Idealized numerical simulation experiment of ice seeding in convective clouds using a bin microphysics scheme

Jiefan Yang ,Hengchi Lei

Key Laboratory of Cloud Physics and Severe Storms, Institute of Atmospheric Physics, Beijing, China

Keywords:Cloud seeding Bin model Idealized numerical simulation

ABsTRACT A 2D axisymmetric bin model is used to conduct idealized numerical experiments of cloud seeding.The simulations are performed for two clouds that differ in their initial wind shear.Results show that,although cloud seeding with an ice concentration of 1000 L-1 in a regime that has relatively high supercooled liquid water can obtain a positive effect,the rainfall enhancement seems more pronounced when the cloud develops in a wind shear environment.In no-shear environment,the change in the microphysical thermodynamic field after seeding shows that,although more graupel is produced via riming and this can increase the surface rainfall intensity,the larger drag force and cooling of melting graupel is unfavorable for the development of cloud.On the contrary,when the cloud develops in a wind shear environment,since the main downdraft is behind the direction of movement of the cloud,its negative effect on precipitation is much weaker.

1.Introduction

The distribution of water resources in China is extremely uneven,and fresh water is one of the most in-demand natural resources given the huge human population.Besides,over 42% of China’s land area is arid or semi-arid,including northwestern,central and western North China,amongst other regions,where there is less precipitation and therefore water shortages.Thus,in recent decades,many operational cloudseeding programs and field campaigns have been carried out (Zeng et al.,1991 ;Hong and Zhou,2006),which is a method being pursued to enhance precipitation in many locations.

Although high-resolution cloud models with bulk microphysics schemes can and have been utilized to manipulate cloud seeding progress under the same meteorological conditions to assess the method’s potential effect,considerable uncertainties in the results remain.It is believed that the seeding parameterization implemented in bulk microphysics schemes has inherent limitations (Xue et al.,2013)owing to the use of assumed hydrometeor size distributions and particle evolution only being tracked in bulk.In contrast,the bin microphysical model is able to simulate this evolution explicitly by tracking individual size categories of particles,and provides explicit simulation of the evolution of individual hydrometeor species.By using this advanced type of microphysical scheme,researchers can examine the interactions between aerosol or silver iodide particles and convective clouds.For instance,Cui et al.(2011) used a 2D axisymmetric,bin-resolved cloud model to examine the impact of aerosol changes on the development of mixed-phase convective clouds and concluded that the accumulated precipitation responds very differently to changing aerosol in marine and continental environments.Xue et al.(2013) evaluated the possible impact of aerosol solubility and regeneration on warm-phase orographic clouds and precipitation using a detailed bin microphysical scheme coupled with a mesoscale model.Recently,the University of Pecs and NCAR bin microphysical scheme was implemented into the mesoscale Weather Research and Forecasting model to study the impact of silver iodide on precipitation formation in winter orographic clouds.

To analyze in depth the dynamic and microphysical effects produced by cloud seeding,this paper adopts a 2D axisymmetric convective cloud model with a detailed bin microphysical scheme (Cui and Carslaw,2006 ;Cui et al.,2006).In this scheme,all particle spectra change completely according to the stochastic collision equation.Moreover,during the simulation,ice crystals and liquid droplets with different masses (or sizes)grow independently.The sounding profile and warm bubble disturbance technology are used to drive the development of the convective cloud model.The main purpose of this paper is to analyze the impact of cloud seeding on a typical convective cloud under conditions with and without wind shear through an idealized numerical sensitivity test.We hope this paper will be helpful in evaluating the effects of cloud seeding.

Fig.1.The initial profile and the cloud microphysical properties at 40 min: (a) initial profile,in which the red and green solid lines represent the dew point and temperature,respectively;(b) wind speed at different levels;(c) microphysical properties without vertical wind shear;(d) microphysical properties with wind shear.The blue line in (c,d) represents the specific content of liquid water (units: g m -3);the red line is the water content of ice crystals and snow (units: g m -3);and the black line represents the water content of graupel.For the area surrounded by the thick blue line,the inner ring represents the area with LWC ＞ 0.5 g m -3,and the outer ring represents the area with LWC ＞ 0.1 g m -3.

2.Model configuration

The microphysical processes and the dynamic structure of the 2D slab-symmetric non-hydrostatic cumulus cloud model used in this study were first described in Reisin et al.(1996a) and Yin et al.(2000).Four hydrometeor types —liquid-phase droplets,pristine ice particles,graupel,and snowflakes (aggregates) —are divided into 34 size (mass) bins.Meanwhile,the evolution of size (mass) distributions during numerical simulation is carried out by solving the stochastic function using an efficient two-moment algorithm as proposed in Tzivion et al.(1987).The grid size of the model is set to 300 m in both the horizontal and vertical direction,while the width and height of the domain are 30 and 12 km,respectively.

The initial thermodynamic profile based on observations from Xingtai City,Hebei Province at 0800 Beijing Time (BT,UTC+8) 29 June 2020 is used (Fig.1 (a)).A series of model runs are conducted to examine the impact of cloud seeding on the development of convective clouds in two distinct environments —with and without wind shear.For the cloud without wind shear,we artificially set the vertical wind field shear to zero to manipulate the no-shear condition.The second type of vertical wind shear field refers to the measured sounding data (Fig.1 (b)),but due to the limitation of the model area,the vertical wind shear is reduced in a certain proportion to maintain the wind field characteristics and the computational stability of the model simultaneously.For each set of numerical experiments,a control (Ctl) case without seeding and a seeding case are simulated for 100 min.To examine the sensitivity to seeding amount,simulations with different seeding ice number concentrations (100—2000 L-1) are conducted.

The existence of supercooled liquid drops is crucial for efficient cloud seeding.Both numerical results and observations show that seeding can increase the number concentrations of ice crystals and aggregates,but this enhancement does not necessarily increase surface precipitation because of the complexity of microphysical processes (Tessendorf et al.,2019).Thus,specific regimes (-4°C and -15°C) with high and low liquid water content (LWC) are seeded to assess the potential effects of LWC on surface precipitation.Besides,the seeding time is also a critical parameter for obtaining positive effects (Reisin et al.,1996b).In this work,40 min after model initiation is considered to be suitable for cloud seeding,since at this time the supercooled LWC reaches its maximum and begins to be consumed by natural ice.The cloud seeding experiments are carried out in this study using a relatively simple method —that is,a specific number of ice crystals are directly added to the supercooled regime.All the diameters of seeding ice crystals are assumed to be the initial diameter of ice particles produced by nucleation,and the corresponding masses and concentrations are added to the first size bin.

3.Results

3.1.Microphysical and dynamic features of clouds

Fig.2.Comparison of changes in microphysical properties between seeding and no-seeding cases: (a,b) 45 min and 50 min profiles of microphysical quantities(units: g m -3) without seeding;(c,d) profiles of microphysical properties at 45 min and 50 min using 1000 L -1 of seeding;(e,f) profiles of microphysical properties at 45 min and 50 min after 2000 L -1 of seeding.The physical meaning is the same as in Fig.1.

Fig.1 (c) is a cross-section of microphysical properties in convective clouds after 40 min of simulation under no-shear conditions.It demonstrates that the peak value of LWC is located in the center region with the most intense vertical velocity,where the maximum LWC can exceed 1.5 g m-3.The supercooled water content above 3500 m shown in Fig.1 (c)is also relatively abundant (see the area surrounded by the thick blue line in the figure).In this area,the thickness of the regime with LWC＞0.5 g m-3extends to 2 km.It can also be seen from the figure that ice and snow crystals are mainly distributed in the middle and upper layers of the cloud,indicating that the phase transformation in the cloud is not sufficient at this time.A part of the raindrops below the 0°C isotherm layer begin to reach the ground,but the value is relatively small (～0.001 g m-3).

The cloud under the wind shear condition has an obvious trend of inclined development (Fig.1 (d)),while the regime of high supercooled water content is mainly on the back side of the cloud’s direction of movement.Compared to the middle of the cloud with abundant LWC,the cloud top may be inappropriate for cloud seeding because it is mainly composed of natural ice-phase particles.It also indicates that the regime with abundant LWC in this cloud is generally larger than that in the no-shear environment (Fig.1 (d)).For example,in the no-shear environment,the maximum horizontal width of LWC＞0.5 g m-3is about 2 km (Fig.1 (c)),while in the wind shear environment it can reach 4 km(Fig.1 (d)) and the height of LWC＞0.5 g m-3can reach 7 km,which is about 1 km higher than in the no-shear environment.

Several comparative experiments (as shown in Table 1) are conducted to assess the seeding effect of convective clouds.For a large number of the simulations carried out in this study,generally,cloud seeding at the central axis of the cloud in the early stage of precipitation formation (40 min after model initiation),with ice crystals of more than 1000 L-1 in the -4°C area (horizontal range: 1 km;vertical height:300 m) and lasting for 4 min,is able to achieve an evident seeding effect(Table 1).For the unseeded case,the area of LWC＞0.5 g m-3(Fig.2 (a,b)) can maintain for about 10 min,indicating it takes time to consume the supercooled water through riming and deposition of natural ice particles.It also demonstrates that the regime with high supercooled liquid water (＞0.5 g m-3) is still located in the middle of the cloud after 45 min of simulation (Fig.2).Even after 50 min of simulation,although a considerable part of the supercooled water in the middle of the cloud is consumed,there are still areas with LWC＞0.1 g m-3,and the LWC in a small part of the area is still greater than 0.5 g m-3.This shows that,although the natural cloud process can also consume supercooled water in the cloud,the speed is relatively slow.

Fig.3.Impact of seeding on the microphysical processes playing a role in graupel formation and vertical wind evolution in the no-shear environment.The plots represent the difference between the seeding and Ctl cases: (a) maximum downdraft;(b) domain-integrated melting of graupel;(c) domain-integrated riming of ice particles;(d) maximum updraft.

Table 1 Seeding parameters and the difference in total rain with respect to the unseeded cloud for IN (Ice Nuclei) seeding under shear and no-shear conditions.Ctl and Ctl_shear denote control cases (without cloud seeding) for no-shear and wind shear conditions,respectively.

Fig.2 (c—f) shows the distribution of microphysical properties after 5 and 10 min of simulation when seeded with 1000 L-1and 2000 L-1of ice crystals at -4°C.After the cloud seeding with 1000 L-1of ice crystals(Fig.2 (c—d)),the supercooled liquid water in the cloud is rapidly consumed,which results in the LWC in the middle of the cloud decreasing to less than 0.1 g m-3within 5 min.Meanwhile,the riming process is significantly enhanced because of the interaction between supercooled droplets and seeded ice crystals (Fig.3 (c)).Compared to the Ctl case,the domain-integrated melting mass of graupel increases by 83% and 140% in cases ST2 and ST3,respectively,5 min after cloud seeding.A vigorous downdraft forms owing to the drag force of strong precipitation and the cooling caused by the melting of graupel.As is shown in Fig.3,in cases ST2 and ST3,the cloud seeding increases the maximum downdraft by about 2 and 1.5 times,respectively,compared to the Ctl case (without seeding) within 3 min after cloud seeding,which in turn results in the formation of low-level warm clouds on both sides of the cumulus (Fig.2 (c—f)).Although this part of the cloud also enhances precipitation via the warm-cloud process,the stronger downdraft hinders the further development of the cloud.

Fig.4.Comparison of microphysical properties between seeding and no seeding cases: (a,b) microphysical quantities (units: g m -3) without seeding after 45 and 50 min of simulation;(c,d) profiles of microphysical properties with 1000 L -1 of seeding after 45 and 50 min of simulation;(e,f) microphysical properties with 2000 L -1 of ice crystal seeding after 45 and 50 min of simulation.The physical meaning is the same as in Fig.1.

The potential influence of seeding on the surface precipitation of no-shear convective clouds takes place mainly via the following mechanisms: (1) the deposition growth (Bergeron process) of seeding ice crystals consumes the water vapor and cloud droplets;(2) the interaction between the cloud and the large supercooled cloud results in the increasing number concentration of graupel in the middle of the cloud(Fig.3 (b));and (3) the drag force,produced by melted graupel particles and the evaporative cooling process (Fig.3 (d)) lifts on both sides of the cumulus to form a low-level warm cloud.

3.2.Convective cloud seeding simulation in a strong wind shear environment

For the convective cloud developed under strong wind shear conditions (Fig.4),the regime with high LWC (the area surrounded by the thick blue line) has a wider range (Fig.4 (a,b)) compared to the no-shear case.After seeding in the initial stage of precipitation (40 min),supercooled liquid water is rapidly consumed (Fig.4 (c—e)).Whether seeding with an ice concentration of 1000 L-1or 2000 L-1,the areas with LWC＞0.1 g m-3are rapidly reduced after seeding,and a large amount of LWC is consumed at the middle and high altitudes through the processes of deposition and riming.Compared with the no-seeding case (Ctl_shear),the seeded clouds form a relatively strong downdraft at a height of lower than 4000 m (Fig.4 (a) vs Fig.4 (c,e)).

The same as in the no-shear environment,cloud seeding enhances the riming process in the wind shear environment,and thus it also leads to an increase in the melting mass of graupel before 50 min of simulation(Fig.5).Fig.5 also illustrates that,due to the horizontal advection of ice crystals in the wind shear environment,a portion of supercooled LWC is not depleted.The difference (ST_shear minus Ctl_shear) in the processes of riming and melting between seeding cases and the Ctl case is smaller than that in the no-shear environment.

Fig.5.Impact of seeding on the microphysical processes plays in a role in graupel formation and vertical wind evolution in the wind shear environment.The plots represent the difference between seeding and Ctl cases: (a) maximum downdraft;(b) domain-integrated melting of graupel;(c) domain-integrated riming of the particles;(d) maximum updraft.

As shown in Fig.6,regardless of the amount of seeding,the maximum downdraft is significantly increased owing to the melting of graupel compared with the Ctl_shear case in the wind shear environment.However,in extreme cases (such as ST3_shear with 2000 L-1of ice crystals),the effect of cloud seeding on the maximum downdraft is slightly weaker than that in the no-shear environment (Fig.6 (a)).Besides,the response of the maximum updraft in wind shear conditions is significantly different compared with that in the no-shear environment.For example,in the ST2_ shear and ST3_ shear cases,it seems that cloud seeding does not hurt the maximum updraft compared to the Ctl_shear case.After seeding with 2000 L-1(ST3_shear),the maximum updraft is increased by 5 m s-1in 5 min.This difference may be attributable to the misalignment between the main downdraft and the updraft in the wind shear conditions (Fig.4).That is,although the cloud seeding enhances the downdraft behind the cloud’s direction of movement,it also promotes the updraft ahead.

In wind shear conditions,the stronger downdraft formed by the cloud seeding divides into two branches near the ground,one of which forms a vertical circulation at the back of the direction of movement of the cumulus cloud,resulting in the formation of a low-level warm cloud in the layer of 1—3 km behind the convective cloud.The uplift of another branch in front of the cumulus cloud also promotes the development of low-level warm cloud in the cumulus cloud itself again,which is also significant in promoting the enhancement of precipitation (Fig.4 (b) vs Fig.4 (d,f)).

Due to the interaction between the thermal and dynamic effects generated by the environmental wind field and microphysics,the change in surface precipitation after seeding in the shear environment is more complex than that in the no-shear environment,but in general the peak value of precipitation after seeding is shifted to an earlier time (Fig.6).At the same time,due to the drag force and evaporative cooling,lowlevel warm clouds are formed behind the direction of movement of the cumulus clouds,resulting in the expansion of the precipitation range after seeding (Fig.5 (c,d)).Fig.6 shows the change in surface rainfall intensity with time.Under the condition of a 1000—2000 L-1ice crystal seeding amount,the surface precipitation increases to a certain extent,while the impact of high-level seeding on surface precipitation is quite limited.

4.Conclusion

This paper reports on numerical simulations conducted to evaluate the potential effects of cloud seeding in environments with and without wind shear.Although the experiments are very limited,some interesting conclusions can be summarized from the results,which we hope will be helpful for some field operations:

(1) Whether in the environment with or without wind shear,ice seeding at -4°C with moderate concentrations (1000 L-1) in the developing stage of the cloud,and in a high supercooled liquid water regime,can have a significant impact on the distribution of surface rainfall.This finding is similar to previous studies,such as Geresdi et al.(2020),who concluded that the seeding efficiency depends on the amount of liquid water.However,our study further shows that the efficiency of seeding also strongly depends on the environmental conditions.It seems that the efficiency of seeding is more pronounced in the wind shear environment,especially for seeding ice concentrations smaller than 1000 L-1.

(2) In the environment without wind shear,a large number of graupel particles are produced by cloud seeding via the processes of deposition and riming.Although melted graupel tends to enhance the rainfall intensity in the early stage of the precipitation,it also increases the downdraft in the center regime of the cloud,which in turn decreases the ice phase processes,such as the riming of ice particles.

Fig.6.Rainfall rate on the ground as a function of time: (a) no seeding;(b) -4°C seeding ice concentration=2000 L -1 ;(c) -4°C seeding ice concentration=1000 L -1 ;(d) ice core concentration=2000 L -1,but at -15°C.

(3) When seeding convective clouds in a wind shear environment,the change in microphysical and thermodynamic fields is more complex compared in a no-shear environment.In this case,an updraft is provoked by the main downdraft along the direction of movement of the cloud.Since the updraft is misaligned with the main downdraft,a continuous supply of water vapor is allowed to maintain the development of the cloud.

Funding

This study was jointly supported by the National Key Research and Development Program of China [grant number 2018YFC1507900]and the National Natural Science Foundation of China [grant numbers 41875172 and 42075192].

Acknowledgments

The authors would like to thank Professor Yan Yin at the Nanjing University of Information Sciences and Technology for his kindness in sharing the 2D slab bin model.