aerosol chemical composition, aerosol precursors, nanoparticle concentrations, precipitation properties, and aerosol and water vapor vertical profiling were provided by collaborators from the United States and Europe, both for the whole observation period and during intensive observation periods (IOPs; Fig. 1). During IOPs, a total of 152 flight hours with Cessna and Skyvan aircraft provided in situ data on trace gases, aerosol particle concentration, and aerosol chemical composition. Traditional meteorological soundings (4 day−1) were enhanced with novel miniaturized solar radiation, charge, and turbulence sensors during two of the IOPs. In addition to the AMF2 radars, precipitation and cloud properties above Hyytiälä were provided by the Finnish Meteorological Institute (FMI) operational dual-polarization weather radar once every 15 min. One IOP was dedicated to documenting snowfall microphysics through a combination of multifrequency (C, X, Ka, and W band) radar, microwave radiometer (MWR), and lidar measurements supplemented by a comprehensive suite of surface-based precipitation observations. For a comprehensive list of observations, see Table SB1.
Both SMEAR II and the AMF2 Aerosol Observation System (Jefferson 2011) performed in situ measurements of aerosols during BAECC, including cloud condensation nuclei (CCN) concentrations and optical properties. Datasets collected by both facilities have been compared for quality assurance and control. Moreover, additional in situ measurements deployed in the same area allow for the assessment of small-scale spatial variability in aerosol concentration and size distribution.
On the regional scale, the representativeness of the datasets will be evaluated by the continuous and coordinated aerosol observations performed at five other SMEAR aerosol observation network sites in Finland and by the four aircraft IOPs. The existing 20-yr-long measurement data from SMEAR II enables the assessment of the representativeness of the data collected during the BAECC period. The utilization of the SMEAR observation network, Aerosols, Clouds, and Trace Gases Research Infrastructure Network (ACTRIS), and Global Atmospheric Watch data provides avenues for expanding the analysis to an even larger extent. The data from SMEAR II are available via Junninen et al. (2009) and from smartSMEAR (2016). AMF2 data are available from the ARM data archive (ARM 2015).
EMISSIONS TO AEROSOLS.
Of all biomes, the boreal forests appear to have the largest biogeophysical effect on the annual-mean global temperature (Bonan 2008). The forest is a net sink for carbon dioxide consumed in vegetation photosynthesis. Furthermore, the boreal environment is a substantial source of biogenic volatile organic compounds (BVOCs), which can affect tropospheric ozone (Atkinson and Arey 2003), and BVOCs are associated with frequent secondary aerosol formation (e.g., Kulmala et al. 2001; Dal Maso et al. 2005; Tunved et al. 2006; Kulmala et al. 2013).
The emissions of BVOCs in the boreal forest arise through evaporation from specialized storage structures (Guenther et al. 1995) and from de novo biosynthesis (Ghirardo et al. 2010). Environmental factors, such as temperature and the amount of photosynthetic active radiation, govern the emissions, but environmental stress factors, such as drought and ozone exposure, may increase or decrease emissions. There is also considerable variation between individual trees in their BVOC emission fingerprints, which, in Scots pine, are composed of a variable mixture of alpha-pinene, beta-pinene, delta-carene, and sesquiterpenes (e.g., Hakola et al. 2006; Bäck et al. 2012). Additionally, new needle development and growth processes can be large sources of BVOCs (e.g., Aalto et al. 2015) associated with spring recovery of the ecosystem (Dal Maso et al. 2009). Approximately 80% of BVOC emissions in a conifer forest stand originate from trees, with the rest from the soil (Aaltonen et al. 2011), and both source strengths were measured with automated cuvettes during the BAECC campaign (Table 1).
The winter temperatures during the BAECC period were abnormally mild, with photosynthesis (net carbon uptake) persisting even in February, leading to increased emissions of BVOCs. This indicated that the potential for biogenic production of precursor gases was high during the spring months. Monoterpenes are the main emitted compound group from coniferous forests around Hyytiälä (Hakola et al. 2006). The shoot-scale enclosure measurements with proton transfer mass spectrometry revealed that, as in many previous years, extremely high emission rates in early spring coincided with the spring recovery period of trees during early–mid-March and that these peaks provide 2–3 times higher monoterpene emissions compared to the rest of spring (Aalto et al. 2015). The maximum emission rates were observed during 11–12 March 2014, with another high-emission period seen during new foliage growth in late May–early June (Aalto et al. 2014). During summer, the highest emissions coincided with high temperatures in early and late summer, whereas the relatively cool weeks around midsummer resulted in lower monoterpene emissions.
The emissions from the biosphere together with their oxidation products lead to new particle formation (Kulmala et al. 2004, 2013). This process provides a major source of particles in clean boreal forests (Fig. 2; Kulmala et al. 2001). Although many precursors (sulfuric acid, amines, and various organic vapors) have been attributed to initial gas-to-particle conversion, the dominant process growing these clusters to climatically relevant sizes is, without doubt, condensation of organic vapors (Kulmala et al. 1998; Riipinen et al. 2011; Zhou et al. 2014). This fact is also underlined by the typical chemical composition observed, which indicates that the majority of particulate mass in natural boreal forests is made up of organic compounds (Jimenez et al. 2009).
Aerosol formation occurred frequently during BAECC (Fig. 2). The concentration of nanoparticles was probed with a suite of scanning particle size magnifiers (PSM; Vanhanen et al. 2011) and the mobile versatile size-analyzing nuclei counter (vSANC; Pinterich et al. 2016) in the size range 1–5 nm at ground level and at 35-m height. Overall, the seasonal cycle of events during BAECC was typical for Hyytiälä with a maximum in spring and a secondary maximum in autumn. The separation of ion-induced and neutral pathways was determined through measurements of both ions and neutral particle concentrations, with the neutral pathway dominating formation rates. Figure 2 displays an example of a new particle formation event starting from sub-3-nm sizes during the morning hours of 23 April. During subsequent days, as the synoptic situation remained unchanged, these particles grew larger in size, leading to a factor of 10 increase in the concentration of CCN-sized particles.
Typically, higher concentrations of precursor vapors enhance aerosol growth (Kulmala et al. 2001; Riipinen et al. 2012). There are indications that some of the organic molecules are also relevant for the initial steps of aerosol formation (Riccobono et al. 2014; Schobesberger et al. 2013a; Kirkby et al. 2016; Tröstl et al. 2016). During BAECC concentrations of condensable vapors, such as sulfuric acid (Petäjä et al. 2009) and extremely low volatile organic compounds (ELVOC; Ehn et al. 2014), were measured with a suite of mass spectrometers. To contrast observations with other available long-term data, proxy variables were developed to capture the temporal variability of the relevant vapor concentrations (Petäjä et al. 2009; Mikkonen et al. 2011). Atmospheric oxidation of the aerosol precursors was probed by measuring atmospheric oxidants, such as ozone, HOx + ROx (R. L. Mauldin et al. 2015, unpublished manuscript), and stabilized Criegee intermediates (Taipale et al. 2014) during the BAECC spring IOP (Table 1).
Based on a Filter Inlet for Gases and Aerosols chemical-ionization mass spectrometer (FIGAERO-CIMS (Lopez-Hilfiker et al. 2014) and thermal desorption CIMS (TDCIMS; Smith et al. 2004) measurements, contributions to particulate mass were deduced from the partitioning of oxidation products from major biogenic VOC precursors such as monoterpenes, isoprene and/or methyl butene-1-ol, and either gas- or particle-phase accretion products of these species. Nitrogen-containing organics were also observed and may be a tracer of anthropogenic NOx influence or may arise from nighttime chemistry involving the nitrate radical. Monoterpene concentrations were quite high (Fig. 2), and FIGAERO data show that monoterpene oxidation products dominated particulate mass during spring. TDCIMS data during spring showed that recently formed nucleation-mode particles contained higher fractions of oxidized hydrocarbons than older particles, supporting the link between highly oxidized gas-phase compounds and new particle formation. However, as spring progressed, FIGAERO observations showed a noticeable enhancement in the contribution of lower molecular weight compounds containing four to five carbon atoms, likely due to the onset of isoprene emissions with the leaf out of broad-leaved trees and increasing temperatures (Yassaa et al. 2012). Overall, a consistent composition appeared, with large contributions from monoterpene oxidation products, dimers, and nitrogen-containing organics, suggesting a role for anthropogenic influence even in this rural site. An important caveat to the last point is that natural soil emissions of NOx or amines may constitute the dominant source of nitrogen-containing organics at this site. This aspect will be assessed as our analysis progresses.
Bulk aerosol chemical composition measurements in the submicron size range with aerosol chemical composition were performed using an Aerosol Chemical Speciation Monitor (ACSM, Aerodyne Research Inc.; Ng et al. 2011). Data coverage during the BAECC campaign was extensive, with a good dataset obtained for studying the seasonality of the aerosol-phase composition. Increased organic contributions during the most photosynthetically active seasons was clearly observed, whereas in wintertime, anthropogenic sulfate reached its maximum, composing up to 25% of the nonrefractory submicron mass.
EXTENDING SURFACE MEASUREMENTS INTO THE VERTICAL.
Based on direct observations inside clouds, Kerminen et al. (2005) showed that secondary aerosols formed from biogenic emissions can be activated to become cloud droplets. However, unraveling the mechanisms by which biogenic aerosol interacts with clouds requires that the vertical profile of aerosol, clouds, and turbulence be fully characterized in tandem. The process can be split into two main steps: the transport of biogenic aerosol from the surface into the boundary layer and free troposphere and the participation in cloud formation.
The vertical profile of aerosol above the SMEAR II site was studied extensively through the use of remote sensing by lidar and by in situ aircraft-based measurements during three flight IOPs. Continuous remotely sensed vertical profiles of aerosol were obtained from a multitude of lidar instruments. These operated quasi continuously throughout the entire campaign period. For the instrumentation and timing of IOPs, the reader is referred to Table 1.
The transport of biogenic aerosol from the surface into the boundary layer is driven by turbulent mixing. This mixing ensures that all air within the turbulent layer is in intermittent contact with the surface, on time scales from 10 to 30 minutes, with the top of this layer termed the mixing-layer height (MLH; e.g., Emeis et al. 2008). The MLH can be defined in terms of the turbulent kinetic energy dissipation rate (Barlow et al. 2011), determined directly from the high-resolution vertically pointing (O’Connor et al. 2010) and scanning (Vakkari et al. 2015) Doppler lidar data provided by FMI at Hyytiälä. The potential for aerosol-layer identification and aerosol typing through a combination of backscatter coefficient and circular depolarization ratio is clearly shown in Figs. 3 and 4. Humid boundary layers, dry elevated layers, and humid elevated layers can all be distinguished in high-spectral-resolution lidar (HSRL) data.
The flight IOPs took place in three seasons: early spring, beginning of summer, and beginning of autumn. The University of Helsinki (UHEL) operated a Cessna FR172F single-engine light aircraft modified to carry aerosol instrumentation as described in Schobesberger et al. (2013b). Figure 3 shows simultaneous measurements with Cessna and ARM HSRL on 2 April 2014. The flight path for this particular day is typical of the route selected during the flight campaigns. It is oriented in the south–north direction and consists of an initial ascent up to 3.5 km and then several legs at different altitudes above and within the boundary layer. In total, 144 flight hours were flown during 33 days with the Cessna, and all flights were in the vicinity of the SMEAR II station. During the flight IOP at the beginning of autumn, FMI operated a Short SC.7 Skyvan, an unpressurized aircraft (Short Brothers and Harland Ltd., Northern Ireland, United Kingdom) owned by Aalto University. Inclusion of chemical composition profiles from Skyvan flights (Fig. 4) permits the correct choice of refractive index when deriving microphysical properties from the lidar profiles. The details of the instrumentation in both aircraft are presented in Table 1.
The microphysical retrievals can be improved through harnessing data from the multiwavelength Raman lidar system PollyXT (Althausen et al. 2009; Engelmann et al. 2016), where available, through collaboration with the European Aerosol Research Lidar Network (EARLINET; Pappalardo et al. 2014), now within ACTRIS. The combination of multiple channels through the use of the “3 backscatter + 2 extinction + 1 depolarization” approach allows aerosol typing (e.g., Mona et al. 2012; Mattis et al. 2004) and the retrieval of aerosol microphysical optical properties through the application of specific inversion algorithms (Müller et al. 1999; Veselovskii et al. 2005). The retrievals have been demonstrated to have the unique ability of providing range-resolved aerosol effective radius and a complex refractive index (e.g., Müller et al. 2007). The addition of sun photometer observations may permit the determination of the aerosol mass concentration profile, including separation of the fine and course components (Chaikovsky et al. 2016; Lopatin et al. 2013; Binietoglou et al. 2015). The range-resolved properties (e.g., aerosol size distribution, refractive index, single-scattering albedo, mass concentration, and water vapor) can then be independently verified at the surface and from the aircraft. The impact of humidity on aerosol depolarization ratios retrieved from lidar can also be examined in detail, with Fig. 4 showing how closely these two parameters can be linked (low depolarization values imply spherical hygroscopic aerosol, high depolarization values imply dry nonspherical aerosol).
Such a dataset will help identify the relative impact of long-range transport of material and local sources, and we will also investigate the mechanisms for dispersion. Besides confirming the remote sensing methods, the in situ data obtained by the aircraft can be used either as input or validating data for different atmospheric models, such as the model to simulate the concentrations of organic vapors, sulfuric acid, and aerosols (SOSAA; Boy et al. 2011). The inclusion of chemical composition (Fig. 4) and Picarro trace gas profiles from Skyvan provides a direct measurement of the impact of the environment on the aerosol properties during their transport from the surface into the boundary layer.
CLOUD VERTICAL PROFILING.
Well-established ARM and Cloudnet (Illingworth et al. 2007) algorithms were used to derive vertical profiles of cloud properties for the entire campaign. The vertical profile of cloud macrophysical properties, including layer boundaries and phase, was obtained from a combination of vertically pointing Doppler cloud radar [Ka-band ARM zenith radar (KAZR) or marine W-band ARM cloud radar (MWACR) when KAZR is unavailable] and lidar (HSRL and ceilometer), following Illingworth et al. (2007). Combinations of various radar wavelengths (including scanning instruments operating at vertical) together with lidars and microwave radiometers were then used to retrieve the cloud microphysical properties, such as water content and flux, size distributions, and ice morphology. Drizzle properties, including median equivolumetric size, number concentration, drizzle liquid water content, and drizzle liquid water flux, were derived using a combination of KAZR (or MWACR) and a ceilometer (O’Connor et al. 2005).
Investigating the influence of biogenic aerosols on clouds requires that both the aerosol and cloud properties can be measured reliably and within the same volume. This poses a challenge since it is difficult to measure aerosol properties within the cloud by remote sensing methods. However, when considering liquid clouds coupled with the boundary layer, it is safe to assume that the vertical column within and below cloud is well mixed, producing well-defined profiles and gradients of atmospheric properties. In a well-mixed column, in-cloud properties can then be related to below-cloud aerosol properties and, in principle, to aerosol properties (such as size distribution, chemical composition, and hygroscopicity) measured at the surface. Figures 3 and 4 both show suitable clouds at the top of the boundary layer that are coupled to the surface (in Fig. 3, suitable clouds are at 1.5–2 km in altitude from 1000 UTC to just after 1200 UTC; in Fig. 4, the intermittent clouds are present all day between 1 and 1.5 km). Liquid clouds in the boundary layer will be classified according to their aerosol source, biogenic, background, or other aerosol source, once these data are available and stratified by their macro- and microphysical properties to document evidence for the influence of biogenic aerosols on clouds.
CLOUDS TO PRECIPITATION.
In Finland, a high-latitude country, the majority of precipitation is initiated by ice-phase processes (Mason 1971). Since part of the BAECC campaign took place during winter, surface-based observations of solid precipitation microphysics, that is, particle size distribution, particle habit, and their physical properties, can be used to infer what cloud-to-precipitation processes take place and guide retrievals based on remote sensing observations. To take advantage of this opportunity, an intensive observation period termed the BAECC Snowfall Experiment (SNEX), focusing on snowfall, was undertaken from 1 February through to 30 April 2014.
During this IOP, more than 20 snowfall events were recorded. During these events, the dual-channel microwave radiometer detected the presence of liquid water more than 80% of the time. Given the presence of supercooled liquid in the majority of the cases, the focus of the IOP was to investigate how remote sensing observations, that is, multifrequency and dual-polarization Doppler radar in combination with lidar and microwave radiometer, can be used to identify ice-phase precipitation processes, namely, the Bergeron process, riming, aggregation, and ice multiplication, and study their evolution. Because a transition from one process to another could happen within a precipitation event, our analysis and data collection strategies were selected to allow investigation of how the environmental conditions influence these transitions, even if they happen on time scales as short as a few minutes.
Accurate and consistent quantitative observations of snow are notoriously challenging (Rasmussen et al. 2012). To ensure the quality of the measurements and to obtain an estimate of uncertainty of retrieved microphysical properties, a comprehensive precipitation measurement setup was established (see sidebar on “BAECC SNEX” for more information). The observations of precipitation intensity and particle size distributions, particle fall velocities and sizes, particle dimensions and shapes in combination with retrievals of snow bulk density, and mass-dimensional relations can be used for classification and characterization of precipitating ice particles and inferences of prevailing precipitation processes. These observations and retrievals were carried out in coordination with radar observations to link surface observations of precipitation microphysics to vertical profiles of meteorological products, as, for example, shown in Kneifel et al. (2015). An example of this analysis applied to a precipitation event that took place on 21 February 2014 is shown in Fig. 5. Furthermore, in the sidebar “Modeling activities for investigating aerosol transport, source attribution, and cloud-to-precipitation processes,” an example of how a numerical model—the Weather Research and Forecasting (WRF) Model—can be used to further understand how microphyiscal processes occurring in a vertical column affect the properties of precipitation at the surface is presented.
The snowfall measurement experiment (BAECC SNEX; Fig. SB1) was a collaborative effort between DOE ARM, University of Helsinki, FMI, NASA, and Colorado State University. The IOP took place from 1 February to 30 April 2014 and was dedicated to documenting snowfall microphysics through a combination of multifrequency (C, X, Ka, and W band) radar, microwave radiometer, and lidar measurements supplemented by a comprehensive suite of surface-based precipitation observations. Combining the multi-instrumental remote sensing and ground-based observations provides a detailed view of snow growth processes, such as condensation growth of ice crystals, aggregation, and riming.
The standard AMF2 surface-based precipitation measurement instruments were supplemented by an array of sensors, given in Table SB1. The operational schedule of the nearest FMI dual-polarization weather radar was modified to include range–height indicator (RHI) scans above the AMF2 location.
To facilitate accurate surface measurements of snowfall properties, a DFIR wind protection (shown in Fig. SB1) for the following instruments was built on site: weighing precipitation gauge, laser disdrometer (OTT Parsivel), and 2D video disdrometer. Because of the duplication of the instruments, consistency of the precipitation microphysics retrievals can be checked and the dataset can also be used to characterize snowfall measurement errors as a function of wind speed. The wind measurements were performed at instrument height inside and outside of the fence and at 10-m height.
During the IOP, more than 20 precipitation events were recorded, with conditions varying from dry to wet snow and particle types ranging from pristine crystals to densely rimed particles and fluffy aggregates.
By combining the detailed observations obtained during BAECC with a range of numerical models, we ultimately aim to understand the role of biogenic aerosols in cloud formation and microphysical processes leading to precipitation. Here, one example is presented in which surface and remote sensing observations are combined with a numerical simulation conducted with the WRF Model, version 3.6.1 (Skamarock et al. 2008).
During BAECC SNEX (see "Clouds to precipitation" section and the "BAECC SNEX" sidebar), six cases of layered cloud that resulted in enhanced surface precipitation occurred. On 21–22 February, a frontal system that lead to multiple cloud layers and snowfall moved east across Finland. This case is analyzed in detail using WRF Model output and observations (some of which are presented in Fig. 5) to determine 1) what microphysical processes are occurring aloft and how these relate to the type and intensity of precipitation observed at the surface and 2) how accurately multiple mixed-phase cloud layers and their microphysical processes are represented by WRF.
The WRF simulation was initialized from the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim) data (Dee et al. 2011) at 1200 UTC 20 February 2014 and run for 72 h. The simulation domain consisted of an outer domain with 30-km grid spacing covering most of northern Europe and an inner nested domain with 10-km grid spacing covering Finland and neighboring sea areas. Boundary layer turbulence was parameterized using the Yonsei University (YSU) scheme, deep convection was parameterized using the Kain–Fritsch scheme, and the Morrison double-moment six-class scheme was used to parameterize the microphysics. The WRF simulation did not capture the thin layer of boundary layer cloud that was present before the arrival of the front, which is, in part, due to model spinup issues, but in general such low-level clouds are poorly represented in numerical weather prediction (NWP) and climate models (Stevens et al. 2007). Associated with the frontal passage were two cloud layers, both of which were captured well by WRF (Figs. SB2 and 5).
The observed radar reflectivity (Fig. 5a) shows that between 2300 and 0000 UTC, precipitation from the upper-cloud layer falls into the lower-cloud layer. The same feature is present in the WRF simulation; however, it occurs 1.5 h later, between 0030 and 0300 UTC (Fig. SB2). At the upper levels, where the temperature is below −40°C, there are large numbers of ice particles (Fig. SB2). These primary ice and snow particles produced in the upper-level seeder cloud fall into the lower feeder cloud and grow by accretion. Observations and model simulations of liquid and ice water path (Fig. SB3) support this as almost all liquid water is removed with the onset of the precipitation falling from the upper cloud into the lower cloud and there is an increase in ice water path (at 2300 UTC in the observations and 0030 UTC in the model simulation). In addition, new ice is generated at the top of the seeder cloud (at temperatures from −20° to −15°C) and hence would be plates or dendrites. This results in two distinct populations that have different fall speeds and therefore aggregation can readily occur. Observations (Fig. 5) show that after 2300 UTC there are more large particles and that the bulk density of snow at the surface decreases, indicative of aggregates. Thus, it can be concluded that a seeder–feeder mechanism occurs and alters the properties of the surface precipitation.
Also of interest in this case is the high concentration of ice particles simulated by WRF between 1930 and 2130 UTC at temperatures between −8° and −3°C, the temperature range where the Hallett–Mossop ice splintering processes is expected to occur (Hallett and Mossop 1974). However, in this case it is unlikely that this ice is produced by this process as there is little or no supercooled liquid present. This potential case of secondary ice production, and others observed during BAECC SNEX, will be the topic of future studies.
Overall, WRF reproduces the bulk aspects of this frontal case very well. There is reasonable agreement in the cumulative precipitation when the entire frontal system is considered, yet WRF simulates the precipitation to be lighter and to occur over a longer period of time than what was observed (Fig. SB3). Furthermore, notable discrepancies between the measured and simulated number concentration of ice/snow particles at the surface were detected (Fig. SB3) as WRF was not able to capture the increase in frozen hydrometeor number that occurred between 2300 and 0000 UTC. This highlights the challenges models face in correctly simulating precipitation amounts and properties of hydrometeors and indicates that microphysical processes are not yet adequately represented in numerical weather prediction or climate models.
During BAECC SNEX, surface precipitation observations were performed by a number of instruments. Given the duplication of most of the precipitation instruments and utilization of instruments with different measurement principles, the consistency of the retrieved snow microphysical properties can be checked. For example, precipitation intensity is measured by two weighing gauges [inside and outside of the Double Fence Intercomparison Reference (DFIR)] and by a total precipitation sensor (Hotplate). Those instruments supplement standard AMF instrumentation. Particle size distributions were recorded by two OTT Parsivel laser disdrometers, two 2D video disdrometers, and a particle imaging package (PIP) designed by the National Aeronautics and Space Administration (NASA), which is the new generation of snow video imager, providing also images of individual hydrometeors (Newman et al. 2009).
Similarly, retrieval techniques that rely on different sets of assumptions were adopted. To characterize precipitating ice particles, bulk density of falling snow, velocity–dimensional (υ–D) relations, areal ratios, and mass–dimensional (m–D) relations must be derived. The bulk density can, for example, be estimated using several methods:
by matching 2D video and/or PIP-based estimates of snowfall rate to the value recorded by the gauges, similar to the techniques described by Brandes et al. (2007);
estimating the bulk density of freshly fallen snow by comparing changes in snow depth to the precipitation accumulation (Power et al. 1964); and
by matching reflectivity values calculated from PSD observations with directly observed reflectivity values (Huang et al. 2010).
An example of retrieved bulk density values from PIP and weighing-gauge measurements is shown in Fig. 5. By combining observed particle area ratios and υ–D relations, the mass-dimensional relations were estimated by following the procedure described by Böhm (1989) and Huang et al. (2015); see Fig. 5 for the bulk density calculated from the retrieved m–D relation. As one can see from Fig. 5, the bulk densities retrieved using two different sets of instruments and methods match rather well, which supports our confidence that such retrievals are possible and meaningful.
SUMMARY AND OUTLOOK.
During BAECC, the physical, chemical, and optical characterization of aerosol particles at the surface was conducted simultaneously with comprehensive cloud and precipitation observations. This was enabled by combining the state-of-the-art capabilities of both the SMEAR II research station at Hyytiälä, Finland, and the AMF2, a mobile research facility, which was brought to Hyytiälä by the ARM Program of the U.S. Department of Energy. This facilitated good opportunities to benefit from the NASA Global Precipitation Measurement (GPM) mission ground validation in surface particle size distribution and water equivalent rate gauges, while the European Commission via ACTRIS transnational access provided resources for gap-filling aerosol physical and chemical measurements as well as cloud observations.
Overall, the BAECC campaign was highly successful and provided a vast 8-month-long dataset obtained during three seasons, including a variety of cloud and precipitation systems. Together with the extensive flight campaigns, and with the comprehensive near-20-yr-long continuous surface-based measurements of aerosols at SMEAR II, the representativeness of the BAECC dataset can be thoroughly evaluated.
The connection between increased aerosol concentration of aerosol particles and cloud properties has not yet been properly quantified because of the challenges connecting in situ aerosol measurements at the ground and cloud properties aloft (e.g., Paasonen et al. 2013 ; Kulmala et al. 2014). With vertical profile measurements during BAECC and suitable modeling tools, the quantification of these connections can now be undertaken. The comprehensive observations available in BAECC enable radiative transfer modeling to estimate the effects of aerosols and clouds on the radiative (e.g. Zieger et al. 2015) energy balance of the entire atmospheric column. With the combination of the measurements obtained during BAECC and the supporting SMEAR network, combined with satellite data, we will be able to evaluate these correlations on a larger spatial scale in the boreal environment.
Expansion of the pointwise measurements to a larger context with satellite-derived data is ongoing (Sporre et al. 2016; Krüger et al. 2016, manuscript submitted to J. Geophys. Res.). The data will enable development of proxy variables that can be expanded into a global perspective (e.g., Kulmala et al. 2011).
The comprehensive data will be utilized in multiscale modeling. For example, the atmosphere column model SOSAA (Boy et al. 2011), combining different emission modules, boundary layer dynamics, and both chemical and aerosol dynamical processes, will be used to investigate the formation, vertical transport, and aging of atmospheric aerosols inside the mixing layer. The model simulations will link our knowledge based on long-term ground observation with the new measurements by the AMF instruments. Valuable sensitivity studies with the Weather Research and Forecasting Model with Chemistry (WRF-Chemistry) with different cloud microphysics parameterizations are possible where the models can be verified against multifrequency radar observations available from AMF2.
Future work will take full advantage of the success of the BAECC campaign and contribute toward the following objectives: 1) to study the interactions between aerosol microphysics and turbulent mixing within the continental boundary layer to understand the transformation of aerosol while being transported from the surface to clouds, 2) to evaluate the impacts of long-range transport and transformation of aerosol on how they can act as CCN, 3) through comprehensive aerosol typing with the various observation methods to determine the roles of the regional and long-range transported aerosol in the formation of CCN-sized particles, 4) to investigate the sensitivity of evolving cloud–precipitation processes on CCN concentrations, 5) to characterize aerosol removal and transformation mechanisms as a function of particle size and precipitation type, and 6) to compare and understand falling-snow precipitation processes and rates to remote sensing retrievals from satellite missions such as GPM.
Finally, the representativeness of the BAECC data will be understood within a long-term perspective through connection to the nearly 20-yr continuous measurement dataset at SMEAR II, unique globally in the sense that nowhere else has comparable measurements that have been performed continuously for such a long time period.
This work was partly supported by the Office of Science (BER), U.S. Department of Energy via BAECC (Petäjä), BAECC SNEX (Moisseev), DE-SC0011791 (Thornton), DE-SC0014469 (Smith), and ASR (Chandrasekar); European Commission via projects ACTRIS, ACTRIS-TNA, ACTRIS2, BACCHUS, and PEGASOS and European Research Council via NANODYNAMITE (616075); and Academy of Finland Centre of Excellence (Project 272041), KONE foundation (Grant 46-6817), Nordforsk via Cryosphere–Atmosphere Interactions in a Changing Arctic Climate, CRAICC, TEKES in the CLEEN/MMEA programme, and the Ministry of Transport and Communication through ICOS-Finland. The BAECC SNEX was also supported by NASA Global Precipitation Measurement (GPM) mission ground validation program. The deployment of AMF2 to Hyytiälä was enabled and supported by ARM. Argonne National Laboratory’s work was supported by the U.S. Department of Energy, Assistant Secretary for Environmental Management, Office of Science and Technology, under Contract DE-AC02-06CH11357. The authors gratefully acknowledge the support of AMF2, SMEAR II, and the BAECC community for their support in initiating the BAECC campaign, its implementation, operation, and data analysis.