Analysis of Rain Drop Size Distribution to Elucidate the Precipitation Process using a Cloud Microphysics Conceptual Model and In Situ Measurement

Raindrop size distribution (DSD) is an important parameter in rainfall research and can be used for quantitative precipitation estimation (QPE) in meteorology and hydrology. DSD also improves the understanding of the uncertainty of cloud microphysical processes (CMPs) such as ice-based and warm rain growth during climate change. Changes in CMPs impact the generation of precipitation. However, the estimation of CMPs based on in situ observation is difficult because of the complexity of microphysics processes, and most previous studies on the CMP involved approximations to predict the types of microphysical processes affecting precipitation generation based on in situ observations performed in real-time. Therefore, we developed a simple method for understanding the CMPs of precipitation generation using a conceptual model of CMPs and in situ observation DSD data. We employed previously observed DSD parameters and a CMP conceptual model of the DSD observation-based microphysical process. As case studies, we applied DSD observation data obtained in Korea and East Asia to estimate the CMPs. For example, the major CMP of megacities was vapor deposition in Beijing (< 1 mm h−1) and Seoul (< 5 mm h−1), as the strong updraft of the urban heat island effect in megacities results in increased liquid water content, leading to the formation of large number of supersaturated clouds at higher altitudes.


Introduction
Raindrop size distribution (DSD) plays a pivotal role in rainfall research and is used for quantitative precipitation estimation (QPE) in meteorology and hydrology as well as to understand uncertainty in cloud microphysical processes (CMPs) during climate change. Studies on DSD have been conducted in different locations worldwide (Bringi et al. 2003(Bringi et al. , 2009). The characteristics of DSDs are greatly affected by the geographical location, climate type, and seasonal variation (Tokay and Short 1996;Seela et al. 2017;Ji et al. 2019;Ma et al. 2019;Cha and Yum 2021).
Previous DSD studies suggested that climate change has altered precipitation around East Asia (Zhang et al. 2014;Li et al. 2016;Ullah et al. 2018), where rain and cloud systems exhibit complex spatiotemporal characteristics (Ding and Chan 2005;Li et al. 2010;Yin et al. 2013;Salah et al. 2020).
Numerous studies have revealed the statistical parameters of DSDs in East Asia (Chen et al. 2016(Chen et al. , 2017Ma et al. 2019;Wang et al. 2021). Ma et al. (2019) observed the DSD using a particle size velocity (PARSIVEL) disdrometer (hereafter referred to as PARSIVEL) in Beijing. Bringi et al. (2003) reported that convective rain in Beijing was neither continental nor maritime because of its specific location and complex topography. Tang et al. (2014) observed many different DSDs in the northern and southern regions of China. Chen et al. (2017) performed statistical analyses of DSD parameters based on 3 years of disdrometer data measured at a high altitude (4,500 m above sea level (asl)) over the Tibetan Plateau. The characteristics of DSD have also been widely examined in Korea (You et al. 2004(You et al. , 2014You and Lee 2015;Suh et al. 2016Suh et al. , 2021Loh et al. 2019).
For example, Loh et al. (2019) and Jwa et al. (2021) used a statistical model to classify DSD parameters as Changma, localized low pressure, and typhoons.
However, most previous studies did not focus on the detailed relationship between DSD and CMPs based on weather types and environmental characteristics (Cotton et al. 2011). In addition, the estimation of CMP through in situ observation is difficult because the microphysics processes are complex, and most researchers approximate CMP to determine the types of processes affecting precipitation generation by performing in situ observation in real-time. Therefore, we developed a simple method for understanding the CMPs of precipitation generation using a conceptual model of CMPs and in situ observation DSD data. We used the CMP conceptual model described by Dolan et al. (2018) along with observed DSD parameters for microphysical processes. Dolan et al. (2018) analyzed 12 global disdrometer datasets across three latitudes using principal component analysis (PCA) and classified global DSD characteristics into six groups. Their results showed that the physical processes shaping the DSD characteristics varied as a function of location.
In this study, we first analyzed the parameters of the DSD statistical model in Korea and East Asia to identify the dominant CMPs in rainfall events using the conceptual model described by Dolan et al. (2018). Second, we analyzed the DSD characteristics, and calculated the normalized gamma distribution for the DSD at a mountainous site in Korea using ground-based disdrometer data. Third, we applied the results of the DSD parameters of the normalized gamma distribution to estimate the CMP as case studies in Korea and East Asia using the conceptual model of the CMPs and in situ observation DSD data.

Data and Methods
The PARSIVEL (OTT MESSTECHNIK GmbH & Co., Kempten, Germany) is a laser-based optical instrument that measures all types of precipitation. The laser sensor of the PARSIVEL produces a horizontal strip of light. The emitter and receiver are integrated into a single protective housing. To determine the particle speed, the duration of the signal is measured. A signal is generated immediately after a precipitation particle enters the light strip and ends when the particle has completely left the light strip. The PARSIVEL was used to collect raindrops at Daegawllyeong (site I in Fig. 1) (37°41′N, 128°45′E; at 845 m asl) in 2010 and 2011. Quality control procedures based on Cha and Yum (2021) method were applied to the PARSIVEL data used to analyze the DSD in this study as follows: 1) The total number of particles was more than 10 drops.
2) The DSD spectra for which non-zero drop counts were recorded in a series of consecutive channels were selected and spectra in which the drop diameter was < 0.56 mm were excluded from the analysis. 3) To remove the wind-effect error, winds with speeds of < 10 m s. −1 were considered (Friedrich et al. 2013) 4) To remove raindrop contamination by mixed-type precipitation, rain and snow data were selected when the temperature was warmer than 5 °C.
The DSD characteristics obtained in this study were compared with those obtained in previous studies in East Asia ( Fig. 1) as case studies. The Köppen-Geiger climate classification (Peel et al. 2007) was used to define the climate at each site in East Asia. As shown in Fig. 1, A (Taoyuan) and B (Kaohsiung) have a humid subtropical climate; F (Palau) has a tropical rainforest climate; D (Beijing) has a hot summer continental climate; C (Naqu) and E (Motuo) have a subarctic climate in a high-altitude region; H (Seoul) has a hot summer continental climate; I (Daegawllyeong) is a mountainous region with a warm summer continental climate; and G (Busan) has a dry winter and humid subtropical climate. Sites D and H are megacity regions affected by the urban climate (Hong et al. 2013(Hong et al. , 2019Ma et al. 2019). To determine the differences in the DSD parameters of different climate zones in Korea, the DSD parameters at site G were compared with those at site I, as site I is in a mountainous region with a warm summer continental climate. Table 1 summaries the characteristics of the DSD observational sites, including the measurement period, site information, and equipment at each observation site.

Calculation of Normalized Gamma Distribution for Estimating DSD Parameters
Most studies of the characteristics of DSDs employed a normalized gamma distribution-based statistical model (Testud et al. 2000;Seela et al. 2017;Ma et al. 2019). The same model was used in the present study because its parameters provide a physical meaning for DSDs.
where D is the drop diameter (mm), N(D) (m −3 mm −1 ) is the number of drops per unit volume per unit size interval, N w (m −3 mm −1 ) is the normalized intercept parameter, μ (unitless) is the shape parameter, and D m (mm) is the massweighted mean drop diameter. D m is used to describe the DSD and is estimated using the following equation: To estimate CMPs, as reported by Dolan et al. (2018), the median volume diameter ( D o , mm) was calculated using the following equation: Two parameters are considered when characterizing rain intensity: the liquid water content (LWC, g m −3 ) and rainfall  (Testud et al. 2000). Testud et al. (2000) found that the LWC has a clearer physical meaning compared with R (as altitude increases, R is subject to vertical air motion and its terminal fall velocity changes with air density).
N w (m −3 mm −1 ) is estimated using the following equation: where, ρ w is the water density. Moment-based DSD estimation of meteorological products is a general method used in radar meteorology. An optical disdrometer is used for the products. The nth-order moment is determined as: The parameters use to describe the DSD were calculated as follows:

Conceptual Model of DSD Observation-Based Microphysical Process
To identify the cause of changes in the DSD parameters, the CMPs were estimated using the conceptual model (DCPM) of the DSD observation-based microphysical process suggested by Dolan et al. (2018). DSD parameters were classified into six groups using PCA, and vertical observation data from the dual-polarization radar were compared with the DSD parameters of the six groups to generate the DCPM. Dolan et al. (2018) described a DCPM based on group determination in log 10 N w -D 0 space ( Fig. 12 in Dolan et al. (2018)). Groups 1, 3, 5, and 6 are characterized by convective precipitation processes, where particles are assumed to increase in size as they are lifted by a convective updraft; after sufficient growth, the particles fall back through the updraft and grow to even larger sizes. Groups 2 and 4 are stratiform precipitation processes, with increasing D 0 and decreasing log 10 N w correlated with a bright band intensity and representing an evolution from melted vapor-grown particles to aggregation and riming processes. Especially, the light rain rates and small drop diameter associate with group 2 by vapor deposition. We used the DCPM to identify the dominant mechanisms in the six groups objectively determined from the surface disdrometers using PCA in log 10 N w -D 0 space.

DSD Parameter Characteristics in Korea
To estimate the DSD parameters at site I, PARSIVEL data from Daegawllyeong (site I in Fig. 1) collected in 2010-2011 were used. The total observation time of rain events was 1,303 h. To evaluate the quality of the PARSIVEL data, we compared the rain rate of the PARSIVEL with that of the Tipping Bucket (TB) rain gauge at site I; the correlation coefficient was calculated as 0.91 (Fig. 2). The standard deviations of the collected observational data using the PARSIVEL and TB rain gauge were 2.07 and 1.93 mm h −1 , respectively. The correlation coefficient calculated by You and Lee (2015) in Korea was 0.99 between the PARSIVEL and TB rain gauge. Our results and those of previous studies in Korea show that PARSIVEL is suitable for studying the DSD. As shown in Fig. 3a, D m decreases with an increase in μ, indicating that the breadth of the DSD decreases when Comparison of the rain rate determined using a particle size velocity disdrometer (PARSIVEL) with that obtained using the tipping bucket (TB) rain gauge from hourly data at site I (Daegwallyeong) from 2010 to 2011 after data quality control (Cha and Yum 2021) the number of small raindrops increases. D m and log 10 N w exhibit an inverse relationship ( Fig. 3b). At the same water content, the total drop concentration decreases when the mean volume weighted diameter size increases. The total drop concentration at the same water content increases with increasing μ (Fig. 3c). Most previous studies showed similar results for the relationship of D m , log 10 N w , and μ (Chen et al. 2016;Ma et al. 2019;Suh et al. 2021;Wang et al. 2021). Thus, the PARSIVEL data were of suitable. Figure 4a, 5a, and 6a show the relative frequencies (%) of the parameters at site I. The mean (standard deviation) values of D m , μ, and log 10 N w were estimated as 1.40 (± 0.43) mm, 3.71 (± 3.46), and 3.62 (± 0.60) mm −1 m −3 . The values of the parameters calculated in this study showed similar ranges as those reported previously (You et al. 2004;You and Lee 2015;Bang et al. 2017).  Figure 4a shows the relative frequency histograms of D m at site I. You and Lee (2015) studied DSD parameters at site G using a Precipitation Occurrence Sensor System (POSS), which is a low-power X-band bi-static radar system, in 2002 and a PARSIVEL in 2012. We compared the DSD parameters obtained in this study with the PAR-SIVEL DSD data collected at site G in 2012 by You and Lee (2015). The mean (standard deviation) value of D m at site G was 1.45 (± 0.40) mm (Fig. 4b) and mean (standard deviation) value of D m at site I calculated in this study was 1.40 (± 0.43) mm (Fig. 4a). The maximum values of D m were approximately 3.8 and 2.6 mm at sites G and I, respectively. The average precipitation particle size was larger in coastal areas than in mountainous areas in Korea (Cha and Yum 2021). As shown in Fig. 4a and b, the relative frequency of D m at site I in the mountainous region had a wider distribution than that at site G in the coastal region because of differences in the CMPs of precipitation drop generation. Figure 5 shows a histogram of μ, the shape of the DSD at both sites. As shown in Fig. 5a and b, the mean μ (standard deviation) values at sites I and G were 3.71 (± 3.46) and 5.99 (± 3.75), respectively. A larger μ value indicates a wider DSD and vice versa (Testud et al. 2000). In Korea, continental mountain clouds at site I tend to generate a large number of small particles, and oceanic clouds at site G generate larger rain drops (Cha and Yum 2021). Figure 6a and b show a histogram of the relative frequencies of log 10 N w at sites I and G. The mean (standard deviation) values of log 10 N w at site I and G were 3.62 (± 0.60) and 3.51 (± 0.45) mm −1 m −3 , respectively.

DSD Parameter Characteristics in East Asia
The mean (standard deviation) values of D m , log 10 N w , and μ in East Asian regions were investigated based on ground observations ( Table 2). The DSD parameters at site I were evaluated as precipitation parameters in the mountainous region in the hot summer humid continental climate.
The values of D m (1.40 (± 0.43) mm) and log 10 N w (3.62 (± 0.60) m −3 mm −1 ) at site I were smaller than those at site H; D m = 1.54 (± 0.32) mm and log 10 N w = 3.98 (± 0.42). That is, mountainous terrain near site I forces airflow to ascend and then initiates the formation of precipitation, as described by Cosma et al. (2002). Therefore, there was a much larger number of smaller particles at site I than at site G; additionally, these particles were smaller at site D m than at site H (Seoul). The value of μ at site G [5.99 (± 3.75)] was higher than that at site I [3.71 (± 3.46)] because the convective rainfall at site G increases the value of μ (You and Lee 2015). The value of μ at site A [6.72 (± 7.11)] and at site B [6.93 (± 7.93)] at the Taiwan observation sites were twofold higher than those at site I. The breadth of the DSD at sites A and B was smaller than that at site I. If the value of μ is > 0, the DSD curve is concave downward; if the value of μ is < 0, the DSD curve is concave upward; and if it is equal to 0, the DSD is exponential (Ulbrich 2001). The value of μ at site F [8.37 (± 8.94)] was approximately two-fold higher than that at site I, [3.71 (± 3.46)]. The breadth of the DSD was narrow because of the convective rainfall in the tropical rainforest climate. Similarly, for sites C and E, which are at high-altitude regions in China, the values of μ were approximately three-fold lower than that at site I because of the higher total aerosol concentration, resulting in a wider breadth of DSD in the high-altitude regions with a continental climate. Although the total aerosol concentration is small in marine climates, many particles can act as cloud condensation nuclei (Kim et al. 2012). Therefore, the DSD varies widely, and the value of μ was relatively low at sites C and E. Figure 7 shows a comparison of the parameters of the DSD statistical model based on the rainfall intensity data obtained using different disdrometers, including PARSIVEL. The rainfall intensity was classified based on the rain rate into the following six categories: 1) 0.1-1 mm h −1 (very light rain); 2) 1-2 mm h −1 (light rain); 3) 2-5 mm h −1 (moderate rain); 4) 5-10 mm h −1 (heavy rain); 5) 10-20 mm h −1 (very heavy rain); and 6) > 20 mm h −1 (extreme rain). To compare the results using different equipment, the two-dimensional video disdrometer (2DVD) and Joss Waldrogel Disdrometer (JWD) observation data reported by Tokay et al. (2001) and POSS observation data reported by You et al. (2004) were used. Comparison of D m by rainfall intensity measured using different instruments (Fig. 7a) revealed similar trends in which D m increased with increasing rainfall intensity. The difference between the 2DVD and other instruments tended to decrease with increasing rainfall intensity, except that the 2DVD showed the highest D m value in very light rain. As observed using the POSS, a sudden decrease in extreme rainfall intensity occurred because of the attenuation effect of strong precipitation (Park et al. 2005;Kestwal et al. 2014). Figure 7b shows the μ values for different instruments. The μ value of POSS is relatively low and that of 2DVD is high, except for extreme rainfall intensity. The mean μ values of PARSIVEL, JWD, and 2DVD are similar. As shown in Fig. 7c, the mean log 10 N w value of 2DVD is high and that of POSS is low. The JWD and PARSIVEL showed similar log 10 N w values with a similar trend according to the change in rainfall intensity for all instruments, except for very light and light rain.

Case Studies for Estimating CMPs in Korea
The CMPs at site I (Daegawllyeong) in Korea were estimated using the DCPM based on the calculated DSD parameters (Fig. 8). To conveniently explain the CMPs used by Dolan et al. (2018), vapor deposition is represented as VD, weak convection as WC, collision and coalescence as CC, icebased and warm rain growth as ICWRG, ice-based growth as Table 2 Mean (standard deviation) rain drop size distribution (DSD) parameter values of D m , log 10 N w , and μ in East Asia; D m is a massweighted mean drop diameter, μ is a shape, and N w is a normalized intercept parameter of the DSD ICG, and aggregation and riming as AR. Figure 8 shows the results of applying D 0 and log 10 N w , observed in this study, to DCPM. The classification ranges of these values were approximated based on DCPM to understand the CMPs. As shown in Fig. 8, the convective precipitation is a more dominant mechanism than stratiform precipitation at site I in Korea. Estimates of the major CMPs at site I were 30.3% for AR, 23% for ICWRG, 12.7% for WC, 11.6% for 'mixed ICG and AR', and 8.5% for 'mixed ICWRG and ICG', as shown in Table 3. The VD were classified as having little effect at site I. Analysis of CMPs at site I showed that most of cold   cloud formation processes, such as AR and ICWRG, contributed to precipitation. In further studies, the classification of CMPs at site I should be improved using more objective and specific statistics methods, as the overlapping CMPs in Fig. 8 are unclear.

Case Studies for Estimating CMPs in East Asia
To estimate the CMPs at other sites in East Asia as case studies, we selected various sites and analyzed the DSD parameter data from previous DSD studies. We could not analyze all CMPs at the selected sites because some DSD parameter data from previous DSD studies were unavailable. Therefore, we approximated one major CMP for case studies using mean DSD parameters such as D 0 and log 10 N w for rainfall intensities reported previously. Figure 9 and Table 4 show the results of classification of the major CMPs based on rainfall intensity at sites A (Taoyuan) and B (Kaohsiung). At site A, the one major precipitation-generating CMP was a VD of < 1 mm h −1 (A1-A4) and 'mixed WC, CC, and ICWRG' of > 2 mm h −1 (A6-A12). Using the criteria described by Janapati et al. (2020), the major CMP at site B was ambiguous at < 1 mm h −1 (B1), as observed at site A. Thus, a large amount of water vapor leads to cloud growth and rain via VD of < 1 mm h −1 (A1-A4), and many complex CMPs affect the precipitation generation at sites A and B in humid subtropical climates. Figure 10 and Table 5 show the major CMPs by rainfall intensity in a subarctic climate in high-altitude regions, where the average elevations of sites C (Naqu) and E (Motuo) are 4,508 and 1,200 m, respectively. The major CMP in these alpine climates was caused by VD < 5 mm h −1 (C1-C3 and E1-E3) and by ICWRG > 5 mm h −1 (C4-C5 and E4-E5). At site D (Beijing, China), VD was the only major CMP at approximately ≤ 1 mm h −1 (D1-D2), and the ICWRG was  Fig. 9 Case study to estimate major cloud microphysical processes using the conception model described by Dolan et al. (2018) at sites A and B; VD (red), ambiguous (brown), and mixed WC + CC + ICWRG (blue). Diamonds and dots represent sites A and B, respectively. The ranges of the rain rate (mm h −1 ) at A1-A11 and B1-B6 are shown in Table 4 (VD: vapor deposition, WC: weak convection, CC: collision and coalescence, and ICWRG: ice-based and warm rain growth)  Fig. 10 Same as Fig. 9 except at sites C, D, and E; VD (red) and ICWRG (blue). Squares, diamonds, and dots represent sites C, D, and E, respectively. The ranges of the rain rate (mm h −1 ) at C1-C5, D1-D8, and E1-E5 are shown in Table 5 (VD: vapor deposition and ICWRG: ice-based and warm rain growth) dominant above this level (D3-D8). At sites C and E, VD was the only major CMP at < 5 mm h −1 (C1-C3 and E1-E3). Therefore, the VD is the most important CMP < 5 mm h −1 at sites C and E based on the presence of numerous supersaturated clouds in alpine climate regions (Hammer et al. 2013). Thus, alpine regions with higher elevation have lower the temperatures, where the air is likely to be oversaturated. Figure 11 and Table 6 show the major CMPs based on rainfall intensity at site F (Palau), a tropical rainforest region. Precipitation in this region is mainly formed through CMPs such as WC, CC, and ICWRG. The results suggest that WC and CC are the major precipitation-generating microphysical processes at site F because of the influence of the tropical climate. The major CMPs by rainfall intensity at site F were WC at 0.1-0.7 mm h −1 (F1-F3), 'mixed WC, CC, and ICWRG' at 0.7-5 mm h −1 (F4-F6), 'mixed WC and CC' at 5-18 mm h −1 (F7-F9), CC at 18-40 mm h −1 (F10-F11), and ICWRG at > 40 mm h −1 (F12). These results demonstrate that the collision-coalescence process is the primary mechanism responsible for producing precipitation and that heavy rain is mainly generated by ice-based and warm rain growth in climates with tropical rainfall. Figure 12 and Table 7 show the major CMPs based on rainfall intensity at site H (Seoul), G (Busan), and I (Daegawllyeong) in Korea. At mountainous site I in a warm summer continental climate, 'mixed ICG and ICWRG' was the major CMP at > 1 mm h −1 (I3-I5), whereas the CMP was ambiguous at < 1 mm h −1 (I1-I2). The megacity's site H, Seoul in Korea, in a hot summer continental climate case such as site D, Beijing in China, was mainly affected by VD at < 5 mm h −1 (H1-H4) and ICWRG at > 5 mm h −1 (H5-H8). The major CMP was ICWRG at > 5 mm h −1 (G2-G4) and ambiguous at < 5 mm h −1 (G1) at the harbor city's site G (Busan) in dry winter humid subtropical climates. In particular, the major CMP of megacities was vapor deposition at site D in Beijing (< 1 mm h −1 (D1-D2)) and site H  in Seoul (< 5 mm h −1 (H1-H4)) because the strong updraft of the urban heat island effect in megacities increased the LWC (Han et al. 2012), with many supersaturated clouds formed at higher altitudes.

Summary and Conclusion
We analyzed the DSD observation data using PARSIVEL in the mountainous regions of Korea in 2010 and 2011. We also determined the characteristics of DSD parameters based on rainfall intensity observed in previous DSD studies by comparing the PARSIVEL data used in this study with those obtained using other disdrometer equipment. We developed a simple method for understanding the CMPs of precipitation generation using a conceptual model of CMPs and in situ observation DSD data. We used observed DSD parameters and the CMP conceptual model of DSD observation-based microphysical processes suggested by Dolan et al. (2018). As case studies, we applied DSD observation data from Korea and East Asia to estimate the CMP. At sites A (Tauyuan) and B (Kaohsiung) in a humid subtropical climate region, the major CMPs were vapor deposition at less than a light rain rate; mixed weak convection, collision and coalescence; and ice-based and warm rain growth at more than moderate rain rates. At site F Palau) in a tropical rainforest climate region, the major CMPs were weak convection, collision and coalescence, ice-based and warm rain growth, and their mixed CMPs. Thus, a larger number of CMPs at site F (Palau) are involved in generating precipitation compared to those at other sites. At sites C (Naqu) and E(Motuo) in highaltitude regions with a subarctic climate, vapor deposition was the major CMP at < 5 mm h −1 because of the presence of many supersaturated clouds in alpine climate regions. Sites D (Beijing) and H (Seoul) have a hot summer continental climate and are megacity sites. The major CMPs at sites D (Beijing) and H (Seoul) were similar to those of the subarctic climate in high-altitude regions in East Asia, such as vapor deposition and ice-based and warm rain growth. Particularly, vapor deposition is the major CMP in megacities such as Beijing and Seoul because the strong updraft of the urban heat island effect in these regions increases the LWC, after which large numbers of supersaturated clouds form at higher altitudes. Ice-based growth is an important CMP at site I (Daegawllyeong), which is in a mountainous area. The major CMP at site G (Busan) is ice-based and warm rain growth, as a maritime cloud affects this site in dry winter humid subtropical climate regions.
Our case studies revealed that CMPs can be estimated using DSD observation data in different climate regions and can improve the classification of CMPs using new techniques such as machine and deep learning. In our further studies, we will apply the conceptual model of CMPs by Dolan et al. (2018) to vertical rain radar data, such as the Micro Rain Radar data, to calculate vertical DSD  Ambiguous   H1   H2  H3  H4  H5   H6   I1 I2  I4  I5  I6   G1   G2  G3  G4 I3 Fig. 12 Same as Fig. 9 except at sites G, H, and I; VD (red), mixed ICG + ICWRG (purple), ICWRG (blue), and ambiguous (brown). Dots, diamonds, and squares represent sites G, H, and I, respectively. The ranges of rain rate (mm h −1 ), G1-G4, H1-H6, and I1-I6, are shown in Table 7 (VD: vapor deposition, ICWRG: ice-based and warm rain growth, and ICG: ice-based growth) parameters. These calculated parameters will be compared with ground-based DSD observation data to evaluate the accuracy of the vertical DSD distribution, and the estimated cloud microphysical model data will be compared with S-band dual-polarization radar observation data in Korea. These studies may improve the estimation accuracy of radar-based precipitation and evaluation of precipitation-generating microphysical processes by using an objective numerical weather model.
Funding This research was supported by the "Research for Weather Modification and Cloud Physics" (KMA2018-00224) project of NIMS/ KMA.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.