A Comprehensive Observational Analysis for the Effects of Gas Cannons on Clouds and Precipitation

The use of human intervention to modify weather in a controlled direction is called artificial weather modification, which mainly includes artificial precipitation/snow enhancement, artificial cloud/fog elimination, and artificial hail prevention (Guo and Zheng ^[1]; Ma et al. ^[2]; Guo et al. ^[3]; Bruintjes ^[4]; Changnon et al. ^[5]; Yao ^[6]). In recent years, due to the needs of hail and drought prevention, air quality improvement, environmental protection, and weather conditions for major events (Ni et al. ^[7]; Zhao et al. ^[8]; Lin et al. ^[9]; Liao et al. ^[10]; Li et al. ^[11]), artificial weather modification technology has been improved greatly (Guo et al. ^[3]). With the rapid advancement of industrialization and urbanization, many problems have emerged along with economic prosperity, such as degradation of the ecological environment and serious atmospheric pollution (Cui ^[12]). Studies have shown that there is a negative correlation between the concentration of atmospheric pollutants and precipitation and that the erosion and removal effects of precipitation play an important role in reducing the concentration of atmospheric pollutants (Mcmullen et al. ^[13]; Ouyang et al. ^[14]; Bei et al. ^[15]). Therefore, it is of great significance to rationally utilize atmospheric cloud water resources and artificial precipitation enhancement to reduce wildfire risk levels, alleviate drought, and restore the ecological environment (Zahra et al. ^[16]; Gao and Huang ^[17]). Currently, rockets, antiaircraft artillery, and aircraft are the main tools for artificial precipitation enhancement (Guo and Zheng ^[1]). Although the maneuverability is high and the technology is mature, the safety precautions for aircraft or antiaircraft artillery are not yet perfect, and many problems remain for severe weather conditions and airspace management (Li et al. ^[18]; Luo et al. ^[19]). Ground burners are mainly used for catalyzing topographic clouds and are generally installed at high altitudes. As early as the 19th century, acetylene cannons were developed in Europe for agricultural production to prevent hail (Morgan and Griffith ^[20]). After 2000, the China Huayun Meteorological Science and Technology Group manufactured gas cannons similar to the European acetylene cannons (Xu ^[21]), providing a good solution for precipitation enhancement and the above problems. Gas cannons are different from traditional precipitation enhancement and hail prevention methods. The shock waves generated during gas explosion continuously disturb the airflow. Supplemented by catalysts, the technique is able to achieve the goal of artificial precipitation enhancement (Wu ^[22]). Compared with traditional precipitation enhancement methods, this method does not require an airspace application, has no risk of damage due to falling objects, and has no restrictions on the operating area. Currently, the gas cannon method is used in Shouxian County of Anhui Province, Hinggan League of Inner Mongolia, and Linfen of Shanxi Province. In general, the use of gas cannons is still in the exploratory stage in China, and field tests are particularly needed (Xu ^[21]; Wu ^[22]).

Does the precipitation enhancement operation increase surface precipitation? How can the investment benefits of gas cannons be increased? To answer these questions, the effects of artificial precipitation enhancement must be evaluated. Objective, scientific, and quantitative evaluation of the effects of precipitation enhancement is an important component of artificial weather modification research. Without evaluation, it is impossible to explain the effects of artificial precipitation enhancement (Guo et al. ^[23]; Huang ^[24]; Cui et al. ^[25]). However, due to the large rate of change in clouds and precipitation, effectively and accurately testing the effects of precipitation enhancement is difficult, and there are uncertainties in evaluation results. Currently, three main methods are used to test the effects of artificial precipitation enhancement: statistical testing, physical testing, and numerical model testing (Wang and Yao ^[26]; Qin et al. ^[27]; Zhou and Yao ^[28]; Xu and Yin ^[29]).

Many scholars in China have conducted studies on the effects of precipitation enhancement using rockets, antiaircraft artillery, and aircraft (Guo et al. ^[23]; Cui et al. ^[25]; Wu et al. ^[30]; Yang et al. ^[31]; Dong et al. ^[32]; Pokharel and Geerts ^[33]; Dong et al. ^[34]). Few studies have focused on evaluating the effects of gas cannonbased precipitation enhancement using dual-polarization radar and disdrometer data. In this study, based on seven precipitation enhancement tests using gas cannons in the northern plains of Zhejiang Province from 2019 to 2020, multisource data were used to analyze the effects of precipitation enhancement from multiple angles. The findings are expected to provide a reference for evaluating the effects of gas cannons on clouds and the precipitation enhancement effects.

Multisource observational data, including those from the National Centers for Environmental Prediction (NCEP) reanalysis, Hangzhou and Huzhou new-generation weather radars, laser disdrometer, ground-based automatic weather station, wind profiler radar and Lin'an C-band dual-polarization radar data, were adopted in this study. The distribution of the observation equipment is shown in Fig. 1. The dotted black box indicates the boundaries of the study area with a size of 60 km × 60 km, which is centered approximately 140 km from the Lin'an C-band dual-polarization radar. The distance between the Hangzhou and Huzhou radars is 63 km, and the magenta line connecting the two radars represents their baseline.

Figure 1.  The distribution of the Hangzhou and Huzhou S-band Doppler radars, Huzhou wind profiler radar (black triangle), Lin'an C-band dual-polarization radar (red star), laser disdrometer (magenta rectangle), gas cannon (red dot) and the 3D wind field retrieval area (black dotted box).

Based on the variational dual-Doppler radar three-dimensional (3D) wind field retrieval of the Hangzhou and Huzhou radars and the polarimetric variables obtained by the Lin'an dual-polarization radar, we analyzed the microphysical processes and the variations in the macro - and microphysical quantities in clouds from the perspective of the synoptic background before precipitation enhancement, the polarization echo characteristics before, during and after enhancement, and the evolution of the fine 3D dynamic structure and the microphysical structure. Prior to the retrieval of the three-dimensional wind field from the dual-radar data, an automated dealiasing algorithm (Zhang and Wang ^[35]) was first employed on radial velocity, followed by 1.0-km horizontal and 0.5-km vertical interpolation of each measured variable to a Cartesian grid. The wind field within an area with a size of 60 km × 60 km × 20 km (as indicated by the dotted box in Fig. 1) was retrieved from the radar data using the variational dual-Doppler wind analysis algorithm (Shapiro et al. ^[36]; Potvin et al. ^[37]). In the evaluation of precipitation enhancement, the radar plane-position indicator (PPI) product is commonly used to analyze the horizontal distribution characteristics of the radar echo near the operating point. To understand the vertical structure characteristics and variations in the radar echo, the radar needs to perform vertical scans in the specified direction and obtain the range-height indicator (RHI) product. Because current radar devices usually adopt the VCP21 volume scan mode, the vertical profile products (RCS, and VCS) can only be obtained through vertical interpolation of the volume scan data (9 elevations) (Zhou et al. ^[38]). The limitation of this approach is the low spatial resolution in the vertical direction, which is not conducive to the analysis of the fine vertical characteristics of the echo. Therefore, in this study, the vertical structure and evolution characteristics of the radar echo near the operating point were analyzed based on the radar quasi-vertical profile (QVP) method (Ryzhkov et al. ^[39]; Kumjian and Lombardo ^[40]), and then the precipitation enhancement effect of the gas cannon was evaluated.

Specifically, the QVP method uses the observational data of a specified elevation angle (> 10°) in the radar volume scan data at a certain time. Based on the average radar echo variables (reflectivity, radial velocity, or polarimetric variables) of all azimuth angles on different distance circles and the radar height measurement formula, the "radar echo variable-height" profile at that time was obtained. Then, the profile of all times in a time series was obtained and plotted to obtain the QVP. Taking the 19.5° elevation observational data in the VCP21 scanning mode as an example (Fig. 2) and without considering the radar antenna height, earth curvature, and atmospheric refraction, the QVP product corresponding to the 3-km height at a certain time is the average values of the radar echo variables on the 16.9-km distance circle. Similarly, the QVP product corresponding to the 10-km height represents the average value of the radar echo variables on the 56.5 km distance circle. Due to the influence of radar beam broadening and the QVP principle, the echo resolution of the low-level QVP is higher and the sampling range (distance circle diameter) is smaller, which can represent the fine structure characteristics of local precipitation. The high-level echoes represent a wider range of precipitation structure characteristics, but due to the expansion of the sampling volume and sampling range, the representativeness of the results is reduced to a certain extent after averaging.

Figure 2.  Example of the principle of the QVP method.

The HY-R gas cannon-based precipitation enhancement device is a new type of equipment that uses shock waves, sound waves, and catalyst technology to interfere with and catalyze local weather. The tank is filled with gas, catalyst, and compressed air in a certain proportion. Igniting the mixed gas in the tank produces deflagration. The deflagration flame is accelerated when passing through the flame duct, forming a shock wave on the front of or around the flame. The detonation wave is modulated in the shock wave generator to strengthen its pressure front. Lastly, the modulated shock wave and catalyst are emitted from the gas cannon barrel. The equipment emits shock waves, strong sound waves and catalysts into the air every 6-10 s (adjustable), thereby continuously modifying the target area by interfering with the flow of air and increasing the number of condensation nuclei. Ultimately, this device can achieve the goal of precipitation enhancement (Xu ^[21]; Wu ^[22]). A total of seven field tests in the northern plains of Zhejiang Province from 2019 to 2020 were selected in this study for analysis. The specific operation information is shown in Table 1.

On May 15, 2019, the immense cold vortex system on the west side of Lake Baikal continued to split into small vortices, driving cold air down the northwest path to the south (Fig. 3a). High and low westerly jets transported water vapor on the periphery of the subtropical high south of 30° N. The maximum water vapor flux of the 850 hPa jet core was 22 g cm^-1 hPa^-1 s^-1. The northern part of Zhejiang Province was located at the front end of the 850 hPa jet and on the south side of the warm shear line (Fig. 3b). In conjunction with the high-altitude weak cold air, these conditions were conducive to precipitation.

Figure 3.  Wind field, water vapor flux and water vapor flux divergence at 14:00 BT on May 15, 2019 at (a) 500 hPa and (b) 850 hPa.

On July 27, 2019, the subtropical high was strong. The 588-dagpm contour extended to 105° E westward, and the northern part of Zhejiang Province was located near the 588-dagpm contour. The high-value water vapor flux zone was in the southwest-northeast direction, mainly in areas from Hubei to Henan. The northern part of Zhejiang Province was located on the inner side of the high-pressure edge. The 700-hPa water vapor flux reached 6 g cm^-1 hPa^-1 s^-1, and there was convergence of the wind speed at 850 hPa, indicating good dynamic conditions (graph not included). On August 4, 2019, the northern part of Zhejiang Province experienced an easterly airflow on the south side of the subtropical high, and there was continuous water vapor from the Northwest Pacific. The 700-hPa water vapor flux reached 6 g cm^-1 hPa^-1 s^-1, and the northern part of Zhejiang was located behind the easterly trough. The dynamic uplift promoted the condensation of water vapor, which was conducive to precipitation. At 8:00 on August 28, 2020 (Beijing Time (BT), applied to all time points hereinafter), the northern part of Zhejiang was between the continental high and the subtropical high. There was a low vortex between the two highs at 700 hPa. There was a southwest monsoon on the west side of the subtropical high, which was conducive to the transportation of water vapor from the South China Sea. There was a water vapor flux convergence center at 850 hPa over the Hangzhou-Jiaxing-Huzhou area, and the water vapor flux divergence in the eastern part of Huzhou reached - 4 × 10^-8 kg cm^-2 hPa^-1 s^-1, the low vortex further developed at 14:00. The northern part of Zhejiang was in the inverted trough on the north side of the low vortex. The southeast wind in front of the trough strengthened, and the convergent center of water vapor flux divergence developed. At 14: 00 on September 15, 2020, the northern part of Zhejiang was located on the north side of the subtropical high. The high-altitude cold vortex moved eastward and guided the cold air to flow southward and meet the warm and humid air at the edge of the subtropical high, which was conducive to the occurrence of convective precipitation. The 700-hPa jet exit was over the northern part of Zhejiang, forming a covergence center over the Jiaxing-Huzhou area, with strong vertical uplift movement. Meanwhile, the water vapor flux reached 20 g cm^-1 hPa^-1 s^-1. In summary, the synoptic condition was conducive to the occurrence of precipitation, enabling the possibility of artificial precipitation enhancement.

The evaluation of artificial precipitation enhancement is a scientific challenge (Wang and Yao ^[26]; Zhou and Yao ^[28]). Statistical methods are often used to evaluate precipitation enhancement in China. The evaluation is generally based on comparisons of catalyzed and noncatalyzed clouds and comparisons of the affected and unaffected areas during the operation (Zeng et al. ^[41]). In this paper, the cells near the operation site were selected as the target clouds. In the windward and crosswind sides of the surrounding area with similar terrain, the cells with consistent cloud type, intensity and echo evolution trend before operation were selected as the contrast clouds. Through the comparison of the two types of clouds, the effects of gas cannons on cloud precipitation were analyzed. Based on the composite reflectivity (CR) product of the Huzhou radar, seven groups of targets and contrast clouds were selected (Fig. 4). Given the small influence area of precipitation enhancement operations by gas cannons, except for the affected area (red box in Fig. 4) and contrast area (black box in Fig. 4) in the 4th and 5th operations, which were 49 km² (7 km×7 km), the other operations encompassed an area of 100 km² (10 km×10 km).

Figure 4.  The target and contrast clouds of the seven precipitation enhancement operations in Nanxun District, Huzhou City from 2019 to 2020 (a1-a3, b1-b3, ... and g1-g3 are the evolutions of CR products before, during and after the operations, respectively). The red boxes represent the target clouds, and the black ones represent the contrast clouds.

Figure 5 shows the time series of reflectivity vertical profile of the target and contrast clouds before, during and after the operations. In the 3rd operation, the target and contrast clouds included all precipitation echoes from cumulo-stratus clouds. The maximum reflectivities before the operation were between 35 and 50 dBZ, and the height of the strong echoes remained around 4 km. In the 4th operation, both the target and contrast clouds included precipitation echoes from stratocumulus clouds, and the maximum reflectivities before the operation were from 25 to 40 dBZ. The other target and contrast clouds were stratus clouds and the associated precipitation, and the echo intensities were all weak before the operation. In general, all target and contrast clouds were comparable because their echo types, intensities, and evolution characteristics were similar before the cannon operation.

Figure 5.  The time series of reflectivity vertical profiles of the target and contrast clouds for the seven precipitation enhancement operations in Nanxun District, Huzhou City from 2019 to 2020 (a1-a7 and b1-b7 represent the target and contrast clouds, respectively). The red and black dash-dotted lines represent the start and end times of gas cannon operations, respectively.

The evolution characteristics of the horizontal (Fig. 4) and vertical (Fig. 5) structures of the reflectivity for the target and contrast clouds show that for the echo intensity, strong echo height, and development trend before operation were consistent, and the echo intensities (especially near the ground) and height of the target clouds were generally higher than for the contrast clouds. Moreover, the echo maintenance times were longer than that of the contrast clouds. However, the intensity and height of the two clouds were close or the target clouds were larger than the contrast clouds after operation.

Figure 6 shows the variation in the CR of the target and contrast clouds during the seven precipitation enhancement operations from 2019 to 2020. Before the 1st and 2nd operations, the echo intensities of the target and contrast clouds were similar and all showed a trend of strengthening first and then weakening. During the operations, the echo intensities of the target clouds in the 1st operation increased rapidly to 38 dBZ, and those in the 2nd operation fluctuated significantly and increased gradually to 32 dBZ. However, the echo intensities of the contrast clouds in the two operations all exhibited little change, and the maximum echo intensities were 20 and 16 dBZ, respectively, much smaller than that of the target clouds. After the operations, the intensities of the target clouds all decreased significantly. Although the intensities of the contrast clouds were maintained, they were still lower than that of the target clouds. Before the 3rd, 4th and 6th operations, the echo intensities and development trends of the target and contrast clouds were almost the same. During the operations, the intensities of both clouds decreased, while those of the target clouds were stable and higher than those of the contrast clouds, with a high range of 0-25 dBZ, 3-21 dBZ and 1-9 dBZ, respectively, and the reduction rate of the echo intensity in the 3rd operation was 63 dBZ h^-1, less than half of that of contrast cloud. After the operations, both of the intensities did not change much, but the intensities of the target clouds in the 4th operation remained much higher than the contrast clouds. The target and contrast clouds in the 5th operation were stable weak stratus clouds. From the start to the end of the operation, the echo intensities of the target clouds rose slowly, while the contrast clouds remained or weakened. In the 7th operation, there was little difference in the intensities and variation trends between target and contrast clouds before, during and after operation, but the intensities of the target clouds remained higher than for the contrast clouds during operation, and the increase (14 dBZ) and peak value (34 dBZ) of the CR were both larger than that of the contrast clouds (10 dBZ and 28 dBZ).

Figure 6.  Variation in the CR of the target and contrast clouds during the seven precipitation enhancement operations from 2019 to 2020 (a_i represents the i-th operation). The dashed and dash-dotted lines represent the start and end times of gas cannon operations, respectively.

Table 2 shows the changes in the CR of the target and contrast clouds during the seven operations. The echo intensity of the target clouds reached a peak 3-82 min after the start, and the CR increased in the range of 2-35 dBZ. For the contrast clouds, the intensity decreased or increased slightly, with a range of - 6-8 dBZ. It can be seen that the gas cannon operation increased the CR and promoted it to reach an earlier peak.

Taking the four operations in 2019-2020 as an example, based on the QVP characteristics of the reflectivity of the gas cannon operation area, the height of the cloud top gradually increased, and the height of the cloud bottom gradually decreased after the operation started on the morning of August 28, 2020. The echo intensity gradually increased (Fig. 7a ①). After the operation started in the afternoon, the height of the cloud top increased, and the echo intensity increased significantly from 16: 00 to 17: 00 BT (Fig. 7a ②). On September 15, 2020, the gas cannon operation area was dominated by stratiform clouds and rainfall. In the vertical profile of the reflectivity, there was an obvious bright band, and the height was at 4.8-5 km. After the start of the operation, the echo intensity increased significantly, the echo top (ET) increased, and the height of the echo bottom decreased. This process continued until approximately 17:00 BT (Fig. 7b③).

Figure 7.  QVP product of reflectivity and radial velocity retrieved from the Huzhou radar of the operation area (the operation period is between the two dotted lines). (a) Reflectivity QVP during 9:01-18:26 BT on August 28, 2020, (b) reflectivity QVP during 13:52-18:45 BT on September 15, 2020, (c) reflectivity QVP during 11:41-18:52 BT on May 15, 2019, and (d) as in (c), but for the radial velocity QVP (positive velocities indicate updrafts, and negative velocities indicate downdrafts).

Figures 7c and 7d show the QVP product of the reflectivity and radar radial velocity during 11:41-18:52 BT on May 15, 2019, respectively. After the operation started, the height of the cloud tops also increased gradually, and the height of the cloud bottoms decreased gradually, with the echo intensities increasing gradually (Fig. 7c④). During the operation, due to the disturbance of the shock waves and sound waves, the radial velocity changed from negative to positive (that is, from downdraft to updraft), which provided favorable dynamic conditions for the formation of precipitation (Fig. 7d ④). After the operation was stopped, the precipitation increased significantly, and as the precipitation fell, there was an obvious negative velocity.

On May 15, 2019, the Hangzhou and Huzhou new-generation Doppler weather radars were used jointly to perform dual-Doppler radar 3D wind field retrieval, and the retrieval results were compared with the Huzhou wind profile radar data to verify the rationality of this method (Xue et al. ^[42]). Fig. 8a and 8b show the horizontal wind time series at different heights above the wind profiler radar detected by the Huzhou wind profile radar (location shown in Fig. 1) and retrieved from the dual-Doppler radar, respectively. The time period in the red box of Fig. 8a is consistent with that in Fig. 8b, and the time intervals in the horizontal axis in Figs. 8a and 8b are both 6 min. The comparative analysis shows that the wind direction and speed of the two results are very close at heights of 1.5-5 km, but the wind speed of the dual-Doppler radar retrieval at heights of 5-6 km is slightly larger than the wind speed observed by the wind profile radar. In general, the horizontal wind fields and changing trends detected by the two methods are basically the same.

Figure 8.  Time series of the horizontal wind fields at different heights detected by the Huzhou wind profile radar and retrieved from the dual-Doppler radar on May 15, 2019. (a) is the wind field detected by the Huzhou wind profile radar during 16:00-20:00 BT, and (b) is the wind field retrieved from the dual-Doppler radar during 17:28-17:58 BT. The time period in the red box in Fig. 8a is consistent with that in Fig. 8b.

Based on the retrieval results of the 3D wind field, the structure of the 3D fine wind field at meso - and microscales were analyzed, and the evolution of the wind field before and after precipitation enhancement using the gas cannons was studied, as shown in Fig. 9. Before the enhancement, the radar echoes were weak, and the horizontal wind field was small (Fig. 9a); after the operation, the radar echo intensity obviously increased, and the horizontal wind field increased as well (Fig. 9b). The vertical structure above the precipitation enhancement area was then compared before (Fig. 9a1 and 9a2) and after the operation (Fig. 9b1 and 9b2). After the operation, the vertical velocity increased, the echo intensity increased, and the echo range expanded. Moreover, the ET increased from 5-6 km to 9-10 km, which was nearly twice as high as that before the operation.

Figure 9.  3D wind fields and radar echo structures of the precipitation area at 12:47 BT and 17:53 BT on May 15, 2019. (a) and (b) are the wind fields and radar echo structures at 12: 47 BT and 17: 53 BT at a height of 2 km, respectively; (a1) and (a2) are the vertical sections along the red and blue dashed lines in (a), respectively, and (b1) and (b2) are the vertical sections along the red and blue dashed lines in (b), respectively. The pink dots are the operation sites.

Then, the role of changes in the vertical velocity and water vapor content in the process of precipitation enhancement was explored. The precipitation formation mechanism was analyzed. The growth of raindrops generally depends on the height and thickness of a cloud, which are affected by meteorological conditions such as water vapor and vertical movement. The calculation equations of the precipitation rate I and cumulative precipitation W are given as follows (Shou ^[43]):

The precipitation rate I is proportional to the vertical velocity w and the water vapor condition δF, and the cumulative precipitation W is proportional to the vertical velocity w, the water vapor condition δF, and the lasting time t. Based on the above results, the vertical velocity w increased, and the echo intensity and height increased after precipitation enhancement, indicating an increase in the water vapor content. Therefore, the precipitation rate I increased after the precipitation enhancement operation. Moreover, the precipitation enhancement extended the life cycle of the echoes (that is, the lasting time t increased), so the cumulative precipitation W increased as well. As a result, the operation achieved precipitation enhancement.

From the precipitation time series of the automatic weather stations and the radar radial velocity of 0.5° elevation angle evolution near the operation area (Fig. 10), precipitation began to appear at 15: 25 BT in Dahongqiao and Shanlian. Local precipitation led to obvious horizontal temperature gradients with opposite directions in the boundary layer. The thermal forcing effect caused the acceleration of the boundary layer airflow (red dashed box in Fig. 10), which in turn strengthened the convergence of the wind speeds in the front of the jet stream, causing the intensity of local precipitation to be further maintained and enhanced and thus forming positive feedback.

Figure 10.  The precipitation time series of the automatic weather stations and radar radial velocity at 0.5° elevation angle evolution near the operation area from 15:10-17:55 BT on May 15, 2019.

Lin'an C-band dual-polarization radar provides many new polarimetric variables (e. g., differential reflectivity Z_DR, specific differential phase shift K_DP, and copolar correlation coefficient ρ_hv) related to the microphysical structure of precipitation (Bringi et al. ^[44]; Liu et al. ^[45]; Wu et al. ^[46]; Ryzhkov et al. ^[47]). Based on the 0° elevation PPI echo before (Fig. 11a1-a4) and after (Fig. 11b1-b4) precipitation enhancement, the area and intensity of the dual-polarization echoes in the operation area (yellow box) increased, indicating that weak surface precipitation appeared and the precipitation area expanded. Based on the evolution of the vertical structure of the dual polarimetric variables (white circles) in the operation area, the thickness of the clouds increased, and the values of the polarimetric variables increased. At a height of 4-7.5 km, the reflectivity Z_H, the Z_DR and the K_DP increased, and the ρ_hv decreased, indicating that the diameter of the precipitation particles and the concentration of raindrops increased after precipitation enhancement (Ryzhkov et al. ^[47]; Zhang ^[48]). Furthermore, precipitation particles were lifted to the height corresponding to 0 to -20℃ by the vertical updrafts; due to the presence of ice crystals, snowflakes, and raindrops, the ρ_hv value dropped (Wu et al. ^[46]; Ryzhkov et al. ^[47]; Snyder et al. ^[49]).

Figure 11.  Variation in polarimetric variables at 12:46 and 17:50 BT on May 15, 2019 (a1-a4 and b1-b4 show the Z_H, Z_DR, ρ_hv, and K_DP at 12:46 BT before the operation and at 17:50 BT after the operation, respectively).

At the same time, Table 3 shows the precipitation data every 5 minutes from the five representative automatic stations with the corresponding time before and after the operation in the operation area. It was found that there was no precipitation at the representative stations before the operation, and a trace amount of precipitation of 0.1mm 5min^-1 occurred at the four stations after the operation, which reflected the positive feedback effect of precipitation enhancement operation.

Based on ground-based disdrometer data, the time series of the disdrometer parameters before, during and after precipitation enhancement were analyzed. The mass distribution is defined as the water mass in a unit volume per unit drop size bin (Zhang ^[48]): $ m(D)=\frac{\pi}{6} \rho_{\omega} D^{3} N(D)$, and N_w is a physical quantity reflecting the concentration of raindrops. The disdrometer data (Fig. 12a) show that the density of raindrops was low before precipitation enhancement, with a maximum raindrop diameter (D_max) of approximately 1.0 mm. The peak of the density spectrum was at a raindrop diameter of 0.5 mm during precipitation enhancement, suggesting that the concentration of small raindrops was large at this time, the mass-weighted diameter D_m value dropped to 0.5 mm, and the peak of the mass distribution (Fig. 12b) was also at a small raindrop diameter (Beard et al. ^[50]). Moreover, there was a significant increase in lgN_w (Fig. 12d). After the rain enhancement operation stopped at 15: 33 BT, the number density of raindrops and the lgN_w value decreased, and the mass-weighted diameter D_m increased slightly. At 15: 55 BT, the raindrop concentration was similar to that during the operation, yet the mass-weighted diameter D_m increased to 1.5 mm, the peak of the mass distribution was at the small to medium raindrop diameter, and the lgN_w value was lower than that during the precipitation enhancement operation, corresponding to a significant increase in the precipitation intensity (Fig. 12c). The above observation results are consistent with the dual-polarization radar results (Zhang ^[48]; Ding et al. ^[51]).

Figure 12.  Disdrometer time series from 12: 00 to 18: 00 BT on May 15, 2019 (red dashed box indicates the precipitation enhancement period). (a) Drop size distribution (filled color indicates number concentration, and black solid line indicates D_m), (b) raindrop mass distribution (filled color represents the water mass in a unit volume per unit drop size bin), (c) rain intensity, and (d) lgN_w.

Automatic weather stations can provide precipitation data every minute. Because the precipitation intensity was not very high during the seven operations and minute precipitation data were dense but indicated no precipitation, we selected the precipitation data of the automatic weather stations every 5 min for comparative analysis (Table 4). The echoes in the 2nd operation were very weak. Although the echo characteristics were slightly enhanced after precipitation enhancement, the precipitation was generally low and localized, and the nearby automatic weather stations did not detect it. For the 3rd and 5th operation, the maximum precipitation intensity of the target clouds all reached 0.6 mm 5 min^-1 and lasted for 40 min. For the contrast clouds, however, there was no precipitation. For the other operations, the maximum precipitation intensity of the target clouds was significantly higher than that of the contrast clouds (2, 3 or 6 times). In addition, the precipitation duration of the target clouds was 1.3 to 9 times that of the contrast clouds.

For the seven gas cannon-based precipitation enhancement operations between 2019 and 2020, multisource observational data, including those from NCEP reanalysis, dual-polarization radar, wind profiler radar, disdrometer, and ground-based automatic weather station, were adopted to analyze the weather background of the precipitation enhancement experiments using gas cannons, the operation time, the kinematics structure and the microphysical processes in the cloud before, during and after the operation, and the precipitation at the ground-based automatic weather stations. The main conclusions are as follows:

(1) The weather conditions were conducive to water vapor condensation on each day of operation, allowing for artificial precipitation enhancement.

(2) According to the echo characteristics and vertical profile changes of the target and contrast clouds, during the operation, the CR and height of the target clouds all increased, and the increases were greater than those of the contrast clouds.

(3) The QVP of the reflectivity and radial velocity of the operation area showed that during operation, the height of the cloud top gradually increased, the height of the cloud bottom gradually decreased, and the echo intensity gradually increased. In the operation area, the airflow changed from downdraft to updraft. After the operation, the precipitation increased significantly, and there was an obvious downdraft.

(4) Using dual-Doppler radar 3D wind field retrieval, the evolution of meso- and microscale 3D fine wind fields in the operation area was analyzed. The results show that the radar echoes were significantly enhanced, the echo range was increased, and the ET increased after the operation. Moreover, the horizontal and vertical wind speeds increased.

(5) The structural evolution of the polarimetric variables revealed that the diameter of precipitation particles and the concentration of raindrops increased after the operation.

(6) Before the operation, the raindrop density was low. During the operation, the concentration of small raindrops and lgN_w value increased significantly, and the D_m value dropped. It can be seen that the small raindrops that could not fall before the operation fell during the precipitation enhancement operation. This resulted in an increase in the concentration of small raindrops and a decrease in the D_m value, indicating that the operation played a certain "lubricating" effect. After the operation, the concentration of raindrops did not change much compared with that during the precipitation enhancement operation, yet the mass-weighted diameter D_m increased, corresponding to an obvious increase in the surface precipitation intensity.

In summary, for weather conditions that are conducive to water vapor condensation, the gas cannonbased precipitation enhancement method has a positive effect on the local evolution of cloud macroscopic and microphysical structures. Further studies will be carried out to understand its mechanism.