Recent laboratory experiments and in situ observations have produced results in broad agreement with respect to ice crystal habits in the atmosphere. These studies reveal that the ice crystal habit at −20°C is platelike, extending to −40°C, and not columnar as indicated in many habit diagrams found in atmospheric science journals and texts. These diagrams were typically derived decades ago from laboratory studies, some with inherent habit bias, or from combinations of laboratory and in situ observations at the ground, observations that often did not account for habit modification by precipitation from overlying clouds of varying temperatures. Habit predictions from these diagrams often disagreed with in situ observations at temperatures below −20°C. More recent laboratory and in situ studies have achieved a consensus on atmospheric ice crystal habits that differs from the traditional habit diagrams. These newer results can now be combined to give a comprehensive description of ice crystal habits for the atmosphere as a function of temperature and ice supersaturation for temperatures from 0° to −70°C, a description dominated by irregular and imperfect crystals. Cloud particle imager (CPI) habit observations made during the Second Alliance Icing Research Study (AIRS II) and elsewhere corroborate this comprehensive habit description, and a new habit diagram is derived from these results.
Since 2004, a comprehensive picture of atmospheric ice crystal behavior has been revealed through both laboratory studies (Bailey and Hallett 2002, 2004, hereafter BH02 and BH04) and field observations utilizing the Spec. Inc. CPI (Lawson et al. 1998), a nonimpacting cloud particle imager. This has been confirmed with observations made during the Second Alliance Icing Research Study (AIRS II) and several additional studies utilizing the CPI (Korolev et al. 1999, 2000; Baker and Lawson 2006, hereafter BL06; Lawson et al. 2006a,b, hereafter LB06a and LB06b). While those who read atmospheric science journals might be aware of these advances, most outside the field are not. Erroneous descriptions of ice crystal habits are currently found on the Internet, and most atmospheric texts have yet to be revised (except for Wallace and Hobbs 2006), misleading those new to the subject who have an interest in or need for accurate habit information.
A review of over 70 years of ice crystal studies reveals a bewildering variety of habit diagrams describing ice crystal shape as a function of temperature and ice supersaturation. The diagrams were drawn from laboratory studies, in situ observations, or combinations of the two. These include combinations of separate laboratory results as can be found in texts such as Pruppacher and Klett (1997) in addition to field observations of snowfall typically obtained at the ground (Magono and Lee 1966) as shown in Fig. 1. Although almost all the habit diagrams agree on the behavior of ice crystals at temperatures between 0° and −18°C, they differ substantially for lower temperatures. Aircraft observations of ice crystals in clouds offer still a different picture of ice crystal habits at temperatures below −20°C, especially with respect to laboratory studies and observations at the ground.
Two significant misconceptions involving ice crystal shape are common in the current habit diagrams. First is the designation of a columnar habit beginning at −20°C. This misconception evolved from an erroneous combination of biased laboratory observations (BH02) and field observations at the ground and in clouds that did not take into account the modification of habit distribution that can occur during sedimentation, combining habits from different temperatures. This led some in the past (Chen and Lamb 1994) to erroneously propose an oscillatory habit behavior, from plates (0° to −4°C) to columns (−4° to −8°C) to plates (−8° to −22°C) and back to columns again (T ≈ −22°C), a misinterpretation that continues to appear in habit descriptions (e.g., Hashino and Tripoli 2007).
Second is an overemphasis on symmetric shapes. While aesthetically appealing and offering a striking subject for photography, the fact is that most ice crystals are defective and irregular in shape to varying degrees and are mostly polycrystalline at temperatures below −20°C. Many are so irregular that they were mistakenly identified as aggregates in past studies. In their classic book on snow crystals, Bentley and Humphries (1931) noted that symmetric snowflakes or dendrites were the rare exception, with poorly formed and irregular branched dendritic crystals making up the overwhelming majority, which anyone with a magnifying glass can confirm on a snowy day. Tape (1994), in his book on halos observed in the Antarctic, noted that, even under the relatively stable growth conditions that produce halo crystals, the vast majority of ice crystals are defective in shape with rough surfaces. The fact is that ice is one of the most disordered crystals in nature and takes on a wide variety of complex shapes and habit themes as a function of temperature, ice supersaturation, and atmospheric pressure. Although they are rarely the idealized hexagonal shapes depicted in the habit diagrams, they have their own aesthetic appeal and are fascinating in their complexity, and it is time for a revised habit diagram that reflects these characteristics and the new revelations concerning habit, temperature, and ice supersaturation.
2. AIRS II and other field studies
Early laboratory studies were performed before the modern era of aircraft observations of ice crystals, which followed the invention of replicators, 2DC, and other imaging probes. One exception to this was the pioneering work of H. K. Weickmann (Weickmann 1945; aufm Kampe et al. 1951) in Germany during the Second World War, who flew an open cockpit plane through clouds ranging from stratiform to cirrus. While holding lacquer-coated shingles in the airstream, Weickmann captured and replicated ice crystals, obtaining a fairly accurate picture of in situ habits as a function of temperature. Unfortunately for the atmospheric community, Weickmann’s work went largely unnoticed while more attention was paid to the work of Nakaya (1954), Kobayashi (1957, 1961), and others, which at times was skewed by experimental artifacts as discussed in BH02. Weickmann’s field observations, together with subsequent laboratory studies (aufm Kampe et al. 1951), indicated the predominance of plates and platelike polycrystalline forms at temperatures near −20°C and colder, with a variable mix of hollow and solid columnar (“prism”) forms, as opposed to the columnar habit designated in many of the currently accepted habit diagrams, although even Weickmann at times misidentified complex polycrystals as aggregates.
In situ observations from aircraft employing replicators suffered from the fact that many of the collected crystals were shattered on impact, except for the more robust solid crystals like thick plates, columns, and compact polycrystals. The 2DC images suffered from poor resolution; hence, habit identification was often ambiguous or incorrect. For example, there are long, narrow, platelike polycrystals, for example, “scrolls” and “spearheads” (related to side planes) which grow between −20° and −40°C (BH04), that likely were misidentified as columns or bullets in pixilated 2DC images. Even with these limitations, it was soon evident that in situ habit observations often differed substantially from the laboratory results for clouds with temperatures below −20°C, casting doubt on the relevance of laboratory ice crystal studies at lower temperatures.
In 2002 the opportunity arose to examine thousands of CPI images gathered by Korolev and Isaac (Environment Canada) from studies of continental, maritime, and arctic stratiform and cumulus clouds [e.g., the First International Satellite Cloud Climatology Project (ISCCP) Regional Experiment Arctic Cloud Experiment (FIRE ACE), AIRS, and the Canadian Freezing Drizzle Experiment (CFDE III)]. Prior to this, the habit diagram in Fig. 2 had already been established from newer laboratory growth studies (BH02; BH04). The CPI images revealed that at times, between −20° and −40°C, bullet rosettes, columns, and mixed rosettes (both bullet and platelike components) appeared with platelike polycrystals (e.g., side planes, crossed plates, assemblages of plates, etc.), while at other times predominantly platelike habits were observed for approximately the same temperature regime. In the absence of nucleation bias (BH02), ice crystal habit is known to be, first, a function of temperature, with ice supersaturation playing a secondary role whose influence increases with increasing relative humidity. Thus, the varying appearance of columnar and platelike forms at the same temperature appeared contradictory. However, when flight log records were compared with crystal images, a pattern emerged. Platelike polycrystals dominated the habit distribution when cloud tops were warmer than approximately −40°C and, additionally, there were few or no overlying cirrus clouds. When cloud top temperatures were brlow −40°C or when overlying cirrus was present, a mix of columnar and platelike polycrystals, including bullet and mixed rosettes, was observed. These observations agreed with the laboratory habit results in BH04 and further indicated that overlying cirrus clouds were often the source of the columnar forms appearing in the platelike growth regime between −20° and −40°C.
In the fall of 2003, the authors participated in the Second Alliance Icing Research Study conducted in the Great Lakes region with the NCAR C-130 and a number of other aircraft. The goal of the study was to obtain meteorological and cloud microphysical measurements of mixed phase clouds with substantial liquid water content, which are responsible for aircraft icing. Mixed phase clouds with temperatures from 0° to −35°C were encountered during flights, as well as colder clouds with temperatures as low as approximately −50°C. Hence, the project provided an excellent opportunity for observing ice crystal habits under a broad range of conditions for comparison with the laboratory results in BH04.
Among the instruments deployed for this project were particle imagers, including a CPI, and both large and standard format cloudscopes provided by the Desert Research Institute (DRI). Also on board were a dual wavelength lidar, two FSSPs, a counterflow virtual impactor (CVI), and cloud condensation nuclei (CCN) counters. Other instruments provided by DRI included a large and small “T-Probe” (Hallett and Vidaurre 2006), an impaction device that measures total water content (TWC) and liquid water content (LWC) in a manner similar to the Nevzorov probe, but with substantially different probe geometries and collection efficiencies. Measurements with many of these instruments have recently been shown to be affected by particle shattering on instrument surfaces, discussions of which can be found in Korolev and Isaac (2005) and Isaac et al. (2006).
On one particular mission (1 December 2003) the C-130 made several passes through the tops of cumulus towers over Lake Huron with cloud top temperature near −37°C. Brief episodes of supercooled droplets were encountered in these towers at temperatures as low as −35°C. Figure 3 shows a sample of lidar data and CPI images obtained while passing through one of these clouds, which is visible at the right side of the lidar figure near 2035 UTC. Because ice crystal habit is first a function of temperature, it can be seen that some of the crystals observed have traveled a long way in updrafts from the regions where they likely nucleated. For instance, the large six-branched stellar plates in the lower right quadrant, partially rimed, are habits that typically nucleate at temperatures above −20°C. Except for a small number of short columns, thick plates, and skeletal plates of varying complexity, the majority of the habits observed were the complex platelike polycrystals commonly observed in the laboratory between −20° and −40°C near water saturation, as indicated in Fig. 2. While a detailed habit frequency analysis was not performed, the observed habit distribution was consistent with those observed in situ by others under similar conditions (Korolev et al. 1999, 2000; BL06; LB06a,b).
While some of the complex platelike assemblages in Fig. 3 contain columnar components (“mixed rosettes”), there were a few rosettes (below center) that may have been true bullet rosettes; however, some of the components appear to be platelike, rendering them mixed rosettes. Visually, the sky above cloud top was clear at the time, and lidar data showed no significant backscattering that would indicate the presence of overlying cirrus that could have been the source of columnar forms falling into these clouds (a weak signal from an altitude of 6–7 km might indicate the presence of subvisual cirrus). These observations essentially confirm the hypothesis that true bullet rosettes and columnar forms, in general, are not likely to nucleate at temperatures higher than approximately −40°C, as in the laboratory, although short columns appear with low frequency throughout the platelike region between approximately −20° and −40°C (BH04).
These habit results are also corroborated by extensive observations of snowfall at the ground in the Arctic by Kajikawa et al. (1980) and in the Antarctic by Ohtake and Inoue (1980), often precipitating in clear skies and after an intrusion of moist air aloft at heights of 1000 to 4500 m. When soundings showed upper air temperatures between −22° and −28°C, thin plates were mainly observed. Between −30° and −45°C, complex side plane crystals in addition to bullets and columns were observed, and between −40° and −55°C, bullet rosettes were typically observed.
Similar results were recently reported, in BL06 and LB06a,b of ice crystals observed in wave, cirrus, and arctic clouds. In these cases, they found that true bullet rosettes were rarely observed when cloud top temperatures were no colder than −40°C; and at warmer temperatures, 95% of rosette-shaped crystals were not true bullet rosettes, containing side planes and other complex growth structures. Examples of similar crystals observed during AIRS II are shown in Fig. 4. However, they did note that true bullet rosettes were observed on rare occasions when air or cloud temperatures were above −40°C (−36°C in one wave cloud and −33°C for crystals collected at the ground at the South Pole), with no apparent overlying cirrus clouds. In the laboratory, bullet rosettes generally nucleate at a temperature right at, or just below, −40°C and at an ice supersaturation intermediate to ice and water saturation, typically around 25% with respect to ice at −40° to −45°C (BH04). Thus, either −40°C is a true habit transition from predominantly platelike to columnar forms or a particular mode of nucleation, differing from the case in BH04, causes true bullet rosettes to appear on rare occasions at temperatures higher than −40°C, although subvisual cirrus “seed” crystals cannot currently be discounted as a source of columnar forms when cloud top or air temperatures are higher than but near −40°C.
It is worth noting, as in LB06a, that supercooled liquid water is rarely observed at temperatures below −37°C. Most droplets that freeze at −20°C are polycrystalline in nature, and from in situ observations of polycrystalline ice habit frequency, this trend increases with decreasing temperature, with polycrystalline bullet rosettes and mixed rosettes continuing to dominate at temperatures below −40°C in the absence of liquid water. Hence, polycrystalline forms are dominant under most conditions below −20°C; however, they exhibit the characteristics of the particular habit regime in which they grow, that is, platelike versus columnar. It should also be noted that in BL06 and LB06a,b the general term “rosette shape” includes bullet rosettes, mixed-habit rosettes, and platelike polycrystals with a rosette shape, so bullet rosette shapes should generally not be used to model crystals at temperatures above −40°C.
3. The new habit diagram
Figure 5 shows the new habit diagram drawn from laboratory results and from CPI images obtained during AIRS II for temperatures from −1° to −50°C (a pictorial glossary of complex polycrystalline forms and a discussion of their supersaturation dependence can be found in BH04). This new habit diagram retains the description of habits from the older diagrams for temperatures above −18°C, that is, from plates (0° to −4°C) to columns (−4° to −8°C) to plates (−8° to −22°C). Crystals at lower temperatures shown in Fig. 5 are taken from −50°C data and are representative of what is confirmed observationally by LB06a for temperatures down to −61°C and in the laboratory (BH04). The habit results for −70°C at the left side of the figure are similar to those at −60°C, characterized by “budding” rosettes, small compact irregular polycrystals, small columns, plates, and “spheroids”—particles that are too small to be resolved in CPI particle images and appear quasi-spherical (discussed in section 5) but are usually faceted in the laboratory. Bullet rosettes are typically not observed in cirrus at −70°C, but do occur on rare occasions as observed during the Costa Rica Aura Validation Experiment (CR-AVE) by Lawson (P. Lawson 2008, personal communication). The absence of in situ observations of well-developed rosettes in comparison with the laboratory results appears to indicate that effective in situ ice supersaturation at −70°C, including ventilation effects, rarely exceeds 40% with respect to ice (RHice = 140%) and is often considerably lower, which is consistent with the majority of measurements in cirrus clouds at low temperatures as concluded by Ovarlez et al. (2002) and as recently measured by Garrett et al. (2005) during the Cirrus Regional Study of Tropical Anvils and Cirrus Layers–Florida-Area Cirrus Experiment (CRYSTAL-FACE). Habits similar to those observed in the laboratory at −70°C were also observed in subvisual cirrus at temperatures of −75°C and colder by Lawson et al. (2008) during CR-AVE, although instrumental measurements of the ice supersaturation at which these habits appeared exhibited large discrepancies (Jensen et al. 2008).
The term “budding rosette” (BL06) is a description of a class of barely developed bullet, plate, or mixed habit rosettes that appear in CPI data and the laboratory for temperatures lower than −30°C or so. Because they are small, it is often unclear whether the “buds” are columns, thick plates, or something more complex. The buds likely follow the general habit as a function of temperature, although mixed habit rosettes are apparent in CPI data and in the laboratory for temperatures from −30° to −40°C, at all but the lowest ice supersaturation, and down to −60°C for ice supersaturation of 10%–25% (Fig. 5, lower left).
4. Modification of columnar and rosette habits precipitating from cirrus clouds
While studying the CPI images provided by Korolev and Isaac, another pattern emerged with respect to bullet rosettes and columns. Rosette “bullets” and columns grown in the laboratory and observed at in situ at temperatures lower than −40°C typically have aspect ratios of 3–4 or greater, except in the case of single columns growing at low to moderate ice supersaturations (below approximately 20%). However, the in situ bullets and columns observed at temperatures between −20° and −40°C were often much more robust and stout in appearance; that is, they were larger but had considerably smaller aspect ratios, similar to those of short columns that occasionally nucleate and grow in this temperature region, although typically at low to moderate ice supersaturation and with relatively low frequency (BH04). Also, at times, some in situ bullets and columns observed in this warmer temperature region have ends that are capped or skirted with platelike structures, and quite often are hollow to some extent. Since such habits and habit details were rarely observed in the laboratory for crystals that nucleated and grew between approximately −20° and −40°C (a regions with a low frequency of columnar forms), there was reason to suspect that these changes were due to the change in growth regime, from columnar to platelike, that would be experienced by a cirrus crystal precipitating to lower altitudes where platelike habits dominate. Hence, a series of experiments was performed in the laboratory with a static diffusion chamber, which simulated this process.
Columns and bullet rosettes were initially grown at cirrus temperatures (T < −40°C) and pressures (P ≈ 300 mb) with ice supersaturation in excess of 25%. Then the diffusion chamber temperature and pressure were slowly increased at a rate approximating the changing conditions of a precipitating cirrus crystal to warmer temperatures (e.g., −30°C < T < −20°C, 400 mb < P < 550 mb). The ice supersaturation was varied from low values up to water saturation to simulate the possible intracloud and intercloud conditions. The results are summarized in Fig. 6. At ice supersaturations approximately intermediate to ice and water saturation, transitioning bullets remained solid or grew plate “caps” on their ends. Near water saturation, bullets became hollow to varying degrees, sometimes growing platelike skirts around the open end, but in all transition cases, aspect ratios decreased as the bullets grew in width, as can be seen in Fig. 6. This is the response of a column growing in a platelike growth regime.
The shift in habit characteristics due to fall is summarized in Fig. 7 with AIRS II CPI images of bullet rosettes and mixed rosettes gathered at various temperatures. As noted in LB06b, most of the rosettes observed at temperatures above −40°C are mixed rosettes with platelike components. Since this is a platelike growth regime, these secondary growth components can have relatively fast linear growth rates compared with columns, especially near water saturation, often overtaking the thickening bullets in size and length.
Another artifact of the change in rosette habit can be seen by comparing “cold” bullet rosettes below −40°C with mixed rosettes at higher temperatures. The true bullet rosettes have bullets that are usually interpenetrating with respect to the core or kernel of the rosette, whereas bullets at warmer temperatures, especially those which have grown near water saturation, have pinched or tapered bases (see Fig. 6). This can be explained as a response to changing growth conditions. Bullets in the platelike growth regime start to grow in width; however, the bases of the bullets are recessed in the rosette structure and hence are shadowed from the level of ice supersaturation experienced by the exposed ends. So, the base of the bullet grows more slowly, resulting in a tapered shape. It is plausible that the growth of bullet rosettes at high ice supersaturation and temperatures below −40°C might also exhibit tapered bullets, although this could be an indication of bullet rosettes that have cycled through the −40°C level owing to sedimentation and updrafts.
5. Additional habit observations
The traditional habit diagrams have generally been in close agreement with respect to the habits for temperatures warmer than the dendrite growth regime, which terminates at approximately −18°C, and it is only this part of the old habit diagrams that is retained in the new one. Below this temperature are two habit regimes that are dominated by polycrystalline forms at nearly all ice supersaturations: a platelike growth regime between −20° and −40°C and a columnar regime between approximately −40° and −70°C. The intermediate ice supersaturation regime reveals a mix of mostly polycrystals with some columns and plates, with bullet rosettes nucleating at ice supersaturations of around 25% for temperatures below −40°C, according to BH04. As described in BH02 and BH04, the habit at low ice supersaturation (bottom of Figs. 2 and 5) forms a subset of ice crystal habits that shows little temperature dependence, except for a small increase in short columnar forms with decreasing temperature. In the habit diagrams, this is often described as a region of “equilibrium shapes,” another mischaracterization of ice crystal shape. At nearly all temperatures below −20°C and for low ice supersaturation, the habit is a mix of plates and short columns (often distorted to some extent) but more frequently composed of compact faceted polyhedra, crystals referred to as “irregulars” or spheroids in BL06 and LB06a,b.
The bottom left region of Fig. 5 describes crystals as compact faceted polycrystals. Small crystals of these types are like the small irregulars in BL06 and LB06a,b with sizes less than approximately 50 μm. Examples of these can be found in the lower left corner and along the bottom of Fig. 5. In the laboratory these occur at very low ice supersaturation (1%–2%) up to around 10%–15%, and smaller versions of these crystals are likely the unresolved spheroids shown in those papers. However, the term “spheroid” should not be misconstrued to imply quasi-spherical or spherical ice particles, a shape that has often been used in past models of high-level cirrus particles. Except for the possible case of frozen solution drops, they are, in fact, compact polyhedra with many small facets and irregular structures on their surfaces. Crystals of this type with sizes of 5–20 μm have been produced in the laboratory with the DRI fall tower at temperatures of around −42°C and are not spherical, as can be seen in Fig. 8. These crystals were grown from 5-μm drops generated with a sonic nebulizer that nucleated, either homogeneously or by contact with background aerosol, while falling 5 m in the refrigerated fall tower into a large cold box. Microscopic images were obtained with the DRI high-resolution cloudscope, which can resolve crystals down to a size of about 3 microns. Most of the particles that look quasi-spherical at lower magnification are faceted or show emerging facets at a size of 5–10 μm, while somewhat larger compact shapes are clearly faceted or are budding rosettes. Very few of the 5-μm particles were frozen spheres.
The laboratory habit observations of BH02 and BH04 are in agreement with the analysis of well over a million CPI images of in situ ice crystals collected from a broad range of temperature, ice supersaturation, and cloud types. The consensus between laboratory and in situ observations has resulted in a comprehensive habit diagram that effectively describes the shapes of atmospheric ice crystals as a function of temperature and ice supersaturation. Knowledge or estimates of cloud top and bottom temperatures in comparison with the new habit diagram can be used to estimate ice crystal habit for many scenarios.
The new habit diagram retains the well-established descriptions of habits from the older diagrams for temperatures above −18°C, that is, from plates (0° to −4°C) to columns (−4° to −8°C) to plates (−8° to −22°C). However, it diverges from the previous habit diagrams for lower temperatures and reveals that the habit is dominated by polycrystals of various forms with two distinct habit regimes: platelike from −20° to −40°C and columnar from −40° to −70°C. The new diagram also emphasizes the fact that most individual crystals are complex, irregular, and imperfect in appearance to some extent, in agreement with Korolev et al. (1999, 2000), and includes single crystals such as plates and columns that are most common at low ice supersaturation. Additionally, the majority of very small crystals growing at low ice supersaturation are compact faceted polycrystals as opposed to spheroids.
The laboratory habit diagrams in Fig. 2 and the top of Fig. 5 were constructed from crystals observed under static growth conditions and match the in situ observations as long as sedimentation or updrafts do not substantially impact the overall habit distribution. This would be the case for crystals whose growth was restricted to either the platelike or columnar growth regimes. It is clear, however, that substantial changes in habit can occur due to vertical motions, and a history of nucleation and changing growth conditions is often preserved in the habit details of modified crystals. Simulations in the laboratory have reproduced a commonly observed transformation in habit that occurs when columns and bullet rosettes precipitate from temperatures below −40°C into the warmer platelike growth regime at temperatures from −20° to −40°C, as often occurs in the atmosphere. Bullet rosettes typically become mixed rosettes with platelike components, with bullets and columns experiencing a reduction in aspect ratio while growing larger and wider—sometimes becoming hollow or capped, depending on ice supersaturation. The transformation of shape and habit for bullet rosettes and other crystal habits due to sedimentation indicates that further laboratory growth studies involving changing growth conditions would add significantly to our knowledge of ice crystal shape and mass in clouds.
This research was supported by the National Science Foundation, Physical Meteorology Program (ATM-9900560) and NASA (NAGS-7973). AIRS II research was sponsored by NSF (ATM-0313581). Fall tower experiments were supported by the U.S. Air Force Office of Scientific Research (F49620-00-1-0215).
Corresponding author address: Matthew Bailey, Desert Research Institute, 2215 Raggio Parkway, Reno, NV 89512. Email: email@example.com