• American Meteorological Society, 2000: Glossary of Meteorology. 855 pp., AMS.

  • Heymsfield, A. J., 1977: Precipitation development in stratiform ice clouds: A microphysical and dynamical study. J. Atmos. Sci., 34, 367381.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J. & , and R. M. Sabin, 1989: Cirrus crystal nucleation by homogeneous freezing of solution droplets. J. Atmos. Sci., 46, 22522264.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J. & , and L. M. Miloshevich, 1993: Homogeneous ice nucleation and supercooled liquid water in orographic wave clouds. J. Atmos. Sci., 50, 23352353.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., , P. C. Kennedy, , S. Massie, , C. Schmitt, , Z. Wang, , S. Haimov & , and A. Rangno, 2010: Aircraft-induced hole punch and canal clouds. Bull. Amer. Meteor. Soc., 91, 753766.

    • Search Google Scholar
    • Export Citation
  • Pruppacher, H. R., 1995: A new look at homogeneous ice nucleation in supercooled water drops. J. Atmos. Sci., 52, 10241933.

  • Rangno, A. & , and P. V. Hobbs, 1986: Deficits in ice particle concentration in stratiform clouds with top temperature < −30°C. Preprints, Conf. on Cloud Physics, Snowmass, CO, Amer. Meteor. Soc., 2023.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D. & , and W. L. Woodley, 2000: Deep convective clouds with sustained supercooled liquid water down to −37.5°C. Nature, 405, 440442.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., 2003: Cirrus cloud iridescence: a rare case study. Appl. Opt., 42, 486491.

  • Sassen, K. & , and G. C. Dodd, 1989: Haze particle nucleation simulation in cirrus clouds, and applications for numerical and lidar studies. J. Atmos. Sci., 46, 30053014.

    • Search Google Scholar
    • Export Citation
  • Shaw, J. A. & , and N. J. Pust, 2011: Icy wave-cloud lunar corona and cirrus iridescence. Appl. Opt., 50, F6F11.

  • WMO, 1975: International Cloud Atlas, Vol. I (revision of the original 1956 version). Manual on the Observation of Clouds and Other Meteors. World Meteorological Organization, No. 407, 155 pp.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Cirrocumulus cloud looking west at 2132 UTC. Airflow and shear are from west to east (bottom to top). Note the transverse waves along the upwind (lower) edge and the longitudinal striations along the wind near the middle and top of the photo. There are several clear “holes,” including one long longitudinal hole or “moat” near the leading edge. (Photo by Carlye Calvin.)

  • View in gallery

    Profiles of (a) temperature and dewpoint, (b) mixing ratio, (c) wind direction, and (d) wind speed, from the 0000 UTC 24 Mar 2011 Denver sounding. The sounding time was a few hours after the cloud sighting in Boulder, which is about 50 km northwest of the launch location. The higher moist layer and the top of the lower moist layer are highlighted in green.

  • View in gallery

    Near-simultaneous photographs of the cloud. (a) Photo (cropped) by Margaret LeMone, taken at ~2122 UTC looking west from a westbound car on Iris Avenue (Fig. 4). Note the presumed distrail in the cloud and contrail west of the cloud, both apparently made by the same aircraft in a wide left turn. The horizontal white lines mark the western edge of the cloud and the top of Mt. Sanitas. Two small trees (dotted white circle) were used as a landmark in photogrammetric calculations. A mark in the windshield appears in front of the right green light. Note that the bottom of the picture as originally cropped was above the traffic lights, but the trees were visible. (b) Photo by Sam Hall, taken at 2124 UTC looking west from the NCAR Foothills Lab (Fig. 4). Most of the apparent contrail, easily visible in Fig. 3a, seems merged with the cloud. The apparent distrail is at top middle. The horizontal white lines mark the western edge of the cloud, the top of Mt. Sanitas, and the horizon.

  • View in gallery

    Map of central and northern Boulder, Colorado, with 20-ft elevation contours on a U.S. Geological Survey quadrangle map. The position of Mt. Sanitas and the two camera locations are marked by crosses. Distance: d1 is between the two photograph locations; d2 is from LeMone's picture to the north-south line through the top of Mt. Sanitas; arrows lie along Iris Avenue. Hall's picture is from NCAR Foothills Laboratory building 0 (FL0).

  • View in gallery

    Time–height cross section at 6-min intervals of returned power from the vertical beam of the Platteville, Colorado, profiler. The color bar at bottom refers to the returned power in decibels. UTC time = MDT + 6 h. The vertical axis refers to (left) height (km) above sea level and (right) the corresponding pressure (hPa) in a standard atmosphere. Note the enhanced reflectivity (light blue) near 8 km from 1700 to 1900 UTC and near 9 km from 2100 to 2215 UTC.

  • View in gallery

    Diagram showing how the height of the cloud was determined. The red plus signs at bottom denote the locations of Mt. Sanitas, LeMone's photo, and Hall's photo along a common horizon.

  • View in gallery

    Infrared temperature as a function of time (UTC) from an upward-looking Heimann KT15.85 (9.6–11.5 micron) sensor at Radiometrics, located 4 km northeast of the NCAR Foothills Lab (see Fig. 4). The small temperature rise around 2100–2200 UTC is consistent with a high, thin cloud overhead. This infrared sensor is a component of a microwave profiling radiometer that provides data to http://madis.noaa.gov; real-time and archived data from a network of radiometer sites are available via http://madis.noaa.gov/rdmtr_stations.html. (Courtesy of Randolph Ware, Radiometrics.)

  • View in gallery

    Cloud base from the Vaisala CL31 ceilometer at the NSF-sponsored University of Colorado Skywatch Observatory. The enhancement of the green backscatter signal at about 7,250 m between about 2100 and 2200 UTC indicates the presence of a cloud. A thicker cloud below 4,000 m AGL was detected near 0000 UTC on 24 Mar. (Courtesy Department of Atmospheric and Oceanic Sciences, University of Colorado at Boulder.)

  • View in gallery

    Cirrocumulus cloud, looking west at 2140 UTC. Note the billows and virga at lower right. (Photo by Carlye Calvin.)

  • View in gallery

    Portion of cirrocumulus cloud, looking south at 1537 UTC, showing an east-west moat of clear air. White rectangle shows area enlarged in Fig. 10. (Photo by Carlye Calvin.)

  • View in gallery

    Close-up of moat with cirrus filaments, corresponding to the region enclosed by the white rectangle in Fig. 9.

  • View in gallery

    Close-up of hole at 1541 UTC, looking roughly along wind. (Photo by Carlye Calvin.)

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 84 84 8
PDF Downloads 46 46 7

A Striking Cloud Over Boulder, Colorado: What Is Its Altitude, and Why Does It Matter?

View More View Less
  • 1 National Center for Atmospheric Research,* Boulder, Colorado
  • | 2 Cooperative Institute for Research in Environmental Sciences, University of Colorado, and NOAA Earth System Research Laboratory, Boulder, Colorado
  • | 3 University Corporation for Atmospheric Research, Boulder, Colorado
© Get Permissions
Full access

Scientific investigation is supposed to be objective and strictly logical, but this is not always the case: the process that leads to a good conclusion can be messy. This narrative describes interactions among a group of scientists trying to solve a simple problem that had scientific implications. It started with the observation of a cloud exhibiting behavior associated with supercooled water and temperatures around −20°C. However, other aspects of the cloud suggested an altitude where the temperature was around −40°C. For several months following the appearance of the cloud on 23 March 2011, the people involved searched for evidence, formed strong opinions, argued, examined evidence more carefully, changed their minds, and searched for more evidence until they could reach agreement. While they concluded that the cloud was at the higher and colder altitude, evidence for supercooled liquid water at that altitude is not conclusive.

CORRESPONDING AUTHOR: Margaret A. LeMone, National Center for Atmospheric Research, Boulder, CO 80307-3000, E-mail: lemone@ucar.edu

*NCAR is sponsored by the National Science Foundation

Scientific investigation is supposed to be objective and strictly logical, but this is not always the case: the process that leads to a good conclusion can be messy. This narrative describes interactions among a group of scientists trying to solve a simple problem that had scientific implications. It started with the observation of a cloud exhibiting behavior associated with supercooled water and temperatures around −20°C. However, other aspects of the cloud suggested an altitude where the temperature was around −40°C. For several months following the appearance of the cloud on 23 March 2011, the people involved searched for evidence, formed strong opinions, argued, examined evidence more carefully, changed their minds, and searched for more evidence until they could reach agreement. While they concluded that the cloud was at the higher and colder altitude, evidence for supercooled liquid water at that altitude is not conclusive.

CORRESPONDING AUTHOR: Margaret A. LeMone, National Center for Atmospheric Research, Boulder, CO 80307-3000, E-mail: lemone@ucar.edu

*NCAR is sponsored by the National Science Foundation

OBSERVATION.

Around 2100 UTC (1500 MDT) on 23 March 2011, in Boulder, Colorado, a spectacular cloud layer, tentatively identified as cirrocumulus, formed overhead and lasted about 90 minutes. At the NCAR Foothills Laboratory at the northeast edge of town, Robert Henson alerted a colleague, photographer Carlye Calvin, who snapped about 30 photos. Margaret LeMone paused to photograph the cloud from the lab parking lot and, later, from her car on the way home, and then at home. In south Boulder, Tom Schlatter snapped a dozen pictures from NOAA's David Skaggs Research Laboratory.

Large sheets of cirrocumulus are rare in Boulder. The cloud layer (Fig. 1) was beautiful and intricate, with very small elements throughout, north–south waves that look like rippled sand along the upwind (lower) edge, and east–west striations near the middle and top of the photo. There are several clear “holes,” including one long east-west hole or “moat” (following the terminology used by Heymsfield et al. in a 2010 BAMS article) near the leading edge.

Fig. 1.
Fig. 1.

Cirrocumulus cloud looking west at 2132 UTC. Airflow and shear are from west to east (bottom to top). Note the transverse waves along the upwind (lower) edge and the longitudinal striations along the wind near the middle and top of the photo. There are several clear “holes,” including one long longitudinal hole or “moat” near the leading edge. (Photo by Carlye Calvin.)

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

During the next few days, LeMone drafted a blog-like article about the cloud formation for UCAR's AtmosNews website (see Acknowledgments). LeMone associated the cloud with a moist layer at around 8,500–9,000 m MSL1 in the Denver sounding (Fig. 2a), where the temperature was −40°C, rather than the top of a much lower moist layer at around 5,000 m, where the temperature was −19°C. She based her opinion on three pieces of evidence. First, the individual cloudlets or billows were less than a degree across (about twice the angle subtended by the full moon) and had no shading, fitting the definition of cirrocumulus in the International Cloud Atlas. Second, the eastern (downwind) part of the cloud appeared fibrous, suggesting that it was at cirrus level. And third, no one observed iridescence, which is usually a reliable indicator of water droplets provided that the droplets are of uniform size. [Iridescence has been observed in cirrus clouds, but only rarely. See, for example, Shaw and Pust (2011) or Sassen (2003).]

Fig. 2.
Fig. 2.

Profiles of (a) temperature and dewpoint, (b) mixing ratio, (c) wind direction, and (d) wind speed, from the 0000 UTC 24 Mar 2011 Denver sounding. The sounding time was a few hours after the cloud sighting in Boulder, which is about 50 km northwest of the launch location. The higher moist layer and the top of the lower moist layer are highlighted in green.

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

A colleague of LeMone's, Charles Knight, found LeMone's draft on a printer and came to talk to her about it. Knight had also witnessed the cloud, in which he noted behavior similar to that described in the Heymsfield et al. BAMS article. Drawing a parallel with the clouds in that article, Knight thought the clouds consisted of supercooled water droplets, with holes and slots developing as the droplets were converted to ice crystals that fell in thin streams of virga. This behavior would put the cloud height at the lower moist layer—around 5,000 m, where the temperature was −19°C. Andrew Heymsfield, another NCAR colleague, looked at Calvin's photos and agreed with Knight. In fact, he disagreed when LeMone interpreted a streak in her cloud photo as a distrail (Fig. 3a, top middle), a narrow, clear corridor in a cloud created by passage of a jet aircraft. In this case, Heymsfield thought the streak was merely a shadow cast by a contrail high above the cloud.

Fig. 3.
Fig. 3.

Near-simultaneous photographs of the cloud. (a) Photo (cropped) by Margaret LeMone, taken at ~2122 UTC looking west from a westbound car on Iris Avenue (Fig. 4). Note the presumed distrail in the cloud and contrail west of the cloud, both apparently made by the same aircraft in a wide left turn. The horizontal white lines mark the western edge of the cloud and the top of Mt. Sanitas. Two small trees (dotted white circle) were used as a landmark in photogrammetric calculations. A mark in the windshield appears in front of the right green light. Note that the bottom of the picture as originally cropped was above the traffic lights, but the trees were visible. (b) Photo by Sam Hall, taken at 2124 UTC looking west from the NCAR Foothills Lab (Fig. 4). Most of the apparent contrail, easily visible in Fig. 3a, seems merged with the cloud. The apparent distrail is at top middle. The horizontal white lines mark the western edge of the cloud, the top of Mt. Sanitas, and the horizon.

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

After the article was posted on 30 March, we took a closer look at available data and searched for new data sources that might be helpful in determining the cloud height.

The Denver radiosonde launched at 0000 UTC 24 March revealed that the two moist layers lay in a very dry deeper layer with considerable vertical shear of the horizontal wind. Dewpoint depressions in the moist layers were of the order of 10°C (Fig. 2a); the total precipitable water was only 3.93 mm. The winds in the upper moist layer (8,659–9,343 m, Fig. 2c,d) were westerly (260°) with an average speed of 59.6 m s−1, and a westerly vertical shear of 0.5 m s−1 per 100 m. The shear through the lower moist layer (4,267–5,115 m) was larger—west-northwesterly (254°) at 0.9 m s−1 per 100 m—but the winds averaged only about 12 m s−1 out of the southwest.

To Henson and Schlatter, the presence of strong winds in the upper moist layer was strong evidence that the cloud layer resided there: both had seen the cloud elements moving rapidly eastward. Also, a few cumulus clouds (not shown), capped by a stable layer near 5,200 m, appeared to be far below the cloud layer in question.

We also consulted infrared imagery from one of NOAA's Geostationary Operational Environmental Satellites (GOES). However, the cloud in question was very thin, with many interstices through which one could see blue sky from the ground. From the satellite's viewpoint, infrared radiation from the cloud was mixed with that coming from the ground and from the atmosphere. Thus, it was neither possible to infer a cloud-top temperature nor the corresponding cloud altitude (Lindsay Daniel, Colorado State University, personal communication).

Two other potentially useful sources of data were vertically pointing lidars operated by NOAA and NCAR in Boulder, and aircraft that sample water vapor at flight level and on ascent and descent. Unfortunately, neither lidar was operating, and no aircraft measurements were available near the time the cloud was observed.

TWO FORTUITOUS PHOTOGRAPHS FUEL DEBATE.

In response to the article, Sam Hall of NCAR forwarded some pictures to LeMone, including one that looked nearly identical to one that LeMone had taken from her car. In LeMone's picture (Fig. 3a), the presumed contrail and distrail are clearly visible. Hall's picture (Fig. 3b), taken from NCAR's Foothills Laboratory at 2124 UTC, showed the contrail mostly obscured by the cirrocumulus layer. If the pictures were taken at the same time, and if the contrail was higher than the cloud, as Heymsfield suggested, then the difference in position could explain why more of the contrail was visible to LeMone. LeMone's camera did not record when her photo was taken; but interpolating between the time she left work and downloaded the photo to her computer, LeMone judged that her photo was taken within about five minutes of Hall's photo.

Using the two images, LeMone sought to determine the cloud height photogrammetrically. Luckily, both cameras were pointing westward across a local landmark, Mt. Sanitas, and rays from the cameras to the western cloud edge were nearly in the same plane (Fig. 4). The distance from LeMone's photo location to the north–south line through the top of Mt. Sanitas (d2) and the distance between LeMone's and Hall's photo locations (d1) were easily measured on a map (Fig. 4). Hall was able to supply accurate positions and elevations for Mt. Sanitas and his office at NCAR's Foothills Lab. He used a theodolite, pointing horizontally, to determine landmarks in the photo at the same elevation as his camera. Such landmarks determine the horizon.

Fig. 4.
Fig. 4.

Map of central and northern Boulder, Colorado, with 20-ft elevation contours on a U.S. Geological Survey quadrangle map. The position of Mt. Sanitas and the two camera locations are marked by crosses. Distance: d1 is between the two photograph locations; d2 is from LeMone's picture to the north-south line through the top of Mt. Sanitas; arrows lie along Iris Avenue. Hall's picture is from NCAR Foothills Laboratory building 0 (FL0).

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

LeMone had to estimate the location of her photo, as it was taken from a moving car. This was difficult because she initially could find only a cropped image that cut out the traffic light in Fig. 3a, leaving only a few branches to help identify the location precisely. Moreover, since only the upper portion of the foothills was visible, LeMone had to estimate the height of Mt. Sanitas relative to her elevation by using features on the mountain as references.

The sidebar, with the help of Fig. 3, describes the geometry LeMone used to estimate the height of the cloud based on the height of Mt. Sanitas and the information about her photo and Hall's. LeMone arrived at a cloud height around two miles above ground level (AGL), or around 4,900 m MSL (Boulder's elevation is 1,650 m). The cloud seemed to correspond to the lower moist layer. LeMone was sufficiently convinced to send her estimate to colleagues on 12 April. The calculation was quick and dirty, but she thought it sufficiently accurate to show that the cloud layer was located closer to 5,000 than 9,000 m. In meters AGL, this amounts to roughly a factor of two in height (3,350 vs. 7,350 m AGL). Heymsfield and Knight were delighted at the findings, but Henson and Schlatter were skeptical.

MORE OBSERVATIONS.

Responding to a request from Schlatter, Douglas van de Kamp of NOAA provided his interpretation of wind profiler data from NOAA's Platteville, Colorado, site, about 45 km ENE of Boulder, on 13 April. Wind profiles for several hours centered at 2100 UTC on 23 March (not shown) were consistent with those from the Denver radiosonde—namely, strong winds of 45 m s−1 or more above 8-km altitude. Maxima in returned power are evident in Fig. 5 near 8 km from 1700 to 1900 UTC and again between 8.5 and 9 km from about 2100 to 2215 UTC, the interval during the cloud observation over Boulder. The profiler also indicated enhanced vertical velocity variance in these high-returned-power areas (not shown). There was enhanced power and vertical velocity variance at heights up to 5 km, but this appears related to the growing planetary boundary layer.

Fig. 5.
Fig. 5.

Time–height cross section at 6-min intervals of returned power from the vertical beam of the Platteville, Colorado, profiler. The color bar at bottom refers to the returned power in decibels. UTC time = MDT + 6 h. The vertical axis refers to (left) height (km) above sea level and (right) the corresponding pressure (hPa) in a standard atmosphere. Note the enhanced reflectivity (light blue) near 8 km from 1700 to 1900 UTC and near 9 km from 2100 to 2215 UTC.

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

According to Van de Kamp, elevated values of returned power are caused more often by moisture fluctuations than by temperature fluctuations, especially at high altitudes, though both are clearly involved. Similarly, the enhanced vertical velocity variance is consistent with the small-scale patterns in the cloud (Fig. 1). On balance, then, the profiler data favored a cloud at the higher rather than the lower altitude, but LeMone noted that high returned power and turbulence at 8.5–9.0 km did not necessarily imply a cloud layer at that height.

In late March, spurred by a lunchtime conversation, LeMone contacted colleagues in NCAR's Research Applications Laboratory (RAL) to see if they could track down the aircraft responsible for the contrail/distrail, thus potentially providing the height of the cloud. In early May, Jason Craig of RAL sent the routes and altitudes of aircraft crossing the Boulder area during the time interval, including one that was making a slow left turn as it flew west, just as LeMone had witnessed. The altitudes fit the high-cloud scenario, but the timing of the turn (2059 UTC) was 20–25 minutes too early, leading to another dead end: with the strong west wind, the resulting contrail/distrail would have been too far to the east.

Schlatter suggested a potential flaw in LeMone's photogrammetric calculations: an assumption one of the reference points used—the western edge of the cloud layer—was stationary. Using refined estimates of the horizon and camera locations, LeMone estimated the horizontal change in cloud-edge position that would be necessary to shift the height estimate from 4,900 m to 8,400 m. She arrived at a shift of 3.2 km. Was this plausible? Perhaps. The layer cloud appeared to be associated with a lee wave: a slight change in the ascent angle of the air trajectories to the east of the Rockies could lead to a large horizontal change in the western edge of the cloud. LeMone's certainty was eroding.

HOW THE HEIGHT OF THE CLOUD WAS DETERMINED PHOTOGRAMMETRICALLY

The first step is to estimate the heights h1 and h2 of the rays pointing from the cameras to the western edge of the cloud, relative to a common reference elevation: the horizontal line corresponding to the elevation of the photographers. Once these heights are known, the quantities a1 and a2 are readily computed from the first set of ratios shown in the figure. Finally, one can solve the two equations a1 = z / x and a2 = z / (xd1) for the two unknowns: the height of the cloud, z, and the horizontal distance from the cloud edge to the Foothills Lab, x.

Finding a1 is straightforward. On the photo itself (Fig. 3b), LeMone drew in the horizon that Hall had determined with a theodolite. The horizon is the lowest of three horizontal lines in the figure. She then measured the vertical distance between the horizon and the horizontal line through the top of Mt. Sanitas. Finally, she measured the vertical distance between the horizon and the western edge of the cloud. The height h1 is equal to the ratio of the larger to the smaller distance times the actual elevation difference between the top of Mt. Sanitas and the known elevation of the camera (surface elevation from Google Earth, plus the height of the camera in the building relative to the Earth's surface).

Finding a2 from LeMone's photo (Fig. 3a) is more difficult. Because the camera was pointing slightly upward toward the contrail, no horizon is visible. Thus it was necessary to return to the site where the photo was taken and take more photos. The horizon, identified as the horizontal line through the point at which the sides of Iris Avenue converged, could then be used as the reference line for measuring the height of Mt. Sanitas. In order for this measurement to apply to Fig. 3a, she used the two trees (circled) as a reference,2 finding that they lie approximately halfway between the horizon and the peak. Using this, she could measure vertically from the peak to the trees, multiply by approximately two, and use that information not only as an estimate of the vertical distance on the picture, but also to draw a horizon line just below the bottom of Fig. 3a. From this line, she could estimate the vertical distance to the cloud edge on the picture, providing the final piece of information needed to estimate h2. The new photos were also used to confirm the location of the photo in Fig. 3a. Corrections were later made for the slope of Iris Avenue and the differences in photographer elevation.

i1520-0477-94-6-788-f11

Diagram showing how the height of the cloud was determined. The red plus signs at bottom denote the locations of Mt. Sanitas, LeMone's photo, and Hall's photo along a common horizon.

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

STILL MORE OBSERVATIONS AND A CONVERGENCE OF OPINION.

By this time, “the cloud” was a frequent topic of discussion among LeMone and her acquaintances at lunch. Upon hearing about the lack of data from vertically looking lidars, a colleague suggested that she contact Randolph Ware, whose firm, Radiometrics, had a vertical sounder located about 4 km northeast of NCAR's Foothills Lab (Fig. 4). Ware sent a plot of data from an infrared (IR) radiometer, operating at wavelengths in the so-called water-vapor window, where absorption by water vapor is minimal (Fig. 6). When the atmosphere is dry and the sky clear (as it was for most of 23 March 2011), the instrument readings are affected by the cold of space. If a visually opaque cloud is present, the instrument records an infrared temperature close to the temperature of the cloud base. When the cloud was overhead, the IR temperature rose from less than 200 to 210 K (−63°C). Such a temperature is reached on the Denver sounding at around 15 km (not shown). This suggests, but does not prove, that the cloud was thin (the cold of space was mixed with the signal from the cloud) and at high altitude.

Fig. 6.
Fig. 6.

Infrared temperature as a function of time (UTC) from an upward-looking Heimann KT15.85 (9.6–11.5 micron) sensor at Radiometrics, located 4 km northeast of the NCAR Foothills Lab (see Fig. 4). The small temperature rise around 2100–2200 UTC is consistent with a high, thin cloud overhead. This infrared sensor is a component of a microwave profiling radiometer that provides data to http://madis.noaa.gov; real-time and archived data from a network of radiometer sites are available via http://madis.noaa.gov/rdmtr_stations.html. (Courtesy of Randolph Ware, Radiometrics.)

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

Ware suggested contacting Julie Lundquist, a faculty member at the University of Colorado, who sent data from the University of Colorado Skywatch ceilometer (Fig. 7), located about 4 km SW of Foothills Lab (Fig. 4). With this additional data, the cloud height was nailed: 7,250 m AGL (8,900 m MSL).

Fig. 7.
Fig. 7.

Cloud base from the Vaisala CL31 ceilometer at the NSF-sponsored University of Colorado Skywatch Observatory. The enhancement of the green backscatter signal at about 7,250 m between about 2100 and 2200 UTC indicates the presence of a cloud. A thicker cloud below 4,000 m AGL was detected near 0000 UTC on 24 Mar. (Courtesy Department of Atmospheric and Oceanic Sciences, University of Colorado at Boulder.)

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

Data continued to come in. On 27 April, Schlatter participated in a blood drive at the NOAA Labs in Boulder. Also at the blood drive was Daniel Wolfe, another NOAA employee. In casual conversation, Schlatter learned that Wolfe spends considerable time 20 km to the east at the Boulder Atmospheric Observatory (BAO). At his office later, Wolfe mentioned to Schlatter a sky camera at BAO that looks west and takes one photo per hour. Two of the photos on the afternoon of 23 March showed the cloud layer over Boulder, one during its formation in place and the other when it was well established. A pyrheliometer at the BAO measured a slight but obvious decrease in direct solar radiation for about an hour starting at 2100 UTC, when the cloud was most evident. This was the only disturbance that day to the normal clear-sky diurnal curve. A ceilometer at the BAO detects clouds up to 7,000 m above ground. It detected no clouds when solar radiation decreased, suggesting that any clouds reducing solar radiation at the ground were higher than 7,000 m—provided the cloud extended that far east.

LeMone's thoughts returned to her photogrammetric calculations. She recalled that the weather website maintained by NCAR's Research Applications Laboratory (http://weather.ral.ucar.edu) included current time-lapse videos of clouds. Checking the site, she noticed historical cloud videos. Could video from 23 March still be available? On 16 May, LeMone emailed Robert Rilling (NCAR Earth Observing Laboratory) to find out. By the 17th, she had the complete video. She and Schlatter examined it carefully and found that the western edge of the cloud shifted very little with time.

Now filled with doubt about her cloud-height estimate, LeMone redoubled her efforts to reconcile it with the newly available data. She used the cloud video to determine that the time of her photograph was about 2122 UTC, about two minutes before Hall's photo. The video also explained the lack of a visible contrail west of the cloud in Hall's picture: the contrail had been blown rapidly eastward relative to the nearly stationary western edge of the cloud, consistent with the strong wind at that level.

In striving to make a more precise calculation of the cloud height, LeMone made two more discoveries:

  • 1) She had used as the horizon the point where the sides of Iris Avenue converged, a valid procedure if the street is horizontal. However, it ascends toward the west (about 20 m per km).
  • 2) She needed to correct for the difference in elevation from which the photographs were taken.

LeMone also double-checked the ratio of the heights of the cloud edge and Mt. Sanitas, noting that the estimates of h1 and h2 are quite sensitive to these. She obtained a range of cloud heights around 6,200–8,600 m AGL (7,850–10,250 m MSL), closer to the height indicated by the ceilometer and other data. Because the lines of sight from the cameras to the cloud edge intersect at a small angle, the cloud-height calculations are quite sensitive to the input measurements.

The midpoint—7,400 m AGL—is close to the cloud height in Fig. 7. This height and the angle subtended by the cloud's upwind edge at Foothills Lab (α1 = 14.7°) indicates an upwind edge 28 km to the west of Foothills Lab. Photographs (not shown) indicate that the cloud layer extended at least as far east, confirming that the cloud layer was above the BAO ceilometer but not detected because of its high altitude.

At the request of a reviewer, Schlatter reexamined GOES images at full resolution. A loop of these images gave only the slightest hint of a cloud, hardly distinguishable from the ground in terms of brightness (albedo) and stretching east–west across central Boulder County east of the foothills. There was a clearer signature in the infrared images. The 11-micron brightness temperature of the cloud was about 10°C lower than the ground temperature. Since the air temperature at cloud altitude was near −40°C, the IR data are consistent with this cloud being very thin. It is probably no coincidence that the upward-looking radiometer indicated a 10°C positive excursion in brightness temperature against the background of space while the cloud was present.

CONFIRMED: A HIGH CLOUD. NOW WHAT?

The 2010 BAMS article by Heymsfield et al. noted that aircraft penetrating water droplet clouds caused conversion of water droplets to ice crystals, which then grew and precipitated out as virga. The holes in the 23 March cloud somewhat resembled the figures in the Heymsfield et al. article. Figures 3a and 3b suggested an aircraft penetration of the cloud. Indeed, LeMone thought she had witnessed the aircraft; but, as noted earlier, its presence and height could not be confirmed from available data. Knight had remembered seeing virga, and he and LeMone found virga when they examined Carlye Calvin's pictures carefully.

To confirm the presence of virga, LeMone systematically examined all of Calvin's high-resolution photos at different magnifications in mid-July. She isolated several pictures that seemed to have virga and sent the list to Schlatter and Henson. Schlatter took a quick look, but remained skeptical about the presence of virga.

We reconvened on 8 February 2012. After examining four or five photos, we agreed that virga was present, along with several other interesting features. Figure 8 shows virga at lower right, falling amid well-defined and two-dimensional billows near the upwind edge of the cloud. Figure 9 shows a long, east–west “moat” that contains cirrus filaments in the middle, which are more obvious in the blown-up portion (Fig. 10).

Fig. 8.
Fig. 8.

Cirrocumulus cloud, looking west at 2140 UTC. Note the billows and virga at lower right. (Photo by Carlye Calvin.)

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

Fig. 9.
Fig. 9.

Portion of cirrocumulus cloud, looking south at 1537 UTC, showing an east-west moat of clear air. White rectangle shows area enlarged in Fig. 10. (Photo by Carlye Calvin.)

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

Fig. 10.
Fig. 10.

Close-up of moat with cirrus filaments, corresponding to the region enclosed by the white rectangle in Fig. 9.

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

The features observed—holes, moats, and virga—were associated with water clouds by the authors of the Heymsfield et al. article. Furthermore, the billows at the lower right in Fig. 8 are solid white, have sharp edges, and are not fibrous—they look like water clouds. The virga at the lower right in Fig. 8 at least vaguely resembles that depicted in Heymsfield et al. Other virga observed, such as that in Figs. 9 and 10, are far more diffuse, but that could be explained in terms of the relatively tiny amounts of water vapor available: the saturation mixing ratio for water vapor at cloud level (−40°C and 300 hPa) is only 0.4 g kg−1, about a quarter of the amount available to produce ice crystals under conditions associated with “hole-punch clouds” (−20°C and 500 hPa). And one must consider the possibility that the temperature at that level could have been higher than −40°C given the time and space displacement between the sounding and the cloud observation.

Could this mean water droplets in a cloud whose temperature was near −40°C? Observational evidence suggests “yes.” Heymsfield and Miloshevich (1993) report observations of supercooled liquid water droplets down to −36°C in orographic wave clouds; they suggest that droplets can exist down to −40.7°C and lower, as do laboratory experiments reported in Pruppacher (1995). Water droplets at temperatures close to −40°C have been reported by Heymsfield (1977, −36°C), and Sassen and Dodd (1989, −37°C, orographic cirrus). Heymsfield and Sabin (1989) show liquid water in cirrus down to −35.3°, though observations are much more frequent in the −30° to −35°C range in their Table 1. This behavior has been attributed to a lack of ice-forming nuclei by Heymsfield (1977) and Rangno and Hobbs (1986). More recently, Rosenfeld and Woodley (2000) presented evidence of liquid water at temperatures down to −37.5°C in the upper parts of deep convection.

However, other features distinguish the cloud from those described in the Heymsfield et al. BAMS article. As noted earlier, no one observed iridescence, which is often but not infallibly used to differentiate ice from water clouds. Furthermore, the cloudlets in the eastern part of the layer were clearly fibrous. Indeed, a close-up of one of the holes, in Fig. 11, shows fibrous billows outside the hole with along-shear fibrous streaming cloudlets likely associated with secondary circulations. More significantly, there are gossamer billows in the holes in Figs. 8 and 11, rather than virga and clear air, in stark contrast to the holes in the Heymsfield et al. clouds. Finally, the time-lapse film shows that the moat in Figs. 9 and 10 originates at the cloud's upwind edge, rather than with the passage of a jet aircraft.

Fig. 11.
Fig. 11.

Close-up of hole at 1541 UTC, looking roughly along wind. (Photo by Carlye Calvin.)

Citation: Bulletin of the American Meteorological Society 94, 6; 10.1175/BAMS-D-12-00133.1

CONCLUSIONS.

The distinctive cloud layer observed over Boulder on the afternoon of 23 March 2011 has been determined to be “high” beyond a reasonable doubt, based on ceilometer cloud-base measurements roughly 4 km SW of Foothills Lab. The high altitude is also consistent with the rapid eastward movement of the cloud elements, given the strong westerlies at that level, as well as measurements from an upward-looking infrared sensor 4 km NE of the lab, ceilometer measurements 20 km to the east, IR satellite images, and the final photogrammetric estimate of cloud height.

With sharply outlined billows, virga, holes, and moats, this cloud resembled clouds described by Heymsfield et al. in a 2010 BAMS article that contained supercooled water at much higher temperatures than suggested here (“near” −40°C). The differences—finer virga, “holes” with barely visible billows, fully fibrous cloudlets in the downwind portion—could indicate different processes in an ice cloud, or they could be consistent with supercooled water with less vapor available.

Whether or not the cloud contains supercooled water, we still identify the cloud layer as “cirrocumulus.” This is consistent with the recognition in the Glossary of Meteorology that “cirrocumulus may be composed of highly supercooled water as well as ice crystals,” undoubtedly in response to some of the observations cited here. Moreover, the Glossary notes the possible presence of virga, as seen in this case.

The complexity of this cloud and its interesting features added to the challenge of determining its altitude. However, that complexity also gave the cloud the very beauty and delicacy that inspired the three authors to examine it in the first place.

This exercise illustrates some important lessons:

  • Photos gathered informally can be invaluable in analyzing cloud features, but it is essential that the photos be carefully referenced in time and space. Originals should be kept along with cropped photos.
  • It is surprising how many sources of relevant data are available if one searches hard enough (though Boulder may be blessed with more sources than most locales).
  • Quick-and-dirty calculations can often lead to erroneous conclusions.
  • Skepticism and (sometimes spirited) debate are important parts of the discovery process.

The process that led from first observation to conclusion illustrates the messy way science works. Our journey initially followed the traditional steps of the scientific method, as often described: making an observation, posing hypotheses regarding the cloud's height, collecting data, consulting the literature and colleagues, and presenting our interpretations to each other. The messiness lay in our rush to conclude that the cloud was high (or low). Our solution to this messiness was an aspect often neglected in descriptions of scientific method: peer review. In our case, the review took the form of questioning each other's results (Knight's skepticism regarding the “high” cloud altitude in LeMone's draft, Heymsfield's agreement with Knight based on Calvin's photographs, Henson and Schlatter's disagreeing with Knight and Heymsfield, Schlatter's questioning LeMone's initial photogrammetric calculations). The friendly tensions thus created were healthy, driving us to gather more data and refine the photogrammetric calculations until the results lined up in a consistent way. Without that interaction, “wrong” conclusions might have resulted. The additional step of formal review affords a chance to catch errors and clarify our message.

ACKNOWLEDGMENTS

This article draws from and builds on two posts on the UCAR online AtmosNews website (https://www2.x.edu/atmosnews/opinion/4616/cloud-remember-part-2-mystery-solved and https://www2.ucar.edu/atmosnews/opinion/4209/cloud-remember-and-mystery-solve). In addition to the people mentioned in the text and captions, many others contributed to the discussion on this cloud. In addition to those mentioned in the narrative, Jielun Sun (NCAR) referred us to datasets we were unfamiliar with, and Alan Scott Kittleman (University of Colorado) educated us about the contents of the CU Atmospheric Observatory website. We wish to acknowledge our internal reviewers, as well as Tammy Weckwerth (NCAR) and an anonymous reviewer for improving the paper through the formal review process. Two of the authors (Schlatter and LeMone) have modest postretirement support from their respective employers, NOAA and NCAR. Henson is supported by UCAR. NCAR and UCAR are supported by the National Science Foundation.

FOR FURTHER READING

  • American Meteorological Society, 2000: Glossary of Meteorology. 855 pp., AMS.

  • Heymsfield, A. J., 1977: Precipitation development in stratiform ice clouds: A microphysical and dynamical study. J. Atmos. Sci., 34, 367381.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J. & , and R. M. Sabin, 1989: Cirrus crystal nucleation by homogeneous freezing of solution droplets. J. Atmos. Sci., 46, 22522264.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J. & , and L. M. Miloshevich, 1993: Homogeneous ice nucleation and supercooled liquid water in orographic wave clouds. J. Atmos. Sci., 50, 23352353.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., , P. C. Kennedy, , S. Massie, , C. Schmitt, , Z. Wang, , S. Haimov & , and A. Rangno, 2010: Aircraft-induced hole punch and canal clouds. Bull. Amer. Meteor. Soc., 91, 753766.

    • Search Google Scholar
    • Export Citation
  • Pruppacher, H. R., 1995: A new look at homogeneous ice nucleation in supercooled water drops. J. Atmos. Sci., 52, 10241933.

  • Rangno, A. & , and P. V. Hobbs, 1986: Deficits in ice particle concentration in stratiform clouds with top temperature < −30°C. Preprints, Conf. on Cloud Physics, Snowmass, CO, Amer. Meteor. Soc., 2023.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D. & , and W. L. Woodley, 2000: Deep convective clouds with sustained supercooled liquid water down to −37.5°C. Nature, 405, 440442.

    • Search Google Scholar
    • Export Citation
  • Sassen, K., 2003: Cirrus cloud iridescence: a rare case study. Appl. Opt., 42, 486491.

  • Sassen, K. & , and G. C. Dodd, 1989: Haze particle nucleation simulation in cirrus clouds, and applications for numerical and lidar studies. J. Atmos. Sci., 46, 30053014.

    • Search Google Scholar
    • Export Citation
  • Shaw, J. A. & , and N. J. Pust, 2011: Icy wave-cloud lunar corona and cirrus iridescence. Appl. Opt., 50, F6F11.

  • WMO, 1975: International Cloud Atlas, Vol. I (revision of the original 1956 version). Manual on the Observation of Clouds and Other Meteors. World Meteorological Organization, No. 407, 155 pp.

    • Search Google Scholar
    • Export Citation

1 Unless otherwise indicated, altitudes will be given relative to mean sea level (MSL).

2 In LeMone's first calculations using the original cropped photo, the reference point was a rock outcrop slightly more than halfway between the two trees and the top of the ridge in the two pictures in Fig. 3, and a range of photo locations was used. The earlier “most probable location” of LeMone's photo was not far from the actual location. Use of the two trees and the correct location increased the cloud-height estimate only slightly compared to the first result with the best location, so the more recent calculation using Fig. 3a is highlighted for brevity.

Save