Statistical characteristics of raindrop size distribution over Western Ghats of India: wet versus dry spells of Indian Summer Monsoon

Abstract. The nature of raindrop size distribution (DSD) is analyzed during wet and dry spells of the Indian Summer Monsoon (ISM) over Western Ghats (WGs) using Joss-Waldvogel Disdrometer (JWD) measurements. The observed DSDs are fitted with gamma distribution, and the characteristic DSDs are studied during the summer monsoon seasons (June–September) of 2012–2015. The DSD spectra show distinct diurnal variation during wet and dry spells. The dry spells exhibit a strong diurnal cycle with two peaks, while the diurnal cycle is not prominent in the wet spells. The observational results reveal the microphysical characteristics of warm rain during both the wet and dry spells. Even though the warm rain processes are dominant over WGs during monsoon, the underlying dynamical processes cause the differences in DSD characteristics during wet and dry spells. In addition, the differences in DSD spectra with different rain rates are also observed during the wet and dry spells. The DSD spectra are further analyzed by separating into stratiform and convective types. Finally, an empirical relation between slope parameter, Λ and shape parameter, μ is derived by best fitting the quadratic polynomial for the observed data during both wet and dry spells as well as for the stratiform and convective types of precipitation. The Λ–μ relations obtained in the present study are slightly different in comparison with the earlier studies.



Introduction
Western Ghats (WGs) is one of the heavy rainfall regions in India. WGs receives a large amount of rainfall (~6000 mm) during the Indian Summer Monsoon (ISM) period (Das et al., 2017, and references therein). Shallow clouds contribute significantly to the monsoon rainfall on the windward side (Kumar et al., 2013;Das et al., 2017;Utsav et al., 2017Utsav et al., , 2019 and deep convection in the leeward side (Utsav et al., 2017(Utsav et al., , 2019Maheskumar et al., 2014) of the WGs. In addition, thunderstorms also function during the monsoon season. They observed that bimodal and monomodal DSD during low and high rainfall rates, respectively. However, their study is limited to brightband and non-brightband conditions only. Harikumar (2016) studied the differences between DSD on the coastal (Kochi) and high altitude (Munnar) station located in the WGs region. He found for a given rain rate, more number of larger size drops are present at Munnar than at Kochi. Das et al. (2017)  They observed bimodal DSD in mid-altitude station and monomodal DSD in the high altitude station.
Their study also confined to stratiform rain only.
There are limited studies of DSDs exist in the WGs region by considering long-term dataset.
This work is the first study to analyze the DSD characteristics by considering the monsoon intra-100 seasonal oscillations (wet and dry spells). The present study brings out the results of a unique opportunity by analyzing a more extensive dataset and also considering the different phases of the monsoon intra-seasonal oscillations in the WGs. With this background, the current study attempted to address the following issues: 1. How do the DSD characteristics vary during wet and dry spells in the WGs region? addition to the above shortcoming, the JWD miscounts raindrops in the lower size bins, specifically for drop diameters below 1 mm (Tokay et al., 2003). An effort has been made to overcome this deficiency by discarding noisy measurements and applying the error correction matrix provided by the manufacturer. To reduce the sampling error arising due to insufficient drop counts at lower rain rates, the rain rates less than 0.1 mm hr -1 are discarded in the present study. During heavy rain, the JWD underestimates the number of smaller drops, known as disdrometer dead time. To account the aforementioned error in the JWD estimates, the rain rates during wet and dry spells are analyzed. It is 135 observed that ~85% (90%) of the rain rates lies below 8 mm hr -1 during wet (dry) spells (figure not shown). By using the noise-limit diagram of Joss and Gori (1976), Tokay et al. (2001) investigated the underestimation of small drops by JWD. They found that 50% of the drops below 0.4 mm cannot be detected by the JWD when the rainfall rate is above 20 mm hr -1 . In the present study, only 4% (1%) of the rain rates exceed 20 mm hr -1 during wet (dry) spells. Hence, the underestimation of small drops by integration period can contribute to numerical fluctuations in DSDs, whereas higher sampling time may 150 miscount actual physical deviations (Testud et al., 2001). Hence, in the present study, we have averaged the JWD measurements into 1 min period to filter out these deviations.
The concentration of raindrops, N(D) (mm -1 m -3 ) at an instant of time is (1) where A is the surface area of observation (50 cm 2 ), t is the integration time, ni is the number of (2) The rain rate (R) and reflectivity ( ) are estimated by assuming that the momentum is entirely due to the terminal fall velocity of the raindrops and the raindrops are spherical and assume Rayleigh The one-minute DSD measurements obtained from JWD are fitted with a three-parameter gamma distribution, as suggested by Ulbrich (1983). The details about the DSDs used in the present 165 study can be found in Das et al. (2017) and Krishna et al. (2017).
The functional form of the gamma distribution assumed for the DSD is expressed as where, N(D) is the number of drops per unit volume per unit size interval, N0 (in m -3 mm -(1+μ) ) is the number concentration parameter, D (in mm) is the drop diameter, D0 (in mm) is the median volume diameter, and µ (unitless) is the shape parameter (Ulbrich, 1983;Ulbrich and Atlas, 1984). The gamma DSD parameters are calculated using moments proposed by Cao and Zhang (2009). Here, 2 nd , 3 rd , and 4 th moments are utilized to estimate the Gamma parameters. This method gives relatively fewer errors 175 compared to other methods (Konwar et al., 2014). The 'n' order moment of the distribution can be calculated as The shape parameter, µ, and the slope parameter, Λ are given by The other parameters, normalized intercept parameter, Nw (in mm -1 m -3 ), mass-weighted mean diameter, Dm (in mm), and liquid water content, LWC (in gm m -3 ), are calculated following Bringi and Chandrasekar (2001).
Apart from JWD, the ERA-Interim (Dee et al., 2011) dataset is also used to understand the 190 dynamical properties responsible for different DSD characteristics during wet and dry spells. The ERA-Interim provides atmospheric data on 60 levels in the vertical from the surface to 0.1 hPa. The ERA-Interim data are available at 3-hourly and 6-hourly intervals. In the present study, temperature (K), and specific humidity (kg kg -1 ) at 700 hPa with a spatial resolution of 0.25 o × 0.25 o at 0000 UTC are considered during ISM of 2012-2015. The specific humidity at 700 hPa infers the amount of water 195 vapour available for the cloud formation over the study region, WGs.
The daily accumulated rainfall collected by the India Meteorological Department (IMD) rain gauge is used to identify the wet and dry spells of ISM. The IMD receives the rainfall accumulations at 08:30 LT (LT=UTC+05:30 hrs) every day. To examine the JWD data quality, the daily accumulated rainfall measured by the JWD is compared with the daily accumulated rainfall collected from the rain gauge. For comparison, JWD rainfall data accumulated at 08:30 LT is calculated for all the days during the monsoon season of 2015. The daily accumulated rainfall collected by rain gauge and JWD above 1 mm is considered for the comparison. A total of 76 days of data is utilized. The non-availability of data for this period may occur either due to maintenance activity or due to non-rainy days. Figure 1 shows the scattered plot of daily accumulated rainfall between JWD and rain gauge. A linear fit is carried out 205 to the scatter plot and is displayed with the grey line in the figure. The correlation coefficient is about 0.99 between the two measurements despite their diverge physical and sampling characteristics. The bias in JWD measured rainfall is about -0.7 mm, and root mean square error is about 2.9 mm. These results suggest that the JWD measurements can be utilized to understand the DSD characteristics during the wet and dry spells in the WGs region.

Identification of wet and dry spells
In the present study, an objective methodology proposed by Pai et al. (2014) is used to identify the wet and dry spells. The IMD generated high-resolution gridded rainfall data using a rain gauge network over the Indian region.
These standardized anomaly time series are used to separate the wet and dry spells. A period in 225 this standardized anomaly time series is marked as wet (dry) if the standardized anomaly exceeded a value of 0.5 (-0.5) for consecutive three days or more (Utsav et al., 2019). Figure 2 shows the standardized rainfall anomalies calculated using eq. (13). Table 1 shows the number of wet and dry days during the study period. It is observed that there is more number of dry days during 2012-2015 monsoon seasons, and July has comparatively more number of wet days. In this work, 44,640 (149,760)

DSD overview-Topographic perspective:
The single point-wise instrument is not sufficient to address the orographic impacts on DSD characteristics. One of the difficulties in studying the effect of orography on DSD properties is the (2016) also found that the GPM underestimates (overestimates) the mean Dm (Nw) during the southwest and northeast monsoons over Gadanki, a semiarid region of India. They showed that the single-255 frequency algorithm underestimates the mean Dm by ~0.1 mm below 8 mm hr -1 , and the underestimation is a little higher at higher rain rates. Whereas in the dual-frequency algorithm, the mean Dm is nearly the same below 8 mm hr -1 but underestimates (~0.1 mm) at higher rain rates. Further, the underestimation is very small for Dm values below 1.5 mm. In the present study, most of the Dm values present below 1.5 mm. Hence, it is reasonable to consider the GPM measurements to have an overview 260 of DSD characteristics over the WGs.
Three locations are selected to understand the rain microphysical processes at different topographic regions in WGs. These locations are the ocean, high altitude cloud physics laboratory (HACPL; located on the windward slope of the WGs), and leeward side of the WGs. The DSD differences in these three sites can partially infer the effect of orography on DSD. Figure 3 shows the distribution of Dm over the ocean, windward, and leeward sides of the WGs. In this plot, the box represents the data between first and third quartiles, and the whiskers show the data from 12.5 and 87.

Results and Discussion
The DSD and rain integral parameters during the wet and dry spells are examined in terms of with diameters less than 1 mm are considered as small drops, with diameters in the range 1-4 mm are regarded as mid-size drops and with diameters above 4 mm are considered as large drops.

Raindrop size distribution during wet and dry spells
The information on the background microphysical processes, which are responsible for 285 precipitation formation in convective and stratiform systems, could be inferred from observed variations in the DSDs at the ground. Figure 4 shows the temporal evolution of normalized raindrop concentration during wet and dry spells, exhibiting distinct diurnal features. The concentration of smaller drops ( Figure 4a) is higher during the dry periods. The higher concentration of small drops in dry spells indicates the predominance of orographic convection over WGs. In the mountain regions, DSDs 290 evolved through warm/shallow rain processes. This warm rain is produced when the upslope wind is stronger, and moisture availability is high (White et al., 2003). In such a situation, the strong orographic wind enhances the growth of cloud droplets via condensation, collision, and coalescence (Konwar et al., 2014). Further, a large number of small raindrops during the dry spells indicate that the breakup and evaporation processes may be more efficient during the dry periods. In the smaller drop spectra, dry 295 spells exhibit a strong diurnal cycle with a primary maximum in the afternoon hours (1500-1900 LT) and a secondary peak in the night time (2300-0500 LT). This diurnal feature is also noted by Utsav et al. (2019) in the 15-dBZ echo top height (ETH) from X-band radar observations during the dry spells.
However, such a diurnal cycle is not present in smaller drops during the wet spells. These smaller drops show a little higher concentration during morning hours (0500-0700 LT), representing the oceanic In the mid-size drops (Figure 4b), the concentration is higher in wet spells compared to dry spells. The higher concentration of mid-size drops during the wet spells are due to the collisioncoalescence process (Rosenfeld and Ulbrich, 2003), and accretion of cloud water by raindrops (Zhang et al., 2008). This result indicates that the congestus clouds are omnipresent during the wet spells. Further,

305
in the mid-size drops, both the spells exhibit a diurnal cycle; however, their strengths are different. The wet spells exhibit two broad maxima, one in the late afternoon (1400-1900 LT) and the other in the early morning (0500-0700 LT) times. The dry spells also show two maxima, one in the late afternoon (1400-1900 LT) as in the wet periods, and the other in the night time (2300-0500 LT). Such a diurnal cycle is also observed in rainfall features over WGs (Shige et al., 2017;Romatschke and Houze, 2011).

310
Shige et al. (2017) found a continuous rainfall with a double-peak structure of nocturnal and afternoonevening maxima in the WGs region. Romatschke and Houze (2011) observed a double peak rainfall pattern in the WGs region. They proposed that the morning peak is related to oceanic convection while the afternoon peak is associated with the continental convection. Figure 5 shows the mean DSDs during wet and dry spells along with the seasonal mean DSD for 315 the study period. Here, N(D) is plotted on a logarithmic scale to accommodate its large variability. In general, the DSDs during the dry spells are narrower than the DSDs during the wet periods. The mean DSDs are concave downward during both the spells. The mean concentration of smaller drops (< 0.9 mm) is higher, and the mean concentration of medium and larger drops is lower in dry periods. An increased concentration in smaller drops and a decrease in medium and larger drops concentration is 320 found in the dry spells compared to the seasonal mean concentration. This indicates the collision and breakup processes, as described by Rosenfeld and Ulbrich (2003) and Konwar et al. (2014). In contrast, low concentrations of smaller drops and an increase in number concentration of drops above 0.9 mm diameter are observed in the wet spells.
To study the differences in DSD during the wet and dry spells with rain rate, the distribution of 325 N(D) is compared at different rain rates, as shown in Figure 6. Here N(D) is plotted on a logarithmic scale. It is evident from this figure that significant differences exist in N(D) from wet to dry spells. The contours are shifted to higher rain rates and higher diameters in the wet spells. It indicates that the midsize drops in the range 1-2 mm are higher in wet spells than in dry spells for the same rain rate. This result is more pronounced in lower rain rates below 10 mm hr -1 . Further, the concentration of raindrops 330 in the range 1-2 mm increases as the rain rate increases between 5-15 mm hr -1 during the wet periods.
At higher rain rates (above 10 mm hr -1 ), the smaller and mid-size drops are higher in the wet spells than in the dry periods. However, this difference decreases gradually as rain rate increases. At above 30 mm hr -1 , both the periods show a similar distribution of N(D) (not shown in the figure). However, in the larger drop diameters above 4.5 mm, the concentration is higher in the wet spells compared to the dry 335 periods in all rain rate intervals (not shown in the figure). The distribution of Dm is broader in the dry spells. The Dm value varies from 0.42 to 4.8 mm, with the maximum occurrence at ~1.2 mm during the wet periods, whereas it ranges from 0.4 to 5 mm, with the 340 maximum appearance at ~0.8 mm during the dry spells. For Dm values < 1 mm, the distribution for the dry spells is higher than for the wet spells. This finding indicates the predominance of smaller drops during the dry spells. The mean value of Dm, along with the standard deviation and skewness, are provided in Table 2. The mean value of Dm is 1.3 mm, and its standard deviation is 0.38 during the wet spells, whereas the mean Dm is 0.9 mm, and its standard deviation is 0.37 during the dry spells. A relatively large number of small drops reduce the Dm value in the dry spells, while the presence of fewer smaller drops and relatively more mid-size drops increases the Dm value in the wet periods. The histograms of log10(Nw) are negatively skewed during both wet and dry spells (Figure 7b). The log10 (Nw) shows an inverse relation with Dm and is varied from 0.52 to 5.11 during the wet spells and 0.50 to 5.43 during the dry periods. The histogram of log10(Nw) peak at 3.9 during the wet periods. The histograms

350
of log10(Nw) shows a bimodal distribution during the dry spells. This bimodal distribution of log10(Nw) peaks at 3.9 and 5. This finding is consistent with the results of Utsav et al. (2019). They analyzed the 0 dBZ echo top heights, which represent the cloud top heights during wet and dry spells. They observed a bi-modal distribution in 0 dBZ echo top height, which peaks at 3 km and 6.5 km during the dry periods. However, in the present study, 10 consecutive 1 min DSD samples are considered to classify the rainfall 375 as stratiform and convective. If the mean rain rate of 10 successive DSD samples is greater than 0.5 mm hr -1 , and if the standard deviation of 10 consecutive DSD samples is less than 1.5 mm hr -1 , then the precipitation is classified as stratiform; otherwise, it is classified as convective. Figure 8 presents the histograms of Dm, log10(Nw), Ʌ, and µ during stratiform rain events in wet and dry spells. The mean, standard deviation, and skewness of these parameters are provided in Table 3.   Table 4.
Several points can be noted from the above discussion: a. The maximum value for mean Dm and the largest standard deviation is found for convective rain in wet spells.
b. The maximum value for log10(Nw) and higher standard deviation are observed during stratiform rain in dry spells. c. A considerable difference is found in the histograms of Dm and log10(Nw) during the stratiform rain in dry and wet periods. However, this difference is small in convective rain.
d. In histograms of Ʌ and µ, the distinct differences exist in stratiform rain during wet and dry spells.

415
The above results indicate that the rainfall over WGs is associated with warm rain processes during both wet and dry spells. The microphysical processes in warm rain include rain evaporation, accretion of cloud water by raindrops and rain sedimentation (Zhang et al., 2008). Giangrande et al.  Figure 10 shows the mean specific humidity (kg kg -1 ) and temperature anomalies (K) at 700 hPa derived from the ERA-Interim reanalysis dataset. In this plot, the colour bar represents the mean specific humidity, and the contours represent the temperature anomalies. This level is chosen, as the temperature anomaly and the availability of moisture at this level aid the growth of active convection. It is observed that the temperature is cooler over the west coast of India (including the study region) in the wet spells compared to that in the dry periods. Further, the mean specific humidity is higher over WGs during the wet periods. The thermal gradient between WGs and surrounding regions and the availability of more moisture favours the growth of active convection in the wet spells. It is known that the vertical velocity during the wet periods is stronger compared to the dry spells (Uma et al., 2012). The strong updrafts aid the growth of cloud liquid water particles and thereby 435 increase the size of the drops. Whereas, positive temperature anomalies in the dry spell can lead to the evaporation of raindrops, which subsequently can break the drops, thereby leading to lesser diameter drops in the dry spell.
The diurnal variation in mean rain rate during wet and dry spells is shown in Figure 11. The mean rain rate is higher during wet periods throughout the day. The relatively lower rain rates are due to 440 the presence of a higher concentration of smaller drops during the dry spells. The diurnal variation in rain rate shows bi-modal distribution during both wet and dry spells. The primary maximum is in the afternoon hours and the secondary maximum present during morning hours. The raindrop concentration increases monotonically (Fig. 4), with an increase in rain rate for all the drop sizes during the dry spells.
This finding indicates that the increase in rain rate is responsible for the rise in both concentration and 445 raindrop size during the dry spells. However, in the wet periods, the concentration of smaller drops is constant throughout the day, and the increase in rain rate is due to the rise in concentration and size of mid-size raindrops. This further indicates that the collision and coalescence processes as well as deposition of water vapour on to the cloud drops, which are responsible for the increase in the concentration (afternoon and early morning hours) of mid-size raindrops during the wet spells. In addition, the raindrop diameter depends on the rain rate, which varies between wet and dry spells. The distribution of Dm during wet and dry spells at different rain rates are shown in Figure 12. For lower rain rates (below 10 mm hr -1 ), the raindrops falling from the cloud tops can grow by deposition of water vapour and accretion of cloud water during the wet spells. This result in larger Dm values during the wet spells compared to dry spells. At higher rain rates (above 20 mm hr -1 ), the Dm distribution remains the 455 same during both the spells. This is due to the equilibrium of DSD by the collision, coalescence, and breakup mechanisms, as described in Hu and Srivastava (1995) and Atlas and Ulbrich (2000). The above analysis indicates that the dynamical mechanisms are different during wet and dry spells, resulting in different DSD characteristics.

460
The gamma distribution function has been widely used in the microphysical parameterization schemes in the atmospheric models to describe various DSDs. However, µ is often considered to be constant. Milbrandt and Yau (2005) found that µ plays a vital role in determining sedimentation and microphysical growth rates. In this context, the microphysical properties of clouds and precipitation are sensitive to variations in µ. Several researchers showed that the value of µ varies during the 465 precipitation (Ulbrich, 1983;Ulbrich and Atlas, 1998;Testud et al., 2001;Zhang et al., 2001;Islam et al., 2012). Zhang et al. (2003) proposed an empirical µ-Ʌ relationship using 2DVD data collected in Florida. They examined the µ-Ʌ relation with different types of precipitation. These µ-Λ relations are useful in reducing the bias in rain parameters from remote sensing measurements (Zhang et al., 2003).
Recent studies have demonstrated the variability in µ-Ʌ relation in different types of rain and at various 470 geographical locations (Chang et al., 2009;Kumar et al., 2011;Wen et al., 2016). Hence, it is necessary to derive different µ-Ʌ relations based on local DSD observations, in particular, over the WGs.
In the present study, an empirical µ-Ʌ relationship is derived for both wet and dry spells. To minimize the sampling errors, the DSDs with a rainfall rate of less than 5 mm hr -1 are excluded. In addition, the total drop counts above 1000 are only considered in the analysis, as proposed by Zhang et 475 al. (2003). Figure 13 shows the µ-Ʌ relation for wet and dry spells, and the corresponding polynomial least-square fits are shown as solid lines. The fitted µ-Ʌ relations for wet and dry spells are given as follows:

480
Similar behaviour is observed for both wet and dry spells, the smaller the value of Ʌ (higher rain rates), smaller is the value of µ. Thus, the DSDs tend to be more concave downwards with the increase in rainfall intensity. This finding suggests a higher fraction of small and mid-size drops and a lower fraction of larger drops, reflecting less evaporation of smaller drops and more drop breakup processes.
However, the fitted µ-Ʌ relation exhibits a large difference for wet and dry spells. Comparing Eq. (14) 485 and (15), one can observe that the coefficient of the linear term is smaller in wet spells than that of dry spells. Hence, for a given value of µ, the dry spells have a higher value of Ʌ compared to the wet spells.
Further, the Dm value is higher during wet spells compared to dry spell for the given rainfall rate due to different microphysical mechanisms as discussed above (Fig. 12). This leads to higher µ values in wet spells compared to dry spells. This result suggests that different microphysical mechanisms during wet 490 and dry spells lead to different µ-Ʌ relations. Hence, it is apparent that the single µ-Ʌ relation cannot reliably represent the observed phenomenon during different phases of the monsoon.
Comparing the µ-Ʌ relations in this study with that obtained from Zhang et al. (2003), the µ-Ʌ relationship of the dry spell has a smaller slope. These differences reveal that the DSD during dry spell have lower values of Dm. This indicates that the underlying microphysical processes in the orographic 495 precipitating systems are different from those observed over Florida in 1998 summer. Further, the µ-Ʌ relationships are derived for convective and stratiform rain for the JWD measurements and are provided in Figure 14. The least-square polynomial fit for convective and stratiform rain is as follows: Convective rain: = 0.0069 2 + 0.576 + 2.42 Stratiform rain: = 0.0022 2 + 0.933 + 1.86 (17)

500
It is observed that the coefficients of the squared and linear term of convective precipitation are smaller than those given by Zhang et al. (2003). Hence, for a given value of µ, the convective precipitation in the present study gives lower values of Λ than that for the convective precipitation from and different types of instruments. To explore the plausible effect of rainfall rate, the µ-Ʌ relations are compared with the previous studies for rain rates below 5 mm hr -1 , and above 5 mm hr -1 (figure not shown). It is observed that, when the rain rates are below 5 mm hr -1 , the shape parameter shows bimodal distribution (above µ=10), especially in the wet spells. In this rain rate region, the first distribution (with lower µ values) is comparable with Chu and Su (2008), and Zhang et al. (2003), whereas the other 515 distribution (with high µ values) is comparable with Seela et al. (2018). Chu and Su (2008) derived the µ-Ʌ relations for rain rates above 1 mm hr -1 , as well as rain rates below 5 mm hr -1 . Hence, the observed differences in µ-Ʌ relation with Chu and Su (2008) could be attributed to the difference in the rain rates.
The second distribution is similar to that observed in the rain rates above 5 mm hr -1 . The slope of the µ-Ʌ relation is higher compared to Chu and Su (2008), and Zhang et al. (2003) in the rain rates above 5 520 mm hr -1 . This result indicates that the wet and dry spells have higher µ values compared to the previous studies for the same Λ values. This represents that, the underlying microphysical processes are different over the complex orographic region, WGs. It can be observed that the Dm values in the present study are higher compared to the previous studies (e.g., Seela et al., 2018). The different Dm distributions lead to different µ values as (Ulbrich, 1983): Thus, the relatively higher values of Dm could contribute to higher values of µ for the same Λ values in the present study. Hence, the differences in the µ-Ʌ relations with previous studies may be related to different microphysical processes (such as collision-coalescence, breakup, etc.) occurring in the rainfall over WGs. In addition, Zhang et al. (2003), Chu and Su (2008) used the 2DVD measurements, whereas, in the present study, JWD data are utilized. The different instruments can have different sensitivities, which can also affect µ-Ʌ relations. The µ-Ʌ relationships derived for the present study are compared with the other orographic precipitations and are provided in Table 5. It is clear that µ-Ʌ relations vary in different types of rainfall and climatic regimes. ii. The DSD over WGs shows distinct diurnal features. The diurnal variation shows that the concentration of smaller drops is higher in dry spells, while the concentration of mid-size drops is higher in wet spells throughout the day.

545
iii. The dry spells exhibit a strong diurnal cycle with double-peak during late afternoon and night time in both smaller and mid-size drops. Whereas, this diurnal cycle is weak for smaller drops in wet spells. iv.
The higher concentration of mid-size and larger drops is observed in wet spells compared to dry spells. The thermal gradient between WGs and surrounding regions, higher availability of water 550 vapour, and strong vertical winds favours the formation of cumulus congestus, which are responsible for the presence of medium size/larger drops during wet spells. vii. An empirical relation is derived between µ and Ʌ during wet and dry spells. The fitted µ-Ʌ relationship for both spells exhibits a significant difference between them. The different microphysical mechanisms lead to different µ-Ʌ relations during wet and dry spells. It is evident from this study that, even though the warm rain is predominant, the dynamical mechanisms underlying the microphysical processes are different, which causes the difference in 565 observed DSD characteristics during wet and dry spells. The distinct features of DSD during the wet and dry spells of the ISM over WGs are summarized in Figure 15.

Author contributions:
UVMK and SKD designed, analyzed, and prepared the manuscript. SKD, UVMK, and UB              represents the specific humidity, and contours represent temperature anomalies. The positive anomaly represents heating, and negative anomaly represents cooling. The black dot represents the observational site.

Fig 11:
Diurnal variation of mean rain rate (mm hr -1 ) during wet and dry spells.