Abstract

During the Cloud and Aerosol Characterization Experiment (CLACE) 2013 field campaign at the High Altitude Research Station Jungfraujoch, Switzerland, optically thin pure ice clouds and ice crystal precipitation were measured using holographic and other in situ particle instruments. For cloud particles, particle images, positions in space, concentrations, and size distributions were obtained, allowing one to extract size distributions classified by ice crystal habit. Small ice crystals occurring under conditions with a vertically thin cloud layer above and a stratocumulus layer approximately 1 km below exhibit similar properties in size and crystal habits as Antarctic/Arctic diamond dust. Also, ice crystal precipitation stemming from midlevel clouds subsequent to the diamond dust event was observed with a larger fraction of ice crystal aggregates when compared with the diamond dust. In another event, particle size distributions could be derived from mostly irregular ice crystals and aggregates, which likely originated from surface processes. These particles show a high spatial and temporal variability, and it is noted that size and habit distributions have only a weak dependence on the particle number concentration. Larger ice crystal aggregates and rosette shapes of some hundred microns in maximum dimension could be sampled as a precipitating cirrostratus cloud passed the site. The individual size distributions for each habit agree well with lognormal distributions. Fitted parameters to the size distributions are presented along with the area-derived ice water content, and the size distributions are compared with other measurements of pure ice clouds made in the Arctic and Antarctic.

1. Introduction

Ice crystal precipitation in various shapes, intensities, and sizes is observed in the cold regions of Earth. In particular, ice crystals under a millimeter in size exhibit strong radiative effects in the thermal infrared as well as in the solar spectrum, and their net effect on the radiative budget depends on their number concentration, size, shape, and altitude (Wendisch et al. 2007; Ehrlich et al. 2008).

In the absence of supercooled liquid water, the growth of ice crystals to precipitation size is most likely dominated by aggregation of smaller ice crystals, which depends on the ice crystal number concentration and temperature (Hobbs et al. 1974). In the polar regions and also at high altitudes, ice crystal precipitation may form in clear, cloudless air, which is referred to as “diamond dust.” Several studies investigated the microphysical and also chemical properties of diamond dust in Antarctica (Hogan 1975; Ohtake 1978; Ohtake and Yogi 1979; Walden et al. 2003; Lawson et al. 2006a) and also in the Arctic region (e.g., Ohtake et al. 1982; Intrieri and Shupe 2004; Domine et al. 2011; Gultepe et al. 2014, 2015). They agree that these small ice crystals tend to form in a layer close to the ground, which is usually a few hundred meters deep and colder than −10°C.

The formation mechanism of diamond dust in the Arctic or Antarctic region is mostly driven by radiative cooling as described in Gotaas and Benson (1965) and Curry et al. (1996). In the high Alpine region, which is the subject of our study, meteorological conditions are of course different from those observed in Antarctica or the Arctic regions. Relative to Arctic/Antarctic locations where ice fog dominates, the vertical wind velocity on the upslope side of the Jungfraujoch in the Swiss Alps is significantly higher, and temperatures at approximately 3600 m MSL in the midlatitudes are—in contrast to the conditions during the polar winter—seldom lower than −30°C. The minimum temperature at Jungfraujoch during the measurement period of this work was slightly above −30°C.

Several studies, for example, Auer et al. (1969), Cooper and Vali (1981), and Rauber (1981), investigated the ice crystal size distributions and the ice crystal origin in orographic clouds. Auer et al. (1969) found a large deficit of ice nuclei when compared with the ice crystal number concentration, which is approximately a factor of 1000 for temperatures above −10°C and a factor of 10–100 for temperatures below −20°C. While Auer et al. (1969) suggested an ice multiplication process with a time constant below 300 s, Cooper and Vali (1981) state that this discrepancy can be explained by condensation freezing and contact freezing of supercooled cloud droplets. Another finding from Fig. 14 of Cooper and Vali (1981) is the presence of pristine short columns/thick plates in an environment with supercooled cloud droplets. The absence of rimed particles is therefore insufficient to argue for the absence of supercooled liquid water.

One finding from Rauber (1981) is the presence of small (6–9 μm) droplets at a high (100–200 cm−3) number concentration if the air mass lifted by the orography has continental characteristics. In their study, the ice crystal size distributions measured with 2D cloud (2D-C) probes on two aircraft tend to appear monomodal when supercooled droplets are observed. In the absence of cloud droplets, the ice crystal size distributions become bimodal, with a first mode around 150 μm and a second mode around 1 mm at an average number concentration between 5 and 20 L−1.

However, in a mountainous environment, it is very common that small snow or hoarfrost particles are lifted up by the wind, reach altitudes up to a kilometer above the surface, and may reside in the air for several hours (Vali et al. 2012). These surface-generated particles may act as seeds for the formation of ice crystals in clouds containing supercooled liquid water in an alpine environment. Ice crystals are the ideal ice nuclei and may start the formation of cloud ice, which is then accelerated by the Wegener–Bergeron–Findeisen process (Korolev 2007). This effect might explain why, in those clouds, there are far more ice crystals observed than ice nuclei (Vali et al. 2012). On the one hand, these surface-generated ice crystals can be considered as artifacts as they appear as a background in any other case of ice crystals observed at mountain sites (Rogers and Vali 1987). On the other hand, they play an important role in the local cloud microphysics and need to be understood in more detail (Vali et al. 2012; Lloyd et al. 2015).

This work differs from other ground-based measurements of ice crystal precipitation as the orographic influence is significant, and the holographic technique allows us to investigate the spatial distribution of hydrometeors on the centimeter scale. However, conditions are also different from those observed in the other orographic cloud systems, for example, Cooper and Vali (1981) and Rauber (1981).

2. Previous studies on ice crystal precipitation in the Arctic/Antarctic region

There are numerous studies of ice crystal precipitation over Antarctica and also over the Arctic region as thermodynamic conditions often allow the formation of ice particles in clear air. In the high mountains, temperatures in winter might get low enough for deposition freezing and ice supersaturated conditions are often observed, either produced by an upslope flow via adiabatic cooling or caused by radiative cooling, probably in the evening and night hours.

Measurements of small ice crystals in pure ice conditions on mountain tops or in the polar regions require both a large sample volume and the ability to measure the particle size with an accuracy of a few microns. A very common way to present microphysical properties of such ice crystals is photomicrography from Formvar replicas. These replicas allow at least a coarse representation of the particle size distribution and habit distribution as they often have poor counting statistics, a low spatial resolution, and no time information. A continuous replicator would be needed to capture particle size distributions versus time. Size ranges from Hogan (1975), Ohtake (1978), Walden et al. (2003), and Domine et al. (2011) could be derived from single photomicrographs by measuring the maximum dimension of each ice crystal. The particle size varies between 10 and 550 μm.

The best quantitative dataset of diamond dust in Antarctica derived from several thousand images from a Stratton Park Engineering Company (SPEC) cloud particle imager (CPI) is found in Lawson et al. (2006a). Similar to Hogan (1975), vertically thin ice clouds were present above the site during the precipitation event. Even though a large fraction of ice crystals is found to be pristine columns and plates, a significant fraction of irregularly shaped particles was detected as well. Ice fog/diamond dust particles usually form via the deposition nucleation process (Gultepe et al. 2014).

Ohtake et al. (1982) obtained airborne measurements of ice crystals from Arctic diamond dust in several layers up to 1 km above the ground near Barrow, Alaska, with a Particle Measurement Systems (PMS) optical array probe, version 1 (OAP-1). At a flight altitude of 900 m above the ground, they measured a particle size distribution with a mode diameter between 50 and 100 μm at a number concentration around 40 L−1. Toward lower altitudes, the number concentration was reduced by a factor of 10, and a secondary mode between 200 and 300 μm appeared. This relatively large diamond dust originated from frozen cloud droplets of stratus fractus, which had developed over open leads. Such an additional supply of water vapor may be responsible for the larger ice crystals in the Arctic when compared with the Antarctic cases found in the literature.

Farther inland, the ice crystal size distributions tend to exhibit a smaller mean particle size but an enhanced number concentration, as observed by Gultepe et al. (2015). With number concentrations around 1–10 L−1, there is little or no effect on visibility. The ice fog events with less than 1 km of visibility shown in Gultepe et al. (2015) occurred at number concentrations above 1 cm−3. In an environment with only marginal orographic effects, ice fog/diamond dust tends to form a layer between the ground and the top of the boundary layer, which is rather homogeneous on the horizontal scale and shows a maximum in the lidar/ceilometer backscatter intensity close to the top of the layer. This is consistent with the vertical profile of the diamond dust observed by Ohtake et al. (1982) as they measured a high concentration of small particles at the top of the diamond dust layer.

Lawson et al. (2006a) also investigated a blowing snow event and precipitation from a deep cloud above the South Pole Station. The blowing snow particles are found to be smaller than the diamond dust and also less abundant. However, it is difficult to compare these results with observations in a mountainous environment as the surface properties of the snow layer and the local wind speed determine size and concentration of these snow plumes. The “complex and side plane” precipitation event in Lawson et al. (2006a) revealed the presence of large rosette-shaped crystals and plate aggregates, which are typical for conditions with an ice supersaturation of approximately 20%–50% at temperatures between −30° and −50°C (cf. Bailey and Hallett 2009).

3. Field measurements at Jungfraujoch

a. The CLACE 2013 field campaign

The Cloud and Aerosol Characterization Experiment (CLACE) took place from mid-January to early March 2013 at the High Altitude Research Station Jungfraujoch (JFJ) in the Bernese Alps, Switzerland (3580 m MSL; 7°59′2″E, 46°32′53″N). This location is far away from industrial sites and other sources of air pollution and is usually in the free troposphere during winter. One of the aims of this field experiment was to investigate ice nucleation and aerosol properties (Schenk et al. 2014; Worringen et al. 2015; Schmidt et al. 2015). Another focus of the experiment was to explore the microphysical properties of liquid, ice, and mixed-phase clouds (Henneberger 2013; Lloyd et al. 2015). This study focuses on ice crystal precipitation.

b. Meteorological instrumentation

Meteorological data are available from two weather stations: the MeteoSwiss WMO station at the summit site and additional instruments from the University of Manchester [including a Metek 3D sonic anemometer and temperature/humidity sensors from Vaisala and Rotronics as described by Lloyd et al. (2015)]. The instruments from the University of Manchester were present during the first two cases of this work. For the last case study, we rely on the measurements at Jungfraujoch from MeteoSwiss. An overview of the instrument location is given in Fig. 1.

Fig. 1.

Location of the instruments during CLACE 2013. Some instruments used in this study and the north–south direction are marked. The main wind direction is from the north-northwest.

Fig. 1.

Location of the instruments during CLACE 2013. Some instruments used in this study and the north–south direction are marked. The main wind direction is from the north-northwest.

To obtain information concerning thin clouds or particle layers, a ceilometer operated by the Swiss Federal Institute of Technology Zurich was used. Retrieved range-corrected backscatter intensities are recorded for range gates/altitude intervals of 10-m length from 10- to 7700-m altitude above the summit site (approximately 11 km above sea level).

c. Cloud hydrometeor instruments

The number concentrations and size distributions of cloud hydrometeors were determined by various instruments: Particles between 2 and 50 μm were measured with a cloud droplet probe (CDP) (CDP-100 from Droplet Measurement Technologies; Lance et al. 2010). The CDP was calibrated using glass beads, with the difference in refractive indexes between liquid water and the calibration beads taken into account. Particle size is calculated from Mie theory and therefore is valid for spherical particles. Irregular particles scatter light differently from Mie theory, which leads to uncertainties in particle size and more so for particles greater than approximately 15 μm.

Larger hydrometeors with a size between 10 and 1300 μm were measured with a 3-view cloud particle imager (3V-CPI) (from SPEC), which consists of a two-dimensional stereoscopic shadow imaging probe (2D-S; Lawson et al. 2006b) and a CPI. The sample area of the 2D-S is 0.64 cm2, which yields a volume sampling rate of 1.02 l s−1 for an aspiration velocity of 16 m s−1 and particles of 100 μm or larger if the depth of field is calculated from the equations mentioned in the SPEC “2D-S post processing using 2D-S view software” documentation (version 1.1). From a minimum size of 10 μm onward, the volume sampling rate increases with the square of the particle maximum dimension until the depth of field is larger than or equal to the distance between the two arms (see Fig. 2, left). For the measurement of the bulk condensed water content in a size range below approximately 50 μm, a particulate volume monitor (PVM; Gerber 1991) is used.

Fig. 2.

(left) Volume sampling rate vs particle size for the 2D-S (blue) and the GipfelHolo instrument (red). The minimum size for a size-independent volume sampling rate of the 2D-S is 100 μm. The error bars were calculated from the bin edges. (right) Optical setup of the GipfelHolo instrument. The window material (sapphire) was chosen for sufficient heat conductivity, not for optical reasons.

Fig. 2.

(left) Volume sampling rate vs particle size for the 2D-S (blue) and the GipfelHolo instrument (red). The minimum size for a size-independent volume sampling rate of the 2D-S is 100 μm. The error bars were calculated from the bin edges. (right) Optical setup of the GipfelHolo instrument. The window material (sapphire) was chosen for sufficient heat conductivity, not for optical reasons.

In addition to the CDP, the PVM, and the 3V-CPI, a Cloud, Aerosol, Precipitation Spectrometer (CAPS) (from Droplet Measurement Technologies; Baumgardner et al. 2001) and an FSSP was deployed next to the other instruments during the campaign. Unfortunately, these instruments did not record valid data during the time frame of the analysis presented in this paper. The instruments described in this and the previous paragraphs were all operated by the University of Manchester.

Another instrument for the measurement of cloud microphysical properties is the holographic cloud probe GipfelHolo from the Institute for Atmospheric Physics, University of Mainz/Max Planck Institute for Chemistry, which is a further development of the setup of Raupach et al. (2006). It uses transmission holography to capture a three-dimensional volume of air containing hydrometeors in a single image. These images, by definition holograms, are recorded at a maximum frame rate of 6 frames per second (fps) with a sample area of 3.6 × 2.4 cm2 and with a depth along the optical axis of 35 cm, leading to a sample volume of 305 cm3. The illumination of the particles is done with a collimated beam of a frequency-tripled neodymium-doped–yttrium–aluminum–garnet (Nd:YAG) laser (FTSS-Q3 from CryLaS GmbH, 355-nm wavelength, 27-μJ pulse energy over 1.5 ns). Each image recorded with a 29 megapixel charge coupled device (CCD) camera (HR29050 from SVS-Vistek GmbH) of 5.5-μm pixel length/width contains the interference pattern from the incident light wave and the waves diffracted by the objects within the sample volume. At a frame rate of 0.5 fps, which was used for all three cases of this study, this yields a volume sampling rate of 0.15 l s−1, which is approximately 1/7 of the volume sampling rate of the 2D-S. A semiautomatic reconstruction routine is used to obtain the in-focus image of each particle and its 3D position (Fugal et al. 2009). The instrument was operated in an open-path configuration without an inlet or pumping/aspiration. The right panel in Fig. 2 shows the main features of the optical design. The physical detection limit of GipfelHolo is 2 pixels (11 μm), but the effective detection limit is determined by the noise level of each hologram and can only be estimated empirically. Because of the noise floor in the reconstructed holograms, we take 50 μm as the effective lower detection limit of GipfelHolo in our studies described in section 4.

After reconstruction and focus estimation, each particle is visually classified into different habits: Ice crystals with smooth facets (little or no irregularity on the edges) and a fourfold/sixfold symmetry are defined as columns/plates (this class contains also bullet-shaped crystals). Particles that appear circular are classified as spheroids. Ice crystals that consist of bullets/columns which are stuck together are identified as rosette shapes. Other ice crystals that consist of multiple smaller ice crystals are classified as aggregates. Particles that do not belong to any other category are classified as irregular. Below a particle size of approximately 50 μm (9–10 pixels), the habit identification is less reliable than for larger ice crystals. A distinction between ice and liquid water is not possible with this instrument, and additional data from the Small Ice Detector, version 3 [SID-3; Ulanowski et al. (2014) and Vochezer et al. (2015)], or similar instruments are needed for the corresponding separation. Examples of particle size distributions, spatial distributions, and the habit classification from single holograms are presented in Fig. 3.

Fig. 3.

Examples of particles measured with the GipfelHolo instrument. (left) Precipitating mixed-phase cloud on 27 Jan 2013. (right) Single event of high ice crystal concentration from surface effects on 9 Feb 2013. For each case, a (top) particle size distribution, (middle) spatial distribution of particles in the sample volume (instrument windows at z = 31 mm and z = 381 mm), and (bottom) particle collage (same color code as in the spatial distribution) is shown. Each dataset is obtained from one single hologram. The scale bars are 100 μm each.

Fig. 3.

Examples of particles measured with the GipfelHolo instrument. (left) Precipitating mixed-phase cloud on 27 Jan 2013. (right) Single event of high ice crystal concentration from surface effects on 9 Feb 2013. For each case, a (top) particle size distribution, (middle) spatial distribution of particles in the sample volume (instrument windows at z = 31 mm and z = 381 mm), and (bottom) particle collage (same color code as in the spatial distribution) is shown. Each dataset is obtained from one single hologram. The scale bars are 100 μm each.

4. Microphysical properties of ice crystals at Jungfraujoch

a. Diamond dust and precipitation observed 2050–2230 UTC 8 February 2013

During the first event, an upper trough centered over central Europe moved slowly southwestward and led to advection of a cold Arctic air mass toward the Alps. As shown in the left column of Fig. 4, air masses from the lower troposphere ascended to approximately 700 hPa and reached ice supersaturation. Diabatic heating of 3–4 K in 6 h is also found for some of the trajectories and could be associated with cloud formation. At the Jungfraujoch, the recorded temperature minimum of this cold period was −27°C on 9 February 2013. There was only light snowfall from deep cumulus clouds on 7 February and early 8 February 2013. From the afternoon hours until approximately 2000 UTC, clear-sky conditions were observed, which were followed by an approximately 50–200-m-deep cloud layer moving across the site from the north (Fig. 5, top left). The cloud layer, at 2050 UTC found between 300 and 500 m above the station, descended continuously to the station altitude at approximately 2210 UTC. Between 2040 and 2100 UTC and also between 2115 and 2120 UTC, enhanced backscatter intensities close to ground level coincide with particle number concentrations >10 cm−3 in the CDP data with a mean volume diameter around 6 μm (Fig. 5, middle left). At the same time, ice crystals were both visually observed and detected by the GipfelHolo and 2D-S instrument (Fig. 5, bottom left).

Fig. 4.

The 24-h backward trajectories from the ECMWF model for (left) 2100 UTC 8 Feb 2013 and (right) 0000 UTC 24 Feb 2013. (top) Map of all trajectories, color coded by pressure. Dots are plotted every 6 h. (middle) Relative humidity over ice along the trajectories with at least 10 hPa ascent in the past 24 h as a function of time and pressure. (bottom) Time series of the change in potential temperature along the trajectories from t = −24 h, with the same selection criterion as for the middle panels. Positive values indicate diabatic heating; negative values indicate diabatic cooling.

Fig. 4.

The 24-h backward trajectories from the ECMWF model for (left) 2100 UTC 8 Feb 2013 and (right) 0000 UTC 24 Feb 2013. (top) Map of all trajectories, color coded by pressure. Dots are plotted every 6 h. (middle) Relative humidity over ice along the trajectories with at least 10 hPa ascent in the past 24 h as a function of time and pressure. (bottom) Time series of the change in potential temperature along the trajectories from t = −24 h, with the same selection criterion as for the middle panels. Positive values indicate diabatic heating; negative values indicate diabatic cooling.

Fig. 5.

(left) From 2050–2230 UTC 8 Feb 2013: (first row) ceilometer backscatter intensity, (second row) particle number concentration from the CDP, (third row) ice crystal number concentration from the 2D-S and the GipfelHolo instruments, and (fourth row) condensed water content from all cloud hydrometeor instruments. (top right) Vertical profiles of T, Td, and Tf [mean and min/max from MeteoSwiss (squares) and Manchester T/RH sensors (stars) at Jungfraujoch—the vertical gap is for ease of reading; profile data are from NOAA Real-Time Environmental Applications and Display System (READY) at 2100 UTC, and sounding data are from Payerne on 0000 UTC 9 Feb 2013], and (bottom right) size distribution of ice crystals from GipfelHolo (various colors) and 2D-S (gray) for the same time period as in the left panels. The concentration (markers with error bars) of the total ice (tot; black), aggregates (agg; blue), columns and plates (col + plt; red), spheroids (sph; green), irregular particles (irr; magenta), and the corresponding lognormal fits (dashed lines) are shown.

Fig. 5.

(left) From 2050–2230 UTC 8 Feb 2013: (first row) ceilometer backscatter intensity, (second row) particle number concentration from the CDP, (third row) ice crystal number concentration from the 2D-S and the GipfelHolo instruments, and (fourth row) condensed water content from all cloud hydrometeor instruments. (top right) Vertical profiles of T, Td, and Tf [mean and min/max from MeteoSwiss (squares) and Manchester T/RH sensors (stars) at Jungfraujoch—the vertical gap is for ease of reading; profile data are from NOAA Real-Time Environmental Applications and Display System (READY) at 2100 UTC, and sounding data are from Payerne on 0000 UTC 9 Feb 2013], and (bottom right) size distribution of ice crystals from GipfelHolo (various colors) and 2D-S (gray) for the same time period as in the left panels. The concentration (markers with error bars) of the total ice (tot; black), aggregates (agg; blue), columns and plates (col + plt; red), spheroids (sph; green), irregular particles (irr; magenta), and the corresponding lognormal fits (dashed lines) are shown.

Below the Jungfraujoch, a deep layer of stratocumulus clouds was located approximately 5 km upstream at an altitude between 2 and 2.5 km above sea level as visually determined by mountain peaks of known height. This feature is consistent with the vertical profile from the Payerne radiosounding at 0000 UTC 9 February 2013 (approximately 80 km WNW of the Jungfraujoch) on 9 February, which reveals an ice supersaturated layer between 2 and 3.4 km above sea level (Fig. 5, top right). Forty-seven percent of the ice crystals measured with GipfelHolo are found in the pristine column and plate fraction with typical aspect ratios between 1 and 2 and a narrow monomodal size distribution with a mean size of approximately 80 μm (Fig. 6, left). The peak level of the diamond dust event (up to 40 ice crystals L−1) continued until approximately 2120 UTC when the CDP signal declined, and also, the ice crystal concentration decreased (Fig. 5, left). The coincidence between high particle number concentrations below 10 μm in the CDP data and higher number concentrations of ice crystals in the 2D-S and GipfelHolo data as well as the orographically induced ascent shown in the trajectory maps suggest droplet freezing as an important mechanism to generate the ice crystals.

Fig. 6.

Collage of ice crystals from (top left) the diamond dust case on 2050–2120 UTC 8 Feb 2013 and (top right) the precipitation case on 2150–2230 UTC 8 Feb 2013 from the CPI. Each scale bar increment is 200 μm. (middle) As in the top panels, but for the GipfelHolo instrument. Each scale bar increment is 100 μm. (bottom left) Size distribution of the diamond dust case from GipfelHolo and 2D-S with the same color code as in the bottom right panel of Fig. 5 and (bottom right) size distribution of the precipitation case.

Fig. 6.

Collage of ice crystals from (top left) the diamond dust case on 2050–2120 UTC 8 Feb 2013 and (top right) the precipitation case on 2150–2230 UTC 8 Feb 2013 from the CPI. Each scale bar increment is 200 μm. (middle) As in the top panels, but for the GipfelHolo instrument. Each scale bar increment is 100 μm. (bottom left) Size distribution of the diamond dust case from GipfelHolo and 2D-S with the same color code as in the bottom right panel of Fig. 5 and (bottom right) size distribution of the precipitation case.

The narrow size distribution shown in the left panel of the second row in Fig. 5 (σ = 2 μm) with a median particle diameter close to 4 μm at an average number concentration of 92 cm−3 would argue for the presence of supercooled liquid water. However, there were virtually no rimed ice crystals found in the spectrum, and the meteorological sensors clearly indicated water subsaturated conditions at the site. A similar size spectrum in particle median diameter, concentration, LWC, and standard deviation was observed in the FSSP data from the 26 November continental airmass cap cloud case of Rauber (1981), where it was attributed to supercooled liquid water.

The number concentration of ice crystals began to increase again around 2150 UTC (from 0.7 to 2.0 L−1on average) as the cloud base descended farther and precipitation was added to the diamond dust. At 2205 UTC, there was again a notable increase of the particle number concentration measured by the CDP with a mean volume diameter around 6 μm. The precipitation particles contain a larger fraction (32% as compared with 9% for the diamond dust) of ice crystal aggregates, which appear clearly as a secondary mode in both the size distribution of the total event from 2050 to 2230 UTC (Fig. 5, bottom right) and in the size distribution from 2150 to 2230 UTC (Fig. 6, right). Relative to the ice crystal images from the diamond dust case (2050 to 2120 UTC; Fig. 6, left), the aggregates from the precipitation event were on average larger (280-μm median size as compared with 196-μm median size for the diamond dust), and some of them appeared to be rimed (Fig. 6, middle right). Even though the number concentration did not change significantly, the ice water content increased by a factor of 5 (see Table 1). Around 2230 UTC, the number concentration in the CDP data increased to more than 100 cm−3 with a median diameter around 12 μm, and also, the PVM showed a significant increase in the condensed water content. At this time, we assume that the cloud base of the precipitating cloud reached the Jungfraujoch.

Table 1.

Fit parameters for particle size distributions of the three days of ice crystal measurements from GipfelHolo using Eq. (1). Besides the fit parameters (95% confidence interval in parentheses), the RMSE, the degrees of freedom adjusted coefficient of determination, the area-derived ice water content from Eq. (2), and the total number of particles N are shown. The total ice size distribution of 8 Feb precipitation is fitted with a bimodal lognormal distribution, and the two modes are denoted as M1 and M2 in the table. The same parameters are also shown for the 2D-S data, if available.

Fit parameters for particle size distributions of the three days of ice crystal measurements from GipfelHolo using Eq. (1). Besides the fit parameters (95% confidence interval in parentheses), the RMSE, the degrees of freedom adjusted coefficient of determination, the area-derived ice water content from Eq. (2), and the total number of particles N are shown. The total ice size distribution of 8 Feb precipitation is fitted with a bimodal lognormal distribution, and the two modes are denoted as M1 and M2 in the table. The same parameters are also shown for the 2D-S data, if available.
Fit parameters for particle size distributions of the three days of ice crystal measurements from GipfelHolo using Eq. (1). Besides the fit parameters (95% confidence interval in parentheses), the RMSE, the degrees of freedom adjusted coefficient of determination, the area-derived ice water content from Eq. (2), and the total number of particles N are shown. The total ice size distribution of 8 Feb precipitation is fitted with a bimodal lognormal distribution, and the two modes are denoted as M1 and M2 in the table. The same parameters are also shown for the 2D-S data, if available.

Considering the thermodynamic background, the steep ascent toward the Jungfraujoch, and a trajectory analysis from ECMWF model data (Fig. 4, left), the most plausible source of the diamond dust particles is frozen cloud droplets and/or water vapor from evaporated cloud droplets deposited on ice nuclei. The maximum ice supersaturation found in the trajectory analysis is between 10% and 20%, which is in agreement with the in situ data from MeteoSwiss—the Manchester humidity measurements show conditions close to ice saturation (Fig. 5, top right). As the average vertical velocity measured with a 3D sonic anemometer ranged between 1 and 2 m s−1, an ascent from Kleine Scheidegg (2061 m MSL, approximately 3 km upstream) to the Jungfraujoch takes on average 12–24 min, which is enough time to grow 30–50-μm ice crystals after nucleation from the vapor phase at approximately 10% ice supersaturation and a temperature around −35°C (Bacon et al. 2003). At least 18% of the observed ice crystals in this work have an optical equivalent diameter (which is defined as the diameter of a sphere that has the same projected area as the particle) at or below 50 μm; 90% of them are short columns and spheroids. Therefore, it is thought that the observed ice crystals originated mostly from frozen droplets as they grow faster to sizes around 100 μm than from ice nuclei. Frozen droplets were also identified as the main source of diamond dust particles in the study from Ohtake et al. (1982) in the Arctic, in contrast to the ice formation driven by deposition freezing as mentioned in Gultepe et al. (2014). These ice crystals described by Ohtake et al. (1982) often developed some irregularities on the micrometer scale. Another interesting observation from Bacon et al. (2003) is that particles that start with some kind of irregularity tend to remain irregular during their entire growth process. This is also confirmed by a study from Korolev et al. (1999) and might explain the presence of irregular particles during diamond dust events like in this study and also in Lawson et al. (2006a).

b. Ice crystals from the surface observed on 9 February 2013

When compared with the previous day, the overall synoptic pattern did not change much. The convergence line with cumulus convection, which caused the precipitation on the previous night, crossed the Alps, and there was no precipitation later on 9 February 2013. In the evening hours, clear-sky conditions were observed at the Jungfraujoch with some distant stratocumulus clouds located approximately 20 km upstream. At 2040 UTC, the presence of ice crystals in clear, cloudless air was revealed by light scattering in the beam of a flashlight. At the same time, the ceilometer backscatter data showed a pattern similar to fall streaks of ice crystals between 0 and 300 m above the Jungfraujoch (Fig. 7, top left) and high backscatter intensities close to the station altitude. These events also showed a signal in the data from the GipfelHolo, the 2D-S instrument, the CDP, and also the PVM (Fig. 7, left). Sounding and model data show ice subsaturated conditions around this time, but the MeteoSwiss humidity sensor indicates slightly ice supersaturated conditions (Fig. 7, top right). There are individual times when the 2D-S and also the GipfelHolo instruments measure number concentrations above 250 L−1 for periods shorter than 10 s. In contrast to the precipitation case of 8 February, the size distribution of the total ice and also the size distributions of the individual habits can be represented well by monomodal lognormal distributions because of aggregates not producing a second larger mode (Fig. 7, bottom right).

Fig. 7.

As in Fig. 5, but for ice crystals produced by surface processes on 2020–2130 UTC 9 Feb 2013.

Fig. 7.

As in Fig. 5, but for ice crystals produced by surface processes on 2020–2130 UTC 9 Feb 2013.

To investigate possible mechanisms that produced the ice, the number concentration per hologram was determined. Of all holograms from this day, 80% are empty, and 90% of the nonempty holograms have number concentrations between 1 and 5 per hologram. Even though there exists a nonzero probability of missed detections in the so-called empty holograms, their contribution to the total ice water content would be only a few percent (even in the worst case), as missed detections are supposedly particles close to the effective lower detection limit of 50 μm. The average number concentration would increase by no more than 20% if 50% of the empty holograms had one particle instead. From this, we argue that missed detections would have only a marginal influence on the results.

To test if their production mechanism might be different from the ice crystals in the densely populated holograms, the total population was divided into two subsets: ice number concentration at or below 5 per hologram and ice number concentration of at least 6 per hologram. In Fig. 8, both size distributions shown next to each other have almost the same width and a similar mean particle size (121 μm for the high-concentration events; 160 μm for the low-concentration events). The collages show very similar particle shapes with aggregates and irregular particles being the dominant fraction (85% for the high-concentration events; 84% for the low-concentration events). As the difference of the median size and the habit fractions are not statistically significant, we conclude that the particle production mechanism is likely the same for all cases from 9 February, independent of time and number concentration.

Fig. 8.

As in the middle and bottom rows of Fig. 6, but for the case of 2020–2130 UTC 9 Feb 2013. (left) Holograms with 1–5 particles per hologram (549 holograms) and (right) holograms with 6 and more particles per hologram (57 holograms). The scales of the ordinate vary.

Fig. 8.

As in the middle and bottom rows of Fig. 6, but for the case of 2020–2130 UTC 9 Feb 2013. (left) Holograms with 1–5 particles per hologram (549 holograms) and (right) holograms with 6 and more particles per hologram (57 holograms). The scales of the ordinate vary.

A few events with ice crystal number concentrations exceeding 100 L−1 occurred during this period. They lasted only a few seconds, and a peak number concentration of 4250 L−1 was reached in one case. The spatial distribution and size distribution of the ice crystals from this event are shown in the right panel of Fig. 3. The spatial clustering as in the middle-right panel of Fig. 3 is likely due to an angular difference between the orientation of the instrument and the wind direction and also turbulence around the instrument.

We can conclude surface processes (blowing snow, hoarfrost detachment, etc.) are a main particle source on 9 February 2013 as 1) ice crystal number concentrations changed by a factor of 100–1000 within seconds and 2) irregular particles and aggregates dominate the number concentration (84%) and also the ice water content (97%) during this event. Some snow plumes from the summit of Mönch have been observed quite frequently during the field campaign, but in general, they did not reach the measurement platform directly as their origin lay at approximately 1-km distance perpendicular to the wind direction. More likely is the breakup of fragile vapor-grown ice crystals on a snow surface beneath the measurement platform. The ice particle concentration from this type of event is much less correlated with the wind speed than would be the case for blowing snow. This finding is discussed in Lloyd et al. (2015), which has a partial overlap of data with this paper [8 and 9 February 2013 are included in the data presented in Lloyd et al. (2015)]. Interestingly, an extended event of moderately enhanced ice crystal concentration (60 particles L−1 on average during 34 s) coincides with a local minimum in wind speed around 2 m s−1. It is possible that the particle samples from both cases of 8 February also contain a fraction of ice particles from the surface, especially in the irregular particle fraction.

c. Cloud ice and snow grains from cirrostratus on 23 February 2013

The third case presented in this work is associated with a warm front moving northward. The backward trajectories in the right panel of Fig. 4 show some air parcels ascending near the JFJ because of orographic forcing (the northeasterly ones starting at a pressure around 900 hPa). But there was also large-scale ascent along the warm front of boundary layer air, which originated from Italy and the Tyrrhenian Sea. These trajectories followed a southeasterly flow and ascended to pressure levels between 650 and 300 hPa. During the ascent, these trajectories reached at least ice saturated conditions and exhibited strong diabatic heating of approximately 16 K in less than 10 h. During the passage of this front, a persistent 22° halo could be observed around the moon, indicating the presence of regularly shaped ice columns or plates. Light ice crystal precipitation started around 2000 UTC and continued during the night. Sounding data from Payerne at 0000 UTC 24 February 2013 reveal the presence of an ice supersaturated layer from approximately 4.4 to 5 km MSL (Fig. 9, top center). Below this layer, a frost point depression of >10°C is found between 3 and 4 km MSL, so ice subsaturated conditions are expected at the site, confirmed by local T/RH measurements from MeteoSwiss. No clouds are visible at lower altitudes during the measurement interval.

Fig. 9.

Data from 23 Feb 2013. (top left) Ceilometer backscatter intensity. (top center) Temperature and humidity profile (the squares represent data from MeteoSwiss at Jungfraujoch station). Potentially unstable regions are highlighted in magenta. (top right) Particle collage from GipfelHolo. (bottom left) Number-weighted, (bottom center) area-weighted, and (bottom right) mass-weighted size distribution from GipfelHolo for each habit.

Fig. 9.

Data from 23 Feb 2013. (top left) Ceilometer backscatter intensity. (top center) Temperature and humidity profile (the squares represent data from MeteoSwiss at Jungfraujoch station). Potentially unstable regions are highlighted in magenta. (top right) Particle collage from GipfelHolo. (bottom left) Number-weighted, (bottom center) area-weighted, and (bottom right) mass-weighted size distribution from GipfelHolo for each habit.

The cloud that produces the snow grains appears as an optically rather thin cloud in the ceilometer backscatter data. The backscatter intensities were approximately two orders of magnitude below saturation and more than one order of magnitude above the noise level. Its appearance differs significantly from those clouds containing large amounts of supercooled liquid water, and we suppose it to be a pure ice cloud for the following reasons: 1) a persistent halo was observed, and 2) the backscatter intensity profile showed patterns similar to fall streaks. The top layer of the cloud in the ceilometer data is very close to the upper edge of the ice supersaturated region in the sounding data (see Fig. 9, top left and top center), but there are also filaments in the backscatter intensity signal at higher altitude.

An examination of the particle habits shows a fraction of 7.5% rosette-shaped particles, which were not observed during the other two cases (Fig. 9, top right). The median of the number-weighted size distribution shown in the bottom-left panel of Fig. 9 is at approximately 270 μm and therefore is much larger than in the previous cases. Another unique feature of this case is the broad size distribution of columnar and plate crystals. Some short columns/thick plates with aspect ratios close to 1 and sizes similar to diamond dust particles are observed simultaneously with columns or plates of several hundred microns in length. Also, many of the large columnar crystals and bullet rosettes show rounded edges. Rounded edges often are the result of sublimation in an ice subsaturated layer below the cloud. This is indicated by the meteorological measurements at the site and the vertical profiles at approximately 4 km MSL.

Column aggregates, bullet rosettes, and similar particle habits usually form in an environment with an ice supersaturation above 10% and temperatures below −25°C. From Bailey and Hallett (2009), we would conclude that these ice crystals need to have their origin in a temperature range between −35° and −45°C and a relative humidity over ice between 110% and 160%. These temperatures were present at altitudes > 3 km above the Jungfraujoch, but the sounding data do not show ice supersaturation in this altitude range. However, there are some thin “filaments” of scatterers visible in the ceilometer data around 1910 and 2100 UTC (Fig. 9,top left), which are found in the upper unstable layer between 5400 and 5800 m MSL. As the temperature and humidity profiles from the Global Data Assimilation System (GDAS) provided by NOAA suggest a very deep ice supersaturated region above the Jungfraujoch, and the sounding data show conditions close to ice saturation, there might be a way to produce short-lived ice supersaturated regions above the unstable layer via dynamic processes in the warm front. The ice supersaturated layer between 4200 and 5000 m MSL might then act as the water vapor source for a further growth of the ice crystals. Our observations in Fig. 9 are comparable to the “complex and side plane” case in Lawson et al. (2006a) w.r.t. the average particle number concentration (between 1.5 and 2 L−1) and the number-weighted median size (between 240 and 270 μm). The mass-weighted size distributions of both studies are dominated by complex aggregates with side planes and also rosette shapes.

5. Discussion

The majority of the number-weighted ice crystal size distributions for the total ice and the individual habits agree well with monomodal or bimodal lognormal distributions (normalized r2 values close to 1; RMSE values between 0.05 and 0.35 L−1). Equation (1) was used to approximate the number concentration of each particle maximum dimension bin D with a number concentration , a mean particle size , a normalization particle size m, and the geometric standard deviation σ similar to Seinfeld and Pandis (2012), their Eq. (8.33):

 
formula

To estimate the ice water content (IWC) from the GipfelHolo and 2D-S particle area, we follow the recommendation of Baker and Lawson (2006) and use their area-to-mass relation (ice mass m in mg; particle area A in mm2):

 
formula

This formula is in good agreement with the true area-to-mass relationship for columnar ice crystals with a maximum dimension between approximately 40 to 200 μm and an aspect ratio between 1 and 2 (Fig. 1 from Lawson and Baker 2006). Fit parameters for the individual size distributions as well as the area-calculated total ice water content from GipfelHolo are given in Table 1.

An intercomparison between the GipfelHolo and the 2D-S instruments (Fig. 10) shows a significant discrepancy between the two instruments. On 8 February 2013, the particle number concentration between 120 and 400 μm from the 2D-S is significantly higher than the number concentration from GipfelHolo. This discrepancy is most prominent in the diamond dust case (Fig. 10, left) and cannot be explained by particles missed or mis-sized in GipfelHolo. A possible reason for differences in the number concentration and distribution width is the spatial displacement of the two instruments. Two factors that might contribute to cloud heterogeneity are the complicated flow conditions due to the orographic forcing and mixing induced by the mountain ridge.

Fig. 10.

Size distributions of (left) the diamond dust case of 8 Feb 2013, (center) the precipitation case of 8 Feb 2013, and (right) the surface ice from 9 Feb 2013. The size distributions for each habit (color) and the total ice (black) from the GipfelHolo instrument and the total ice from the 2D-S instrument (gray) are shown. (top) Number-weighted, (middle) area-weighted, and (bottom) mass-weighted size distributions. The particle mass of both the 2D-S and GipfelHolo data was derived from the particle area using Eq. (2).

Fig. 10.

Size distributions of (left) the diamond dust case of 8 Feb 2013, (center) the precipitation case of 8 Feb 2013, and (right) the surface ice from 9 Feb 2013. The size distributions for each habit (color) and the total ice (black) from the GipfelHolo instrument and the total ice from the 2D-S instrument (gray) are shown. (top) Number-weighted, (middle) area-weighted, and (bottom) mass-weighted size distributions. The particle mass of both the 2D-S and GipfelHolo data was derived from the particle area using Eq. (2).

The IWC derived from the particle size distributions differs significantly (0.6 mg m−3 for GipfelHolo and 1.1 mg m−3 for the 2D-S; see Table 2) and is mainly caused by the larger number of particles around 200 μm. Especially in the precipitation case (Fig. 10, center), the 2D-S detects slightly fewer particles at sizes greater than 400 μm. This leads to the fact that the mass-weighted median diameter from GipfelHolo (575 μm) is almost twice as high as the mass-weighted median diameter from the 2D-S (315 μm).

Table 2.

Comparison of microphysical quantities from the CDP, PVM, 2D-S, and GipfelHolo (GH) instruments. LWC, IWC, volume-weighted median diameter (MVD), number-weighted median diameter (MD) and number concentration (NC) are compared. The values in parentheses denote the maximum if the quantity is NC, LWC, or IWC. In case of MD or MVD, the values in parentheses denote the first and third quartile of the size distribution. For the case of 23 Feb precipitation, only data from GH are available.

Comparison of microphysical quantities from the CDP, PVM, 2D-S, and GipfelHolo (GH) instruments. LWC, IWC, volume-weighted median diameter (MVD), number-weighted median diameter (MD) and number concentration (NC) are compared. The values in parentheses denote the maximum if the quantity is NC, LWC, or IWC. In case of MD or MVD, the values in parentheses denote the first and third quartile of the size distribution. For the case of 23 Feb precipitation, only data from GH are available.
Comparison of microphysical quantities from the CDP, PVM, 2D-S, and GipfelHolo (GH) instruments. LWC, IWC, volume-weighted median diameter (MVD), number-weighted median diameter (MD) and number concentration (NC) are compared. The values in parentheses denote the maximum if the quantity is NC, LWC, or IWC. In case of MD or MVD, the values in parentheses denote the first and third quartile of the size distribution. For the case of 23 Feb precipitation, only data from GH are available.

The case of 9 February 2013 (Fig. 10, right) shows the opposite when compared with the first two cases. The number concentration from the 2D-S is less than the number concentration from GipfelHolo by almost a factor of 6. However, the area- and mass-weighted size distributions have a very similar mode diameter and distribution width. The difference in particle number concentration might at least partially be caused by the flow conditions: On this particular day, wind direction and wind speed varied significantly and quickly such that the inlet rotator was not aligned with the wind when a plume of ice crystals reached the site. As this ice cloud was most likely produced by surface effects, we expect significant inhomogeneities in particle number concentration on a scale of a few meters.

A comparison of the habit fractions from CPI images and GipfelHolo (Fig. 11) shows reasonable agreement between the two instruments. For the diamond dust and surface ice cases, the relative abundance of each habit from one instrument is within the statistical uncertainty from the counting statistics of the other one. However, the total number of particles in the CPI images was much lower than in the GipfelHolo particle images, and therefore, the error bars are significantly larger. For the precipitation case, there was significantly more pristine ice (columns/plates) in the CPI data when compared with GipfelHolo. We conclude from Fig. 11 that the habit segregation applied to particle images from both instruments is reasonably consistent.

Fig. 11.

Histogram of the particle habit fractions from (top) the CPI and (bottom) the GipfelHolo instruments for particles greater than 100 μm. The error bars represent statistical uncertainties from the total number of particles in each habit class. The spheroid fraction is not shown here as the number of spheroids greater than 100 μm is too low for reliable statistics.

Fig. 11.

Histogram of the particle habit fractions from (top) the CPI and (bottom) the GipfelHolo instruments for particles greater than 100 μm. The error bars represent statistical uncertainties from the total number of particles in each habit class. The spheroid fraction is not shown here as the number of spheroids greater than 100 μm is too low for reliable statistics.

The median number-weighted particle size of diamond dust observed at Jungfraujoch lies somewhere in between the measurements from Antarctica and from the Arctic (Fig. 12, top left). Considering the temperature in the lowest layers of the measurement site, there is not much temperature dependence of the particle size to be found. The studies from the past, which are mentioned in section 2, suggest that Antarctic diamond dust tends to be smaller in size than Arctic diamond dust. This difference in particle size would be plausible because of the higher availability of water vapor from open leads in the Arctic, but this finding is not statistically significant because of the low number of cases.

Fig. 12.

(top left) Size distribution comparison of studies on diamond dust from Antarctica (blue), the Arctic region (red), and this work (black) vs temperature in the lowest layer. (top right) Comparison of the different cases from Lawson et al. (2006a) with this work. These cases are diamond dust (blue), ice from surface processes (red), precipitation (black), and a mixture of precipitation and diamond dust (magenta). (bottom left),(bottom right) As in the top right panel, but for particle size vs ice water content on the left and particle size vs particle number concentration on the right. The symbols denote the median of each size distribution, the boxes denote the inner 50%, the thick lines denote the inner 90%, and the error bars contain the outliers. Temperatures are either obtained from radiosonde data or from measurements at the site. The inner 90% interval is not shown for studies where the size range is derived from a single replicator slide.

Fig. 12.

(top left) Size distribution comparison of studies on diamond dust from Antarctica (blue), the Arctic region (red), and this work (black) vs temperature in the lowest layer. (top right) Comparison of the different cases from Lawson et al. (2006a) with this work. These cases are diamond dust (blue), ice from surface processes (red), precipitation (black), and a mixture of precipitation and diamond dust (magenta). (bottom left),(bottom right) As in the top right panel, but for particle size vs ice water content on the left and particle size vs particle number concentration on the right. The symbols denote the median of each size distribution, the boxes denote the inner 50%, the thick lines denote the inner 90%, and the error bars contain the outliers. Temperatures are either obtained from radiosonde data or from measurements at the site. The inner 90% interval is not shown for studies where the size range is derived from a single replicator slide.

In comparison with the diamond dust case in Lawson et al. (2006a), we found a larger fraction of larger ice crystals, which might be due to a surface-generated ice background and/or a greater vertical depth of the cloud layer above. The average particle number concentration in this work (1.7 L−1 with GipfelHolo; 2.1 L−1 with the 2D-S) is lower by a factor of 50, and the ice water content (0.6 mg m−3 for GipfelHolo and 1.1 mg m−3 for the 2D-S) is lower by a factor of 20, than in Lawson et al. (2006a) (see Fig. 12, bottom). However, the pristine column and plate fraction contributes to almost 50% in both studies (if weighted by number concentration). The ice crystals in this study are larger (number-weighted median diameter of 85 μm and mass-weighted median diameter of 207 μm as compared with 77 μm and 105 μm, respectively), and the aggregates and irregular particles contribute to 79% of the ice water content in this study (43% in Lawson et al. 2006a).

A direct comparison of the blowing snow from Lawson et al. (2006a) and the ice crystals from surface processes at Jungfraujoch (Fig. 12, top right) shows very little deviation from a lognormal distribution for the inner 90% and the inner 50% of the size distributions. However, the ice crystals from blowing snow in Lawson et al. (2006a) are much smaller (median diameter: 57-μm number weighted; 67-μm mass weighted) than the surface-generated ice at Jungfraujoch (128-μm number-weighted median diameter and 381-μm mass-weighted median diameter). Another difference is found in the ice water content [4.6 mg m−3 in this work as compared with 0.1 mg m−3 in Lawson et al. (2006a); see Fig. 12, bottom left] and the number concentration (5.8 vs 1.5 L−1; see Fig. 12, bottom right). This finding might be explained by the higher temperature in the Jungfraujoch region, which would lead to a faster growth of hoarfrost crystals under conditions near ice saturation.

The size distribution from the precipitation case right after the diamond dust (the magenta markers in the top-right panel of Fig. 12) has a median size that is very similar to that of the surface-generated ice. But we should keep in mind that the precipitation case has a true bimodal size distribution in the GipfelHolo data, with small diamond dust particles in the primary mode and larger aggregates in the secondary mode (Fig. 6, right), which is not represented in Fig. 12. A bimodal size distribution was also found by Ohtake et al. (1982) for Arctic diamond dust with a “nucleation mode” around 75 μm and an “aggregation mode” around 250 μm, which is similar to our observations (primary mode at 80 μm; aggregate mode around 270 μm). It might be just a statistical correlation between this work and Ohtake et al. (1982), but it is also possible that the same physical process (aggregation) led to the bimodal size distributions in both cases.

The case of 23 February 2013 with precipitation particles from an ice cloud above has a lot in common with the “complex and side plane” case of Lawson et al. (2006a) (see Fig. 12, top right and bottom row): particle number concentration (1.6 vs 1.8 L−1), ice water content (5.2 vs 4.4 mg m−3), fraction of columns and plates (15% vs 21% number weighted; 6% vs 11% mass weighted), and median particle size (265 vs 301 μm number weighted; 429 vs 411 μm mass weighted). The dominant fraction of the ice water content is found in the aggregates and rosette-shaped particle habit class (88% in this work; 70% in Lawson et al. 2006a). From the habit diagram in Bailey and Hallett (2009), we would conclude that the formation temperature for these ice crystals was significantly lower than the ambient temperature at the site and probably associated with the filaments observed on 23 February 2013.

6. Conclusions

On 8 February 2013, we measured ice crystal size distributions and cloud droplet spectra from two ice crystal precipitation events similar to diamond dust at the Jungfraujoch observatory. In contrast to the definition of diamond dust by Intrieri and Shupe (2004), supercooled liquid water was likely present in this case as indicated by narrow (σ ≈ 2μm) monomodal size distributions of small (m) particles. The condensed water content was on average higher by almost a factor of 10 than the ice water content (8.5 vs 0.63 mg m−3), but the ice crystal habit distribution and particle sizes (m) are similar to the observations of diamond dust at South Pole Station by Lawson et al. (2006a) (m). The major contribution to the total ice water content of the diamond dust case is given by the aggregates, which provide approximately 56% of the total IWC. However, the number concentration of columns, plates, and spheroids is greater by a factor of 5 than the number concentration of aggregates. The presence of spheroids and the observed cloud conditions below the Jungfraujoch suggest frozen droplets as a main source of the diamond dust particles in this case.

The second event of 8 February 2013, labeled as the precipitation case, showed an additional well-defined secondary mode of large ice crystal aggregates with a mean size around 270 μm, which contribute to 33% of the number concentration and 86% of the total ice water content.

Surface-generated ice differs much from precipitation and diamond dust: 84% of the number concentration comes from irregular particles and aggregates. The IWC of columns, plates, and spheroids is almost negligible for the total IWC, with a contribution of less than 3%. Isolated events of very high particle number concentration (>4000 L−1) do not show a correlation with the ambient wind speed. Hoarfrost detachment is thought to contribute more to the total IWC than blowing snow as it is less sensitive to wind speed (Lloyd et al. 2015). These dense ice particle clouds do not follow a uniform random distribution in space, which is partially caused by the angle of the background flow w.r.t. the sample volume of the holographic instrument and turbulence around the instrument. The entire holographic dataset is corrected for particle shattering [as defined in Korolev et al. (2013)], which occurred only in 1 of more than 10 000 holograms.

Precipitation and cloud ice from a cirrostratus cloud is dominated by large plate aggregates and rosette-shaped ice crystals in terms of IWC (87% of the total IWC). From Bailey and Hallett (2009), we expect the origin of these ice crystals at much lower temperatures than measured at the site. This assumption is in agreement with the radiosounding data and the vertical profile from NOAA as potentially unstable ice supersatured layers exist at temperatures between −35° and −50°C. Rounded edges of some larger ice crystals suggest the presence of an ice subsaturated layer below the cloud base.

The habit classification applied to the holographic data and the CPI images shows consistent results for particle sizes greater than 100 μm despite poor counting statistics of the CPI data. A comparison of the size distributions of 8 February 2013 from the 2D-S and the holographic instrument shows a concentration enhancement in the 2D-S data by a factor of 2–10 in a size range between 120 and 400 μm, which cannot be explained by sizing or detection issues of GipfelHolo. The exact reason for this discrepancy is still unclear; local inhomogeneities of the ice crystal precipitation might be part of the explanation. On 9 February 2013, the 2D-S data show a significantly lower number concentration and ice water content when compared with GipfelHolo, which is likely due to rapid changes in the wind direction.

Even if the spheroids from GipfelHolo were liquid droplets despite the low temperatures close to −30°C, their contribution to the total water content would be negligible in all four cases shown in Table 1, and the assumption of liquid instead of frozen droplets would not change much of the results. However, the 3–20 μm particles in the CDP data are likely supercooled liquid water droplets because of their narrow size distribution, and their contribution to the total water content is significant.

Temperature measurements at the site agree well with both the sounding data and the NOAA GDAS vertical profile. More critical is the humidity measurement: the sensors from the University of Manchester and the MeteoSwiss WMO station do not agree whether ice subsaturated conditions or ice supersaturated conditions were present at the site. This uncertainty complicates detailed conclusions about the ice crystal origin.

Ceilometer data provide a good overview of the cloud conditions above the site. Even though phase determination is not possible with this measurement type, the vertical structure and time evolution of patterns with enhanced backscatter intensities can help to determine the type of cloud. Some questions about the cloud conditions below the site could have been addressed with an additional ceilometer present at Kleine Scheidegg below the Jungfraujoch.

More case studies of ice crystal precipitation in the Alpine region are needed to provide a better characterization of the microphysical properties and links to the possible origin, including blowing snow and hoarfrost detachment. The primary mode of small ice crystals around 80 μm occurred not only in the diamond dust case of this study but also in almost every case shown by Lloyd et al. (2015). This might be a feature general to ice clouds or perhaps some peculiarity of the Jungfraujoch region. Also, more studies on diamond dust in various places are needed to obtain statistically robust information about typical ice crystal sizes, habits, and concentrations.

Acknowledgments

This work is funded in part by the German Research Foundation (DFG) under SPP 1294 (AMOSIL) and also by an Advanced Grant of the European Research Council (ERC), project 321040 (EXCATRO). Thanks go to Ulrich K. Krieger from ETH Zurich for providing the ceilometer data plus analysis software and also to Matthias Voigt from IPA Mainz for calculating the ECMWF trajectories. We thank the Swiss Meteorological Service (MeteoSwiss) for their permission to use the meteorological data from Jungfraujoch and from the Payerne sounding. We also thank the National Oceanic and Atmospheric Administration (NOAA) for providing the model data used in the vertical profiles. Thanks go also to the International Foundation High Altitude Research Stations Jungfraujoch and Gornergrat (HFSJG) for providing the support for experiments like CLACE 2013. Also, we thank Maria and Urs Otz as well as Joan and Martin Fischer for their assistance during the field experiment. Thanks are given to the three anonymous reviewers for their helpful comments to improve this paper.

REFERENCES

REFERENCES
Auer
,
A. H.
,
D. L.
Veal
, and
J. D.
Marwitz
,
1969
:
Observations of ice crystal and ice nuclei concentrations in stable cap clouds
.
J. Atmos. Sci.
,
26
,
1342
1343
, doi:.
Bacon
,
N. J.
,
M. B.
Baker
, and
B. D.
Swanson
,
2003
:
Initial stages in the morphological evolution of vapour-grown ice crystals: A laboratory investigation
.
Quart. J. Roy. Meteor. Soc.
,
129
,
1903
1927
, doi:.
Bailey
,
M. P.
, and
J.
Hallett
,
2009
:
A comprehensive habit diagram for atmospheric ice crystals: Confirmation from the laboratory, AIRS II, and other field studies
.
J. Atmos. Sci.
,
66
,
2888
2899
, doi:.
Baker
,
B.
, and
R. P.
Lawson
,
2006
:
Improvement in determination of ice water content from two-dimensional particle imagery. Part I: Image-to-mass relationships
.
J. Appl. Meteor. Climatol.
,
45
,
1282
1290
, doi:.
Baumgardner
,
D.
,
H.
Jonsson
,
W.
Dawson
,
D.
O’Connor
, and
R.
Newton
,
2001
:
The cloud, aerosol and precipitation spectrometer: A new instrument for cloud investigations
.
Atmos. Res.
,
59–60
,
251
264
, doi:.
Cooper
,
W. A.
, and
G.
Vali
,
1981
:
The origin of ice in mountain cap clouds
.
J. Atmos. Sci.
,
38
,
1244
1259
, doi:.
Curry
,
J. A.
,
J. L.
Schramm
,
W. B.
Rossow
, and
D.
Randall
,
1996
:
Overview of Arctic cloud and radiation characteristics
.
J. Climate
,
9
,
1731
1764
, doi:.
Domine
,
F.
, and Coauthors
,
2011
:
The specific surface area and chemical composition of diamond dust near Barrow, Alaska
.
J. Geophys. Res.
,
116
,
D00R06
, doi:.
Ehrlich
,
A.
,
M.
Wendisch
,
E.
Bierwirth
,
A.
Herber
, and
A.
Schwarzenböck
,
2008
:
Ice crystal shape effects on solar radiative properties of Arctic mixed-phase clouds—Dependence on microphysical properties
.
Atmos. Res.
,
88
,
266
276
, doi:.
Fugal
,
J. P.
,
T. J.
Schultz
, and
R. A.
Shaw
,
2009
:
Practical methods for automated reconstruction and characterization of particles in digital in-line holograms
.
Meas. Sci. Technol.
.
20
,
075501
, doi:.
Gerber
,
H.
,
1991
:
Direct measurement of suspended particulate volume concentration and far-infrared extinction coefficient with a laser-diffraction instrument
.
Appl. Opt.
,
30
,
4824
4831
, doi:.
Gotaas
,
Y.
, and
C. S.
Benson
,
1965
:
The effect of suspended ice crystals on radiative cooling
.
J. Appl. Meteor.
,
4
,
446
453
, doi:.
Gultepe
,
I.
, and Coauthors
,
2014
:
Ice fog in Arctic during FRAM–Ice Fog project: Aviation and nowcasting applications
.
Bull. Amer. Meteor. Soc.
,
95
,
211
226
, doi:.
Gultepe
,
I.
, and Coauthors
,
2015
:
A review on ice fog measurements and modeling
.
Atmos. Res.
,
151
,
2
19
, doi:.
Henneberger
,
J. F.-W.
,
2013
: Mountain-top in-situ observations of mixed-phase clouds with a digital holographic instrument. Ph.D. dissertation, ETH Zurich, 125 pp., doi:.
Hobbs
,
P. V.
,
S.
Chang
, and
J. D.
Locatelli
,
1974
:
The dimensions and aggregation of ice crystals in natural clouds
.
J. Geophys. Res.
,
79
,
2199
2206
, doi:.
Hogan
,
A. W.
,
1975
:
Summer ice crystal precipitation at the South Pole
.
J. Appl. Meteor.
,
14
,
246
249
, doi:.
Intrieri
,
J. M.
, and
M. D.
Shupe
,
2004
:
Characteristics and radiative effects of diamond dust over the western Arctic Ocean region
.
J. Climate
,
17
,
2953
2960
, doi:.
Korolev
,
A.
,
2007
:
Limitations of the Wegener–Bergeron–Findeisen mechanism in the evolution of mixed-phase clouds
.
J. Atmos. Sci.
,
64
,
3372
3375
, doi:.
Korolev
,
A.
,
G. A.
Isaac
, and
J.
Hallett
,
1999
:
Ice particle habits in Arctic clouds
.
Geophys. Res. Lett.
,
26
,
1299
1302
, doi:.
Korolev
,
A.
,
E. F.
Emery
,
J. W.
Strapp
,
S. G.
Cober
, and
G. A.
Isaac
,
2013
:
Quantification of the effects of shattering on airborne ice particle measurements
.
J. Atmos. Oceanic Technol.
,
30
,
2527
2553
, doi:.
Lance
,
S.
,
C. A.
Brock
,
D.
Rogers
, and
J. A.
Gordon
,
2010
:
Water droplet calibration of the cloud droplet probe (CDP) and in-flight performance in liquid, ice and mixed-phase clouds during ARCPAC
.
Atmos. Meas. Tech.
,
3
,
1683
1706
, doi:.
Lawson
,
R. P.
, and
B. A.
Baker
,
2006
:
Improvement in determination of ice water content from two-dimensional particle imagery. Part II: Applications to collected data
.
J. Appl. Meteor. Climatol.
,
45
,
1291
1303
, doi:.
Lawson
,
R. P.
,
B. A.
Baker
,
P.
Zmarzly
,
D.
O’Connor
,
Q.
Mo
,
J.-F.
Gayet
, and
V.
Shcherbakov
,
2006a
:
Microphysical and optical properties of atmospheric ice crystals at South Pole Station
.
J. Appl. Meteor. Climatol.
,
45
,
1505
1524
, doi:.
Lawson
,
R. P.
,
D.
O’Connor
,
P.
Zmarzly
,
K.
Weaver
,
B.
Baker
,
Q.
Mo
, and
H.
Jonsson
,
2006b
:
The 2D-S (stereo) probe: Design and preliminary tests of a new airborne, high-speed, high-resolution particle imaging probe
.
J. Atmos. Oceanic Technol.
,
23
,
1462
1477
, doi:.
Lloyd
,
G.
, and Coauthors
,
2015
:
The origins of ice crystals measured in mixed-phase clouds at the high-Alpine site Jungfraujoch
.
Atmos. Chem. Phys.
,
15
,
12 953
12 969
, doi:.
Ohtake
,
T.
,
1978
:
Atmospheric ice crystals at the South Pole in summer
.
Antarct. J. U.S.
,
13
,
174
175
.
Ohtake
,
T.
, and
T.
Yogi
,
1979
:
Winter ice crystals at South Pole
.
Antarct. J. U.S.
,
14
,
201
203
.
Ohtake
,
T.
,
K.
Jayaweera
, and
K.-I.
Sakurai
,
1982
:
Observation of ice crystal formation in lower Arctic atmosphere
.
J. Atmos. Sci.
,
39
,
2898
2904
, doi:.
Rauber
,
R. M.
,
1981
: Microphysical processes in two stably stratified orographic cloud systems. M.S. thesis, Dept. of Atmospheric Science, Colorado State University, 151 pp.
Raupach
,
S. M. F.
,
H. J.
Vössing
,
J.
Curtius
, and
S.
Borrmann
,
2006
:
Digital crossed-beam holography for in situ imaging of atmospheric ice particles
.
J. Opt.
,
8A
,
796
806
, doi:.
Rogers
,
D. C.
, and
G.
Vali
,
1987
:
Ice crystal production by mountain surfaces
.
J. Climate Appl. Meteor.
,
26
,
1152
1168
, doi:.
Schenk
,
L. P.
, and Coauthors
,
2014
:
Characterization and first results of an ice nucleating particle measurement system based on counterflow virtual impactor technique
.
Atmos. Meas. Tech. Discuss.
,
7
,
10 585
10 617
, doi:.
Schmidt
,
S.
, and Coauthors
,
2015
:
In-situ single submicron particle composition analysis of ice residuals from mountain-top mixed-phase clouds in central Europe
.
Atmos. Chem. Phys. Discuss.
,
15
,
4677
4724
, doi:.
Seinfeld
,
J. H.
, and
S. N.
Pandis
,
2012
:
Atmospheric Chemistry and Physics: From Air Pollution to Climate Change.
Wiley, 1232 pp.
Ulanowski
,
Z.
,
P. H.
Kaye
,
E.
Hirst
,
R. S.
Greenaway
,
R. J.
Cotton
,
E.
Hesse
, and
C. T.
Collier
,
2014
:
Incidence of rough and irregular atmospheric ice particles from Small Ice Detector 3 measurements
.
Atmos. Chem. Phys.
,
14
,
1649
1662
, doi:.
Vali
,
G.
,
D.
Leon
, and
J. R.
Snider
,
2012
:
Ground-layer snow clouds
.
Quart. J. Roy. Meteor. Soc.
,
138
,
1507
1525
, doi:.
Vochezer
,
P.
,
E.
Järvinen
,
R.
Wagner
,
P.
Kupiszewski
,
T.
Leisner
, and
M.
Schnaiter
,
2015
:
In situ characterization of mixed phase clouds using the Small Ice Detector and the Particle Phase Discriminator
.
Atmos. Meas. Tech.
,
9
,
159
177
, doi:.
Walden
,
V. P.
,
S. G.
Warren
, and
E.
Tuttle
,
2003
:
Atmospheric ice crystals over the Antarctic Plateau in winter
.
J. Appl. Meteor.
,
42
,
1391
1405
, doi:.
Wendisch
,
M.
,
P.
Yang
, and
P.
Pilewskie
,
2007
:
Effects of ice crystal habit on thermal infrared radiative properties and forcing of cirrus
.
J. Geophys. Res.
,
112
,
D08201
, doi:.
Worringen
,
A.
, and Coauthors
,
2015
:
Single-particle characterization of ice-nucleating particles and ice particle residuals sampled by three different techniques
.
Atmos. Chem. Phys.
,
15
,
4161
4178
, doi:.

Footnotes

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).