1. Introduction
Blowing snow is quite common over mountains (e.g., Vionnet et al. 2013) and in high-latitude regions such as Antarctica (e.g., Mahesh et al. 2003; Palm et al. 2011). Most studies of blowing snow have focused on the resulting spatial snow redistribution, which is important for watershed hydrology (e.g., MacDonald et al. 2009). Other blowing snow studies are motivated by reduced visibility and snow drifts over roads and runways (e.g., Schmidt 1982; Nakai et al. 2012). Several surface snow models have been developed, sometimes coupled with land surface schemes, to numerically simulate snowpack dynamics over complex terrain, including wind-driven snow redistribution (Lehning et al. 2006; Liston and Elder 2006; Yang and Yau 2008; MacDonald et al. 2009; Bernhardt et al. 2010; Vionnet et al. 2013; Winstral et al. 2002; Chung et al. 2011). The snow erosion, drifting, and sedimentation are parameterized as a function of wind speed and other atmospheric, terrain, vegetation, and snowpack characteristics. These models are not coupled to an atmospheric model, that is, the lofted ice particles do not interact with cloud processes.
This study focuses on blowing snow as a source of ice crystals in supercooled boundary layer clouds, and thus as a natural mechanism of glaciogenic seeding of clouds that are coupled to the underlying surface, such as shallow orographic clouds or lake-effect clouds. Blowing snow ice particles (BIPs) typically hover close to the surface, and BIP concentrations drop off rapidly with height (Schmidt 1982). But the release of snow from tree canopies, and especially boundary layer separation at even small terrain crests, allow the mixing of BIPs over a greater depth. BIPs sublimate quickly in clear, dry conditions. They likely are far more common and mixed over a greater depth when they are not readily seen [i.e., during snow storms with otherwise poor visibility due to concurrent snowfall or shallow cloudiness, especially in mountains (e.g., Vionnet et al. 2013)]. This raises the question of the significance of ice crystal lofting from the surface to introduce ice in near-surface supercooled liquid water (SLW) clouds, and to enhance snow growth there. That is the motivation of this study.
The three hypotheses we aim to test are the following: 1) ice particles lofted from the surface under strong wind are detectable using cloud radars and/or lidars; 2) these BIPs contribute to the glaciation of shallow SLW clouds over mountains; and 3) disproportionally high concentrations of small ice particles during snowstorms at a mountaintop station in the Rocky Mountains are mostly due to BIPs and not to ice splintering due to supercooled droplets impinging on rimed surfaces on the ground.
This paper is organized as follows. A literature survey of BIP characteristics is presented in section 2. Data sources are listed in section 3. Examples of blowing snow as seen by airborne profiling radar and lidar are shown in section 4. Ice particle concentrations and size distributions measured on a mountaintop are compared against airborne measurements above the mountain in section 5. The findings are discussed in section 6, and summarized in section 7.
2. Blowing snow properties
Starting at a threshold wind gust speed, BIPs may become mobile at the surface of the snowpack, creeping, rolling, and leaping (“saltation”) (Yang and Yau 2008). As the wind speed increases, BIPs leap farther and fracture in the process. While BIPs are mostly confined to the lowest 1 m AGL (Schmidt 1982), they may become lofted by upward gusts, by turbulent vortices, and by boundary layer separation along snow edges that occur at a range of scales, from small surface irregularities, to roof edges, to terrain cliffs. In dry air these small ice crystals quickly sublimate (Yang and Yau 2008).
The threshold 2-m wind speed for BIPs in freshly fallen snow is 7–10 m s−1, with a weak trend toward lower threshold speeds at lower air temperatures (Dery and Yau 1999). Factors other than wind speed control the existence and concentration of BIPs, including snowpack properties (history of precipitation and snow transformation) and atmospheric conditions (temperature, surface radiation budget, turbulent kinetic energy) (e.g., Vionnet et al. 2013). Schmidt (1984) recorded size distributions of BIPs with an optical counter capable of detecting particles as small as 60 μm. The mean (mode) particle size, measured at various heights up to 1 m above the snow surface, was ~150 (~100) μm (i.e., not much larger than the size threshold for this instrument). Using a similar, although more resolved, optical array probe, Nishimura and Nemoto (2005) show a similar size distribution in the lowest 1 m, although in a colder environment. Both studies took place over flat terrain and not in cloud. Neither study included measurements above a few meters above ground level (AGL).
Blowing snow mostly has been studied in a dry environment where it sublimates quickly. Yet when BIPs are mixed into a cloudy (ice supersaturated) environment, they would be expected to grow. And during snowfall, snowflakes approaching the earth’s surface almost all fracture into small crystals upon collision with vegetation or with the snow-covered ground at wind speeds exceeding 5 m s−1 (Sato et al. 2008), producing an additional source of ice particles from the ground up. In the presence of cloud near the ground, BIPs may be mixed among SLW droplets, glaciating the cloud at temperatures well above typical heterogeneous freezing temperatures (which depend on the type of ice nuclei). Or in a mixed-phase cloud, BIPs may simply increase the ice crystal concentration and thus facilitate snow growth. Some evidence for the possible importance of this mechanism has been given in Geerts et al. (2011), Kristovich et al. (2012), and Vali et al. (2012). Further evidence is given in this paper.
3. Data sources
a. Cloud radar and cloud lidar
This study primarily uses data collected during the 2010/11 Cloud Property Validation Experiment (StormVEx)/Colorado Airborne Multiphase Cloud Study (CAMPS) (Mace et al. 2010). The primary data sources for this study were collected aboard an aircraft, the University of Wyoming King Air (UWKA), and at Storm Peak Laboratory (SPL), located on Mount Werner above Steamboat Springs, Colorado, at an elevation of 3208 m (Fig. 1). The UWKA carried a 94-GHz (3 mm) Doppler radar with zenith- and nadir-pointing antennas, the Wyoming Cloud Radar (WCR),1 and a polarization backscatter lidar, the Wyoming Cloud Lidar (WCL) (Wang et al. 2012).
Terrain map of south-central Wyoming and north-central Colorado, showing the UWKA tracks relevant to this study, and a few measurement sites. The grid-shaped flight pattern (black lines) was flown on 13 Jan 2011 around Steamboat Springs. The two flight legs highlighted as red and yellow arrows are discussed in the text. Two flight legs over the Sierra Madre on other days are shown as white lines.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
A BIP signature is difficult to detect with radars and lidars for several reasons. As mentioned in section 2, the BIP particle diameter is O(100) μm, requiring a sensitive millimeter-wave radar. The BIP concentration well above the surface probably is orders of magnitude less than the typical concentration of droplets in any cloud, thus lidar detection is difficult as well. Any remote sensor must have a high resolution near the ground. Detection of a BIP signature becomes far more difficult in the presence of concurrent larger falling snowflakes (which will dominate the radar signal) and/or in the presence of cloud (which attenuates the lidar signal).
The nadir antenna WCR and WCL data are especially useful to study BIPs, because data are available close to the underlying terrain surface or forest canopy. Over flat terrain, at typical pulse width, WCR data are unaffected by ground clutter down to ~40 m AGL. Sidelobes may contaminate weak echoes up to ~100 m AGL over steep terrain. BIPs and other particles smaller than ~600-μm scatter in the Rayleigh regime at 94 GHz (W band), thus a sensitive cloud radar (such as the WCR) should be able to detect BIPs in clear air and in an initially liquid-only cloud. In clear air the WCL may obtain a highly depolarized return from BIPs, because of their nonspherical shape, but in a SLW cloud the lidar signal will be dominated by cloud droplets, as mentioned above.
The WCR zenith- and nadir-pointing antenna Doppler velocities are processed to remove the motion and 3D attitude variations of the aircraft, as well as contamination by the horizontal wind when the antennas are not perfectly vertically oriented. The resulting field, referred to as WCR vertical velocity, approximates the hydrometeor vertical motion with an accuracy of ±1 m s−1 (Damiani and Haimov 2006).
b. In situ particle probes
The UWKA also carried several in situ cloud microphysics probes, including a Cloud Imaging Probe (CIP) from Droplet Measurement Technologies (DMT) and a Forward Scattering Spectrometer Probe (FSSP) during StormVEx/CAMPS. At SPL, a DMT CIP probe was mounted 1.5 m above the roof of the building, which itself was 3–4 m above the snow surface. The probe was mounted on a freely rotating vertical axis and pointed into the wind by means of a tail vane. A 3D sonic anemometer was also mounted on this assembly to obtain the assembly-relative wind vector at high frequency. This wind was used to compute the true airspeed into the CIP. The same mount also included an FSSP, measuring particles between 0.5 and 47 μm in diameter. Data from the two CIP probes (at SPL and aboard the UWKA) are processed slightly differently, by DMT and UWKA software, respectively. But both processing techniques use the same depth of field correction, the same rejection criteria, and the same adjustment of the sampling volume for larger particles (Korolev et al. 1998). The CIP records sizes in 62 bins, ranging from 12.5 to 1537.5 μm, with a 25-μm size interval. As pointed out by Korolev et al. (1998), diameter assignment and thus concentration for particles smaller than ~4 times the diode size (i.e., smaller than ~100 μm for the CIP) is less certain than that for larger particles, and it is highly uncertain for particles smaller than twice the diode size (<50 μm in this case). Therefore, particles in the first two CIP bins (size < 62.5 μm) are ignored. Both SPL and the UWKA operated similar FSSP probes. These are not 2D optical array probes, thus it is impossible to distinguish between cloud droplets and small BIPs. Therefore, BIP concentrations are estimated from the two CIPs only, and BIPs < 62.5 μm are not counted.
Aside from the StormVEx/CAMPS project in the winter of 2010/11, the UWKA also flew over the Sierra Madre in Wyoming (Fig. 1) in early 2012, as part of the AgI Seeding Cloud Impact Investigation (ASCII) (Geerts et al. 2013). Most of the ASCII flights dealt with artificial glaciogenic seeding of orographic clouds, using silver iodide (AgI) nuclei, but two flights focused on BIP detection (Kristovich et al. 2012). Surface meteorological measurements were collected from automated weather stations at SPL during StormVEx/CAMPS and from Dixon and Battle Pass during ASCII (Fig. 1). Several other instruments operated at Battle Pass during ASCII, including a Cloud Particle Imager (CPI), which images particles at a resolution of 10 μm (Lawson et al. 2006). The CPI was mounted on a scaffold about 4 m above the snow surface in a protected area surrounded by trees. It was pointed into the wind by means of a pivoting vane. Radiosondes were released from an upstream site during each UWKA flight (i.e., from Steamboat Springs during StormVEx/CAMPS and from Dixon during ASCII; Fig. 1).
4. Airborne radar and lidar examples of blowing snow events
a. 13 January 2011
In the morning of 13 January 2011 a thin stratus cloud deck was present over Steamboat Springs upwind of the Park Range (Fig. 1). Radiosonde data from Steamboat Springs (Fig. 2a) indicate a valley inversion in the lowest 50 mb (1 mb = 1 hPa), where the wind was almost calm. A saturated layer is present near mountaintop level (~660 mb), where the wind is ~15 m s−1. This layer is capped by a second inversion starting at 630 mb, about 400 m above mountaintop level. A WCL transect from a flight leg just north of Steamboat Springs (Fig. 1) indicates two thin cloud layers over the valley (Fig. 3c). The height range of these layers (3.2–3.9 km MSL) spans the saturated layer in the sounding. The upper cloud layer (at 3.9 km MSL, temperature ~−11°C, labeled “upper SLW cloud” in Fig. 3c) does not contain any ice particles, because no echo is present in the WCR transect (Fig. 3a) and because the WCL depolarization ratio is very low (Fig. 3d). The lower cloud layer (3.3 km MSL, rising to 3.7 km just upwind of the crest, temperature ~−10°C, labeled “lower SLW cloud” in Fig. 3c) does contain some ice, according to the higher WCL depolarization ratio values, but the ice particles are small since the WCR reflectivity is quite low upwind of the mountain (~−30 dBZ, which is close to the WCR noise level at this range). The base of this lower cloud over the mountain is not evident from WCL data, due to signal extinction at higher levels, but ceilometer data at Steamboat Springs indicate that this cloud is thin: the ceilometer cloud-base height is 3.1 km MSL at the time of the flight leg in Fig. 3. This cloud is recorded only intermittently by the ceilometer, suggesting that it is broken or sometimes very thin. The sounding suggests that the upstream air is nearly saturated down to 700 mb (3.0 km); thus, the top of the Park Range in this transect likely is in cloud.
Skew T–logp diagrams at (a) 1500 UTC 13 Jan 2011, (b) 1615 UTC 18 Jan 2012, (c) 1521 UTC 29 Feb 2012, and (d) wind profile from the three soundings. The rawinsonde for (a) was released from Steamboat Springs, and those for (b),(c) were released from Dixon (Fig. 1).
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
WCR and WCL transect for an along-wind flight leg between 1626 and 1636 UTC 13 Jan 2011 over the Park Range, highlighted as a red line in Fig. 1. The wind direction is from left (west) to right (east). (a) WCR reflectivity, (b) WCR hydrometeor vertical velocity, (c) WCL backscatter power, and (d) WCL depolarization ratio. The UWKA flight level is at 4.3 km MSL on the west side, ascending to 4.6 km MSL on the east side. Note that the color scheme for the WCR hydrometeor vertical velocity is centered at −1 m s−1, to account for an assumed ice particle fall speed of 1 m s−1. The terrain profile is added as a white line in all panels. The sampling times are shown on the lower abscissa; the corresponding distance is shown on the upper abscissa. The cloud features labeled in (c) are discussed in the text.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
The WCR reflectivity transect indicates that very light snow falls from this cloud over the Park Range. Ice particles remain present in the trapped lee waves downwind of the mountain, but most sublimate before reaching the ground (Fig. 3a). The increase in depolarization ratio in the cloud layer from west to east near and just above the peak of the Park Range (Fig. 3d) suggests increasing glaciation as the flow crosses the mountain. This glaciation may be the result of injection of BIPs. The greater reflectivity observed by the WCR near the surface just upwind of the mountain crest may in part be due to BIPs, especially where echoes first develop near the surface, near x = 30 km in Fig. 3a.
Another zonal transect flown some 15 min earlier and farther north, where the Park Range is higher and thus more likely immersed in cloud, is shown in Fig. 4. This flight track is from west to east, allowing a good view of the SLW stratus cloud upwind of the crest and the lee wave cloud from aboard the UWKA (Fig. 5). Again there are two thin cloud layers upstream of the mountain (Fig. 4c). In this transect the lower cloud layer is ice free as well (lack of WCR-measured reflectivity, see Fig. 4a). It is not clear from the WCL data whether the entire layer between 3.2 (or the terrain) and 4.0 km MSL is cloud filled upstream and over the mountain (roughly between 10 < x < 40 km). Since the WCL signal is largely attenuated by the top layer, the cloud depth is not clear, but the ceilometer cloud base is ~3.1 km, hence, it is likely that the lower SLW cloud (labeled in Fig. 4c) continues farther east and intercepts the mountain.
As in Fig. 3, but for the leg flown between 1610 and 1618 UTC, shown as a yellow line in Fig. 1. The dashed line in (a) and (c) shows the approximate lower edge of the forward camera view aboard the UWKA in Fig. 5a.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
Forward camera view along the flight track analyzed in Fig. 4, viewing the cloud transition across the Park Range from the west at (a) 1610:30, (b) 1611:30, (c) 1612:00, and (d) 1612:30 UTC.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
WCR echoes develop within this cloud, starting near x = 30 km, ~7 km upwind of the crest (Fig. 4a). The near-surface WCR reflectivity gradually increases over the high terrain toward x = 40 km, and then suddenly increases to −5 dB in what appears to be a hydraulic jump near x = 41 km. This cloud jump is also evident in the forward camera (Fig. 5): it appears as a dark cloud region since its upslope part is not illuminated by the sun (which is in the southeast at this time of the day). Hydraulic jumps have been observed regularly in the lee of major terrain crests using the same profiling cloud radar [e.g., Fig. 12 in Vali et al. (2012); Fig. 9 in Geerts et al. (2015); French et al. (2015)]. The dramatic increase in reflectivity near x = 41 km may be the result of the injection of BIPs, growing in the hydraulic updraft. Reflectivity values continue to increase in the trapped lee wave, indicating snow growth, and in fact snow almost reaches the ground in the downstream valley at the time of the transect, according to the WCR reflectivity profile (Fig. 4a).
We do not have visible evidence of BIP plumes in this case, because the Park Range itself is not visible from flight level during either transect due to intervening cloud (Fig. 5), but WCR and WCL data lend support to the first two hypotheses stated in section 1, although not unambiguously. Therefore, we examine another case, on 18 January 2012, which gives unambiguous evidence for the first hypothesis.
b. 18 January 2012
Snow had fallen over the Sierra Madre on 17 January 2012, with surface temperatures of at most −4°C at Dixon. Strong westerly winds (>10 m s−1) blew at the surface the next morning, especially just north of the Sierra Madre, which is one of the windiest regions in Wyoming in winter (Randolph 2012), in particular when a strong west–east 800-mb height gradient is present, as was the case on 18 January.
A radiosonde released from Dixon at 1615 UTC reveals a deep, dry well-mixed boundary layer above a surface-based stable layer (Fig. 2b). Unlike places farther north, Dixon is somewhat protected in a valley upwind of the Sierra Madre (Fig. 1), hence, the stable surface layer in this valley. The well-mixed PBL is capped by an inversion starting near 4 km AGL, well above mountain level, hence, the flow over the mountain is unblocked. The UWKA flies a track from northwest (NW) to southeast (SE), approximately along the Sierra Madre crest line (Fig. 1). During this flight track the wind speed at Battle Pass (Fig. 1) is 20–23 m s−1 from the west-southwest (WSW) (~253°), gusting to 32 m s−1. The surface wind is almost normal to the flight track, but the low-level wind shear is roughly along the track (Fig. 2b). The terrain is readily visible from flight level during this flight: Fig. 6 shows pockets of blowing snow separating from terrain crests.
Blowing snow separating from a local crest north in the Sierra Madre, as seen from the UWKA at 1609:21 UTC 18 Jan 2012, along a track shown in Fig. 1. The view is to the southeast. The photograph was taken by D. Kristovich.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
The WCR reflectivity transect for this track (Fig. 7a) reveals several plumes of BIPs, up to ~250 m AGL. Their reflectivity is generally higher toward the ground, indicating that the plumes originate from the ground and sublimate aloft. The BIP plumes tend to occur near the highest terrain. Some rise, but most descend (Fig. 7b). A detailed mapping of the flight track over the terrain (not shown) indicates that the sinking motion occurs mostly in the lee of the terrain crests where boundary layer separation took place. A zoomed-in view of the two deepest plumes (Figs. 7e–h) shows that both plumes (labeled A and B) clearly reach higher than the surrounding terrain, possibly associated with hydraulic jumps as in Fig. 4b. Plumes A and B are associated with large horizontal vorticity, both with downdrafts in excess of 6 m s−1 on the downshear side (Fig. 7f). Presumably this vorticity is derived from the ambient wind, whose along-track profile can be inferred from the wind profile shown in Fig. 2d: in the lowest 1 km AGL the wind shifts from 210° (SSW) to 265° (W), implying a north-northwest (NNW) shear, roughly along the flight track (Fig. 1).
(a)–(d) As in Fig. 3, but between 1608 and 1614 UTC 18 Jan 2012 over the Sierra Madre (Fig. 1). (e)–(h) As in (a)–(d), but zoomed-in on the two deepest BIP plumes.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
The fact that the plumes can be seen by eye (Fig. 6) and can be captured by WCL (at least plume B, Fig. 7g) indicates a rather high concentration of ice particles. The WCL depolarization ratio (Fig. 7d) is high in plume B, with values around 0.10–0.15, higher than values below 0.05 in the SLW stratus (Figs. 3d and 4d). These high values (in the absence of high backscatter power) confirm that the scatterers are not droplets. None of the BIP plumes along this transect reach flight level (3.6 km MSL), so there is in situ no information about BIP size distribution or concentration.
c. 29 February 2012
The above example shows that the WCR/WCL can detect BIP plumes (first hypothesis). To test the second hypothesis (that these BIP plumes can contribute to the glaciation of shallow orographic SLW clouds), we examine data from a third flight. Several tracks were flown northwest of the Sierra Madre on 29 February 2012. The terrain rises only very slightly toward the Continental Divide in this area (Fig. 1). In the 24 h before this flight 4.2 mm (water-equivalent) of snowfall was recorded at Battle Pass. Forward and downward camera footage aboard the UWKA confirms that the ground was covered with at least some fresh snow in the vicinity of the flight track. The 10-m wind speed at Battle Pass averages 16 m s−1 at the time of the flight track (~1700 UTC). The Dixon sounding (Fig. 2c) indicates a well-mixed PBL that becomes nearly saturated just below 700 mb (~2.9 km MSL), and a saturated layer near 3.8 km MSL. The surface-based lifting condensation level (LCL) is 2.6 km for this sounding. The sounding and station data at Dixon and at Rawlins (the first weather station north of the track) suggest that surface air temperature below the flight track was at most −3°C.
Data from one of the transects are shown in Fig. 8. A thin altostratus cloud is present near flight level (3.8 km MSL), extending over the plains to the west. This cloud (labeled “ice” in Fig. 8c) contains almost no liquid water, according to both UWKA in situ probes and the WCL. It contains a rather high concentration (10–30 L−1) of small ice crystals. These crystals are mostly horizontally oriented near x = 24 km (high depolarization ratio and low backscatter power, Figs. 8c and 8d), and assume a more random orientation across the crest, near x = 34 km (lower depolarization ratio and higher backscatter power). This ice cloud subsides below flight level in subsident flow across the divide, part of an upstream-tilting gravity wave (Fig. 8b). A very shallow orographic stratocumulus layer containing droplets emerges near the west end of the track (labeled “SLW Sc” in Fig. 8c). This broken cloud layer is evident from flight level (Fig. 8, forward camera insert), and, according to WCL data (Fig. 8c), has a cloud top at 2.9 km MSL at the west end of the track, thickening and rising to 3.2 km MSL near the Continental Divide crest (x = 32 km). This liquid cloud contains enough droplets in some sections to totally attenuate the lidar signal (Fig. 8c). The depolarization ratio typically increases to high values when this happens (Fig. 8d) due to multiple scattering (Wang et al. 2012).
As in Fig. 3, but for 29 Feb 2012, north of the Sierra Madre (Fig. 1). A forward camera snapshot taken from the left (west) side of the transect is shown between (b) and (c).
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
The stratocumulus (Sc) cloud is within the well-mixed PBL (Fig. 2c) and contains turbulent eddies (Fig. 8b). WCR vertical velocity variations were analyzed as a function of height AGL following the method described in Geerts et al. (2011) to confirm that these variations indeed are turbulence, in this case shear driven. The shallow cloud is not seeded by the altostratus aloft, at least not upwind of the crest (Fig. 8a). Given the high cloud-top temperature (−11°C) and the lack of seeding from aloft, it is likely that this cloud is seeded by BIPs from below. BIP plumes are not evident from the forward camera imagery (Fig. 8), but the flight scientist in the cockpit noted observing streaks of BIPs over the partially snow-covered landscape. Reflectivity values gradually increase from the western margin (where they are below the noise threshold) toward the divide (where they reach −5 dBZ), indicating ice crystal introduction and growth, hence, the Sc cloud becomes mixed phase (Fig. 8c). The first detectable WCR echoes form not near the surface, nor near cloud top, but somewhere in between, possibly because most BIPs sublimate in the dry lower PBL (Fig. 2c). Those that are lofted above the cloud base (~2.6 km) are able to start growing. Light snowfall first reaches the ground some 15 km (~20 min) downwind of the first WCR echoes.
Most particles sampled by the CPI at Battle Pass during this flight are small, fractured ice particles, O(100) μm in diameter. They generally have smooth edges, without a clear crystal growth habit (Fig. 9). These features are anticipated for BIPs (Schmidt 1984; Nishimura and Nemoto 2005). Occasionally larger ice crystals are encountered, in the shape of columns, plates, or dendrites. These probably are cloud-borne crystals.
Example CPI images from Battle Pass (Fig. 1) during the 29 Feb 2012 flight.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
5. Airborne versus mountaintop observations of ice particles
a. Ice particle size distributions as a function of wind speed
To further explore the presence of BIPs we compare simultaneous in situ ice particle data from a mountain site and from the free atmosphere above, using data collected in StormVEx/CAMPS. This project was designed to examine microphysical properties of orographic clouds using in situ data at flight level and at SPL. For the objectives of this project, blowing snow conditions were considered an undesirable contamination. Therefore no intensive operating periods (IOPs) were conducted under strong winds. The strongest IOP-average surface wind at SPL was <10 m s−1. But half of the 12 CAMPS IOPs with the UWKA flying overhead had average surface winds above 6.5 m s−1 (Table 1), in theory sufficient for at least some BIPs in the presence of fresh snow on the ground.
Summary of the 12 StormVEx/CAMPS cases used in this study. The flight period is the time the UWKA was flying near SPL. The number of data points is the number of samples when the aircraft was within the white circle shown in Fig. 1, and ice crystals were present both at flight level and at SPL (located at 3.2 km MSL). The data rate is 1 Hz, hence, the number of data points is the aggregate sample period (s). The last four columns summarize weather at SPL (mean values for the duration of the flight period). The relative humidity (RH) is relative to water.
We now compare CIP ice particle concentrations and size distributions at SPL to those at flight level, on average 1.7 km above SPL, for all 12 IOPs (Fig. 10). The comparison is based on a series of time sections when the UWKA was within 20 km from a point 10 km west of SPL (white circle in Fig. 1). This westward displacement is intended to account for the typical horizontal displacement of falling snow. The comparisons are simultaneous, in other words the orographic snowfall is assumed to be in steady state. The flight time within the white circle in Fig. 1 comprises a good fraction of the total UWKA flight time on station, since most flights followed a grid pattern similar to that shown in Fig. 1. Times when either CIP was not functioning or measured <0.5 particles per liter over a 1-min period are excluded. The resulting cumulative data period (in seconds) for each IOP, and average weather conditions at SPL during these periods, are shown in Table 1. The CIP comparison is based on an average of 43 min of data per IOP, and at least 20 min. The SPL surface temperature was cold (−7° to −17°C) and the relative humidity was high, >92% relative to liquid water (Table 1).
Particle size distributions for all 12 CAMPS cases with overlapping airborne and ground-based (SPL) CIP measurements, and with the UWKA flying within the white circle in Fig. 1. The date (YYYYMMDD), average 10-m wind speed at SPL, and the composite data period are shown in the top-right corner of each panel. The CIP total particle concentration at SPL and aboard the UWKA is shown in the bottom-left corner.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
In the absence of blowing snow, ice particle concentrations at flight level may differ significantly from those at the mountaintop. Supercooled droplets, upon freezing, may splinter into numerous ice crystals. This mechanism (Hallett and Mossop 1974) would result in more ice crystals at SPL than at flight level, which occurs on some flights (Fig. 10). On the other hand, aggregation would lead to fewer, but larger ice crystals at SPL. Given the relatively short sample periods, and the high variability of ice crystal concentrations (along the flight track and at SPL), significant time–space mismatch issues exist in the comparisons shown in Fig. 10. But there is a strong tendency for large concentrations of small particles on windy days, for instance two of the three highest ice particle concentrations at SPL occur during the two flights with the highest average surface wind speed (9 m s−1), and on both days this is disproportionally the result of small particles (smaller than 200 μm). On the less windy days, the two size distributions tend to match better in value or slope, and concentrations tend to be more comparable. We suspect that most small particles on windy days are BIPs.
Blowing snow ice particle amplification (BIPA) as a function of wind gust speed, for the 12 CAMPS IOPs. The regression line is shown.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
b. Another mechanism for ice crystal production from the ground
Other than by blowing snow, ice crystals could also be introduced from near the earth surface when cloud droplets splinter upon impacting rimed vegetation. This is the Hallett–Mossop (HM) mechanism, but in this case droplets impact stationary rimed surfaces, rather than rimed crystals falling through a water cloud. This hypothesis has been mentioned in Geerts et al. (2011) and Vali et al. (2012). The upwind slope of the Park Range is mostly covered by deciduous trees at lower elevation and conifers at higher elevation. Secondary ice crystal generation by riming is most effective under a specific range of impact velocities, temperatures, and drop size distributions (Pruppacher and Klett 1997). We now examine these three factors for the 12 CAMPS IOPs.
First, near-surface wind speeds at SPL averaged between 2 and 9 m s−1 during these IOPs (Table 1). Such wind speeds are higher than typical impact velocities of falling rimed particles (i.e., the fall speed of rimed particles), but even at such speeds ice splinter formation may occur (Mossop 1985), and weaker winds are likely to occur within the vegetation canopy than at SPL, which is located on an open ridge. Second, the temperature criterion is more restrictive: secondary ice crystal generation by riming rapidly ceases at temperatures below −8°C (e.g., Griggs and Choularton 1986). This eliminates all but one IOP (Table 1), although more IOPs may satisfy this condition at a lower elevation, below SPL, but still above the cloud base. The third criterion is restrictive as well: large droplets are needed for ice splinter formation, larger than about 20 μm, depending on crystal habit (Wang and Ji 2000). Clouds in only three of the 12 IOPs had an average concentration of large droplets (>20 μm) of at least 20 cm−3, according to FSSP data collected at SPL, and those three IOPs are too cold at the level of SPL. All this suggests that the HM ice multiplication mechanism is an unlikely candidate to explain the observed relatively large concentration of small ice crystals at SPL during some flights (Fig. 10). This confirms the third hypothesis (stated in section 1), at least in part.
c. Further analysis of particle probe data at Storm Peak Laboratory
To further test this hypothesis, specifically that disproportionally high concentrations of small ice particles at SPL are due to the HM process on rimed surfaces, CIP and FSSP data at SPL are examined for the entire StormVEx/CAMPS field phase, between 15 December 2010 and 31 March 2011, irrespective of UWKA overpasses. Snowfall occurred at SPL during 40 days of the campaign. We further narrow the analysis period, requiring that the CIP concentration exceeds 0.1 L−1 (hourly averages). CIP and FSSP data then are averaged over 5-min intervals, matching the time resolution of meteorological data at SPL. For the comparison in Fig. 12, data are filtered further to include only periods when the wind speed is less than 5 m s−1 (no blowing snow), and the RH relative to water at least 95% (cloud present). An additional filter for Fig. 12a is that the SPL surface temperature is between −8° and −3°C (the second HM criterion listed above). For Fig. 12b an additional condition is that there are droplets larger than 20 μm (the third HM criterion).
Particle size distributions measured by the CIP at SPL during CAMPS for all times that specific conditions discussed in the text are satisfied. (a) The blue (red) line is the particle size distribution when NFSSP(D>20 μm) > 20 cm−3 [NFSSP(D>20 μm) < 1 cm−3], where NFSSP(D>20 μm) is the concentration of droplets larger than 20 μm, according to the SPL FSSP. (b) The blue (red) line is the particle size distribution when −8° < T < −3°C (T < −10°C), where T is the air temperature at SPL. Also shown is the cumulative period over which these conditions are satisfied (h) and the CIP concentration across all sizes.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
The blue (red) line in Fig. 12a indicates particle size distributions when the HM mechanism is possible (highly unlikely), based on the third HM criterion. Note that the FSSP may confuse small ice particles (such as BIPs) for droplets larger than 20 μm (it is impossible to tell because the FSSP is not an imaging probe), except that the weak wind condition excludes BIPs. If the ice multiplication mechanism on rimed trees is active, then the blue line should lie above the red line in Fig. 12a. It does not. Similarly, the blue (red) line in Fig. 12b is the particle size distribution when temperatures are suitable (too cold) for the HM mechanism to be active. So the blue line should lie above the red line, especially for small particles (<200 μm). Again, it does not, at least not substantially. Thus, both comparisons shown in Fig. 12 indicate that the HM mechanism on rimed trees as a source of ice crystals from the ground is unlikely to be significant.
Further evidence suggesting that this mechanism is not active comes from Fig. 13. The common conditions for this plot are snowfall (CIP concentration > 0.5 L−1), cloud presence (RH relative to water >95%), no BIPS (wind speed < 5 m s−1), and large droplets present [
Droplet size distributions measured by the FSSP at SPL during CAMPS when NFSSP(D>20 μm) > 20 cm−3. The blue (red) line is the size distribution when −8° < T < −3°C (T < −10°C).
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
d. Further evidence for BIPs during winter storms at Storm Peak Laboratory
Finally, we test the other half of the third hypothesis (i.e., whether disproportionally high concentrations of small ice particles at SPL are associated with conditions suitable for BIPs). The 5-min resolution data for the same 40 days with snowfall during the StormVEx/CAMPS campaign is filtered for the following conditions: RH > 95% (cloudy) and CIP concentration > 0.1 L−1 (falling snow or BIPs). The distinguishing condition between the two CIP particle size distributions in both panels of Fig. 14 regards wind speed: the blue curve applies to conditions windy enough for blowing snow (>10 m s−1), and the red curve applies to conditions too quiescent for blowing snow (<5 m s−1). (The transition region between 5 and 10 m s−1 is ignored to ascertain presence or absence of blowing snow.) The wind speed exceeds 10 m s−1 during 15.4 h (Fig. 14a, for temperatures below −5°C; that is, 17% of the time during snow storms). These represent conditions not sampled during the CAMPS IOPs (section 5a).
Particle size distributions measured by the CIP at SPL during CAMPS for all times that specific conditions were satisfied. Distinguishing conditions relate to surface wind speed and temperature at SPL.
Citation: Monthly Weather Review 143, 12; 10.1175/MWR-D-15-0241.1
The ice particle concentration is an order of magnitude larger under windy conditions than under quiescent conditions at SPL, both for a broad temperature range (Fig. 14a) and for a subset with colder conditions (Fig. 14b). Larger particles are more common as well during windy conditions, probably because stronger cross-mountain flow leads to heavier snowfall (e.g., Jiang and Smith 2003; Colle 2004). But the BIPA is larger than unity for both pairs of distributions in Fig. 14 (BIPA = 2.2 for Fig. 14a and BIPA = 1.6 for Fig. 14b). Note that here BIPA is defined as in Eq. (1), but comparing strong wind against weak wind conditions, instead of surface versus aloft. Thus small particles <200 μm are disproportionally more common under strong wind, most of them likely BIPs.
6. Discussion
Blowing snow ice particles, with a typical size O(100) μm, have a very small fall speed, at most a few tens of centimeters per second (Heymsfield 1972; Locatelli and Hobbs 1974; Pruppacher and Klett 1997). Therefore, such particles readily can be lofted in updrafts. BIPs may be lofted over great depths in complex terrain by several mechanisms. These include boundary layer separation as shown in Figs. 7 and 6, PBL turbulence as shown in Fig. 8, and hydraulic jumps as illustrated in Fig. 4. Other possible lofting mechanisms in complex terrain not illustrated here include orographic convection and horizontal lee vortices (Voigt and Wirth 2013).
Several studies have documented pure ice clouds near the surface under windy conditions. Schlenczek et al. (2014) examined several episodes of pure ice clouds on a very steep mountain (Jungfraujoch in Switzerland) and infers from the irregular habits and fractured crystals that in many cases BIPs dominate, rather than “diamond dust,” which mainly results from deposition freezing due to local ice supersaturation under very cold conditions (e.g., Lawson et al. 2006). Palm et al. (2011) used a spaceborne lidar to identify blowing snow conditions and report that on occasion a blowing snow signature can be seen as high as 500 m AGL over Antarctica, especially near the ice dome margin where strong winds prevail. Ice crystals sublimate slowly under the very cold conditions observed around Antarctica, so it is not surprising they can be seen over great depth.
We have shown that BIPs are detectable by profiling cloud radars and/or lidars (first hypothesis), and that disproportionally high concentrations of small ice particles near the surface of a mountain during snow storms are mostly due to BIPs, at least at SPL (third hypothesis). The evidence for the second hypothesis, regarding the ability of BIPs to seed supercooled clouds from the ground up, is more tentative. The importance of BIPs in orographic precipitation probably is underappreciated because BIPs cannot be readily observed or measured in orographic clouds: BIPs cannot be separated from the typically larger falling snow particles, which overwhelm any BIP radar signal, and BIPs are far less numerous than cloud droplets, which overwhelm any BIP lidar signal. A BIP signature separate from falling snow may be evident in optical array probe data collected near the ground (as in Fig. 9). To explore BIP lofting into orographic clouds, ice particle imaging devices need to be installed on a tall tower, or on a very steep mountain like the Jungfraujoch. Flight-level data are rarely useful because the lowest flight level in cloud over complex terrain typically is well above the BIP source region. At that level BIPs may have transformed into more regular crystal habits or aggregates.
We hope that this study will stimulate the development of a parameterization for the wind-driven aerial injection of ice crystals from the surface up, for use in numerical models. This parameterization can build on existing snow drift parameterizations, which are driven by surface wind speed, snowfall history, surface roughness, surface temperature, and solar radiation. Without better data, such parameterization will be poorly constrained, but hopefully it can demonstrate the efficiency of snowfall from shallow supercooled clouds over mountains. Of course, this parameterization may lead to an overproduction of orographic precipitation from deeper clouds, but that may be because simulations without BIP parameterization did well for the wrong reason.
7. Conclusions
Airborne cloud radar and cloud lidar data are presented to demonstrate the presence of blowing snow above complex terrain in Colorado and Wyoming. Evidence is given for BIPs becoming lofted by boundary layer separation behind terrain crests and by hydraulic jumps. Evidence is presented also that BIPs may lead to the glaciation of shallow supercooled orographic clouds, and thus to snow growth and snowfall.
Further evidence for the presence of BIPs near the surface over a snow-covered mountain comes from a comparison between snow size distributions measured at SPL versus those measured aboard an aircraft sampling the same orographic cloud: under windy conditions the concentration of small ice crystals at SPL (cf. to that at flight level) is much larger than the same ratio for larger ice crystals. Finally, the analysis of snow size distributions at SPL under different weather conditions also strongly points to blowing snow as the likely reason for the large concentrations of small ice crystals sometimes observed at SPL. This analysis further shows that a Hallett–Mossop ice splintering mechanism near the ground (i.e., the collision of supercooled droplets with rimed surfaces, such as on trees) is unlikely to be effective at SPL.
We recommend the development of a parameterization for the wind-driven aerial injection of ice crystals from the surface, for use in numerical models.
Acknowledgments
The ASCII campaign is funded by the National Science Foundation Grant AGS-1058426. This work also received funding from the Wyoming Water Development Commission and the U.S. Geological Survey, under the auspices of the University of Wyoming Water Research Program. Jeff French assisted with the processing and interpretation of the CIP data. We thank Linnea Avallone (CAMPS lead scientist) and Jay Mace (StormVEx lead scientist) for the UWKA and SPL data used in this study. Opinions expressed are those of the authors and do not represent those of the funding agencies or universities.
REFERENCES
Bernhardt, M., G. E. Liston, U. Strasser, G. Zangl, and K. Schulz, 2010: High resolution modelling of snow transport in complex terrain using downscaled MM5 wind fields. The Cryosphere, 4, 99–113, doi:10.5194/tc-4-99-2010.
Chung, Y.-C., S. Bélair, and J. Mailhot, 2011: Blowing snow on Arctic sea ice: Results from an improved sea ice–snow–blowing snow coupled system. J. Hydrometeor., 12, 678–689, doi:10.1175/2011JHM1293.1.
Colle, B. A., 2004: Sensitivity of orographic precipitation to changing ambient conditions and terrain geometries: An idealized modeling perspective. J. Atmos. Sci., 61, 588–606, doi:10.1175/1520-0469(2004)061<0588:SOOPTC>2.0.CO;2.
Damiani, R., and S. Haimov, 2006: A high-resolution dual-Doppler technique for fixed multi-antenna airborne radar. IEEE Trans. Geosci. Remote Sens., 44, 3475–3489, doi:10.1109/TGRS.2006.881745.
Dery, S., and M. K. Yau, 1999: A climatology of adverse winter-type weather events. J. Geophys. Res., 104, 16 657–16 672, doi:10.1029/1999JD900158.
French, J., S. Haimov, L. Oolman, V. Grubisic, S. Serafin, and L. Strauss, 2015: Wave-induced boundary layer separation in the lee of the Medicine Bow Mountains. Part I: Observations. J. Atmos. Sci., doi:10.1175/JAS-D-14-0376.1, in press.
Geerts, B., Q. Miao, and Y. Yang, 2011: Boundary-layer turbulence and orographic precipitation growth in cold clouds: Evidence from profiling airborne radar data. J. Atmos. Sci., 68, 2344–2365, doi:10.1175/JAS-D-10-05009.1.
Geerts, B., and Coauthors, 2013: The AgI Seeding Cloud Impact Investigation (ASCII) campaign 2012: Overview and preliminary results. J. Wea. Mod., 45, 24–43.
Geerts, B., Y. Yang, R. Rasmussen, S. Haimov, and B. Pokharel, 2015: Snow growth and transport patterns in orographic storms as estimated from airborne vertical-plane dual-Doppler radar data. Mon. Wea. Rev., 143, 644–665, doi:10.1175/MWR-D-14-00199.1.
Griggs, D. J., and T. W. Choularton, 1986: A laboratory study of secondary ice particle production by the fragmentation of rime and vapour-grown ice crystals. Quart. J. Roy. Meteor. Soc., 112, 149–163, doi:10.1002/qj.49711247109.
Hallett, J., and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249, 26–28, doi:10.1038/249026a0.
Heymsfield, A., 1972: Ice crystal terminal velocities. J. Atmos. Sci., 29, 1348–1357, doi:10.1175/1520-0469(1972)029<1348:ICTV>2.0.CO;2.
Jiang, Q. F., and R. Smith, 2003: Cloud timescales and orographic precipitation. J. Atmos. Sci., 60, 1543–1559, doi:10.1175/2995.1.
Korolev, A. V., J. W. Strapp, and G. A. Isaac, 1998: Evaluation of the accuracy of PMS optical array probes. J. Atmos. Oceanic Technol., 15, 708–720, doi:10.1175/1520-0426(1998)015<0708:EOTAOP>2.0.CO;2.
Kristovich, D. A. R., B. Geerts, Q. Miao, L. Stoecker, and J. M. Ritzman, 2012: Airborne cloud radar and lidar observations of blowing snow during the ASCII project: A possible natural cloud seeding mechanism. 15th Conf. on Mountain Meteorology, Steamboat Springs, CO, Amer. Meteor. Soc., 16.4. [Available online at https://ams.confex.com/ams/15MountMet/webprogram/Paper210260.html.]
Lawson, P., B. A. Baker, P. Zmarzly, D. O’Connor, Q. Mo, J.-F. Gayet, and V. Shcherbakov, 2006: Microphysical and optical properties of atmospheric ice crystals at South Pole Station. J. Appl. Meteor. Climatol., 45, 1505–1524, doi:10.1175/JAM2421.1.
Lehning, M., I. Volksch, D. Gustafsson, T. A. Nguyen, M. Stahli, and M. Zappa, 2006: ALPINE3D: A detailed model of mountain surface processes and its application to snow hydrology. Hydrol. Processes, 20, 2111–2128, doi:10.1002/hyp.6204.
Liston, G. E., and K. Elder, 2006: A distributed snow-evolution modeling system (SnowModel). J. Hydrometeor., 7, 1259–1276, doi:10.1175/JHM548.1.
Locatelli, J. D., and P. V. Hobbs, 1974: Fall speeds and masses of solid precipitation particles. J. Geophys. Res., 79, 2185–2197, doi:10.1029/JC079i015p02185.
MacDonald, M. K., J. W. Pomeroy, and A. Pietroniro, 2009: Parameterizing redistribution and sublimation of blowing snow for hydrological models: Tests in a mountainous subarctic catchment. Hydrol. Processes, 23, 2570–2583, doi:10.1002/hyp.7356.
Mace, J., and Coauthors, 2010: STORMVEX: The Storm Peak Lab Cloud Property Validation Experiment science and operations plan. U.S. Department of Energy Tech Rep. DOE/SC-ARM-10-021, 45 pp.
Mahesh, A., R. Eager, J. R. Campbell, and J. D. Spinhirne, 2003: Observations of blowing snow at the South Pole. J. Geophys. Res., 108, 4707, doi:10.1029/2002JD003327.
Mossop, S. C., 1985: Secondary ice particle production during rime growth: The effect of drop size distribution and rimer velocity. Quart. J. Roy. Meteor. Soc., 111, 1113–1124, doi:10.1002/qj.49711147012.
Nakai, S., and Coauthors, 2012: A Snow Disaster Forecasting System (SDFS) constructed from field observations and laboratory experiments. Cold Reg. Sci. Technol., 70, 53–61, doi:10.1016/j.coldregions.2011.09.002.
Nishimura, K., and M. Nemoto, 2005: Blowing snow at Mizuho station, Antarctica. Philos. Trans. Roy. Soc. London, 363A, 1647–1662, doi:10.1098/rsta.2005.1599.
Palm, S. P., Y. Yang, J. D. Spinhirne, and A. Marshak, 2011: Satellite remote sensing of blowing snow properties over Antarctica. J. Geophys. Res., 116, D16123, doi:10.1029/2011JD015828.
Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation.Kluwer Academic, 954 pp.
Randolph, G., 2012: Variability of the wind resource in Wyoming. M.Sc. thesis, Dept. of Atmospheric Science, University of Wyoming, 126 pp.
Rogers, D. C., and G. Vali, 1987: Ice crystal production by mountain surfaces. J. Climate Appl. Meteor., 26, 1152–1168, doi:10.1175/1520-0450(1987)026<1152:ICPBMS>2.0.CO;2.
Sato, T., K. Kosugi, S. Mochizuki, and M. Nemoto, 2008: Wind speed dependences of fracture and accumulation of snowflakes on snow surface. Cold Reg. Sci. Technol., 51, 229–239, doi:10.1016/j.coldregions.2007.05.004.
Schlenczek, O., J. P. Fugal, K. N. Bower, J. Crosier, M. J. Flynn, G. Lloyd, and S. Borrmann, 2014: Properties of pure ice clouds in an alpine environment. 14th Conf. on Cloud Physics, Boston, MA, Amer. Meteor. Soc. [Available online at https://ams.confex.com/ams/14CLOUD14ATRAD/webprogram/Manuscript/Paper250304/Schlenczek_et_al_2014_AMS_extended_abstract_klein.pdf.]
Schmidt, R. A., 1982: Vertical profiles of wind speed, snow concentration, and humidity in blowing snow. Bound.-Layer Meteor., 23, 223–246, doi:10.1007/BF00123299.
Schmidt, R. A., 1984: Measuring particle size and snowfall intensity in drifting snow. Cold Reg. Sci. Technol., 9, 121–129, doi:10.1016/0165-232X(84)90004-1.
Vali, G., D. Leon, and J. R. Snider, 2012: Ground-layer snow clouds. Quart. J. Roy. Meteor. Soc., 138, 1507–1525, doi:10.1002/qj.1882.
Vionnet, Y., G. Guyomarc’h, F. Naaim Bouvet, E. Martin, Y. Durand, H. Bellot, C. Bel, and P. Puglièse, 2013: Occurrence of blowing snow events at an alpine site over a 10-year period: Observations and modelling. Adv. Water Resour., 55, 53–63, doi:10.1016/j.advwatres.2012.05.004.
Voigt, M., and V. Wirth, 2013: Mechanisms of banner cloud formation. J. Atmos. Sci., 70, 3631–3640, doi:10.1175/JAS-D-12-0353.1.
Wang, P. K., and W. Ji, 2000: Collision efficiencies of ice crystals at low-intermediate Reynolds numbers colliding with supercooled cloud droplets: A numerical study. J. Atmos. Sci., 57, 1001–1009, doi:10.1175/1520-0469(2000)057<1001:CEOICA>2.0.CO;2.
Wang, Z., and Coauthors, 2012: Single aircraft integration of remote sensing and in situ sampling for the study of cloud microphysics and dynamics. Bull. Amer. Meteor. Soc., 93, 653–668, doi:10.1175/BAMS-D-11-00044.1.
Winstral, A., K. Elder, and R. E. Davis, 2002: Spatial snow modeling of wind-redistributed snow using terrain-based parameters. J. Hydrometeor., 3, 524–538, doi:10.1175/1525-7541(2002)003<0524:SSMOWR>2.0.CO;2.
Yang, J., and M. K. Yau, 2008: A new triple-moment blowing snow model. Bound.-Layer Meteor., 126, 137–155, doi:10.1007/s10546-007-9215-4.
