The 24 June 2006 Boulder hailstorm produced very heavy precipitation including disklike hailstones that grew with low density. These disklike hailstones, 4–5 cm in diameter, are unusual, and some of them appear to have accumulated graupel while aloft. A large amount of very fine-grained slush was left on the ground along with the hail. The hail and the great amount of slush suggest that most of the hydrometeor growth in the cloud was by low- or very-low-density riming. Consistent with that, the radar data suggest that the storm updraft had substantially depleted liquid water content. There is evidence that low-density hydrometeor growth within storms may be considerably more frequent than is commonly suspected.
Part I of this two-part paper (Schlatter et al. 2008) describes the meteorological conditions surrounding a June 2006 Boulder hailstorm, the bulk aspects of the storm’s precipitation, and its radar history. The storm was severe with respect to its appearance on radar and its heavy hailfall, and yet the instability did not seem conducive to very strong updrafts and did not justify a severe storm watch, though a warning was issued on the basis of the radar images. This enigma led to part of the interest in the storm, but another part of the interest lay in the unusual aspects of the precipitation at the ground. Some of the individual hailstones were extremely unusual, and the existence of a large quantity of slush on the ground after the storm passed, at the location with the most detailed observations, was unique in our experience. Overall, the precipitation exhibited a surprising dominance of low-density or very-low-density riming that most likely relates to the (presumably) weaker updrafts than would be expected for a severe hailstorm.
Low-density riming itself is not a novel phenomenon and has been incorporated into hail-growth representations within storm models (Ziegler et al. 1983; Farley 1987). As has been written (Farley 1987), “the general assumption of high particle density is not valid for much of an ice particle’s growth history, particularly for those continental clouds that produce precipitation primarily through ice processes. . . ,” as is the case on the high plains, in Colorado, and elsewhere. Thus, in retrospect this storm perhaps should not have been as surprising as it was, but phenomenologically it was certainly unusual.
We start with a detailed description of the precipitation, emphasizing the hailstones collected by one of the authors (Knight).
2. Observations at the time of the hailfall
At the main collection site, 7th Street and University Avenue in Boulder (the westernmost asterisk in Fig. 3 of Part I), hail commenced at roughly 1800 mountain daylight time (MDT). Hailfall had been anticipated from the unusual preceding quiet, the very dark sky, and perhaps muted thunder, which produced the feeling of an impending, severe storm at this site—“the lull before the storm.”
The very first precipitation consisted of infrequent, “large” hail that was visible as bursts of fragments from impacts on a brick-paved patio. There were none of the distinctly solid-sounding impacts that usually are heard even before seeing the first hail, and there was no collectable ice on the patio. The word large above is set in quotation marks because it was impossible to judge the size of the hail from the bursts of these small fragments. However, hailstones several centimeters in diameter could be seen impacting and remaining at least partly intact on the lawn. Ten of these were collected in individual, plastic sandwich bags and put into the freezer before the precipitation became very heavy. At that time the larger hailstones (“stones” seems an inappropriate description for this kind of hail) became mingled with smaller hail and slush on the ground, and the falling hail made it too painful for observers to continue the collection. Despite the softness of the hail, its impacts were substantial. It stripped leaves and large amounts of twigs and small branches from the trees and decimated flowers, but caused no local property damage otherwise. Six of the 10 hailstones will be illustrated below, but during collection it was noted that they were roughly disk shaped, up to 4–5 cm in their longest dimension, and all were somewhat cupped—convex on one side and concave on the other. The outer surfaces on the convex sides felt quite firm. The concave sides tended to be very lumpy for the most part. The immediate visual impression was that some of the lumps looked like individual, conical graupel that had aggregated on the disklike “host.” This was a surprising and most unexpected occurrence, as will be noted below, and care was exercised not to disturb the upper, lumpy but concave sides while placing the hailstones in plastic bags and then into the freezer.
These larger, soft hailstones may have lost parts of their substance colliding with the lawn, and the collisions may have deformed them, though in handling they seemed rigid. It cannot be said with complete confidence that their characteristics as collected were very much like their characteristics during fall. However, since they were not accompanied by much small hail and the collections were made before significant amounts of small hail fell, there is nearly complete confidence that no aggregation occurred on the ground. Likewise, the distinctly harder convex sides provide fairly strong confidence that the somewhat cupped shapes of the disks were not caused by collision with the lawn. The lawn area was sheltered from wind, but the hail was falling with a significant horizontal component of velocity: very roughly, about 30° from the vertical, from the north.
Very heavy precipitation commenced a few minutes after the onset of the large hail, greatly decreasing visibility even across the street, 10–15 m away. More hail was collected after the precipitation stopped, and the strikingly unusual feature at that time was the character of the precipitation layer remaining on the ground. It was about 4 cm deep and consisted of slush containing occasional pieces of hard, white ice that one would normally call hailstones, around 1 cm in diameter. These were very distinct from the grayish, almost liquid slush and constituted perhaps 10% of the total volume. The visual contrast resulted from the fact that the slush contained very few air bubbles and thus appeared gray in comparison with the “hailstones,” which were white because of light scattering from the air bubbles. (See Fig. 1 in Part I, especially at the bottom center and right of the photo.) The slush was very fine-grained and homogeneous, both visually or tactually. Despite the fact that it had to have been an accumulation of individual hydrometeors, it did not appear that way. In the dozens of times we sampled hail from the ground in eastern Colorado, we had never before seen slush on the ground with the hail, nor have we heard this reported. Actually, it is questionable whether it would be reported even if it were fairly common. Also, unless such a “hail–slushfall” was very heavy, as in the present, rare case, the slush would melt quickly on the ground.
A number of the smaller, hard, white nuggets of ice within the slush were collected and sectioned. These may have been small portions of larger hydrometeors with very soft outsides, but most of the heavy precipitation was certainly very-low-density rime soaked with liquid water. There is no way of telling the size of the biggest hydrometeors during the “deluge” stage of the precipitation.
The collections were taken from the home freezer to the cold room the following day and sections were made and photographed the following week.
Three other hail collections from the same storm were made by meteorologists in Boulder, at locations shown by black stars on Fig. 3 of Part I, by J. Brown, the third author, and H. Bluestein. The four collections in Boulder give a consistent picture of the hailstones, and here the first, most extensive, and freshest collection provides most of the illustrations. Regarding that collection, the first author especially regrets not having observed more purposefully the collisions of the larger hail with the lawn, as his attention was more focused on collecting and preserving individual hailstones. Also, the value of trying to do something more with the slush was not appreciated soon enough. Carefully sampling portions of it and freezing them solid would have been useful for verifying small crystal size and perhaps for seeing inhomogeneities within it.
3. Aspects of hailstone growth relevant for interpreting the hailstones
Hailstone growth is called dry or wet depending upon a hailstone’s temperature. Growth occurs by accretion of supercooled water droplets, which commence freezing when they touch the ice, releasing heat of fusion that warms the hailstones above the cloud temperature. With increasing accretion rate (increasing cloud water content), the hailstones are warmer up to the limit of 0°C, past which not all of the accreted water can freeze. Up to that limit the growth is termed dry (there are complications because neither the accretion rate nor ventilation are uniform over the hailstone surface), and past that limit the unfrozen liquid accumulates and the growth is “wet.” The accumulated liquid may flow over the surface and shed into the airstream, in which case hailstone growth is much like icicle growth, or if the cloud temperature is far enough below freezing the ice grows as a mesh of crystals including the liquid, and the hailstone growth is termed spongy (List 1959; Macklin 1961).
There is a wide spectrum of possible growth from very dry to very spongy at subzero cloud temperatures. It is not possible to be very quantitative about where the hailstones from the Boulder storm fit in this spectrum, but it is possible to reach firm qualitative conclusions from knowledge of the structures of natural and artificially grown hailstones. The two features used for the deductions are the air bubble structures and the crystal textures of the hailstones. The familiar hailstone layering mostly represents transitions between environments of dry growth, generally characterized by bubbly ice that appears white in scattered, reflected light, and wet growth, for which the ice is transparent or nearly so.
Starting with the observation that the hail was mostly soft when it fell, one knows that the only ways for hail to be soft are at the two ends of the spectrum of growth conditions described above: either spongy growth with quite a lot of included water, or very dry growth, which produces low-density rime. Rime with low density is familiar as graupel—usually rimed snow crystals. It may occur at much larger sizes (Pflaum 1984) and is familiar in small hail (Hallett 1965; Knight and Knight 1968, 1973), where it often has absorbed liquid water and then frozen solid, so it has become both hard and dense.
Both spongy hail and soaked, low-density rime are hard and dense after collection and storage, but it is easy to distinguish the two origins qualitatively from the air bubble structures alone (Knight and Knight 1968; Pflaum 1984). The hailstones from the Boulder hailstorm all indicate that the growth was entirely dry, ranging from low- to high-density riming, with only one example even approaching wet growth and no spongy growth at all.
In this paper “low-” and “very-low-” density hail are always used in reference to hail as it grows within cloud, before soaking in liquid water, and the terms are largely qualitative because there is no way to quantify them. A theoretical low-density limit in ideal riming is 0.17 g cm−3, or about 81% void space (Buser and Aufdermaur 1973), assuming that all the water droplets impact traveling in the same direction and freeze instantly as spheres upon touching the deposit. There are no relevant experimental data that we know of, so all that we have to estimate the original density is the air bubble content and structure. We use very low density for the cases when soaking expels virtually all of the interstitial air, leaving the ice translucent after freezing. Low density then refers to rime with lower permeability, such that in the soaking process a lot of air is trapped and the ice ends up very bubbly and opaque. High density refers to rime that is impermeable, having small air bubbles layered in response to small changes in the cloud environment. This ice ranges from opaque to nearly transparent as the growth approaches wet. This classification connects the appearance of the ice to its original density in a way that is qualitative but useful. We might guess that very low means roughly 0.1–0.4 g cm−3, low density about 0.4–0.7, and high >0.7.
Crystal size in hailstones is a direct though crude indicator of air temperature during growth (Levi and Aufdermaur 1970). This is relevant in supporting the interpretations of low-density growth. Other things being equal, lower air temperature favors lower density because the droplets freeze more quickly when they collide and spread out less. At the fall velocities of hailstones, this is important because the more the droplets spread out the denser the deposit will be. The ice interpreted here as forming with low or very low density has very small crystal grains.
Figure 1 illustrates air bubble structures within an ordinary, fairly large, high-density hailstone. (It is seen with a bright background, so the light areas are bubble free, and darkness signifies air bubble content.) The conical internal part in Fig. 1a, the embryo, exhibits the transition from very low-density growth at the small, upper end of the cone to low-density growth at the very dark, bubbly, bottom end of the cone during the period when the embryo fell at a more-or-less constant orientation. Then, the hailstone started to tumble (Knight and Knight 1970b), with its perimeter, seen in Fig. 1b, growing faster than the right and left sides of the elliptical cross section seen in Fig. 1a. The bubble-free ice where the growth was fastest (the thickest growth layer) is a result of wet but not spongy growth, and parts of the slowest-growth layers in Fig. 1a, especially at the right and left sides of the ellipse, look like low-density growth.
4. The fresh collection of smaller “hail” that was embedded in slush
Figure 2 shows air bubble and crystal structures of five of the hailstones collected from the slush just after the hailfall ceased. Any very soft, slushy parts that formed originally as very-low-density rime would be missing, having merged with the surrounding slush. The original, preserved lowest-density rime is found especially in the very top portions of the hailstones in Figs. 2b–d, the top portions representing the earliest growth, where the riming occurs at lowest impact velocity. The few air bubbles are distributed haphazardly. In contrast, almost all of Fig. 2e shows the original, as-grown air bubble structure, reflecting the fine details of the changes in time of its growth environment as it grew as nearly solid ice. The interiors of the others show mostly plentiful but rather chaotic air bubbles, which we interpret as low-density rime that was somewhat permeable so that liquid could enter in but air bubbles would be trapped in the process. These areas are interspersed with more structured air bubbles indicating high-density riming.
These five hailstones are also shown between crossed polarizing filters, revealing the crystal size (Fig. 2). All portions interpreted as originating as very-low-density rime have small crystal size. This implies that the air temperature where that rime accumulated must have been below −15°C. Most parts with high-density riming have large crystals, but the correlation is not perfect.
5. The larger, disk-shaped hailstones
In retrospect, the most significant aspect of the precipitation may actually be the immense amount of slush rather than the larger hailstones, but it was the unusual character (near uniqueness) of the bigger hailstones that stimulated the initial interest in the storm’s precipitation. Figures 3 –8 contain photos of different aspects of 6 of the 10 hailstones collected in the earliest phase of the hailfall. As noted above, all 10 gave the impression while being collected that one side was convex and the other concave. Most had lumps that were especially prominent on the concave side, while the convex side appeared lumpy on a smaller scale but was relatively uniform and quite firm. Panel a in each of Figs. 3, 4 and 6 –8 shows the rounded, convex side, while each panel b shows the concave side, and the single exterior photo in Fig. 5 shows the concave side. (The concavity of the concave side is not evident in some of the cross sections, but the contrast between the two sides was very striking while collecting and handling the hailstones, and seeing them in three dimensions.) Most of the hailstones were sectioned first perpendicular to the plane of the disk and purposely through the centers of one or more of the lumps. Then, the two halves were frozen back together and the reconstituted hailstone was sectioned again, parallel to the plane of the disk. The sections perpendicular to the plane of the disk are presented with their convex sides toward the bottom of the page. For Fig. 6 the first section was parallel to the plane of the disk, which was quite thin, and there was not enough material remaining to make another section worthwhile.
Figure 3 shows an exceptionally thick disk, but otherwise it is typical in its lumpiness both in the exterior views and as reflected in the “clumpy” air bubble structure. The views of thin sections (Fig. 3c and 3d) are with diffuse backlight so that air bubbles show up dark. The larger black spots are air bubbles that intersect the section surface and were filled with ice powder during section preparation. Note particularly in Fig. 3d (but also at the left end of Fig. 3c) the rather distinct concentric disposition and radial elongation of the air bubble clumps. This is interpreted as being from an object that grew by collecting supercooled droplets, not by accumulating graupel. The ice is entirely fine-grained, except for a central area a little less than 1 cm in diameter that is coarse-grained (not shown: polarized light views are not included in Fig. 3). Note how much less radial and concentric structure is present in the air bubbles than is present in most hail (e.g., Fig. 1).
Figure 4 is much like Fig. 3 except in being much more distinctly asymmetric end to end and a much thinner disk. One other of the 10 hailstones collected, but not shown, is very similar to this one. In Fig. 4b the dark portion is dark partly because of the lighting (using two diffuse light sources above and to the right and the left, with the camera looking straight down and the hailstone resting on a black background) and perhaps partly because the ice at the concave surface is less bubbly than that at the convex surface.
The hailstone in Fig. 5 was a thin disk with very pronounced lumps on the top side. While being collected, at least two of these lumps were noted to resemble conical small hail that had aggregated onto the disk. The thin section was made through these two lumps, and they still appear to have been individual, small, conical hailstones. The polarized-light view Fig. 5c shows all fine-grained ice (like the hail in Figs. 3 and 4), but in the section parallel to the plane of the reconstituted disk (not shown) there is an area of large crystals located at the center.
The general public often furnishes descriptions of large hail as being agglomerations of smaller hail. Our response to this until now has been that this is impossible: when hailstones collide in free fall they bounce off one another, and the appearance as agglomerations is always due to lobe structure that developed during growth (e.g., Knight and Knight 1970a). The interior structures of large hailstones have always demonstrated this: That the lumps represent a growth shape instability, not agglomeration. The hailstone in Fig. 5 and several others in the collection are the first hailstones we have ever seen where agglomeration does appear likely. In Fig. 5 especially, the two prominent lumps identified in Fig. 5b look like conical graupel with their symmetry axes parallel to the surface of the hailstone but with their apexes pointed in opposite directions. Agglomeration of soft hail is possible, perhaps especially below the freezing level where the very-low-density rime becomes slushy. Wet particles may adhere with the aid of surface tension, especially if they are soft, cushioning the collision.
The exterior photos of the hailstone in Fig. 6 show the embryo from the concave side at the right-hand side of Fig. 6b. The thin sections (Figs. 6c and 6d) show again the somewhat clumpy, chaotic air bubbles, but also reveal that the embryo is a conical piece of graupel composed of large crystals (Fig. 6d).
Figure 7 is distinguished by having a very prominent lump near the center of the concave side of the disk, which is seen in the polarized light section to have distinctly contrasting, large crystals while the disk portion consists entirely of small crystals. This may be the embryo of the hailstone, or it may have aggregated to the disk.
Figure 8 is perhaps most comparable to Fig. 5, showing lumps on the concave side. Some of these look as if they were once individual, conical hailstones. The section in Figs. 8c and 8d was cut to intersect the prominent lumps seen in Fig. 8b, and again the crystal sizes are nearly all small.
Figure 9 shows sections of four hailstones that were not as well collected and preserved, but that show that the collection in Figs. 3 –8 is not unique. Figures 9a–d are plain (upper) and polarized light (lower) views of sections across disklike hailstones collected from the ground at the southern edge of Boulder about one-half of an hour after the storm had passed. They both have distinct embryos on what look like the concave sides of the disks. Figures 9e and 9f are sections from two disklike stones from a different storm: in Lakewood, Colorado, on 27 August 2002 (collected by C. Birkenheuer). That day was notable for storms producing great quantities of hail (as in the 24 June storm), and these were picked up and saved because of their odd shapes. The hailstone sections in Fig. 9 are quite similar to most of the group represented by Figs. 3 –8, in their disk shapes, their mostly small crystal size except for the embryo portions, and in their asymmetry: The embryos are more or less central but on one side of the disk.
We discount the possibility that these are just soft hailstones flattened into disks by collision with the ground. The disks seemed much too stiff to have been deformed plastically to that extent less than a minute before collection. Thus, the working assumption is that the fresh examples in Figs. 3 –8 represent fairly well what the hailstones were like before collision with the ground, except possibly for losing an unknown amount of very soft slush from their exteriors.
Disk-shaped hailstones have been noted in the literature many times (e.g., Smyth et al. 1999) often being called oblate, rather than disklike. The degree of oblateness is highly variable, in our experience. Figure 1 is one example. The disk-shaped hailstones from the 2006 Boulder storm are different from the usual oblate hailstones in two respects. First, they grew as low-density rime according to the present interpretation, while all of the previously described ones probably were high density, and second, they exhibit much more growth on one side than on the other, unlike the example in Fig. 1.
The usual, high-density, oblate hail has been explained by symmetrical tumbling motions of the falling hail during growth (Knight and Knight 1970b; Garcia-Garcia and List 1992), but the extreme one-sidedness of the low-density, oblate hail might require a different tumbling geometry from that shown in Garcia-Garcia and List (1992, Fig. 3). Their suggested motion has the gyration axis horizontal and there is rotation about the disk axis as well. The hail shown here in Figs. 3 –9 would be much easier to explain if the gyration axis were vertical and the gyration angle not far from 90°, so that the growth on the disk rim predominates, along with almost completely one-sided growth.1
Experiments with free-falling models of different shapes, sizes, and densities could help elucidate the falling motions, but the experiments are difficult (e.g., Stewart and List 1983). Experiments with model systems show complicated variations of freefall tumbling motions that depend upon many parameters (Stringham et al. 1969).
6. Concluding discussion: How common is low-density hail?
Part I of this two-part series contains a description of an unusual hailstorm, its meteorological setting, and its radar history. Part II, this paper, has related what we know about the unusual hail that it produced. The main stimulus for putting these reports together was the recognition of the novelty of the event together with the detailed hail observations and the exceptionally complete supporting data. It was fortuitous that this storm caught the attention of individuals equipped to recognize and react to the opportunity. The storm produced a lot of hail that grew with low or very low densities, including some up to about 5 cm in “diameter” (but far from spherical) with features not reported previously. Since the interesting aspects of this storm most likely would have gone unremarked upon had it not occurred where and when it did, it is worth revisiting the possibility that low-density growth of medium-sized hail might actually be a common occurrence aloft, within clouds. This is by no means a new thought. Lozowski et al. (1991) reference a number of others who have expressed this opinion, and Farley (1987, quoted above) points out correctly that it is to be expected. But despite it not being a new thought, it may not be a common thought.
a. How common is low-density hail growth?
The disk-shaped hailstones in one other storm (Fig. 9) at least show that the Boulder storm is not unique, but there is evidence of other cases of low-density growth, and there are arguments that such growth may in fact be quite common aloft. Hail that grew with low density may be fairly common at the ground, but only rarely recognized. We have talked with a number of reliable sources who witnessed large hail splattering upon impact. One similar report from the Boulder storm was of tennis-ball-sized hail that “exploded” on impacting a roof. Another occurrence was observed in Fort Collins, Colorado, on 29 April 2001, by the third author, whose notes mentioned “small snowballs 1–1.5 inches in diameter . . . splatting on the sidewalk and bouncing in the grass . . . [which] appeared to be made up of smaller rounded graupel-like pellets fused together.”
Hail can be soft because of either spongy or low-density growth, and it is clear that in the Boulder storm the reason is low density. After analyzing the Boulder storm, most of the reports of soft hail seem likely to be further evidence of low-density growth rather than spongy growth. The hailstones from the Boulder storm not only showed no structural evidence of spongy growth, they showed almost no evidence even of wet growth.
One direct indication of low-density hail comes from penetrations of thunderstorms by the T-28, the armored, instrumented aircraft operated by the South Dakota School of Mines and Technology. A hail spectrometer, a device similar to a 2D probe but much larger, recorded sizes and concentrations of large hydrometeors. Smith et al. (1980) reported hydrometeors with diameters of up to 5 cm, present in concentrations that would have produced a much stronger radar echo than that which was measured if they had been normal hail. Smith et al. concluded, “We therefore believe that the particles are giant low-density graupel of some sort,” continuing to remark that “they do not appear in the precipitation at the ground so they must break up somehow during descent.” This was in storm penetrations in both Colorado and Oklahoma. These observations are compatible with the precipitation from the Boulder hailstorm. Unfortunately, there is no record of hydrometeor size during the “deluge” stage of the hailfall that produced the great volume of slush in the Boulder storm.
Are conditions conducive to very-low-density hail expected to be common within hailstorms? The answer is, probably, “yes.” First, very-low-density hail growth is physically possible, though it has not been studied in nearly as much detail as high-density growth. According to Pflaum’s (1984) report of his experiments, he could obtain “density ranging from 0.1 to 0.8 g cm−3 . . . at ambient temperatures of −10, −17, and −25°C.” He did not report the quantitative dependence of rime density upon the growth parameters, but for the low-density growth his cloud probably had a median volume droplet radius of about 5 μm (he used that and 12 μm for his two cloud settings), and he used wind velocities up to 35 m s−1.
The general problem in understanding how storms produce hail has usually been to explain how high-enough supercooled water content is maintained for long-enough times within strong-enough updrafts for the hail to grow. For very-low-density hail, however, the favorable combination of conditions includes low-enough supercooled water content so that the hailstone temperature remains well below the freezing point, combined with small droplet size, low cloud temperature and more moderate updraft strength because the fall speeds are lower. These conditions evidently were satisfied in the Boulder storm. We see no reason to think that conditions for low-density growth are uncommon—in fact, this combination of conditions should be common. The exceptional feature of the Boulder storm for producing a lot of soft hail, some relatively large, may have been the horizontal extent of the updraft [indicated by the large bounded weak echo regions (BWER; see Part I, and Fig. 10 here], which would tend to provide long residence times for growth. An exceptionally broad updraft implies exceptionally low entrainment, so the supercooled cloud water content would likely be too high for very-low-density growth unless depleted by smaller hydrometeors: presumably the large concentration of low-density graupel represented later by the thick layer of slush at the ground.
Examining the radar echo structure and history of the Boulder storm with the above reasoning in mind, we find evidence that the supercooled cloud water content probably was substantially depleted. As was related in Part I, there were prominent BWERs in the storm. These are confidently interpreted as the strong updrafts, based both on a lot of previous research and confirmed in this case by Doppler radar data showing radial convergence at low levels and divergence at upper levels associated with them.
Figure 10 shows six successive PPIs of the storm reflectivity structure that illustrate the development in time of BWERS, all from 5.3° elevation angle scans. The elevations at the vault in the figure range from about 8.5 km MSL at first (roughly −30°C in the environment, or −25°C within adiabatic cloud, based upon the red parcel ascent curve in Fig. 9 of Part I), to about 7.5 km at the end. One of the two BWERs identified in Fig. 18f of Part I is seen in the first panel of Fig. 10 (2230 UTC), and the other in the second and third panels (2335 and 2339 UTC). These are noteworthy in the present context for having developed within the storm’s reflectivity core by local weakening of the radar echo, and for having their weak echo largely within the range of 30–40 dBZ: weak only being relative to their surroundings. The later BWER, fully developed at 2349 and 2354 UTC, formed south of the storm’s echo core by the radar echo extending outward and encircling it, and contained a much weaker echo. (Note that the earlier BWERs are not resolved as well as the later one by the radar data, so that their echo strength may be somewhat overestimated, but the potential overestimation appears not to be enough to account for the difference observed.) More important, perhaps, is the fact that the later BWER developed by precipitation growth around the edges of an updraft impulse consisting of air containing little or no precipitation, whereas the earlier ones represent updrafts probably forming within air that already contained substantial amounts of precipitation.
The cloud water content in the updrafts of the earlier BWERs was no doubt substantially less than adiabatic, because of the relatively strong radar echoes within them, mostly 30 dBZ and greater, which were also present at lower scans; not shown. The time of the heavy hailfall in Boulder was about 2345–2355 UTC, when values of reflectivity over 65 dBZ in the 0.5° elevation angle scan were located over the hail collection sites. This was about 10 min after these BWERs. Any hail related to the last BWER may have fallen in the unpopulated area between Boulder (B in Fig. 10) and Golden, Colorado. Harder, more damaging hail might have been expected from it according to this argument, but neither an informal survey long after the time of the storm, nor insurance records, provided any evidence of more damaging hail there.
In the context of the question of how common low-density growth within clouds might be, one might also ask how often radar shows new turrets in multicell storms rising from regions of substantially strong radar echo. In the experience of the first author, this is very common, in fact it is the rule rather than the exception in Colorado and vicinity. A striking example of a strong convective impulse generated within precipitation-filled air is illustrated in Fig. 11 of Knight et al. (2004); not shown here. In that case the preexisting echo was quite uniform, so that the new development was clearly defined. The cloud water for growth of the hydrometeors causing the strong and rapid increase of the radar echo intensity must have resulted from cloud formation within air already containing precipitation-sized hydrometeors. If a strong, precipitation-free updraft rises into precipitation-filled air, the divergence at its top tends to reject precipitation, and the updraft could literally punch a hole up through preexisting radar echo.
It appears likely that very-low-density growth of hail is more common than has been suspected, but further aircraft measurements within storms could test this.
Microphysical processes of hydrometeor formation and growth within convective storms are already complicated, but environments containing very-low-density hail several centimeters across add more complexity yet. Collisions between these particles or between these and ordinary hail would doubtless produce fragments both small and large. This thought recalled evidence of hailstones breaking in midair. A storm at night in Grover, Colorado, on 11 June 1973 left drifts of golf-ball-sized and larger, high-density hail that did a great deal of damage to property. Many hailstones were collected (by N. Knight) and 194 of them were sectioned soon after the storm and photographed to provide a record of hailstone variability in a severe hailstorm. The structural variability was extreme, with hailstones varying from almost entirely bubble free (wet but not spongy growth) to mostly dry with some low-density growth. Figure 11 shows examples of the several percent of the hailstones that grew from large fragments of other hailstones. It had been puzzling why these would break in midair, but very-low-density growth now supplies an appealing explanation for hail being weak enough to be fractured by midair collisions. The other possible explanation, fracturing caused by the expansion upon freezing of included, confined water (this is seen fairly often), looks less likely for the hailstones in Fig. 11.
The radar data of the Boulder storm, as shown particularly in Fig. 10, suggest the importance of the unsteadiness of the airflow in the hail formation process. BWERs have been associated in the literature with a steady-state airflow model of supercell hailstorms, stemming from analyses of long-lived supercells by Browning (1963) and Browning and Foote (1976). Here, on the other hand, there are very pronounced BWERs but they do not last long. The airflow is unsteady, and it may well be that the unsteadiness itself is the most important factor. The convective impulses that produced the first BWERs originated within air containing precipitation while the one that produced the later BWER originated in air nearly free of precipitation. That factor alone could be of major importance in hail production, since that initial precipitation not only provides potential embryos for hail, but it also decreases the supercooled cloud water content below the adiabatic value. Hard hail versus soft hail versus no hail may often depend upon just where an updraft impulse started with respect to the location of the precipitation already produced.
This observation is not new except perhaps in placing an increased emphasis upon the beginnings of the updraft impulses. However, it may represent a less restrictive and more meaningful way of characterizing hail events than the supercell–organized multicell dichotomy that developed out of observations in the National Hail Research Experiment (Browning 1977). The usefulness of that distinction is compromised by the fact that, as illustrated by the Boulder storm, very few actual storms closely resemble either of the idealizations.
These speculations may be worth keeping in mind as more data become available. The WSR-88D (NEXRAD) radar network in the United States has made it possible to obtain radar data on many hailstorms with temporal and spatial resolution adequate for research purposes. When the network is equipped with polarization capability these data will be even more powerful for hailstorm studies, though the difficulty of usually not having adequate ground truth regarding hailfall will remain.
Finally, the Boulder hailstorm provides an example of a storm that looks very severe on radar but probably had exceptionally weak updrafts for the larger hail produced. The low density of the growing hail aloft would decrease the fall speeds appreciably (reduced by about 30% if the density is reduced by half). Also the bigger, disklike stones could have substantially increased drag coefficients, depending upon their falling behavior, reducing the fall speed even more. These lower fall speeds help to reconcile the presence of well-developed BWERs with the not-very-unstable sounding. The softness of the hail was undoubtedly a big factor in the small amount of property damage from the storm, but for agricultural damage that would have made little difference for many crops. The densities were high near the ground, so the impact energies were still high.
Helpful comments from John Brown and Paul Field are gratefully acknowledged, as are comments from the referees.
* The National Center for Atmospheric Research is sponsored by the National Science Foundation
+ Additional affiliation: Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado
Corresponding author address: Dr. Charles A. Knight, NCAR, P.O. Box 3000, Boulder, CO 80307-3000. Email: firstname.lastname@example.org
Visualize a rod with a disk at its end, fixed at right angles to the rod. The disk represents the hailstone. The rod is suspended from its upper end with a universal joint, the kind that allows the rod to swing free but not rotate (like a ship’s compass, mounted on gimbals). Now visualize the rod not hanging straight down but at an angle of (say) 20° from the horizontal, and swinging around the pivot point in a great circle while maintaining the 20° angle. One side of the disk always faces up and the other down, and the lowest point on the rim travels once around the rim every revolution. A hailstone tumbling in this fashion would grow uniformly around the rim and on the side of the disk facing downward but very little on the side facing upward. Finally, shift the fixed point of reference to the center of the disk, and that is the visualized, gyrating mode. One aspect of this gyration can be seen by spinning a coin on its edge on a tabletop and observing as the spin decays.