The Preliminary Application of Spectral Microphysics in Numerical Study of the Effects of Aerosol Particles on Thunderstorm Development

Abstract

Progress in numerical models and improved computational capabilities have significantly advanced our comprehension of how aerosol particles impact thunderstorm clouds. Yet, much of this research has focused on employing bulk microphysics models to explain the impacts of aerosol particles acting as cloud condensation nuclei (CCN) on electrical activities in thunderstorm clouds. The bulk thunderstorm models use mean sizes of particles and terminal-fall velocities. This causes calculation deviation in the electrification simulation, which in turn leads to deviations in the simulation of lightning processes. Developing this further, we established a three-dimensional high-resolution cloud–aerosol bin thunderstorm model with electrification and lightning to provide more accurate microphysics and dynamic fields for studying electrical activities. For evaluating the impacts of aerosol particles, specifically CCN, on the properties of continental thunderclouds, aerosols from both clean and polluted continental environments were selected. Cloud simulations indicate that droplets develop a narrower spectrum in polluted continental conditions, and weakened ice crystal growth increases the number of small ice crystals compared to clean conditions. Smaller droplets and ice crystals result in less effective riming and decreased graupel concentration and mass. Consequently, a significant decrease in large ice particles leads to a weakened process of charge separation under conditions of pollution. As a direct result, there is about a 43% reduction in lightning frequency and a delay of approximately 5 min in the lightning process under polluted conditions.

1. Introduction

Studies have extensively documented the effects of aerosol particles on both the dynamics and microphysical properties in clouds [1,2,3]. The lightning activities in thunderstorms are the product of dynamic and microphysical processes, which have a strong correlation with aerosol particles. Therefore, the impacts of aerosol particles on lightning activities are one of the most important issues when studying atmospheric electricity.

Aerosol particles, by serving as CCN, affect the dynamic and microphysical evolution of clouds, which can subsequently change lightning activity. Various observational research efforts have highlighted the critical impact of aerosol particle concentrations on lightning activity [4,5]. For instance, research during the COVID-19 pandemic, particularly in India, has shown that decreased aerosol concentrations are linked to reduced lightning frequency in major cities, demonstrating the significant impact of aerosols on storm electrification [4,6]; furthermore, the studies by Thornton et al. (2017) [7] and Wang et al. (2023) [5] indicate that an increase in aerosols can enhance lightning activity, emphasizing their crucial role in microphysics and electrification processes.

The impacts of aerosol particles on lightning activities are related to factors such as their chemical and physical characteristics, meteorological conditions, and geographical and spatial factors. Therefore, the relationship between aerosol particle concentration and lightning activity determined from observations remains highly uncertain. Due to the limitation of detection methods, numerical models have become a very useful tool for investigating the effects of aerosol particles on electrical activity in thunderstorms.

In the past ten years, aerosol and electric processes have been coupled with cloud models to establish thunderstorm models, and corresponding research has been performed. In these thunderstorm models, cloud evolution is described by microphysical schemes. These microphysical schemes have developed in two distinct directions: bulk parameterization and spectral (bin) parameterization. Almost all existing thunderstorm models are based on the bulk microphysics parameterizations. These models have shown that compared to a low-concentration aerosol background, electrification strength is larger, and lightning activity is more active in thunderstorm clouds at higher concentrations [8,9,10]. To verify whether the conclusion that increased aerosol concentrations could enhance electrical activity is universal, Mansell et al. (2013) [11] and Tan et al. (2017) [12] have explored the aerosol effects through a wide range of CCN concentrations. Their findings revealed that the relationship between aerosol concentration and the processes of electrification and lightning is not a simple linear one. In their research, higher CCN concentrations were initially linked to more intense lightning activity, yet subsequent enhancements began to diminish the electrical process. Hence, the impacts of aerosol particles on electrical activity remain highly uncertain in the thunderstorm models available. This uncertainty is most likely because the microphysics approach is limited. The modeling of thunderstorm clouds is quite dependent on cloud microphysical parameterizations. Depending on the description of the hydrometeors, cloud microphysical parameterizations can be divided broadly into two categories: bulk parameterization and spectral (bin) parameterization.

In bulk microphysical parameterization, the particle size distribution of the hydrometeors is assumed as an empirical function. The bulk parameterizations in cloud models have also improved greatly over time. Nevertheless, most current cloud models continue to employ bulk microphysical parameterizations, essentially established by Kessler (1969) [13]. A bulk parameterization defines the hydrometeor spectrum and usually calculates prognostic equations for the mixing ratios of each type [3,14,15,16,17,18]. This bulk parameterization is straightforward in concept and efficient in computation.

Nevertheless, bulk parameterization does not specifically address questions about variations in the size or concentration of hydrometeor particles. For instance, aerosol-cloud interactions and their effects on the dynamic, microphysical, and electrical properties of clouds cannot be studied. Besides the mixing ratio, a two-moment bulk parameterization also forecasts the number of particles. This type of parameterization relies on crucial assumptions, including the activation of the CCN, particle spectral shape, and the mean terminal-fall velocities of particles [12,19,20].

Cloud microphysical parameterizations are crucial for the simulation of thunderstorm cloud electrical processes. The electrical activity in thunderstorms is closely associated with particle collisions, that is, the terminal-fall velocity and size of particles [11,21,22]; however, the bulk thunderstorm models use mean sizes and terminal-fall velocities, ignoring the change in their velocities due to their particle size and dynamic characteristics—which are far from the movement of natural particle groups—resulting in deviations in the calculation of particle collisions. In summary, the research results will deviate even if aerosol modules, electrification, and discharge modules are coupled to the bulk cloud model. The mechanism by which aerosols influence electrical activity through changing the dynamic and microphysical properties in thunderstorms urgently needs to be further studied using more sophisticated cloud microphysics schemes.

A spectral microphysical parameterization uses numerous size bins, ranging from dozens to hundreds, to accurately depict the size spectrum of CCN and other hydrometeor particles [23,24,25,26,27]. This approach explicitly computes cloud droplet activation in a bin model, which is indispensable for studying aerosol effects. Compared to bulk parameterization, bin microphysical parameterization is considerably more sophisticated, and these bin models—currently recognized as the most theoretical and realistic numerical simulation scheme—serve to validate and adjust the bulk scheme models [14,20]. The dynamic and microphysical processes of clouds in thunderstorms are simulated as accurately as possible; these may be important conditions for investigating the effects of aerosol particles on lightning activities. Therefore, it is necessary and urgent to use spectral microphysical cloud models that better represent realistic atmospheric conditions to explore the role of aerosol particles in thunderstorm clouds.

An increasing number of spectral microphysical parameterizations have been incorporated into cloud models, and spectral microphysical cloud models have been established. These models have proven successful in exploring the effects of cloud microphysics on the spatial distribution of rainfall, replicating the characteristics of stratiform clouds and their radiative effects, as well as simulating cloud seeding and cloud chemistry [19,28,29,30]. However, a spectral thunderstorm model suitable for studying the relationship between aerosol particles and electrical activity has not yet been reported. The main reason for this lack is the computing power that such a model requires; the discharge process usually needs to solve the Poisson equation iteratively to obtain the spatial potential distribution, which requires a large amount of processing time; therefore, almost all thunderstorm models have chosen to use the bulk thunderstorm models, which are computationally efficient. Recent breakthroughs in computational capabilities and parallel processing technologies have enabled the use of the spectral thunderstorm model, making it possible to establish a regional mesoscale spectral model of thunderstorms based on the Weather Research and Forecasting model (WRF) dynamic framework. This advancement allows for an in-depth exploration of the physical mechanism of the influence of aerosol particles on electrical activity in thunderstorms.

We coupled a self-developed spectral microphysical parameterization, electrification, and lightning parameterization into the WRF dynamic framework to establish a three-dimensional high-resolution cloud–aerosol bin thunderstorm model with electrification and lightning. The main purpose of our study is to couple a detailed spectral microphysical parameterization to the WRF4.0 dynamic framework to establish a three-dimensional spectral (bin) model that provides a more accurate microphysical background field for electrical processes. Based on this, we investigate the impacts of aerosol particles on microphysics and lightning using WRF4.0 with spectral microphysics. This research not only provides a theoretical basis for establishing and improving regional mesoscale thunderstorm forecasting models based on bulk parameterization but also provides a scientific basis for the early warning of and protection from thunderstorm-based disasters.

2. Brief Description of the Model

This work couples advanced spectral microphysical parameterization, electrification, and lightning parameterization into the WRF model (version 4.0) to establish a three-dimensional high-resolution cloud–aerosol thunderstorm model with spectral microphysics.

2.1. Microphysics

We coupled an advanced spectral microphysics scheme to the WRF4.0. The bin microphysical scheme used in this study was developed at Tel Aviv University [31]. The liquid-phase chemical processes proposed by Yin et al. (2005) [32] were included. Four different hydrometeor species are taken into account: droplets, ice crystals, graupel, and snowflakes. These particles are categorized into 44 mass bins, where the mass of particles in the k-th bin, m_k, is calculated as m_k = 2m_k−1. The mass ranges from 0.1598 × 10⁻¹³ to 0.17468 × 10⁻³ kg for both liquid and ice phases, corresponding to diameters from 3.125 to 8063 μm for droplets and 3.23754 to 8540 μm for ice particles. The aerosol particles are divided into 44 bins, with radii from 0.001 to 15.75 μm, focusing only on their role as CCN. This study takes into account only one type of CCN, simplifying the aerosol complexity by considering only ammonium sulfate as the CCN component.

Liquid-phase processes include activation, condensation, evaporation, stochastic collision coalescence, and breakup. Ice-phase processes include nucleation, multiplication, sublimation, and interactions between ice particles (ice–ice and ice–drop) such as coagulation, accretion, or riming, melting of ice particles, and sedimentation of drops and ice particles. In this study, the process of ice nucleation is dependent on both temperature and supersaturation relative to ice, according to the description by Meyers et al. (1992) [33]. Graupel and ice crystals can form through drop freezing, as described by Bigg (1953) [34]. This study integrates the Hallet–Mossop secondary ice production process, as described by Mossop in 1978 [35]. Snow formation from aggregation is included, as described by Yin et al. (2000) [32]. To facilitate comparisons with previous studies, ice crystals exceeding 400 μm in size are transformed into snow, following the approach proposed by Morrison and Gettelman (2008) [36].

2.2. Electrical Processes

Explicit noninductive electrification and three-dimensional branched-lightning parameterization were incorporated into the WRF model.

2.2.1. Electrification Parameterization

There is a general consensus that noninductive charge separation is agreed upon as the primary mechanism for electrification in thunderstorms [11,12]. Experiments in laboratory settings indicate that this charge separation mainly results from the rebounding collisions among graupel and ice crystals. The charge separation rate ($Q_t$) can be expressed as follows:

$$Q_t=K_{gi}\times n_g\times n_i\times \delta q$$

$$K_{gi}=\frac{\pi }{4}{\left(D_i+D_g\right)}^2∆V_{gi}\times {\mathsf{\epsilon }}_{gi}$$

where, $n_g$ and $n_i$ are the concentrations of graupel and ice crystals, respectively; $\delta q$ represents the charge separated per rebounding collision; $K_{gi}$ represents the collision separation kernel; $D_g$ and $D_i$ represent the graupel and ice crystal diameters, respectively; $∆V_{gi}$ is the relative velocity of graupel and ice crystals; and ${\mathsf{\epsilon }}_{gi}$ is the collision separation rate.

This model incorporates the noninductive electrification scheme developed by Saunders et al. (1991, 1998) [22,37]. Laboratory studies by Saunders et al. (1998) [37] have shown that both the polarity and amount of the charge separation ($\delta q$) resulting from graupel and ice particle rebounds vary according to temperature, the effective liquid water content (ELWC), dimensions of the ice particles, and the relative velocity of the colliding ice particles. The polarity of the charge transferred during collisions between ice crystals and graupel can vary with temperature. Laboratory studies have shown that at lower temperatures, the polarity of the charge transferred may reverse, causing graupel to gain more negative charge, which contributes to the complex vertical charge structure in thunderstorms. The average charge separated per collision in the scheme is described as follows:

$$\delta q=k_q{D_{ice}}^m{\left|V_g-V_i\right|}^{n}q\left(R\right)$$

where $k_q$, $m$, and $n$ are constants; $D_{ice}$ is the ice crystal diameter; $V_g$ and $V_i$ are the terminal-fall velocities of graupel and ice particles, respectively; and $q\left(R\right)$ is described in the following equation.

$$q\left(R\right)=\left\{\begin{array}{c}0, R=R_{crit}\\ 6.74\times \left(R-R_{crit}\right), R>R_{crit}\\ 3.9\times \left(R_{crit}-0.1\right)\times \left(4\times {\left[\frac{R-\frac{\left(R_{crit}+0.1\right)}{2}}{\left(R_{crit}-0.1\right)}\right]}^2-1\right),\\ 0.1 \mathrm{g} · {\mathrm{m}}^{-2} · {\mathrm{s}}^{-1}<R<R_{crit}\end{array}\right.$$

where R denotes the rate of rime accretion, calculated as the product of the effective liquid water content and graupel fall velocity ($R=ELWC\times V_g$). Saunders et al. (1998) [17] defined the zones of positive and negative charging for graupel using a critical curve $R_{crit}$, indicating that graupel carries a negative charge at lower values of R and a positive charge at higher values of R.

2.2.2. Lighting Parameterization

The parameterization of lightning discharge was parameterized based on Mansell et al. (2002) [38] and Tan et al. (2006, 2014) [39,40]. The concept of the bidirectional leader was applied in the parameterization: from the point of initiation, positive and negative leaders extend outward in opposite directions, with the electric field around them inducing the charge density along the lightning channel. Lightning initiation involved the runaway electron threshold for the break-even field, followed by the propagation of bidirectional channels—where one end carries positive charge and the other negative—through a stochastic step-by-step process. The intracloud lightning (IC) leaders did not extend to the ground, and a height threshold (1.5 km or 6 grid points above ground) was utilized to classify a flash as cloud-to-ground lightning (CG) (includes +CG and −CG). Details of the parameterization were described by Mansell et al. (2002) [38] and Tan et al. (2006) [39].

3. Results

3.1. Verification

Since we have integrated the spectral microphysics scheme into the WRF model for the first time, it is essential to verify that our new thunderstorm model with this spectral microphysics scheme not only simulates clouds more effectively but also exceeds the performance of the existing bulk microphysics schemes in WRF. The bulk microphysics scheme chosen for comparison is the Morrison two-moment scheme. This scheme provides a detailed description of microphysical processes, including the direct effects of aerosols on the concentrations of cloud droplets and ice crystals [36]. By tracking both the mass and the number of different hydrometeor particles, this provides a more accurate simulation of how aerosol particles influence cloud microphysics and the formation of precipitation. The Morrison scheme is suited for scientific research.

To evaluate the accuracy of the new thunderstorm model, simulations were performed and compared with the observational data of the heavy storm event that occurred from 0000 UTC on 20 July 2021 to 0000 UTC on 21 July 2021, in the Henan province. From 17 July to 21 July 2021, Henan province in China experienced extreme rainfall. On 20 July 2021, Zhengzhou encountered an unprecedented hourly precipitation of 201.9 mm, setting a national record, with the cumulative rainfall for the event reaching 720 mm, surpassing the annual average. This event led to over 302 fatalities and economic losses amounting to USD 17.7 billion. This study presents the predictability of this extreme rainfall event using the WRF 4.0. The meteorological background field used in the model is from the National Centers for Environmental Prediction (NCEP), known as FNL. The FNL reanalysis data provide initial and boundary conditions for cloud simulations. The atmospheric state is initialized by reading meteorological information from the FNL dataset. The FNL reanalysis data have a spatial resolution of 1° × 1° grid, a temporal resolution of 6 h, and are available on 50 vertical levels from 1000 hPa to 50 hPa (https://rda.ucar.edu/datasets/ds083.2/, accessed on 1 July 1999). These data products are obtained through the advancement of remote sensing technology and assimilation technology to provide high-quality, spatially uniform, and temporally continuous global meteorological data. The model utilizes Shuttle Radar Topography Mission (SRTM) data for terrain information. The high-resolution SRTM data used in this study significantly improve the accuracy of the WRF model. The data come from a global remote sensing mission led by the National Aeronautics and Space Administration (NASA). As part of this mission, it captured Earth’s topography at 1 arc-second (30 m) for over 80% of the Earth’s surface. Using synthetic aperture radar and interferometry, SRTM has collected one of the most accurate digital elevation models of Earth. The domains and validation configurations of the model are illustrated in Figure 1 and

Table 1. The initial parameter settings for the model.

Parameter	Description
Nesting	Two-way nesting
Horizontal grid spacing	Domain 1: 6 km Domain 2: 2 km
Grid points	Domain 1: 133 × 100 Domain 1: 138 × 85
Timestep	Domain 1: 24 s Domain 2: 8 s
Top pressure	50 hPa
Cloud microphysics	Morrison two-moment bulk microphysics self-coupled spectral (bin) microphysics
Cumulus	Kain–Fritsch
Planetary boundary layer	MYJ
Surface layer	Mellor–Yamada–Janjic (Eta) TKE
Land surface	RUC
Longwave radiation	CAM
Shortwave radiation	CAM

.

Firstly, we compared the simulated composite reflectivity with observational data. The observed composite reflectivity data are primarily provided by the Henan Meteorological Service. The data are obtained through remote sensing technology. Figure 2 illustrates the comparison of composite reflectivity from observation and simulation results. In the simulations, the results from the Morrison two-moment scheme significantly deviate from the observations. The Morrison two-moment scheme predicted smaller but more scattered convective cells compared to the self-coupled spectral (bin) microphysical scheme (Figure 2b,c). Despite some differences compared to the observations, the simulation from our new thunderstorm model with the spectral microphysics scheme accurately captures the locations with high reflectivity values (greater than 30 dBZ) in agreement with the observations (Figure 2a,c).

Subsequently, we compared the cumulative precipitation. The precipitation data come from the operational observation network of the China Meteorological Administration. Hourly precipitation data from 932 hydrological and 10,352 meteorological rain gauge stations were used to generate daily gridded precipitation data from 19 to 21 July 2021 with a latitude–longitude grid of 0.01° resolution in the region of 30°N–38°N, 109°E–117°E. In areas where observation sites are sparse or data is lacking, remote sensing technology is used to supplement the observational data. Hourly and daily data from each station were compared with the data from surrounding stations. Comparing the cumulative rainfall, it is found that there is a notable difference between the simulation results of the bulk microphysical scheme and the observations (Figure 3a,b). In contrast, the model with the spectral microphysical scheme cloud more accurately reproduces the spatial distribution of the rainbands, particularly in successfully capturing the location of the intense precipitation center (Figure 3a,c). Therefore, the model cloud accurately depicts the evolution of clouds, providing an accurate background field for the subsequent electrical activity of thunderstorm clouds.

3.2. Simulation

3.2.1. Initial Condition

In this study, the initial aerosol spectra were derived from field observations carried out at Huangshan Mountain and in Nanjing City [41,42], representing typical continental clean and polluted conditions, respectively. The aerosol distribution was characterized by overlaying three lognormal distribution functions; the spectrum was characterized by overlaying three lognormal distribution functions:

$$\frac{dN}{dln{r}_n}={\displaystyle \sum }_{i=1}^3\frac{n_i}{{\left(2\pi \right)}^{1/2}log{\sigma }_{i}ln10}\exp \left(-\frac{{\left[log(r_n/R_i)\right]}^2}{2{\left(log{\sigma }_i\right)}^2}\right)$$

where $r_n$ represents the aerosol radius; $n_i$ denotes the aerosol concentration within mode i; ${\sigma }_i$ is the geometric standard deviation; and $R_i$ refers to the geometric mean radius for mode i. $N$ signifies the total number of aerosol particles. Sun et al. (2012a, 2012b) [43,44] have detailed these distribution parameters in their research.

Table 2. Parameters of the aerosol particle distribution.

	Mode	ni (cm−3)	Ri (μm)	log σi
	1	588.6	0.0083	0.0841
clean background	2	1425.2	0.0167	0.1983
	3	882.1	0.094	0.2371
	1	40,000.9	0.0168	0.2615
polluted background	2	8781.2	0.052	0.185
	3	66.7	0.129	0.1646

lists the characteristics of aerosol distribution. Additionally, the initial aerosol concentrations decrease with height, as described by Yin et al. (2000) [32]:

$$N\left(z\right)=N\left(z=0\right)\times \exp \left(-\frac{z}{z_s}\right)$$

where the scale height, denoted as z_s, is specified to be 2 km; $N\left(z\right)$ represents the ground aerosol concentration.

To assess the CCN effects on thunderstorm properties, the experiments were divided into two groups: one with a continental clean background (C_N) and another with a continental polluted background (P_N). Aerosol measurements are from Huangshan Mountain and Nanjing City in Eastern China (Figure 4). Based on the analysis of aerosol chemical composition, the soluble material content was estimated to be 34% for the clean condition and 34.3% for the polluted condition [45,46]. Cases C_N and P_N describe the thunderstorm development under conditions of clean and polluted continental aerosol particles.

An idealized sounding was used for the environment (see Yau, 1980 [47]). For all cases, the initial thermodynamic settings were based on a theoretical temperature and dew point profile, illustrated in Figure 5. This profile is typical of deep convective atmospheric stratification, with high humidity throughout the atmosphere. To initialize the simulations, a pulse of heat that produced a 3 K perturbation with a radius of 500 m was introduced at the center of the domain. The horizontal grid of the model was 48 × 48, with a horizontal resolution of 250 m. Each simulation has 60 vertical layers. The simulation duration lasted for 35 min, with a time step of 2 s.

3.2.2. Microphysical Processes

For the analysis of continental thunderstorms, two cases, C_N and P_N, were simulated. Figure 6 displays the distributions of droplet number concentration and mass mixing ratio after 11 min of simulation, corresponding to the cloud initiation stage in C_N and P_N. With an increase in CCN in the P_N case, there was a noticeable rise in both the concentration and the mass mixing ratio of droplets. This finding is consistent with our previous understanding [48]: as the CCN further increases, the concentration and mass of droplets increase because more aerosol particles can act as CCN and nucleate into droplets under polluted conditions. Figure 7 shows the number density distribution functions of the droplets at the center of the cloud at an altitude of 1800 m (just above the cloud base) at 11 min in C_N and P_N. In P_N, the nucleated droplets have a considerable effect on the production of more numerous but smaller cloud droplets and develop a narrower droplet spectrum under the same meteorological conditions, consistent with observational and modeling research on the effects of aerosols on thunderstorms (Figure 7) [49,50,51]. As the thunderstorm became more polluted, the growth rate of droplets was slower due to their increased concentration and the intensified competition for the available supersaturation in thunderstorms.

The development of ice particles in thunderstorms is intricately linked to the droplets. The impacts of CCN concentration on liquid-phase processes will subsequently impact the growth of ice particles. Therefore, to examine the effect of CCN concentration on ice-phase processes, we display the distribution of number concentration and the mass mixing ratio of ice particles at 22 min (the vigorous stage)—as well as the corresponding number density distribution functions of the ice particles at the center of the cloud—at 8.5 km height in C_N and P_N cases.

Figure 8 illustrates the difference in concentration and mass mixing ratio of ice crystals. In comparison to the C_N case, a higher number concentration and a lower mass mixing ratio of ice crystals are produced in the P_N case. The findings for the number density distribution functions of ice crystals in the two cases are presented in Figure 9. As illustrated in Figure 8 and Figure 9, the comparatively larger concentration and smaller mass of ice crystals in the P_N case correspond to more small ice crystals than in the C_N case. The spectra of ice crystals were broadened toward small ice crystal particles (Figure 9) in the P_N case. The enhanced liquid water content as a higher concentration of CCN promoted the nucleation of ice crystals, so the number concentration of ice crystals increased. However, in the P_N case, the ice crystal growth by deposition and sublimation will decrease relatively due to competition with the increased number of droplets and ice crystals, resulting in an increase in the available water vapor. Consequently, the mass mixing ratio of ice crystals will decrease, and the number of small ice crystals will increase. Additionally, the weakened growth of ice crystals promoted very inefficient aggregation, leading to an increased number of small ice crystals with increased CCN.

The differing CCN concentrations will cause significant changes in the microphysical processes of the graupel particles. A comparison between the graupel in C_N and P_N cases is given in Figure 10. In both cases, the number concentration and the mass mixing ratio of the graupel particles decreased as the CCN increased. The spectrum in the P_N case was much narrower than that of the C_N case (Figure 11). In the cloud microphysical process, graupel particles formed mainly through the riming process between the ice crystals and droplets. In the P_N case, the decrease in riming between ice crystals and droplets is notable due to more numerous but smaller droplets and ice crystals, which are less efficient in forming graupel particles. Thus, due to the lower riming efficiency with increased CCN, the number concentration and the mass mixing ratio of graupel particles were reduced, and the spectrum of the graupel particles became narrower.

3.3. Electrification

Considering the collision rate or relative falling rate between particles, the total spatial charge density in thunderstorms is primarily influenced by the ice crystals and graupel particles. During the initial (pre-lightning) stages of thunderstorm electrification, Figure 12 illustrates a comparative analysis of total charge density between the P_N and C_N cases at T = 22 min. The total charges in the two cases show similar inverted triple charge structures. A rise in the concentration of CCN reduced the total charge density. The reduced charge density in the P_N case can be attributed to the decreased concentrations of larger ice particles. Based on the noninductive charging mechanism, a reduction in the concentrations of large ice particles (graupel) coupled with a rise in smaller ice crystals can substantially weaken the charge separation process. This results in a lower charge density when CCN concentrations are higher. The simulation findings emphasize the significant impact of CCN on altering the intensity of charging.

3.4. Lightning

The total lightning frequency, a key characteristic of thunderstorm clouds, is tightly linked to microphysical processes (e.g., Tan et al., 2014 [40]; Guo et al., 2017 [17]). Significant variations exist in the microphysics generated in two simulations using two different aerosol distributions; therefore, aerosol particles could significantly contribute to the formation of lightning.

Figure 13 provides a detailed comparison of lightning frequency. The total lightning frequencies were 39 and 17 in the C_N case and P_N case, respectively. The lightning processes are delayed by about 5 min. The data showed a correlation between increased cloud pollution and reduced lightning frequency, accompanied by delays in the lightning process. The process of weakening lightning is directly associated with electrification. As CCN concentrations increased in the P_N case, the significant decrease in large ice particles weakened the charge separation mechanism, resulting in a lower lightning frequency and a postponed lightning process.

4. Discussion

Currently, our research (Section 4) primarily uses idealized cases (WRF-Les) that focus on specific atmospheric dynamic processes or meteorological phenomena, without requiring real terrain or meteorological data. Idealized cases simplify environmental conditions and do not account for the complex sources and transmission of aerosols. They lack detailed interactions between aerosols and cloud formation processes, such as the chemical and radiative properties influenced by aerosols. For the comprehensive and accurate study of aerosol impacts on cloud properties, future research will employ real-case (WRF-Real) simulations that consider meteorological data, terrain, and detailed aerosol source information. These methods will better understand the dynamic and complex impacts of aerosols on clouds and weather systems in realistic settings.

5. Conclusions

The impacts of aerosol particles, serving as CCN, on microphysics and electrical processes have been investigated utilizing a three-dimensional high-resolution cloud–aerosol bin thunderstorm model. It provides a detailed description of the intricate processes in both the warm and ice phases, which yields a more reasonable background field for electrical processes. The models were driven by continental clean and polluted aerosol particle distributions, respectively. The primary findings of the research are summarized below:

(1)

The droplets nucleated in the continental polluted background considerably affected the production of a greater number of smaller cloud droplets and led to a narrower droplet spectrum. In polluted conditions, a decline in the mass mixing ratio of ice crystals and a rise in the number of smaller ice crystals are indicated, primarily attributed to weakened ice crystal growth processes. An abundance of smaller droplets leads to less effective riming, decreasing both the concentration and mass mixing ratio of graupel particles, resulting in a narrower spectrum;

(2)

Based on the noninductive charging mechanism, the decreasing number of large ice particle (graupel) concentrations and the increasing number of small ice crystals can significantly impair the charge separation mechanism, resulting in a reduced charge density with increasing concentrations of CCN. These results suggest that CCN significantly impacts the strength of charging;

(3)

As CCN concentrations increase in the P_N case, the dramatic decrease in large ice particles weakens the charge separation effectiveness. This results in a roughly 43% decrease in lightning frequency and a delay of about 5 min under polluted conditions.

The three-dimensional bin thunderstorm model addresses a detailed description of sophisticated warm and ice-phase processes; this model provides a more accurate microphysical and dynamic background for electrical activity. In future work, we will use the advantages of this model to further explore the effects of aerosol size on thunderstorm microphysics and lightning.