In this experiment, a snow cloud system was selected on November 29, 2019 for research about the response of mixed-phase cloud microphysical properties to cloud seeding, and aircraft detection and cloud seeding were carried out over Xingtai, Hebei Province. The King Air-350 aircraft operated by Hebei Weather Modification Office was used for the observation of clouds with an airborne detection system on board, including cloud particle detection system, meteorological measurement system, Nevzorov hot wire liquid water content (LWC), total water content (TWC) probe, and global positioning system. The cloud particle detection system in this study consists of fast cloud droplet probe (FCDP), cloud imaging probe (CIP), and cloud particle imager (CPI). The FCDP probe can obtain the concentration of cloud droplets in the size range of 2–50 μm in radius, with uncertainty of approximately 20% (Faber et al., 2018; Lance et al., 2010). CIP can detect large cloud particles in the size range of 25–1,550 μm, and the CPI probe is able to obtain high-resolution images of cloud particles. The LWC and the TWC were obtained by the Nevzorov hot wire LWC/TWC probe, and the uncertainty was about 15% (Emery et al., 2004; Korolev et al., 2013). CPI uses a CCD camera to capture a high-resolution image of cloud particles, which can distinguish water drops from ice particles. Vertical profiles of temperature and humidity were obtained using the meteorology monitoring system called Aircraft-Integrated Meteorological Measurement System. Table 1 shows these main detection equipment used in this study, along with their detected cloud variables and measurable ranges. Cloud seeding and the detection of cloud microphysical properties were performed by the same aircraft. Also noted is that the silver iodide flames used in this study contain AgI of 1.8 g per stick.

The cloud particle size spectra observed in this study were obtained by FCDP and CIP probes. Considering that most of the ice crystal particles in stratiform clouds are larger than 100 μm in diameter, we assume that cloud particles larger than 100 μm are ice crystals, which can be obtained by CIP probe of 100–1,550 μm, and that the particles smaller than 100 μm are cloud droplets, which can be obtained by FCDP (Korolev and Field, 2015). Considering the higher supersaturation condition of the ice surface than the liquid droplet surface in the mixed-phase cloud, ice crystals larger than 25 μm can grow rapidly in 1 min. Thus, our assumption in general that the FCDP observed particles are all droplets in the cloud examined here should be reasonable. In order to obtain the changes of cloud particle spectra between before and after catalysis in a more detailed manner, the sampling frequency of all instruments used in this study is 1 Hz. Before using the data, we performed quality control on the observed data based on the instrument operating status and true air speed values.

The cloud microphysical parameters used in this study, including the effective radius of cloud particles r _e, number concentration of cloud particles N _ci , and ice water content (IWC) of clouds measured by Nevzorov-LWC/TWC IWC_Nevzorov are calculated as follows: r e = ∫ r 3 n ( r ) d r ∫ r 2 n ( r ) d r = ∑ N c i r i 3 ∑ N c i r i 2 (1) N d = ∫ n ( r ) d r = ∑ N c i (2) I W C N e v z o r o v = T W C N e v o z r o v − L W C N e v z o r o v (3) where n(r) is the droplet size distribution with number concentration of n at size r, and N _ci and r _i are the droplet number concentration and the particle size radius of the i-th bin of the instrument, respectively. Nevzorov hot wire LWC/TWC probe can only obtain liquid and total water content, and IWC_Nevzorov is calculated by subtracting LWC_Nevzorov from TWC_Nevzorov. Due to the potential uncertainty of each instrument, the first and last bins of FCDP and the first three bins of CIP were discarded and not used, which was similar to our previous studies (Yang et al., 2019; Zhao et al., 2019; Dong et al., 2020). The whole size range of cloud particles at 3–1,550 μm was obtained by combining the observations of FCDP and CIP. The time used in this study were all converted to local time (UTC + 8).

Weather background was analyzed using 1° × 1° NCEP/NCAR reanalysis data at 14:00 LT on November 29, 2019, and the flight seeding and observation areas were marked in Figure 1. There was a zonal circulation with two troughs and a ridge at 500 hPa (Figure 1A) at mid-high latitudes over Asia. There was also an obvious westerly trough near Hetao, and the detection area was dominated by a southwest airflow in front of the trough. The trough line at 700 hPa (Figure 1B) was located from the Mongolian Plateau to the Hetao area. The flight observation area was under the control of a southwest airflow in the south of the trough with obvious warm moisture advection, making the relative humidity close to saturation. The warm–humid air from the southwest and the cold air from the north converged, causing the formation of a large cloud system at the east of Hetao and the north of Shanxi. The cloud system gradually affected Hebei during its eastward movement. In this process, the atmosphere over the observation area was relatively stable, and the cloud system was dominated by stratus clouds. On the day of November 29, 2019, the relative humidity above the Xingtai site (Figure 1C) that lies within the observation area was relatively low at below 3,000 m, and the wet layer was mainly concentrated in 3,000–5,000 m, which was consistent with the detection results of the aircraft.

The aircraft took off from Zhengding Airport at 14:50 LT, lifted to 2,100 m, and then leveled over Xingtai, which was located in south-central Hebei Province. After arriving at the detection area, the aircraft first descended to a height of 600 m and then conducted vertical detection of the macro- and micro-properties of cloud by flying from the bottom to the top during the period 15:52–16:18 LT, with an altitude range of 600–7,000 m and a hovering radius of 10 km. After knowing the vertical structure of cloud properties, the aircraft was lowered from a height of 7,000–4,550 m from 16:19 to 16:34 LT. Then, cloud seeding was carried out over Xingtai along the same circular trajectory at 16:37 LT at an altitude of 4,550 m, where the supercooled liquid water content was relatively abundant, and a total of 14 AgI bars were lighted. After cloud seeding, the flight continued to detect the cloud macro- and micro-physical properties in order to know the effect of cloud seeding.

Figure 2 shows the vertical distribution of stratus cloud microphysical properties along with the temperature and relative humidity profiles. It can be seen from Figure 2A that good humidity conditions existed above the height of 2,500 m, which was consistent with the results at around 16:00, as shown in Figure 1C. The corresponding heights of 500 and 700 hpa are about 5,600 and 3,000 m, respectively, and the temperature values in Figure 2A are approximately the same as those in Figure 1A and Figure 1B. As shown in Figure 2A, the vertical detection area of the aircraft was all above the height of 0°C layer, and the temperature decreased with the increase of altitude. However, there were two thin temperature inversion layers near the cloud base and the cloud top, respectively. The temperature at the lower inversion layer varied from -8.2 to −5.4°C with heights from 2,100 m to 2,355 m, and the temperature at the upper inversion layer varied from -13.5 to −10.2°C with heights from 4,600 to 4,800 m. Unlike temperature, relative humidity gradually increased with height, reached saturation in the cloud region, and gradually decreased with height above the cloud top.

It was generally believed that when the number concentration obtained by FCDP (N _C__FCDP) was greater than 10 cm⁻³ and LWC ≥ 0.01 g·m⁻³ at the same time, the corresponding location of the aircraft was determined to be in the cloud (Hobbs, 1991). The dramatic changes of N _C__FCDP with height, as shown in Figure 2B, indicated that the vertical distribution of stratus cloud properties varied greatly. The maximum N _C__FCDP appeared in lower clouds at heights of about 3,220 m, with the peak value of 192 cm⁻³. The number concentration of ice crystals (N _C__CIP) showed a growing trend with height within clouds and reached the maximum value of 10 L⁻¹ at the top of the cloud, as shown in Figure 2C. The maximum LWC of 0.15 g/m³ appeared near the top of the cloud (Figure 2D), which was similar to the results found by Zhao and Lei (2014). IWC was on the same order of magnitude as LWC, with peak values of up to 0.1 g/m³ at heights of around 4,000 m. Figure 2 shows that both liquid droplets and ice crystals were detected at altitudes from 3,500 to 4,700 m, indicating the mixed-phase status at this height range, which was also the optimal part for cloud seeding due to the existence of supercooled liquid droplets.

After knowing the vertical properties of the cloud system, the flight leveled at the height of 4,550 m, which was rich of supercooled liquid water, to obtain the horizontal microphysical properties of the cloud. The corresponding flight time was 16:34-16:43 LT, and two laps were flown along the same trajectory due to the wind speed in the detection area being less than 5 m/s. We believe that the cloud properties do not change much in a short time. To make sure that the LWC obtained during this experiment was reliable, Figure 3 shows a comparison of LWC in the cloud obtained by two different instruments of FCDP and Nevzorov hot wire, which were denoted as LWC_FCDP and LWC_NEVZOROV, respectively. LWC_FCDP and LWC_NEVZOROV had a good consistency on time series (Figure 3A), and the correlation coefficient between them reached 0.81 (Figure 3B), indicating that the LWC detection results had high reliability and accuracy. Note that the LWC from Nevzorov hot wire is slightly smaller than that from FCDP, which might be associated with the potential ice crystals with sizes smaller than 100 μm. LWC_FCDP was used in the following sections to represent the liquid water content during the aircraft observation period.

Figure 4 presents the cloud droplet number concentration (N _C__FCDP; Figure 4A), the cloud droplet effective radius (r _e) measured by FCDP (Figure 4B), and the liquid water content (LWC_FCDP, Figure 4C) during the level flight process at a height of 4,550 m. The flight observation started in a counter-clockwise direction from point A. Figure 4 presents the observation results from 16:34 LT (the observation started from point A) to 16:39 LT (the flight returned to point A for the second time after one lap). The cloud seeding and detection were carried along the same circular trajectory. Even though the AgI flame bars were lighted at 16:37 LT, the aircraft still detected the unseeded area before the end of the lap. Differently from the general expectation that cloud microphysical properties vary weakly at a horizontal layer for stratus clouds (Wang et al., 2013), N _C__FCDP, r _e, and LWC_FCDP varied a lot horizontally. The N _C__FCDP ranged from 70 to 248.4 cm⁻³, corresponding to which the cloud droplet r _e ranged from 43.71 to 10.2 μm. In addition, the large amount of small cloud droplets implied that condensation, rather than collision, was likely the main way of droplet growth. By comparing the horizontal distribution of N _C__FCDP, r _e and LWC_FCDP along the flight track, it was clear that high N _C__FCDP corresponded to low r _e and LWC_FCDP and high LWC_FCDP corresponded to high r _e. This suggested that while small droplets had a great contribution to number concentration, LWC in clouds was mainly contributed by large droplets.

Not only cloud microphysical properties but also the cloud particle phase varied greatly along the flight track at the selected cloud layer. From 16:34 to 16:37 LT (A-B-C in Figure 4A), the concentration of cloud droplets was relatively high, while r _e and LWC in the cloud were relatively small. The cloud particle images obtained by CPI during this period are shown in Figure 5A. In this period, the particles are mostly spherical droplets with a small amount of ice crystals. The effective radius of the particles is relatively small, so the concentration of LWC_FCDP was low, although N _C__FCDP was at a high level. From 16:37-16:39 LT (C-D-A in Figure 4A), the number concentration of cloud droplets was relatively low, with an average of 152.24 cm⁻³. The cloud particle image (Figure 5B) showed that, at this time, cloud particles were mostly in ice phase. Large ice crystals coexist with small ice crystals in irregular shapes, which might come from the shattering of large crystals. The differences of microphysical properties in the clouds at different horizontal locations were clear, which reflect the serious heterogeneity of stratus cloud properties when they were in a mixed phase.

Figure 6 shows the temporal variation of cloud microphysical properties along with the flight height and environmental temperature. Note that the time periods shown in Figure 6 include those before and after cloud seeding. The flight made the observations of vertical profiles during the period from 15:52 to 16:18 LT at heights of 600–7,000 m. Then, the altitude of 4,550 m near cloud top, where LWC (supercooled droplets) was relatively abundant, was chosen for cloud seeding at 16:37 LT, with a total of 14 AgI bars lighted. The cloud seeding and detection were carried along the same circular trajectory. Even though the AgI flame bars were lighted at 16:37 LT, the aircraft still detected the unseeded area before the end of the lap. The aircraft returned to the position where the cloud seeding started at 16:43 LT, and the AgI flame bars got extinguished at 16:58 LT. Then, the aircraft started the observation of cloud microphysical properties after seeding along the same trajectory until 17:05 LT. The concentration of N _C__CIP and N _C__FCDP were 7.1 L⁻¹ and 160.3 cm⁻³, on average, after two circles of detection in horizontal flight, which showed no significant difference from that in the vertical detection stage. As presented from the particle images, the cloud droplets and ice crystals coexisted before cloud seeding (Figure 5A and Figure 5B). The concentration of N _C__FCDP (Figure 6B) decreased from 160.3 to 129 cm⁻³, on average, in about 7 to 8 min after seeding. When the plane arrived at the initial seeding position after a circle of flat flight, the concentration of N _C__FCDP reduced obviously to 88.94 cm⁻³ in minimum, and the horizontal spatial distribution became more uniform. LWC showed obvious regional differences before seeding, and these differences were no longer evident during the seeding process. After seeding (16:58–17:05 LT), LWC presented a significant decreasing trend, which indicated that the silver iodide ice nuclei were introduced into the clouds, and the supercooled water in the cloud was consumed through the Bergeron process. It can be seen from Figure 6C that the concentration of ice crystals at 4,550 m was distributed relatively uniformly before seeding, and the concentration fluctuates after seeding. The concentration of N _C__CIP increased slightly after seeding, and several peaks appeared (for example, N _C__CIP reached 52.2 L⁻¹ at 16:53:59 LT and 149.3 L⁻¹ at 17:00:52 LT). The concentration of ice crystals increased from 7.1 to 10 L⁻¹ on average. The maximum N _C__CIP reached 785.5 L⁻¹ especially in some specific regions. In other studies in central and southern Hebei Province, it was also found that the concentration of ice crystals in some areas increased to hundreds after seeding. Compared with the results of this study, the concentrations of ice crystals are of a similar magnitude (Dong et al., 2020; Dong et al., 2021). Moreover, Figure 6E shows that the proportion of LWC decreased significantly from 67 to 50% after seeding, indicating that part of the LWC was consumed. As a result, the temperature decreased to a certain extent, instead of increasing, due to latent heat released by condensation as we usually thought.

Figure 7 shows the cloud particle size distribution measured by FCDP and CIP in the region with abundant LWC before (16:34–16:43 LT) and after seeding (16:58–17:05 LT). Before seeding, the cloud showed a bimodal particle size distribution. The concentration of cloud particles showed a peak value of about 44.10 cm⁻³ in the size range of 4.5–6 μm (droplets) and another peak value of 0.195 cm⁻³ in the size range of 350–375 μm (ice particles). After seeding, the cloud showed a unimodal particle size distribution. The spectrum of small cloud particles measured by FCDP shifted to the right, and the peak value appeared in the range of 12–14 μm. Compared with cloud particle size distribution before seeding, the peak concentration was slightly decreased, which was about 29.28 cm⁻³. The concentration of small cloud particles (1.5–4.5 μm) on the left side of the peak decreased by about 1 order of magnitude after seeding, but the concentration on the right side of the peak (6–50 μm) increased by about 1 order of magnitude. The changes of particle size distribution of the ice crystals measured by CIP were more obvious. The peak value of ice crystal number concentration at 350–375 μm disappeared after seeding, and the concentration of ice crystals at 100–675 μm increased significantly by about 1–2 orders of magnitude. Particularly, the ice crystal concentration at 100–125 μm increased from 2.73 × 10⁻⁴ to 7.08 × 10⁻² cm⁻³. Moreover, cloud particle images showed that flaky or columnar ice crystals appeared after seeding. From the analysis above, it can be concluded that the aircraft cloud seeding in this study presented a remarkable effect on mixed-phase cloud microphysical properties.