a. Natural cloud characteristics and atmospheric conditions

During this event, a slightly conditionally unstable atmosphere was observed between 4 and 4.5 km at temperatures T between −12° and −18°C (Figs. 3a–b, blue lines, all heights above mean sea level, unless otherwise indicated). Southwesterly flow < 20 m s⁻¹ was observed between the surface and 4 km veering to southerly flow between 5.5 and 7 km where winds increased from 20 to 38 m s⁻¹ (Fig. 3c). Enhanced wind shear of >0.02 s⁻¹ occurred between the surface and 2 km and between 4 and 4.5 km. Prior to seeding, cloud top was > 8 km MSL. This deeper cloud split into two layers, with seeding occurring in the lower cloud layer, which had a top between 4 and 4.5 km. During seeding, the cloud top of the lower layer descended to about 3 km as shallower clouds moved in from the west. Within the shallower cloud layer, tops varied as much as 1 km along a single flight leg. Clouds in which seeding lines were observed had −13° < T < −15°C with cloud top at 4–5 km.

View Full Size

Vertical profile of (a) temperature, (b) equivalent potential temperature, (c) wind speed, and (d) wind direction from the sounding at Crouch closest to the seeding line observations at 1600 UTC 19 Jan (red lines), 0000 UTC 20 Jan (blue lines), and 1600 UTC 31 Jan (green lines). The horizontal gray dashed line represents the height of PJ. The range of cloud-top heights as the seeding lines pass through the ROD are indicated by unfilled boxes in (a); the range of UWKA flights is indicated by filled boxes (color coded by day) associated with the cloud-top range.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

The UWKA flew repeated legs in-cloud within 1 km of cloud top at −11° < T < −14°C. Natural cloud conditions were determined prior to seeding, and during and after seeding outside the seeding lines. Clouds were dominated by SLW, with leg-averaged, LWCs ranging from 0.1 to 0.2 g m⁻³ and maxima ranging between 0.3 and 0.4 g m⁻³. Cloud droplet concentrations were < 30 cm⁻³. Supercooled drizzle with 50 < D < 100 μm was observed in isolated pockets, similar to natural conditions in other cases from SNOWIE (Tessendorf et al. 2019; Majewski and French 2020). Natural ice particle concentrations were < 1 L⁻¹ and often < 0.1 L⁻¹. The only significant concentrations of ice observed at flight level were within seeding lines and over the highest terrain more than 40 km downstream of PJ.

b. Seeding operations and evolution of the seeding lines

The seeding aircraft flew six legs (legs A–F) between 1620 and 1737 UTC (Fig. 2). The reflectivity plumes and microphysical signatures developing from those six legs will be referred to as lines A′–F′. Legs A and B were flown at cloud top (4.1 km) with a mean temperature $\overline{T}$ = −14°C (Fig. 4a). Legs C–F were flown at 4.4 km MSL at $\overline{T}$ = −16°C (Fig. 4a). Legs B–F were flown on a track downwind from A (Fig. 2). Leg A commenced at 1620 UTC. The LWC ranged from 0 to 0.4 g m⁻³ with $\overline{\text{LWC}}$ = 0.11–0.17 g m⁻³ along legs A and B, with lower amounts ($\overline{\text{LWC}}$ = 0.08–0.09 g m⁻³) along legs C and D, mostly confined to the southeastern end of the track (Fig. 4a, Table 1). Negligible SLW ($\overline{\text{LWC}}$ < 0.008 g m⁻³) was observed during legs E and F. Both BIP and EJ flares were deployed during all legs (Table 1, Fig. 4a).

View Full Size

Location of flares (upper plot) and 10-min averaged liquid water content (lower plot) measured by the seeding aircraft for each seeding flight leg on (a) 19, (b) 20, and (c) 31 Jan. The direction of the flight leg is indicated by gray arrows in the upper panels. Circles and lines respectively indicate the locations of the EJ and BIP flares. Data were averaged over 10 min in the lower panels. Numbers in parentheses indicate the mean temperature during the seeding flight leg.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

A line of enhanced reflectivity (Z_e > 15 dBZ_e, with surroundings from −30 to +10 dBZ_e) was first observed at 1647 UTC by the PJ Doppler on Wheels (DOW) radar on the southern edge of the radar observational domain (ROD). Calculations of transport time of the seeding material by the mean wind showed that this line originated at A (and hence will be referred to as A′) where EJs and BIPs were deployed 28 min earlier (Table 1). Line A′ continued to move northeastward with the mean wind, passing over PJ at 1710 UTC (Fig. 5, and Movie S1 in the online supplemental material). A second line of enhanced reflectivity (Z_e > 15 dBZ_e with surroundings from less than −30 to +10 dBZ_e), associated with leg B, was later observed on the southeastern side of the ROD at 1705 UTC, 15 min after the southern portion of B was flown (Fig. 5a). Recall that the seeding aircraft moved its flight track eastward after A (Fig. 2). As a result, A′ and B′ appeared as two parallel lines of enhanced reflectivity with A′ upwind of B′. By the time A′ and B′ propagated through the ROD by 1812 UTC, and moved over higher terrain, the lines had widened and near surface Z_e within the lines enhanced to >25 dBZ_e. Snowfall on the ground was first observed at 1715 UTC and continued through 1812 UTC (Fig. 3 in Friedrich et al. 2020).

View Full Size

Combined maximum Z_e between the surface and 1 km AGL from the PJ and SB DOW radars on 19 Jan at (a) 1710, (b) 1719, (c) 1729, and (d) 1746 UTC. Radar locations are indicated by the star symbols, and the maximum range of 50 km is indicated with a circle centered around each radar. UWKA legs 6–8 are highlighted as a red dashed line; the position of the UWKA aircraft at the radar times is highlighted by an aircraft symbol. Seeding aircraft legs are indicated as black dashed lines. Seeding lines A′ and B′ associated with legs A and B are highlighted.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

The UWKA WCR first detected A′ at 1650 UTC, ~12 km upwind of PJ on UKWA leg 4. At this time, the ice particle plume associated with A′ had a maximum Z_e of 5 dBZ_e as compared with Z_e < −10 dBZ_e in surrounding regions. Line A′ was 1–2 km wide and located at 4–5 km MSL (Fig. 6). Over the next 30 min, A′ grew to 3–5 km wide in the upper portion of the cloud and Z_e increased to 10 dBZ_e. The plume of Z_e associated with A′ extended from cloud top to the surface by 1719 UTC, 8 km downwind of PJ (Fig. 6, leg 6). As precipitation descended to the ground, wind shear caused the near-surface portion of the Z_e plume to lag the upper-level portion so that A′ appeared tilted (Fig. 6, legs 7–8). During UWKA legs 6–10, near-surface Z_e increased from 5 to 15 dBZ_e as A′ passed over higher terrain. Line B′ was observed by the WCR for the first time at 1717 UTC, 15 km downwind of PJ between 3.5 and 4.5 km MSL (Fig. 6, leg 6). Similar to A′, snow began to fall out during the next 30 min. Within B′, Z_e reached a maximum 15 dBZ_e as it propagated over higher terrain. In both A′ and B′, Z_e at 3–4.5 km remained between 5 and 15 dBZ_e for up to 80 min after the seeding material was released, indicating continued particle nucleation and growth in the upper portion of the cloud. Both lines became tilted once the precipitation plume descended below ~3 km as a result of wind shear between the surface and higher altitudes.

View Full Size

Evolution of Z_e from the WCR during UWKA legs 4–10 on 19 Jan. Times given indicate the beginning and end of each leg, and arrows indicate flight direction for that leg. In all cases, wind is from left to right. A clear signal from seeding line A is detected in all seven legs. Seeding line B is detectable in the last five legs. Thick, solid lines contour regions of enhanced reflectivity caused by seeding. Dashed lines show similar regions of increased reflectivity from seeding that are interspersed with areas of natural reflectivity enhancement. The white belt is the WCR blind zone centered at flight level.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

Continuous lines of enhanced reflectivity were not identified from seeding legs C–F. These lines would have been expected to form upwind of A′. Cross sections from the WCR for UWKA legs 6–10 showed a significant decrease in cloud top, from 4.5 km during the time of legs A–B to <3 km when legs C–F were flown. Seeding material from flares released during legs C–F at 4.4 km would not have reached the cloud top and, therefore, no lines developed.

c. Ice initiation and particle growth

During legs 4–10, the UWKA made repeated passes through A′ and B′, 0.5 to 1 km below cloud top with −11° < T <−14°C, near or below the level at which the AgI was released (Fig. 6). Seven and five passes were respectively made through A′ and B′, ranging from 5 to 95 min after the seeding occurred. On leg 4, 20 min after seeding occurred, the UWKA was 1 km below cloud top at T = −11°C and the WCR detected A′ above flight level with Z_e ~ 5 dBZ_e. The Z_e at flight level was −15 dBZ_e, indicating the precipitation in A′ had not yet descended to flight level (Fig. 7c, leg 4). The cloud at flight level consisted of liquid cloud droplets of 20–30 μm in diameter D and was nearly devoid of ice particles (Figs. 7a,b; leg 4). The LWC was between 0.01 and 0.05 g m⁻³ directly beneath the seeding line and as high as 0.08 g m⁻³ a few kilometers downwind. During leg 5, 10 min later at the same level, measurements within A′ showed that the seeding line had descended through the flight level and reflectivity had increased to 10 dBZ_e (Fig. 7c, leg 5). Consequently, the concentration of cloud droplets (D < 50 μm) had been depleted and the concentration of larger, ice particles (D > 100 μm) had increased by two orders of magnitude (Figs. 7a,b; legs 4–5). The observed LWCs within A′ at this time were < 0.01 g m⁻³ and ice water contents (IWC) were > 0.10 g m⁻³. A marked and rapid transition from liquid dominated to ice dominated cloud in A′ had occurred at this level in just 10 min. Few pristine ice crystals were observed just after this transition during leg 5 and most of the ice appeared to be irregularly shaped, suggesting some amount of riming had occurred (Fig. 7b, leg 5). Another five passes through A′ were made over the ensuing 60 min in legs 6–10, and in all cases the observed LWC remained < 0.01 g m⁻³ (Fig. 7c, legs 6–10) and the concentration of hydrometeors D < 50 μm remained < 1 cm⁻³. However, the concentration of larger ice particles continued to exceed, by two orders of magnitude, the concentration observed during the earliest pass and in regions just upwind and downwind of A′. By 50 min after seeding, two-dimensional stereo probe (2DS) images confirmed the presence of dendritic crystals within A′ (Fig. 7b, leg 6). Dendritic crystals were observed in all subsequent legs within A′, interspersed with irregular shaped ice (Fig. 7b, legs 7–10).

View Full Size

(a) Hydrometeor size distributions measured by the in situ probes on the UWKA in line A′ during UWKA legs 4–10 on 19 Jan sampled between 1640 and 1816 UTC. (b) Particle images from the 2DS from seeding-line passage are shown. The frame size is 1.6 mm from top to bottom as indicated in the top frame; the labels on the right indicate the UWKA leg number and the duration between the release of the seeding material and sampling by the UWKA. (c) Vertical profiles of Z_e from the WCR in seeding line A for each of the UWKA legs shown in Fig. 5. Each box is 8 km wide and 4 km tall [leg number and time are on the bottom, as in (b)]. The gray bar indicates that portion of the UWKA track that was in the seeding line for each of the seven passes and used for the size distribution in (a). Maximum LWC observed at flight level within the seeded line is indicated on the top.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

Observations within B′ suggest a similar microphysical evolution (Fig. 8). During UWKA legs 4–5, at 5 and 15 min after seeding, no reflectivity signature was detected by the WCR (Fig. 8c, legs 4–5). For these legs, we estimated the expected location of B′ (had it been detected) using the AgI release time, location along B, and assumed that B is being advected with the mean wind at flight level. During leg 4, the UWKA measured LWC in this region was 0.03 g m⁻³. A significantly smaller amount of LWC was observed in leg 5 and likely resulted from a local lowering of the cloud top. LWCs just a few kilometers on either side of this were between 0.03 and 0.08 g m⁻³ (not shown). During both legs 4–5, the clouds at flight level were nearly devoid of ice and dominated by relatively small supercooled liquid droplets (Fig. 8b, legs 4–5). B′ was detected by the WCR with a reflectivity of 5 dBZ_e during leg 6, 30 min after seeding (Fig. 8c, leg 6). LWC within B′ at flight level was 0.01 g m⁻³, particle images revealed some rimed ice, and particle size distributions showed a decrease in concentration of cloud droplets and an increase in larger hydrometeors (Figs. 8a–c, leg 6). Of the four remaining legs 7–10, only during leg 9 was the LWC > 0.01 g m⁻³ within B’ (Fig. 8c, legs 7, 8, and 10). During this leg, 75 min after seeding, cloud LWCs up to 0.03 g m⁻³ were observed, over higher terrain, nearly 60 km downwind of where the AgI was released. These higher LWCs were collocated in regions of ice hydrometeors concentrations of 2–3 L⁻¹ (Fig. 8a, leg 9) resulting in particle images that appear more irregular and hence more rimed than was observed throughout much of A′. For observations in both A′ and B′, total ice concentrations with D > 100 μm never exceeded 21 L⁻¹. Also, once ice was initiated within the A′ and B′, size distributions for ice hydrometeors with D > 1 mm (Figs. 7a and 8a) were remarkably consistent across all penetrations. Crystals of this size likely resulted from aggregation of smaller, dendritic ice crystals leading to the development of an exponential size distribution of these larger particles (Field and Heymsfield 2003)

View Full Size

As in Fig. 7, but for line B′ on 19 Jan.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

d. Snow growth and fallout

As the seeding lines passed through the ROD, ice crystals and snow continued to grow, generating snowfall on the ground (Fig. 9). Between the first detection of A′ by the PJ radar at 1654 UTC (15 min after seeding) and the time A′ passed over PJ at 1706 UTC (36 min after seeding), Z_e and Z_dr increased from 3 to 7 dBZ_e and 0.6 to 1.8 dB, respectively (Fig. 9a). Between 25 and 40 min after seeding, the values of Z_e < 10 dBZ_e and 1 < Z_dr < 1.5 dB near cloud top (3–4 km; −8° < T < −12°C) suggest that snow most likely grew through water vapor deposition (Ryzhkov and Zrnic 1998; Moisseev et al. 2009; Kennedy and Rutledge 2011; Bechini et al. 2013; Schneebeli et al. 2013). Within 40 min after seeding, weak up- and downdrafts (1–2 m s⁻¹) were observed within A′ by the WCR (Fig. 10, UWKA leg 6). A change in dual-polarization variables started 48 min after seeding (1718 UTC) and lasting for 15–20 min (Fig. 9). Near cloud top (3–4.5 km MSL), pockets of enhanced K_dp (>1° km⁻¹) and Z_dr (0.6–1.2 dB) with updrafts up to 1.5 m s⁻¹ were present. The updrafts near cloud top (Fig. 10, leg 8) may be associated with additional orographic lift that A′ experienced moving up the North Folk Range between 1724 and 1748 UTC after passing PJ (Fig. 9).

View Full Size

RHI composite between 1654 and 1742 UTC along the flight track at 39° azimuthal direction on 19 Jan showing dual-polarization variables for (a) line A′ and (b) line B′ with K_dp on the top plot, Z_dr in the middle plot, and Z_e in the lower plot. Terrain is indicated in dark-gray shading The cone of silence and the area below the lowest radar beam are indicated in light-gray shading. PJ is shown as a star symbol, and the North Folk Range is highlighted. Temperatures were derived from the nearest sounding at 1600 UTC at Crouch (location shown in Fig. 2). Radar times are indicated in the top plot; minutes after seeding are in the middle plot. Red shading indicates the altitude range of the UWKA flight tracks.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Similar to Fig. 6, but for the evolution of near-vertical Doppler radial velocity from the WCR during UWKA legs 6–8 on 19 Jan. Blue shades indicate upward motion, and red shades indicate downward motion. Note that the color scale has been shifted by 1 m s⁻¹ to account for an expected nominal terminal fall velocity of the main scatterers. In this context blue and red regions respectively indicate areas of upward-moving and downward-moving air. Black lines contour regions of enhanced reflectivity caused by seeding.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

The dendritic growth layer (DGL) can be identified from dual-polarization observations as an enhancement in Z_dr and K_dp, reduced ρ_hv, and a strong vertical gradient in Z_e (e.g., Hogan et al. 2002; Kennedy and Rutledge 2011; Lamb and Verlinde 2011; Andrić et al. 2013; Bechini et al. 2013). These changes occur because dendrites enable rapid aggregational snow growth as the crystal branches more readily interlock (Pruppacher and Klett 1997). The enhancement in K_dp and Z_dr near cloud top occurred within −10° < T < −15°C where vigorous growth of dendritic ice particles is expected (Takahashi et al. 1991; Fukuta and Takahashi 1999). Enhanced dendritic growth near cloud top, starting at about 36–48 min after seeding, was also supported by images of particles from the 2DS that showed more dendritic crystals during leg 6 (50 min after seeding) than during leg 5 (30 min after seeding; Fig. 7b). In fact, this enhancement of K_dp (>1° km⁻¹) and Z_dr (0.6–1.2 dB) around 3.8 km or −13°C was observed along the length of A′ as the line encountered higher terrain between 1724 and 1748 UTC 54–78 min after seeding (not shown). Z_dr and K_dp ranged mainly between −7 and 0 dB and between 0° and 0.5° km⁻¹, respectively, between 1654 and 1712 UTC, while positive Z_dr and K_dp up to 1.5 dB and 1.6° km⁻¹ were observed after 1718 UTC. Within the DGL, steady weak updrafts up to 1.5 m s⁻¹ were observed (Fig. 10, leg 7). These updrafts were likely associated with additional orographic lift that A′ experienced as it passed the peak of the North Folk Range at 1730–1742 UTC (Figs. 9a and 10, legs 7–8). As the clouds associated with A′ passed through the ROD, reflectivities remained between 0 and 9 dBZ_e at cloud top (3–4 km MSL). While no information of AgI concentration downwind of PJ is available, it is hypothesized that ice initiation continuously occurred as unactivated residual AgI was transported farther downward and updrafts, associated with higher terrain, provided SLW during local orographic ascent. As precipitation descended to the ground, wind shear caused the near-surface portion of A′ to lag the upper-level portion so that A′ appeared tilted (Fig. 9a). The snow associated with A′ reached the ground within 42–48 min after seeding with Z_e > 10 dBZ_e, K_dp < 0.5° km⁻¹, and downward V_r of 3 m s⁻¹ below 2 km MSL (Fig. 10). These values suggest heavily rimed particles.

Similar evolution and microphysical processes were observed within B′ (Fig. 9b). Line B′ was first observed at 1706 UTC (~20 min after cloud seeding) at 4–4.5 km MSL downwind of PJ, but still upwind of the North Fork Range (Fig. 2). As B′ passed over the upwind side of the North Fork Range, Z_e and Z_dr increased from 1 to 10 dBZ_e and from 0.3 to 1.8 dB, respectively. Passage of the line through the DGL was first observed in the dual-polarization variables at 1730 UTC between 3.5 and 4.5 km MSL with K_dp > 0.8° km⁻¹ and Z_dr > 1 dB in the upper part of the cloud (around 3.8 km or −13°C). Within the DGL, steady weak updrafts (<1 m s⁻¹), associated with orographic lift similar to A′, were observed (Fig. 10, leg 8) between 1730 and 1742 UTC. Similar to A′, Z_e remained between 0 and 9 dBZ_e at cloud top as B′ passed through the ROD suggesting, together with the steady updraft shown in Fig. 10, continuous ice initiation. The snow associated with B′ reached the ground about 36 min after seeding with Z_e > 5 dBZ_e, K_dp < 0.5° km⁻¹, and V_r of 1 m s⁻¹ (Fig. 10) close to the surface (1–2 km AGL).

To further quantify these changes in microphysical processes, we conducted a quantitative analysis by calculating the mean values of dual-polarization parameters at each height and time step associated with A′ (Fig. 11). Within the DGL (~3.3–4.1 km MSL), $\overline{Z_{\text{dr}}}$ steadily decreased from −0.2 to 0.6 dB at 4.1 km MSL (T = −15°C) to −0.4–0 dB at 3.3 km MSL between 1705 to 1806 UTC. The $\overline{K_{\text{dp}}}$ decreased only slightly within the DGL between 1705 and 1806 UTC from 0.5° to 0.6° km⁻¹ at 4.1 km to 0.4°–0.6° km⁻¹ at 3.3 km. A peak in $\overline{K_{\text{dp}}}$ of 0.6° km⁻¹ at 4 km MSL was observed between 1741–1753 UTC. $\overline{Z_e}$ increased below the DGL toward the surface to 10–15 dBZ_e most likely the result of aggregation and riming as snow fell toward the surface. A change toward more aggregated and rimed particles at the surface is seen in decreasing $\overline{{\rho }_{\text{hv}}}$ changing from 0.99 at 1705 UTC to 0.98 between 1734 and 1806 UTC but increased slightly afterward. The enhanced snow growth, most likely related to riming, aggregation, and rapid dendritic growth above the higher terrain between 1718 and 1748 UTC, resulted in higher accumulated snowfall over the North Folk Range and Salmon River Mountains relative to the other areas downwind of PJ between 1705 and 1718 UTC (Fig. 3 in Friedrich et al. 2020).

View Full Size

(a) PPI of Z_e at 2.8 km MSL on 19 Jan. Each plot is 50 km × 100 km. Mean dual-polarization variables are analyzed for line A′ indicated within the black box. Color bars for Z_e and terrain are shown in Fig. 4. Vertical profiles of mean (b) Z_e, (c) Z_dr, (d) K_dp, and (e) ρ_hv as a function of time (color coded). Horizontal lines indicate −10° and −15°C temperatures from the 1600 UTC sounding at Crouch. Red shading indicates the altitude range of the UWKA flight tracks.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

a. Natural cloud characteristics and atmospheric conditions

At cloud top (3.5–4 km), a neutral to slightly conditionally unstable atmosphere with −17° < T < −13°C was observed (Figs. 3a,b). Predominantly southwesterly flow of <10 m s⁻¹ occurred up to 2.5 km changing to westerly flow between 2.5 and 6 km (Fig. 3c). Layers of wind shear > 0.02 s⁻¹ were observed between 2 and 5 km MSL. The clouds formed principally over the higher terrain and did not extend far upwind of the mountain.

The UWKA flew a total of 10 legs, with seven flown in cloud. The presence of extensive pockets of supercooled drizzle with D > 100 μm, resulted in moderate icing conditions requiring that the UWKA fly three legs above cloud top. All in-cloud legs were within 1 km of cloud top at −11° < T < −14°C. Leg-averaged, in-cloud LWCs ranged from 0.1 to 0.2 g m⁻³ with maximum LWC along a leg ranging from 0.45 to 0.6 g m⁻³. Concentrations of cloud droplets were less than 30 cm⁻³. Observed natural ice concentrations were generally less than 0.1 L⁻¹ except in isolated pockets over higher terrain, where concentrations at flight level were between 1 and 5 L⁻¹.

b. Seeding operations and evolution of the seeding lines

The seeding aircraft flew eight legs (legs A–H) on a constant flight track between 0003 and 0129 UTC at an altitude of 4.1 km for legs A, F, and G, and 3.8 km for legs B–E and H (Fig. 2). $\overline{\text{LWC}}$ between 0.04 and 0.28 g m⁻³ and $\overline{T}$ = −14°C were observed along nearly the entire seeding aircraft flight track during seeding legs D, E, and H (Fig. 4b, Table 1). While legs A, B, F, and G were flown mainly (89%–100% of the flight leg) above the cloud to avoid heavy icing, with $\overline{T}$ from −14° to −15°C, 59%–82% of legs C–D were also flown in cloud at T = −14°C, albeit with lower LWC values (0.11 < $\overline{\text{LWC}}$ < 0.18 g m⁻³). As a result, only EJ flares were released during legs A, F, and G. BIP and EJ flares were used in legs B–E and H (Table 1, Fig. 4b).

Reflectivity plumes from all eight seeding legs appeared as a zigzag seeding signature pattern of lines A′–H′ (Fig. 12, Movie S2 in the online supplemental material). Background (natural) reflectivity values ranged between −30 and −10 dBZ_e throughout the event. Lines A′ and B′ initially consisted of circular areas or “dots” of Z_e > 10 dBZ_e with background values from −30 to −5 dBZ_e and were first observed by the PJ DOW radar between 0030 and 0045 UTC ~5 km downwind of A–B at 2.6 km MSL (supplemental Movie S2). The Z_e dots were associated with ice particles forming from individual EJ flares deployed during A and the beginning of B (Fig. 4b). Their 3–4 km spacing matched the average 3-km distance between flare drops. As A′ and B′ moved downwind, C′, D′, and E′ appeared between 0045 and 0124 UTC as semicontinuous lines, a result of continuous ejection of AgI through both BIP and EJ flares. Lines F′ and G′ later appeared as Z_e dots between 0124 and 0143 UTC, the result of using only EJ flares along F and G. Line H′, the final line, appeared between 0142 and 0152 UTC. As the lines propagated through the ROD, they broadened quickly and merged with other lines because of the low wind speed (<10 m s⁻¹) and shear up to 0.02–0.03 s⁻¹ between the surface and 4 km MSL. Precipitation accumulated mainly over the higher terrain and within the ROD (Fig. 3 in Friedrich et al. 2020).

View Full Size

As in Fig. 5, but Z_e from PJ radar for 20 Jan at (a) 0047, (b) 0116, (c) 0143, and (d) 0211 UTC.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

The UWKA flew along the NW edge of seeding lines A′–H′ and intersected the lines on their northernmost extent (Fig. 12) within a region of natural (background) Z_e between −15 and −10 dBZ_e (Fig. 13). Because lines A′–H′ were arranged in a zigzag pattern, the UWKA intersections appeared as line pairs. Line pair A′B′ was first detected by the WCR at 0032 UTC during UWKA leg 6, 3 km upwind of PJ and 21 min after seeding (Fig. 13). At this time, pair A′B′ was confined to between 3 and 4 km MSL and was 1.5 km wide with a maximum Z_e of 0 dBZ_e. Line pair A′B′ generated Z_e plumes that extended to the surface within 18 min (Fig. 13, leg 7) and precipitated out (Fig. 13, leg 8) 35 min after pair A′B′ was first detected by the WCR. Line pair C′D′ was detected by the WCR at 0047 UTC (Fig. 13, leg 7), 15 min after the seeding aircraft turned from seeding leg C to leg D. By 0108 UTC, Z_e plumes reached the surface. By leg 9, pair C′D′ was 8 km wide at 3.5 km MSL and contained Z_e of 10 dBZ_e from cloud top to the surface. Pair C′D′ was still detectable during the UWKA’s last leg (Fig. 13, leg 10), but the maximum Z_e had decreased to 5 dBZ_e and the width at 3.5 km narrowed to 2.5 km. Line pair E′F′ was not detected by the WCR during leg 8 but was clearly visible on leg 9, with Z_e plumes reaching the surface 20 min later on leg 10. Similarly, line pair G′H′ was detected during UWKA leg 9 at 0126 UTC, just 5 min after the seeding aircraft turned from legs G and H. The Z_e at this time was from −10 to −5 dBZ_e at 3.5 km MSL, only 5 dB greater than the natural cloud. Twenty-three minutes later (Fig. 13, leg 10), Z_e within pair G′H′ had increased to 5 dBZ_e and extended from cloud top to 2 km MSL.

View Full Size

As in Fig. 6, but showing UWKA legs 6–10 on 20 Jan. Since the UWKA passed through the northwest end of the seeding lines, pairs of lines show up as a single intersect and are therefore labeled as pairs.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

Line pairs A′B′, C′D′, E′F′, and G′H′ were detected by the WCR 30, 15, 30, and 5 min after seeding occurred, respectively. For pairs A′B′ and E′F′, previous UWKA legs at 2 and 12 min after seeding, respectively, failed to detect a line. In all cases, lines were initially 1–2 km wide, and rapidly grew in width, with pairs C′D′ and E′F achieving maximum width of 6–8 km 50–55 min after seeding. For pairs C′D′, E′F′, and G′H′, the seeding lines were discernible during the last leg flown by the UWKA (Fig. 13, leg 10) from 0145 to 0149 UTC. Note also that the seeding lines drifted east of the UWKA flight track, since the UWKA flight track was not oriented directly along the mean wind (Fig. 2).

On the basis of the seeding aircraft track and the ambient winds, all seeding lines should have passed over PJ. The PJ MRR observed three distinct groups of seeding lines (Fig. 14). The first group between 0100 and 0114 UTC was related to pair C′D′. Pairs E′F′ and G′H′ passed over the MRR between 0128 and 0131 UTC and 0154 and 0159 UTC, respectively. A maximum Z_e of 10 dBZ_e and V_r = −1 m s⁻¹ were observed. As indicated by both the MRR and WCR measurements, none of the snow generated by seeding reached the surface at PJ but rather fell to the surface downwind.

View Full Size

Vertical profile of Z_e (color coded) and Doppler velocity (black lines; m s⁻¹) observed by the MRR at PJ on 20 Jan between 0045 and 0200 UTC. Seeding lines C′–H′ are labeled. Blue shading indicates the altitude range of the UWKA flight tracks.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

c. Ice initiation and particle growth

The UWKA flew 5 legs (legs 6–10) in which seeding lines were detected. Due to moderate icing conditions encountered on this day, only legs 7, 9, and 10 were in-cloud while legs 6 and 8 were above cloud. During the three in-cloud legs, three passes were made through line pair C′D′, two passes were made each through line pair E′F′ and line pair G′H′, and only one pass was made through line pair A′B′. Passes through the seeding lines were made 5–75 min after seeding occurred and within 500 m of cloud top at the T = −12°C level. Observed LWCs within the seeding lines ranged from < 0.01 to 0.143 g m⁻³ in pair C′D′ (Fig. 15d), 0.01 to 0.047 g m⁻³ in E′F′ (Fig. 15e), and 0.016 to 0.034 g m⁻³ in line pair G′H′ (Fig. 15f). For all particle size distributions measured at flight level within the seeding lines and 30 min or more after seeding, a distinctive “tail” of larger (ice) particles, D > 100 μm, was evident with mean concentrations of 1–4 L⁻¹ (Figs. 15a–c). In only two cases, in pair G′H′ at 5 min after seeding and in pair C′D′ at 15 min after seeding (Fig. 15a, leg 7; Fig. 15c, leg 9), was the tail absent; note that during leg 7 in pair C′D′, 15 min after seeding, the UWKA passed underneath the level of the seeding line. The next in situ observation of pair C′D′ was not made until 60 min after seeding during leg 9, by which time particles several mm in diameter were observed (Fig. 15a, leg 9). In all cases, once ice formed, concentrations of ice particles with D > 100 μm never exceeded 26 L⁻¹ at flight level. The Z_e at cloud top began to decrease with time, as the snow precipitated out of the cloud, despite the clouds appearing to contain significant amounts of available SLW.

View Full Size

Hydrometeor size distributions measured by in situ probes on the UWKA in (a) lines C′D′, (b) lines E′F′, and (c) lines G′H′ corresponding to UWKA legs 7–10 on 20 Jan (cf. Fig. 7). Vertical profiles of Z_e from the WCR for (d) lines C′D′, (e) lines E′F′, and (f) lines G′H′ shown in (a)–(c). Each box is 9 km wide and 5 km tall. The gray bar indicates that portion of the UWKA leg from which the size distributions were constructed. The labels below each image indicate the UWKA leg number and the duration of the gray bar. Labels above indicate the maximum LWC observed within the gray bar.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

d. Snow growth and fallout

We were unable to analyze each seeding line separately in the PJ radar data as the eight seeding lines propagated slowly toward the northeast and started to merge quickly, in particular along the northern and southern turning points of the seeding aircraft (Fig. 16a). Instead, we divided the area that the seeding lines moved through into four northwest–southeast-oriented, 8-km-wide, and 60-km-long boxes with box 1 (box 4) representing the earlier (later) stage of the seeding lines’ evolution. Widespread snowfall over more than 2000 km², indicated by the area with Z_e > 15 dBZ_e in Fig. 16b, was primarily observed in box 2 and box 3 between 2 and 3.5 km MSL between 0120 and 0215 UTC. The largest area of snowfall (>3000 km²) was observed at 0123 UTC in box 2 mainly associated with line pair C′D′ about 30–60 min after seeding (Fig. 16b). The largest area in box 2 (>3000 km²) was observed around 2.4–2.7 km MSL and was slightly higher, between 2.5 and 3 km MSL, in box 3 (Fig. 16b). The decrease in area below 2.5 km MSL across all boxes was mainly related to complete and/or partial beam blockage of the radar beam, which was more severe with distance from the radar (particularly by box 4) and, therefore, might not represent realistic snowfall conditions.

View Full Size

As in Fig. 11, but (a) PPI of Z_e at 2.8 km MSL for four radar times on 20 Jan. ROD is divided into four 8 km × 60 km boxes (boxes 1–4) indicating the analysis area. Each panel is 50 km × 60 km. Vertical profiles of (b) area with Z_e > 15 dBZ, (c) $\overline{Z_e}$, (d) $\overline{K_{\text{dp}}}$, and (e) $\overline{Z_{\text{dr}}}$ as a function of time (color coded). Horizontal lines indicate −5°, −10°, and −15°C temperatures from the 0000 UTC sounding at Crouch. Gray shading indicates height levels that might be partially affected by radar beam blockage. Blue shading indicates the altitude range of the UWKA flight tracks.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

The magnitude of mean dual-polarization parameters ($\overline{Z_e}$, $\overline{K_{\text{dp}}}$,and $\overline{Z_{\text{dr}}}$) indicate snow growth with time (Figs. 16c–e). $\overline{Z_e}$ increased from ~10 dBZ_e at 4 km to ~15 dBZ_e at 2.5 km during the time of maximum snowfall (0100–0230 UTC). In addition,$\overline{Z_e}$ increased starting at 0034 UTC and peaked at around 15 dBZ_e at and around 2.5 km between 0123 and 0151 UTC. Similar to $\overline{Z_e}$, $\overline{K_{\text{dp}}}$ showed very little temporal and vertical change. During the main snowfall, $\overline{K_{\text{dp}}}$ ranged between 0.3° and 0.5° km⁻¹ with temporal variations of ±0.1. A slight increase in $\overline{K_{\text{dp}}}$ within box 2 and box 3 occurred around 3.5 km MSL where T = −13°C. This might be an indication of dendritic growth. Between 0100 and 0230 UTC, $\overline{Z_{\text{dr}}}$ remained slightly higher (0.5–1 dB) within the DGL (2.5–4 km MSL) relative to the surface. This indication of dendritic growth was consistently observed in box 2 and box 3 and sporadically in box 1 and box 4. As dendritic growth can lead to rapid snow formation and fallout, the question still remains as to why the seeding lines precipitated out so much faster and farther upwind in comparison with 19 January. This will be explored in section 6.

a. Natural cloud characteristics and atmospheric conditions

Below 2.5 km, a stable boundary layer was observed with westerly winds > 8 m s⁻¹ and T of about 0°C (Figs. 3a,b). Above this stable layer, westerly winds increased from 8 to 40 m s⁻¹ at 5.5 km and above (Fig. 3c). Wind shear layers (>0.03 s⁻¹) occurred between the surface and 5.5 km MSL. The atmosphere was stable with T decreasing from −5°C at 2.5 km to −35°C at 8 km.

Cloud tops with −13° < T < −15°C were steady, between 4.8 and 5.2 km through the entire event, rarely varying more than a 100 m over a flight leg. The UWKA never penetrated more than 150 m below cloud top. Early legs identified severe icing conditions, with widespread presence of supercooled drops with D > 100 μm and some observations of supercooled drops exceeding 150 μm in diameter. The limited in situ observations on this day reveal near-cloud-top LWCs up to 0.4 g m⁻³ with droplet concentrations ranging from 20 to 30 cm⁻³.

b. Seeding operations and evolution of the seeding lines

Because of severe icing and strong winds, the seeding aircraft only flew two legs (legs A–B) deploying BIP and EJ flares continuously (Fig. 4c; Table 2) on a constant flight track between 2040 and 2105 UTC at 4.9 km MSL. Both legs were flown mainly in cloud (> 95%; Table 1) with westerly winds at about 30 m s⁻¹ and $\overline{T}$ = −13°C. LWC was < 0.5 g m⁻³ with $\overline{\text{LWC}}$ ranging between 0.23 and 0.24 g m⁻³ (Fig. 4c).

Two parallel lines, A′ and A″, with Z_e > 15 dBZ_e (in a background of < 10 dBZ_e) separated by about 5 km emerged from the first seeding leg (line A) at 2105 UTC, as observed by the DOW radars (Fig. 17 and Table 1; also see Movie S3 in the online supplemental material). Line A′, farther downwind, had a more continuous pattern, consistent with the burning of BIP flares. Line A″, upwind of A′, consisting of distinct comma-shaped areas of Z_e, related to individual EJ flares, which dropped to lower altitudes into a weaker wind regime below the seeding aircraft flight level. The AgI from these EJ flares was vertically distributed over a depth of about 820 m below flight level, while BIP flares burned as a horizontal line at flight level. Considering shear of >0.03 s⁻¹ between 4.8 and 4.9 km MSL, AgI from BIP and EJ flares were advected at a different wind speed causing the separation into two parallel lines. Line A′ (associated with BIPs) broadened rapidly from 3–5 to >10 km wide between 2114 and 2124 UTC (Fig. 18). Line A″ (associated with EJs) also broadened from <1 km to about 5 km wide but remained much narrower than A′. Wind shear caused the upper part of A′ and A″ to propagate faster than the lower part creating forward tilted seeding lines. Line A′ reached the surface with Z_e > 15 dBZ_e at about 2114 UTC and continued to snow out mostly on the east side of the ROD (Friedrich et al. 2020). Areas with Z_e > 20 dBZ_e at the surface were still observed as A′ moved out of the ROD (Figs. 17d, 18b). Snowfall with Z_e > 20 dBZ_e associated with A″ started to reach the surface at 2124 UTC.

View Full Size

As in Fig. 11, but for Z_e observed by the SB DOW radar on 31 Jan at (a) 2110, (b) 2122, (c) 2134, and (d) 2146 UTC.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

View Full Size

(a) Vertical west–east cross section of Z_e from the UWKA flight legs 6–9. Times and flight direction are indicated. Terrain is shown in black. (b) West–east RHI scan along flight track on 31 Jan observed by the PJ DOW radar at 2114, 2124, 2130, and 2142 UTC. UWKA flight legs 6–9 and the position of the aircraft are indicated as a red line and red aircraft symbol. Lines A′ and B′ indicate the BIP flares, and lines A″ and B″ are the EJ flares. Dark-gray shading indicates topography; lighter-gray shading indicates approximated radar coverage.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

The same pattern of two parallel lines of Z_e > 15 dBZ_e in a background of <10 dBZ_e emerged after seeding leg B. The northern part of B′ and B″ was first observed at 2122 UTC, with B′ quickly merging with A″ at 2129 UTC (Fig. 17c). At 2134 UTC, most of the lines had merged or had left the ROD.

c. Snow growth and fallout

The first seeding lines were observed about 30 min after seeding. Initially, echoes associated with EJ flares were smaller in size and occurred downwind of the BIP flares (Fig. 19a). Since both EJ and BIP flares merged within 10–15 min of radar detection and quickly moved out of the ROD, we analyzed spatiotemporal dual-polarization variables combined within both seeding legs. Dual-polarization variables indicated a rapid increase in snowfall and increase in the size of the seeding lines over 36 min. Below 3.5 km MSL, $\overline{Z_e}$ increased with time from ~7 dBZ_e at 2110 UTC up to 15 dBZ_e at 2134 UTC (Fig. 19b). Over the same time, $\overline{Z_e}$ between 4 and 4.5 km MSL increased from ~7 to 13 dBZ_e. Note that at and after 2134 UTC parts of the seeding lines already moved out of the ROD causing $\overline{Z_e}$ to decrease after 2134 UTC (Fig. 19b). An increase in $\overline{Z_e}$ with decreasing height was primarily observed after 2129 UTC indicating a rapid increase in particle diameter. Note that the increase in $\overline{Z_e}$ with decreasing height in line A′ was already observed at 2114 UTC and later occurred persistently in all seeding lines (Fig. 18b). Little change in $\overline{Z_{\text{dr}}}$ with decreasing height was observed between 2117 and 2134 UTC; except for a slight increase in $\overline{Z_{\text{dr}}}$ of 0.3 dB with decreasing height at 2110 UTC (Fig. 19c). However, a peak in $\overline{Z_{\text{dr}}}$ was observed at 2141–2146 UTC at about 4.3 km (−13°C) with an increase in $\overline{Z_e}$ below 4 km indicating the possibility of dendritic growth. While $\overline{K_{\text{dp}}}$ also remained relatively constant with decreasing height, a slight increase in $\overline{K_{\text{dp}}}$ was observed at 2110 and 2146 UTC in the DGL (Fig. 19d). $\overline{{\rho }_{\text{hv}}}$ profiles showed higher values at earlier (2110 UTC) and later times (2141–2146 UTC; Fig. 19e). All profiles showed a decrease of $\overline{{\rho }_{\text{hv}}}$ with decreasing height indicating a broadening of different hydrometeor types and shapes.

View Full Size

As in Fig. 11, but showing (a) PPI of Z_e at 2.8 km at six radar times on 31 Jan. Each plot is 50 km × 95 km. Vertical profiles of mean (b) Z_e, (c) Z_dr, (d) K_dp, and (e) ⍴_hv over all seeding lines. Horizontal lines indicate −10° and −15°C temperatures from the closest sounding at 1600 UTC sounding at Crouch. Green shading indicates the altitude range of the UWKA flight tracks.

Citation: Journal of Applied Meteorology and Climatology 60, 7; 10.1175/JAMC-D-20-0206.1

• Download Figure
• Download figure as PowerPoint slide

a. Impact of AgI amounts

Ice nucleation efficiency of AgI has been explored in experimental and theoretical studies (e.g., DeMott et al. 1983; DeMott 1995; Boe and DeMott 1999; Xue et al. 2013; Marcolli et al. 2016). Ice nucleus (IN) size and concentration has been identified as controlling ice formation, together with temperature, water vapor saturation, and cloud droplet, which will be discussed in the following sections. Particle size distribution generated by burning AgI flares depends on updraft strength with larger-sized particles occurring during weaker updrafts (DeMott et al. 1983). Since IN size and number concentration observations are not available, we chose to use the total AgI mass as a proxy ice nuclei production acknowledging that the same mass of AgI can lead to different size and number concentration under varying environmental conditions. Ultimately, the ice particle concentrations observed serves as a direct measure of how many ice nuclei actually activated within the seeding plumes.

A total of 445 g of AgI from EJ and BIP flares was released on 20 January from eight seeding legs over approximately 82 min, while on 19 and 31 January, only 20% and 40% of the amount released on 20 January (87 and 178 g AgI), respectively, was distributed over two flight legs in about 19–26 min (Table 3). Despite releasing only 40% as much AgI on 31 January relative to 20 January, the amount of LESnow produced on 31 January was 29% more than was produced on 20 January. Further, the amount of LESnow produced on 20 January was only 2 times that amount produced on 19 January, despite releasing about 5 times as much AgI. Clearly, more AgI did not necessarily produce more LESnow, hinting that atmospheric conditions might play an essential role in the amount and distribution of snowfall. For 1 g of AgI released, 1901 m³ of total LESnow was generated on 31 January, 1409 m³ on 19 January, and 542 m³ on 20 January. This implies that environmental conditions must have played an important role in the amount and distribution of snowfall produced through seeding.

b. Impact of LWC and T along the seeding track

Ice yield also depends temperature and LWC (e.g., DeMott et al. 1983; DeMott 1995; Boe and DeMott 1999; Xue et al. 2013; Marcolli et al. 2016). In a cloud chamber experiment, Boe and DeMott (1999) quantified the number of nuclei generated per gram of AgI as a function of temperature and LWC for BIP flares. In this experiment, the yield of ice crystals increases as T decreases from −5.5° to −10.2°C with constant LWC. As T remains constant at −6° and −10°C, more yield was found when LWC = 0.5 g m⁻³ rather than 1.5 g m⁻³.

The T along the seeding track only fluctuated by 1°–2°C between the days with −14°C during legs A and B on 19 January, from −15° to −14°C on 20 January, and −13°C on 31 January (Fig. 4). $\overline{\text{LWC}}$ > 0.23 g m⁻³ was observed along both seeding legs (A and B) on 31 January and two seeding legs (E and H) on 20 January, while all other in-cloud legs (A–D) on 20 January and every leg on 19 January had $\overline{\text{LWC}}$ ranging from 0.04 to 0.18 g m⁻³ (Table 1). Although the observations do not reveal information on how much AgI was activated, legs D, E, and H on 20 January had the highest measured $\overline{\text{LWC}}$ on that day and lines D′, E′, and H′ showed higher Z_e values (peak at 30 dBZ_e) relative to A′, B′, and G′ with Z_e peaks of <20 dBZ_e (Fig. 12). Lines D′, E′, and H′ also persisted longer (1–2 h) than the other lines on 20 January where snow fell out before 1 h (Fig. 12). Lines D′, E′, and H′ had higher total AgI discharge (>63.2 g per leg) relative to other legs on this day (30.8–66.8 g), with the exception of C′, which totaled 66.8 g of AgI. These observations are consistent in that lines with higher LWC and greater mass discharge of AgI persisted longer and with higher Z_e values than the other lines on this day.

On 31 January, the $\overline{\text{LWC}}$ was similar for both legs (0.24 and 0.23 g m⁻³ for legs A and B, respectively). Also, for both legs the discharge of AgI was comparable (94.8 vs 83.8 g). The resulting seeding lines from BIP (A′ and B′) and EJ flares (A″ and B″) had similar peak Z_e, respectively, with higher values associated with the BIP-related lines (30 dBZ_e for A′ and B′ and 20 dBZ_e for A″ and B″). Seeding lines persisted as they both advected through the entire ROD. $\overline{\text{LWC}}$ for the two legs (A and B) on 19 January was 0.11 and 0.17 g m⁻³. The amount of AgI released on these legs was nearly the same (44.8 and 42.6 g). Both legs had a similar maximum Z_e (30 dBZ_e) and persisted for a similar length of time (45 min).

Within a single day, our observations suggest that higher $\overline{\text{LWC}}$ along the seeding track and greater amounts of AgI release correspond to lines with greater Z_e that persist longer. However, this relationship does not necessarily hold when comparing across days. Lines associated with higher $\overline{\text{LWC}}$ along the seeding track (19 January: A and B; 20 January: C, D, E, and H; 31 January: A and B) show similar peak Z_e of 30 dBZ_e. However, the width of the seeding line and the persistence within the ROD differs. The widths of the seeding lines on 19 and 20 January are similar, whereas snow fell out rapidly over a smaller area producing 51% more snow on 20 January relative to 19 January (Fig. 3 in Friedrich et al. 2020). On 30 January, seeding lines were wider, covering the largest areas (2410 km²) and the largest amount (339 540 m³) of snowfall among the three cases. This implies that $\overline{\text{LWC}}$ along the seeding track plays an important role for ice initiation and formation of the seeding lines. Yet, enhanced riming might determine how fast snow falls out and wind speed and shear determines AgI dissemination and transport across the ROD.

c. Impact of LWC downwind of the seeding track

While it is important to consider the amount of the LWC along the seeding track, one should also consider how the amount and persistence of LWC downwind of the seeding track may influence the evolution and persistence of seeding lines. The ability for the UWKA to obtain in situ measurements downwind of the seeding track varied by day, due mainly to the presence of supercooled drizzle and its impact on airframe icing. On all three days, cloud droplet concentrations were < 30 cm⁻³ and mean cloud droplet diameters ranged from 20 to 30 μm (Table 3). UWKA-measured in-cloud $\overline{\text{LWC}}$ was 0.1–0.2 g m⁻³ on all three days. However, on 31 January the UWKA conducted only a few flight legs in cloud, and those were always within 150 m of cloud top, whereas on 19 and 20 January flight legs penetrated deeper into the cloud, typically 500–1000 m below cloud top. Maximum LWCs observed along legs were greatest on 20 January (0.45–0.6 g m⁻³) and least on 19 January (0.3–0.4 g m⁻³). From the UWKA measurements available from 31 January, LWCs appeared to be steadier through the length of the legs than on 19 and 20 January.

On 19 January, supercooled drizzle drops were observed in isolated pockets and seldom exceeded 100 μm in diameter. On 20 January, drops with D < 150 μm were observed and occurred more extensively than on 19 January, often along one-third to one-half of a UWKA flight leg. The most supercooled drizzle was observed on 31 January. During the few cloud penetrations made by the UWKA, supercooled drizzle was widespread, with 100 < D < 200 μm.

Across the three days, the greatest amount of LESnow produced through seeding occurred on the day with the largest and most widespread occurrence of supercooled drizzle (31 January). One might conjecture that LWC was also more widespread on this day hence leading to greater drizzle production. However, the inability of the UWKA to penetrate more than 150 m below cloud top made comparison between days difficult. It is clear that the day with the least supercooled drizzle and the lowest LWC along the UWKA flight legs downwind of the seeding track (19 January) produced the smallest amount of LESnow through seeding.

d. Impact of wind shear

The question remains as to why 29% more total LESnow accumulated on 31 January relative to 20 January despite similar values of LWC and AgI released and why seeding lines precipitated out faster on 20 January than on 19 January. Although the observations do not provide information on AgI dispersion and spatiotemporal AgI concentration, we hypothesize that strong shear leads to more efficient dispersion of AgI within supercooled clouds resulting in rapid and efficient precipitation formation, which can be tested in future modeling work. Shear at and below seeding level was ~0.01 s⁻¹ greater on 20 and 31 January than on 19 January (Table 3) resulting in more efficient dispersion of AgI. In particular, the rapid decrease in wind speed with decreasing height (40–30 m s⁻¹ between 4 and 5 km) on 31 January led to a separation of the BIP and EJ flares, which was not observed on 19 and 20 January. This separation of flares shown in the Z_e fields further suggested that AgI was activated and distributed over a much larger area (Figs. 17, 18) than on 19 and 20 January (Figs. 5, 6, 12, and 13). The efficient AgI dispersion on 31 January, in combination with greater LWC along and downwind of the seeding track and most supercooled drizzle observed, contributed to the largest area covered by snowfall (2410 km²), highest peak snowfall at 0.25 mm, and highest total accumulations of 339 540 km² on 31 January as compared with 19 and 20 January (1838–2327 km²; 0.14–1.5 mm; 123 220–241 260 m³; Table 3). Interestingly, on 31 January, reflectivity plumes associated with BIP flares resulted in qualitative larger seeded areas (larger area of Z_e > 0 dBZ_e) relative to EJ flares (Fig. 18).

e. Impact of wind speed

While shear affects AgI dispersion, stronger winds will transport ice particles produced through seeding farther downwind. Winds at the seeding level were strongest on 31 January, with seeding lines remaining only 30 min in the ROD, with not all snow reaching the surface inside the ROD. Conversely, winds were weakest on 20 January and on this day the seeding lines mostly reached the surface within the ROD, approximately within 40 km downwind of the seeding legs. On 19 January, winds were 5–8 m s⁻¹ stronger than on 20 January and 12–25 m s⁻¹ weaker than on 31 January. LESnow on 19 January was almost equally distributed over the ROD, whereas on 20 January snowfall mainly accumulated over an area one-half of the size of that on 19 January (Fig. 3 in Friedrich et al. 2020). Wind speed, therefore, played a role in the residence time of seeding lines within the ROD and the resultant distribution of snowfall.

f. Impact of ice particle growth mechanisms

In situ observations of crystal concentrations and habits were made on both 19 and 20 January. On both days, the approximate time between the release of AgI and the development of a seeding line with Z_e > 5 dBZ_e was 15–30 min. After seeding lines were detected by the UWKA, ice particle concentrations remained, on average, between 2.5 and 8 L⁻¹ on 19 January and slightly less (1–3.8 L⁻¹) on 20 January. Also, IWC within seeding lines ranged from 0.1 to 0.48 g m⁻³ on 19 January and 0.1–0.27 g m⁻³ on 20 January. Despite these lower values on 20 January, more LESnow was produced and the lines precipitated out faster on this day.

As noted earlier, LWC measured by the UWKA was greater on 20 January and supercooled drizzle was more prevalent. This may have resulted in more riming. Indeed, images of ice crystals from the UWKA suggest this to be the case, leading to more rapid fallout. Unlike 20 January, the radar returns on 19 January persistently maintained strong echoes (Z_e > 5 dBZ_e) near cloud top. Evidence of dendritic growth in the upper part of the cloud was continuously observed as the seeding lines passed through the ROD. We hypothesize that ice initiation continuously occurred as unactivated residual AgI was transported farther downwind and updrafts, associated with encountering higher terrain, provided SLW as a result of local orographic ascent. This likely aided in the persistence of the seeding lines on 19 January as compared with 20 January.

g. Impact of snow growth mechanisms

Snow growth mechanisms were similar for all three cases. Dendritic growth was observed in the upper part of the clouds where −10° < T −15°C as the seeding lines passed through the ROD. Snow growth related to riming and aggregation occurred closer to the surface, based on radar polarization signatures. The largest increase in Z_e with decreasing height was observed on 31 January (6.25 dBZ_e over 1 km), the day with the greater LWC along and downwind of the seeding track, most supercooled drizzle, and the largest LESnow. On 20 January, Z_e increased by 4.6 dBZ_e over 1 km with decreasing height, while 3 dBZ_e over 1 km was observed on 19 January. These increases occurred close to the surface below the dendritic growth zone. Snowfall at the surface was first observed 12 min after seeding on 31 January and 40–45 min on 19 and 20 January (Table 3; Friedrich et al. 2020). Rapid fallout of snow, highest LWC, and highest Z_e gradient led to the conclusion that heavy riming must have occurred on 31 January. Riming most likely also occurred on 19 and 20 January, but to a lesser degree. In comparing 19 and 20 January, it is seen that snow fell out faster on 20 January, the day with higher LWC, extensive regions of supercooled drizzle droplets with 50 < D < 150 μm, and more AgI release (445 vs 87.4 g).

Cloud-top heights were the highest on 31 January and lowest on 19 January (Table 2). The primary impact of cloud-top height is to affect the residence time of snow in the air prior to impacting the mountain when seeding is conducted near the cloud top. Given similar winds, longer residence times shift the snow farther downwind across the target area.