Microphysical Characteristics of Precipitation during Pre-monsoon, Monsoon, and Post-monsoon Periods

HTML

--> --> -->

1. Introduction

As one of the most fundamental microphysical features of precipitation, the raindrop size distribution (RSD) is of great significance to exploring microphysical processes (Hu and Srivastava, 1995; Rosenfeld and Ulbrich, 2003; Niu et al., 2010; Konwar et al., 2014; Zhang et al., 2017; Li et al., 2018), evaluating cloud water conditions (Wu and Liu, 2017), improving the accuracy of precipitation estimation from radar and satellite data (Yuter and Houze, 1997; Lee and Zawadzki, 2005; Chapon et al., 2008; Liao et al., 2014; Huang et al., 2018), and optimizing microphysics parameterization schemes in numerical models (Milbrandt and Yau, 2005; Williams et al., 2014; McFarquhar et al., 2015).
The broad applications of the RSD necessitates its widespread investigation around the world (Sauvageot et al., 1995; Bringi et al., 2003; Radhakrishna et al., 2009; Tenório et al., 2012; Jayalakshmi and Reddy, 2014; Gatlin et al., 2015; Suh et al., 2016; Wen et al., 2017; Seela et al., 2018; Wen et al., 2019; Zhang et al., 2019). The RSD characteristics over a wide range of continental locations have been extensively studied for many years. In recent years, there has also been an increasing interest in studying the RSD of marine precipitation (Ushiyama et al., 2009; Gatlin et al., 2015; Thompson et al., 2015; Krishna et al., 2016; Seela et al., 2017). The microphysical characteristics of clouds over the ocean and land are quite different from each other in terms of cloud droplets, updraft velocities, and subcloud processes, which may result in significant differences in the RSD characteristics (Rosenfeld and Ulbrich, 2003). Based on both direct disdrometer observations and indirect dual-polarized radar retrievals from a wide range of climatic regimes, Bringi et al. (2003) identified two categories—namely, a “maritime-like” cluster and a “continental-like” cluster—for convective rainfall. The “maritime-like” cluster is characterized by a high concentration of small drops, while the “continental-like” cluster is characterized by a low concentration of large drops. Ulbrich and Atlas (2007) demonstrated a higher median volume diameter (D₀) but lower normalized intercept parameter (N_w) in continental storms than in maritime storms. Tenório et al. (2012) indicated that small raindrops (D ≤ 2 mm) are more dominant in maritime rain than in continental rain in northern Brazil. A comparison study of RSD between the equatorial Indian (Gan Island) and West Pacific (Manus Island) oceans was performed by Thompson et al. (2015), and they reported higher N_w values and more small-to-medium drop diameters at open ocean locations than at continental locations. Recently, the RSD characteristics of rainfall in Taiwan and Palau were analyzed by Seela et al. (2017). They found higher D_m but lower lgN_w values in Taiwan than in Palau for rainfall with the same rain rate.
The Asian monsoon has long been known to play significant roles in the global climate system, the global tropical seasonal circulations, and interannual changes (Webster et al., 1998). The South Asian (Indian) monsoon and the East Asian monsoon are two subsystems of the Asian monsoon, both of which are closely linked but have their own characteristics. The RSD varies not only between land and sea, but also between different monsoon regions (Kouz et al., 2006). In the past several years, the characteristics of clouds and precipitation during different onset periods of the South Asian monsoon have been extensively studied at several Indian stations (Rao et al., 2001; Reddy and Kozu, 2003; Chakravarty et al., 2013; Konwar et al., 2014; Das et al., 2017). For instance, Reddy and Kozu (2003) reported that raindrops in southwestern monsoon precipitation are usually larger than those in northeastern monsoon precipitation at Gadanki. Radhakrishna et al. (2009) found higher concentrations of small drops in northeastern monsoon than that in southwestern monsoon. Moreover, the RSDs in three different periods, i.e., the pre-monsoon, monsoon, and post-monsoon periods, have also been studied using data measured at two stations in India (Chakravarty and Raj, 2013; Chakravarty et al., 2013). The results showed that, for a given rain rate, large raindrops are dominant in the post-monsoon season, but small- and medium-sized drops are more dominant in the monsoon season at Pune. However, at Kolkata, large raindrops are dominant in the pre-monsoon season, whereas small drops are more dominant in the monsoon season.
The onset and movement of the East Asian summer monsoon usually leads to the establishment of a quasi-stationary subtropical front, i.e., the mei-yu (or baiu) front, in East Asia. The mei-yu front brings persistent heavy rainfall along the Yangtze?Huaihe River basin and over the southern part of Japan. Although the RSD characteristics during the baiu period have been previously studied in Japan (Bringi et al., 2006; Oue et al., 2010), few studies have focused on the RSD during the mei-yu season in China. For instance, Chen et al. (2013) revealed that lower concentrations of raindrops with a high median diameter of precipitation can be observed in China than in Japan. Wen et al. (2016) noticed significant variation in the RSD before, during, and after the mei-yu season. They demonstrated that the average D_m decreased from the pre- to the post-mei-yu period, while the N_w value of precipitation in the post-mei-yu period was larger than that in the other two periods. More recently, Wen et al. (2019) investigated the seasonal variations of RSD in East China. They observed a relatively larger raindrop diameter and the highest concentration of raindrops during the summer season than the other three seasons.
The South China Sea (SCS) summer monsoon is the link between the South Asian summer monsoon system and the East Asian monsoon. The establishment of the SCS summer monsoon is anterior to the South Asian and East Asian monsoon, which is also an indicator of the northward movement of the Asian monsoon (Chen and Chen, 1995; Lau and Yang, 1997; Jiang and Qian, 2000). The characteristics and mechanisms of the SCS summer monsoon eruption and its influence on the East Asian rain band have been studied comprehensively since the South China Sea Monsoon Experiment in 1998 (Wang et al., 2001; Lin et al., 2004; Wang and Qian, 2009). The thermodynamic environment of the SCS is quite different before and after the onset of the SCS monsoon, which may lead to considerable variations in the spatial and temporal characteristics of cloud and precipitation microphysics (Chang and Chen, 1995; Li et al., 2009, 2013). Jiang and Qian (2000) demonstrated that the rainy season in the SCS mainly occurs in the north and central parts of 10°N. Chen et al. (2005) noticed both a spatial and seasonal variation of rainfall distribution in the SCS. Furthermore, the specific area of precipitation before and after the SCS summer monsoon were also found to be very different (Song et al., 2010). With the onset of the monsoon period, convective rain became deeper and the bright band altitude of stratiform cloud became higher (Song et al., 2010). However, due to the lack of observational datasets, up to now, little is known about the characteristics of RSD during the different periods of the SCS monsoon season over the SCS.
Based on the onboard Particle Size Velocity (Parsivel) disdrometer dataset collected during three marine surveys conducted in 2012, 2014, and 2016, we investigate the characteristics of the RSD during pre-monsoon, monsoon and post-monsoon periods over the SCS in this paper. Brief descriptions of the instruments, data and methods used in the present study are provided in section 2. Section 3 presents the properties of RSD in different monsoon periods. As an application of the RSD research, several relations used in quantitative precipitation estimation (QPE) are also reported in section 4. Possible mechanisms for the seasonal variability of RSD are discussed in section 5. The major conclusions are summarized in section 6.

2. Materials and methods

2.1. Datasets

-->

2.1. Datasets

The RSD dataset presented in this study was collected through a Parsivel laser particle spectrometer manufactured by OTT Messtechnik, Germany (L?ffler-Mang and Joss, 2000). This instrument has been proven to be a rational scientific instrument for measuring the general strength of precipitation whilst simultaneously measuring the fall speed and size of precipitation particles (Tokay et al., 2013). In 2011, OTT launched a new generation of Parsivel instruments—namely, Parsivel². With a new laser sensor, the OTT Parsivel² achieves more accurate measurements of raindrop size and rainfall velocity, far exceeding the original version (Tokay et al., 2014). Figure 1a shows the installation of the instrument. The locations of the observational rainfall events used to construct the raindrop dataset are depicted with three rectangles in Fig. 1b. The first period of observation was performed in region 01 (12°?19°N, 111°?114°E) from 1 September to 31 October 2012; the second period of observation was performed in region 02 (20°?22°N, 113°?120°E) from 3 May to 20 July 2014; and the third period of observation was in region 03 (18°?21°N, 112°?116°E) from 1 April to 20 June 2016.

Figure1. (a) Field view of the Parsivel instrument installed on the roof of the boat. (b) Locations of the rainfall events used to construct the raindrop dataset. The superimposed rectangles in (b) correspond to region 01 (12°?19°N, 111°?114°E), region 02 (20°?22°N, 113°?120°E), and region 03 (18°?21°N, 112°?116°E).

Along with the disdrometer observations, sounding observations (Vaisala RS92-GDP) were also conducted twice daily [0800 LST (Local Standard Time) and 2000 LST] during the voyages. Additionally, we also used FY-2E satellite data downloaded from the China National Satellite Meteorological Center (NSMC), reanalysis data from NCEP?NCAR, and cloud data from SPRINTARS (Spectral Radiation-Transport Model for Aerosol Species) (Takemura et al., 2005). FY-2E is a geostationary meteorological satellite located approximately 35 800 km above the equator at a longitude of 105°E and is operated by the China Meteorological Administration. Black body temperature (TBB) data products of FY-2E available at a spatial resolution of 0.1° × 0.1° and a 1-h temporal resolution were used. The NCEP?NCAR reanalysis data are recorded four times a day at a spatial resolution of 1° × 1°. Based on these datasets, this study attempts to understand the dynamic and thermodynamic characteristics of clouds associated with the rainfall over the SCS.
Based on the SCS monsoon index published by the China National Climate Center (https://cmdp.ncc-cma.net/Monitoring/monsoon.php), the data were arranged into three different periods—namely, the pre-monsoon (March to May), monsoon (June to September), and post-monsoon (October to December) periods. In the pre-monsoon period, the direction of the zonal wind over the SCS starts to change and the water content also increases. Accompanying the onset of summer monsoon over the SCS, the warm and wet southwesterly flow in the low latitudes strengthens and the convective activity intensifies remarkably over the SCS. During the post-monsoon period, the dry and cold air from the north moves southward to the coastal areas of South China and the northern part of the SCS, resulting in significant changes in the thermal properties over the SCS. These thermodynamic differences are expected to further affect the microphysical characteristics of precipitation.

2

2.2. Data-quality issues

-->

2.2. Data-quality issues

Some measurement error sources, such as strong winds, fall margins, and splashing effects, can induce the misclassification of raindrops by the Parsivel disdrometer. Therefore, quality control procedures were applied to the initial raindrop spectrum data. First, samples with one-minute raindrop values less than 10 or a rain rate (R) value less than 0.1 mm h^?1 were considered to be instrument noise and excluded from the samples (Tokay et al., 2013). Second, raindrops with diameters greater than 6 mm or fall speeds 60% greater than or less than the Beard (Beard, 1976) empirical speed?diameter relationship for rain were eliminated. Third, the effective sampling cross section of the Parsivel was defined as
where D_i is the ith size, L represents the beam length (180 mm) and W represents the width (30 mm). In this paper, the sampling area is adjusted according to Eq. (1) when calculating the RSD quantities containing S.

2

2.3. Methods

-->

2.3. Methods

The Parsivel disdrometer only provides the number of drops (n_i,j) in the ith size (D_i) and jth velocity (V_j) bins. The number concentration of raindrops per unit volume per unit size interval N(D_i) (m^?3 mm^?1) can be calculated by
where ΔD_i is the width of the ith size bin and V_j is the measured velocity of the jth velocity bin. Δt is the sampling duration, which was set to 1 min in this study.
Several bulk parameters related to the RSD, including the radar reflectivity factor Z (mm⁶ m^?3), the rain rate R (mm h^?1), and the total concentration of raindrops N_t (m^?3), are usually utilized to describe the microphysical characteristics of precipitation, and these parameters can be calculated by the following equations:
The first two size bins were not used because of their low signal-to-noise ratios. Characteristic diameters, including the mean diameter D_a (mm), the median volume diameter D₀ (mm), and the mass-weighted mean diameter D_m (mm) were also used to describe the properties of RSD, and these parameters can be expressed by:
The normalized intercept parameter N_w (mm^?1 m^?3), which is related to the rainwater content W and D_m, can be calculated as:
where ρ_w is the density of water (1 g cm^?3). The observed RSD was fitted using the gamma function (Ulbrich, 1983), which can be expressed as
where N₀ (m^?3 m^{?1 ? μ}), μ and λ (mm^?1) represent the intercept parameter, the shape factor, and the slope parameter, respectively. The moment method with the second, fourth, and sixth moments was chosen to calculate these three parameters (Cao et al., 2010).
The nth-order moment is defined as:
where $ \eta = M_4^2/\left( {{M_2}{M_6}} \right)$

3. Results

3.1. Average characteristics of microphysical rainfall parameters

-->

3.1. Average characteristics of microphysical rainfall parameters

Table 1 shows several averaged RSD parameters for the three periods, including the rain rate R, radar reflectivity Z, average diameter D_a, mass-weighted mean diameter D_m, median diameter D₀, average maximum diameter D_max, total number concentration of raindrops N_t, and normalized intercept parameter N_w. In addition, the contribution of raindrops with diameters less than 1 mm to N_t (N₁ / N_t), R (R₁ / R) and Z (Z₁ / Z) is also given, which can be expressed by:

Period	No. samples (min^?1)	R (mm h^?1)	Z (dBZ)	D_a (mm)	D_m (mm)	D₀ (mm)	D_max (mm)	N_t (m^?3)	lg(N_w) (mm^?1 m^?3)	N₁ / N_t (%)	R₁ / R (%)	Z₁ / Z (%)
PRM	220	5.42	32.31	0.89	1.40	0.93	2.40	904.29	3.80	81.11	7.86	1.44
MON	293	12.25	38.61	0.92	1.75	1.47	3.16	834.67	3.72	71.93	5.12	0.65
POM	282	4.86	32.27	0.84	1.45	1.07	2.54	631.02	3.66	75.65	8.53	0.94
All	795	8.39	35.21	0.88	1.55	1.23	2.82	754.19	3.71	76.74	6.29	0.83

Table1. Averaged microphysical parameters of RSD for the pre-monsoon (PRM), monsoon (MON), and post-monsoon (POM) periods. These parameters include the rain rate R, radar reflectivity Z, average diameter D_a, mass-weighted mean diameter D_m, median diameter D₀, average maximum diameter D_max, total number concentration of raindrops N_t, and normalized intercept parameter lg(N_w).

According to Table 1, the total average number concentration of raindrops and the percentage of raindrops with diameters less than 1 mm in the pre-monsoon period is 904.29 m^?3 and 81.11%, respectively. This suggests that small raindrops (D ≤ 1 mm) are prevalent in the pre-monsoon period. The average diameter D_a, mass-weighted mean diameter D_m, and average maximum diameter D_max in the monsoon period are the largest in the three periods. Thus, medium- and large-sized drops are expected to be prevalent during the monsoon period. This feature can also be deduced from the smallest contribution of raindrops with diameters less than 1 mm in the monsoon period. The post-monsoon period has the lowest average drop concentration N_t and the smallest average diameter D_a. This indicates the significant contribution of small raindrops to precipitation in this period, which can also be deduced from the largest R₁ / R ratio (8.53%). In addition, the larger mass-weighted mean diameter D_m and median diameter D₀ in the post-monsoon period than those in the pre-monsoon period suggests that there may also be some big raindrops in this period.
The median volume diameter D₀ in the monsoon period is 1.47 mm, which is very close to the value of 1.5 mm measured during the Australian monsoon season as reported by Penide et al. (2013). This similarity may be related to the influence of both of them by the warm and humid air currents from the ocean. Compared with the precipitation of the pre-monsoon, summer monsoon, and post-monsoon seasons observed at the tropical coastal station of Thiruvananthapuram (Sreekanth et al., 2017), the total concentration of raindrops N_t and the mass-weighted mean diameter D_m during the same season over the SCS are both much larger. The average maximum diameter D_max of precipitation in the pre-monsoon period is close to that observed in Taiwan (Seela et al., 2017), but the mean D_max in the other two seasons is both larger. The N_w values of precipitation in the pre-monsoon and monsoon periods are very close to the values during autumn and winter in East China, whereas they are smaller than those in summer and spring (Wen et al., 2019). The D_m values in the SCS are much larger than those in East China, which indicates there are fewer small drops but more large drops observed in precipitation over the SCS. The average N_t of precipitation over the SCS is higher than the summer precipitation over northern China, whereas the average D_m over northern China is much larger (Wen et al., 2017). Considering the also smaller median diameter D₀ over the SCS, we infer that small drops are more abundant over the SCS than over northern China in the summer season. This might be attributable to the sufficient water vapor supply in the SCS summer monsoon period over the SCS.

2

3.2. Composite raindrop spectra

-->

3.2. Composite raindrop spectra

Figure 2 presents the average raindrop spectra for the three different monsoon periods. The concentration of small drops (D ≤ 1 mm) is the highest in the pre-monsoon period, and the concentration of medium-sized raindrops (1?3 mm) is the highest during the monsoon period. The post-monsoon period has the smallest concentration of raindrops, with a diameter below 3 mm. However, the concentration of raindrops with diameters greater than 4 mm is higher during the post-monsoon period than those during the other two periods. This further confirms the presence of several big raindrops in the post-monsoon period.

Figure2. Spectra for the pre-monsoon (PRM), monsoon (MON), and post-monsoon (POM) precipitation periods.

The concentration of drops observed during the SCS summer monsoon period is higher than that during the South India summer monsoon season as reported by Chakravarty et al. (2013). Compared to the raindrop spectrum reported by Wen et al. (2016), also using a Parsivel disdrometer, the concentration of raindrops with diameters greater than 3 mm is higher in the SCS summer monsoon period.
To further discern the RSD differences among the three periods, the observed RSDs were further divided into six classes with respect to rain rate (mm h^?1)—namely, R ≤ 2 mm h^?1, 2 < R ≤ 5 mm h^?1, 5 < R ≤ 10 mm h^?1, 10 < R ≤ 20 mm h^?1, 20 < R ≤ 40 mm h^?1, and R > 40 mm h^?1.
Table 2 shows the statistical values of RSD parameters in the six rain-rate classes. With an increase in rain rate, there is an increase in N_t and D_a for precipitation in the pre-monsoon and monsoon periods. The mean D_m values all increase with the rain rate in the three periods. The lg(N_w) observed in the pre-monsoon period tends to increase, whereas it tends to decrease in the post-monsoon period. From low-intensity precipitation to high-intensity precipitation, no significant change in lg(N_w) values is observed in the monsoon period. Overall, the values of RSD parameters N₀ and λ tend to decrease with the increasing rain rate during the three periods. For the shape parameter μ, it decreases with the rain rate increasing in the pre-monsoon period. However, there is no orderliness to the change of parameter μ with rain rate in the monsoon and post-monsoon periods.

	Class (mm h^?1)	No. samples (min^?1)	R (mm h^?1)	N_t (m^?3)	D_a (mm)	D_m (mm)	lg(N_w) (mm^?1 m^?3)	lg(N₀) (m^?3 m^{?1 ? μ})	μ	λ (mm^?1)
PRM	R ≤ 2	91	1.10	396.67	0.74	1.12	3.60	8.12	9.97	13.18
	2< R ≤5	43	3.16	455.40	0.81	1.43	3.63	4.97	4.54	5.81
	5 < R ≤ 10	35	7.02	626.00	0.84	1.56	3.68	4.60	3.55	4.57
	10 < R ≤ 20	28	13.79	1759.2	0.85	1.72	3.98	4.77	2.92	4.24
	20 < R ≤ 40	18	31.16	2317.8	0.88	2.05	3.96	4.36	2.19	3.09
	R > 40	5	50.09	3242.2	0.90	2.08	4.13	4.21	1.70	2.75
MON	R ≤ 2	70	0.89	850.90	0.67	1.16	3.81	6.88	6.88	10.43
	2 < R ≤ 5	45	3.89	777.16	0.81	1.45	3.79	5.06	3.90	5.67
	5 < R ≤ 10	84	7.75	848.16	0.94	1.69	3.68	4.69	4.05	4.86
	10 < R ≤ 20	55	14.03	946.87	0.98	1.91	3.69	4.48	3.63	4.17
	20 < R ≤ 40	28	26.91	1065.9	1.06	2.17	3.71	4.25	3.83	3.67
	R > 40	11	52.39	1881.1	0.95	2.47	3.73	3.97	2.81	2.76
POM	R ≤ 2	103	1.05	543.95	0.73	1.20	3.60	5.79	5.33	8.36
	2 < R ≤ 5	72	3.09	714.25	0.86	1.51	3.62	5.62	6.22	7.47
	5 < R ≤ 10	55	7.41	720.28	0.95	1.74	3.64	4.93	5.65	5.64
	10 < R ≤ 20	27	13.82	667.83	1.00	2.09	3.52	4.25	4.37	4.15
	20 < R ≤ 40	19	31.60	701.80	1.23	2.70	3.35	3.64	4.78	3.37
	R > 40	6	41.66	390.74	1.45	3.05	3.20	3.10	6.42	3.24

Table2. Averaged microphysical parameters for each of six rain-rate intervals of precipitation in the pre-monsoon, monsoon, and post-monsoon period. Except for the parameters in Table 1, the RSD parameters including intercept parameter lg(N₀), shape parameter μ, and slope parameter λ are also presented in Table 2.

As shown in Fig. 3, with the rain rate increasing, the width of the RSD spectra in different monsoon periods broadens. The concentration of drops with diameter less than 2 mm increases in the pre-monsoon period but decreases in the post-monsoon period. For R < 10 mm h^?1, the disagreement among the three curves diminishes with the rain rate increasing. For R ≥ 10 mm h^?1, the disagreement among the three curves increases with the rain rate increasing. It is clear that there is a sharp increase in the small raindrop concentration during the pre-monsoon period. The concentration of large (small) drops is the highest (lowest) during the post-monsoon period. This pattern is consistent with the largest mean D_m for the same rain rate classes in the post-monsoon period and the highest N_w in the pre-monsoon period for R ≥ 10 mm h^?1, as shown in Table 2.

Figure3. Averaged RSDs for each of the six class intervals in the pre-monsoon (green), monsoon (red) and post-monsoon (blue) periods.

The value of rain rate is usually proposed to be a discriminator for types of precipitation. Tokay and Short (1996) took R = 10 mm h^?1 as a critical value for the classification of rain types. Thus, the previous analysis might indicate the differences in RSD for precipitation with different types, which will be analyzed further below.

2

3.3. RSD variations in stratiform and convective precipitation

-->

3.3. RSD variations in stratiform and convective precipitation

In the present study, the classification criteria of stratiform and convective precipitation based on rain rate and its standard deviation by Bringi et al. (2003) were adopted. For 10 consecutive 1-min RSD samples, samples with R ≥ 0.5 mm h^?1 and σ_R ≤ 1.5 mm h^?1 were classified as stratiform rain and samples with R ≥ 5.0 mm h^?1 and σ_R ≥ 1.5 mm h^?1 were convective rain.
Figure 4 compares the RSDs of stratiform and convective rainfall in the three periods. As depicted in Fig. 4a, the stratiform rainfall in the pre-monsoon period is characterized by a high concentration of small raindrops with diameter below 1 mm. The stratiform rainfall in the monsoon period has the highest concentration of raindrops with diameter between 1 mm and 5 mm. The concentration of raindrops in different particle sizes during the post-monsoon period is almost the minimum. In Fig. 4b, for convective rainfall, the highest concentration of drops with diameter less than 2 mm occurs in the pre-monsoon period. The concentration of drops with diameter greater than 4 mm for convective rainfall is the highest in the post-monsoon period. Relatively, the drops with diameter between 2 mm and 4 mm in monsoon precipitation are much more abundant.

Figure4. Variations of mean raindrop concentration with rain type for (a) stratiform and (b) convective precipitation. PRM, MON, and POM stand for the pre-monsoon, monsoon, and post-monsoon periods, respectively, while-Str and -Con denote the results from stratiform and convective precipitation, respectively.

Table 3 shows the statistical values of RSD parameters for stratiform and convective rainfall during the three periods. For stratiform rainfall, the mean N_t, mean D_a, and mean lg(N_w) are all the largest in the pre-monsoon period. The mean D_m is the largest in the monsoon period, followed by the post-monsoon period. The three gamma shape parameters of stratiform rainfall are all the largest in the pre-monsoon period, but are the smallest in the monsoon period. In comparison with the stratiform precipitation of the summer and winter seasons observed over northern Taiwan (Seela et al., 2018), larger D_m but smaller lg(N_w) are observed over the SCS. However, their mean gamma distribution parameters, μ and λ, are very close. The stratiform precipitation during the three periods over the SCS has a higher number concentration of raindrops N_t and larger mass-weighted mean diameter D_m than that over northern China (Wen et al., 2017). This suggests that RSDs for stratiform precipitation over the SCS have a higher concentration of bigger-sized drops than those over northern China.

Rain types	M	R (mm h^?1)	N_t (m^?3)	D_a (mm)	D_m (mm)	lg(N_w) (mm^?1 m^?3)	lg(N₀) (m^?3 m^{?1 ? μ})	μ	λ (mm^?1)
PRM-Str	170	3.00	836.42	0.90	1.43	3.65	5.27	4.72	6.60
MON-Str	188	6.07	639.51	0.84	1.51	3.59	4.63	3.36	5.01
POM-Str	203	2.83	479.37	0.78	1.45	3.52	4.77	3.81	5.90
PRM-Con	50	17.35	1575.6	0.94	1.70	4.02	5.03	3.86	4.84
MON-Con	105	18.49	940.02	1.00	1.90	3.77	4.76	4.55	4.75
POM-Con	79	17.20	635.30	1.06	2.00	3.61	4.91	6.62	5.74

Table3. Averaged microphysical parameters for convective and stratiform rain types. The stratiform and convective precipitation types are denoted as “Str” and “Con”, respectively.

For convective rainfall, the largest (smallest) N_t and lg(N_w) have also been found in the pre-monsoon (post-monsoon) period. However, the average D_a and D_m of post-monsoon convective rainfall are the largest, which indicates the occurrence of large raindrops in this period. The slope parameter λ is also much larger in the post-monsoon period than in the other two periods. The average number concentration of raindrops N_t for convective precipitation in the post-monsoon period is very close to the value observed in a coastal city of South China, Yangjiang (Tang et al., 2014). However, the mean D_m [lg(N_w)] of precipitation during the post-monsoon period is larger (smaller) than that in Yangjiang. The difference may be related to the existence of large raindrops in the post-monsoon period over the SCS.

2

3.4. Distributions of D_m and N_w

-->

3.4. Distributions of D_m and N_w

To further reveal the characteristics of RSD, the distributions of D_m and lg(N_w) are analyzed in this section. The mean lg(N_w) versus D_m values, with standard deviation, for the stratiform and convective rain types are depicted in Fig. 5. The maritime-like and continental-like convective clusters reported by Bringi et al. (2003) are also shown in the figure. In addition, Fig. 5 also provides comparative results from other studied regions. From Fig. 5, we can see that the mean D_m and N_w of stratiform rainfall in the three periods are near the right-hand side of the “stratiform line” given by Bringi et al. (2003). Relatively, the D_m (N_w) of stratiform rainfall in the monsoon (pre-monsoon) period is slightly larger.

Figure5. Average value of lg(N_W) versus average D_m (along with ±σ standard deviation bars) for stratiform and convective rainfall during the three periods. The two outlined squares represent (left) the maritime and (right) continental types of convective systems reported by Bringi et al. (2003). The dashed line is the result for their stratiform rain. The suffixes -Str and -Con denote the results from stratiform and convective precipitation, respectively. The blue triangles indicate results observed in other geographic locations.

Convective precipitation in the pre-monsoon period is very close to the “maritime cluster”, whereas convective precipitation in the post-monsoon period is closer to the “continental cluster”. The convective monsoon precipitation is not similar to either maritime or continental convective precipitation. The figure shows a clear shift in the RSD from the “maritime” to the “continental” cluster from the pre-monsoon to post-monsoon period.
Figure 5 also provides comparative results from other studied regions. The stratiform and convective precipitation over the SCS show larger D_m but smaller N_w in comparison with those observed in East China (Wen et al., 2016) and Taiwan and Palau (Seela et al., 2017). The average N_w (D_m) of stratiform precipitation over the SCS is slightly smaller (much larger) than that over northern China reported by Wen et al. (2017). The average D_m and N_w of stratiform precipitation over the SCS are both much larger those in the South Korean coastal region (Suh et al., 2016) (not shown in the figure). However, the average D_m and N_w of convective precipitation observed in the South Korean coastal region are just located between the values of the pre-monsoon convective precipitation and monsoon convective precipitation. This similarity may be related to the fact that all the three observation areas are located in coastal areas, where the precipitation is affected by warm and humid air currents from the sea.

4. QPE

One of the key RSD applications is QPE, because it can provide proper relations for rainfall retrieval algorithms (Lee and Zawadzki, 2005; Chapon et al., 2008; Chen et al., 2017). In this section, the power-law relationship Z = AR^b, widely used in radar meteorology, is firstly studied based on the observed datasets. Then, several empirical relationships used in polarimetric radar are also derived. Finally, the relationships of D_m?Z_ku, D_m?Z_ka, and lg(N_w)?D_m, used in the construction of the dual-frequency precipitation radar (DPR) rainfall retrieval algorithm, are also fitted, to improve the inversion accuracy of marine precipitation with GPM.

2

4.1. Z?R relations

-->

4.1. Z?R relations

Figure 6a presents scatterplots of radar reflectivity factor Z and rain rate R, as well as the results of Z?R fittings by using the least-squares method. The results show that the coefficient A is the largest (smallest) and the exponent b is the smallest (largest) in the pre-monsoon (post-monsoon) period. There is a clear inverse relationship between the coefficient A and exponent b in the three periods. Testud et al. (2001) noted that the Z?R relationship could be approximately set to Z ∝ $ N_{\rm{w}}^{ - 0.5}{R^{1.5}}$

, which means that the coefficient A is proportional to $ N_{\rm{w}}^{ - 0.5}$

if exponent b is close to 1.5. The changes in coefficient A during the three periods are shown to be consistent with the previous statistical results of lg(N_w) (as in Table 1). Uijlenhoet et al. (2003) proposed that the exponent b in the Z?R relation can reflect the different mechanisms of the formations of precipitation particles. Thus, the difference in the values of exponent b infers that the microphysical processes occurring during different monsoon periods are different (Seela et al., 2017). Comparing the current results with those obtained in Taiwan and Palau, the coefficient A over the SCS is slightly higher. The exponent b during the pre-monsoon period is very close to that in Taiwan and Palau (Seela et al., 2017). The coefficient A and exponent b in the pre-monsoon period are also close to the values of southwest monsoon precipitation at Gadanki, reported by Rao et al. (2001). Compared with the values of southwest monsoon precipitation observed at another tropical station, Kadapa, which is located in the southern part of India (Jayalakshmi and Reddy, 2014), the coefficient A of precipitation during the three periods over the SCS is all larger. The exponent b in the pre-monsoon period is slightly smaller than that at Kadapa, while the exponent b in the monsoon period and post-monsoon period are both slightly larger. These comparisons further illustrate the need to study the localized Z?R relationships.

Figure6. (a) Scatterplots of the Z?R values for precipitation during different monsoon periods and the fitted power-law relationships, Z = AR^b, where A is the coefficient and b is the exponent, derived from the least-squares method. (b) Scatterplots and the fitted power-law relationships of the total stratiform (black dashed line) and convective (red solid line) rain types observed in the SCS. The relation for mei-yu convective rain, Z = 368R^1.21 (Chen et al., 2013); the standard NEXRAD Z?R relationship, Z = 300R^1.40 (Fulton et al., 1998); and the tropical Z?R relationship, Z = 250R^1.20 (Rosenfeld et al., 1993) are also shown (green, blue and yellow lines, respectively).

The NEXRAD weather radar quantifies the default equation for the precipitation using Z = 300R^1.40. However, Rosenfeld et al. (1993) recommend Z = 250R^1.20 to estimate the precipitation in the tropics. Scatterplots between Z and R for the total stratiform and convective precipitation over the SCS are presented in Fig. 6b with cross symbols and gray dots, respectively. The results of correlation fitting are also given in this figure. The coefficient A is larger and exponent b is smaller for stratiform precipitation than convective precipitation, which indicates the need for different relationships to estimate different types of precipitation. The Z?R relationship for stratiform precipitation over the SCS has a larger coefficient A but smaller exponent b than those observed at Kadapa (Jayalakshmi and Reddy, 2014) and Gadanki (Rao et al., 2001). This difference may be related to the higher D_m but lower N_w of stratiform precipitation over the SCS than those at the two Indian stations.
The A and b values of the Z?R relationship for convective precipitation over the SCS are 384 and 1.38, respectively, which are similar to those of the standard relationship, Z = 300R^1.40, but differ from those of the standard tropical relationship, Z = 250R^1.20. The tropical relation underestimates the intense precipitation over the SCS. The coefficient A of precipitation over the SCS is very close to that observed during the mei-yu season, whereas the exponent b in the SCS is much bigger. Compared to the Z?R relationship derived from summer monsoon at Zhuhai (Z = 498R^1.3), the convective precipitation over the SCS shows a smaller coefficient A (Zhang et al., 2019). But, their exponent b is much closer, which may result from the fact that they are both affected by the warm and wet currents from the tropical ocean region. Compared with the convective precipitation observed in Taiwan and Palau (Seela et al., 2017), the coefficient A in the SCS is also higher, but their exponent b values are very close, which may be related to the fact that they are all in the western Pacific region. The coefficient A and exponent b for convective precipitation in the SCS are both larger than those during southwest monsoon convective precipitation at Kadapa (Jayalakshmi and Reddy, 2014). These results reconfirm previous opinion that there is a need to adopt modified Z?R relations in the estimation of rainfall in different regions (Seela et al., 2018).

2

4.2. Polarimetric radar applications

-->

4.2. Polarimetric radar applications

In addition to conventional weather radar, polarimetric radar has also been used in rainfall estimation for several years (Zhang et al., 2001; Cao et al., 2010; You et al., 2014). Cao and Zhang (2009) demonstrated that polarimetric measurements, such as differential reflectivity Z_DR and specific differential phase K_dp, can provide valuable information caused by DSD variability, hail contamination, beam blockage, and other sources of error, which is conducive to resolving problems. To improve precipitation estimation with polarimetric radar, several empirical relationships, such as R(Z_H, Z_DR), R(K_dp), R(Z_H), and R(Z_DR, K_dp), also needed to be constructed (Zhang et al., 2017).
In this paper, the widely used R(Z_H, Z_DR) relationship was derived, which can be expressed as follows:
where a is the coefficient. b and c are the exponent for the horizontal polarization Z_H (mm⁶ m^?3) and the differential reflectivity Z_DR (dBZ), respectively.The two parameters were calculated from the observed DSDs using the T-matrix scattering technique described by Zhang et al. (2001), as follows:
where λ is the radar wavelength, K_w is the water dielectric factor, and f_hh,vv(D) is the backscattering amplitudes of a raindrop for horizontally and vertically polarized waves, respectively. The frequency of the radar’s electromagnetic wave is assumed to be 2.85 GHz (S-band). The raindrop temperature was assumed to be 20°C in this study, and the raindrops were also assumed to follow the axis ratio relation proposed by Beard and Chuang (1987):
In addition, for comparison purposes, we defined two parameters—the normalized mean bias (NB) and normalized standard error (NSE), which can be expressed by:
where R_DSD and R_radar are 1-min precipitation values from the DSD data and the calculated radar parameters based on the T-matrix simulation, respectively.
The results are shown in Table 4. As shown in Table 4, the coefficient a in R(Z_H, Z_DR) decreases from the pre-monsoon to the post-monsoon period, whereas exponents b and c in R(Z_H, Z_DR) increase from the pre-monsoon to the post-monsoon period. It also clearly shows that both the NB errors and the NES errors are much smaller for precipitation estimated with R(Z_H, Z_DR) relationships than Z?R relationships, during the same period. The results indicate that the R(Z_H, Z_DR) estimator has an advantage over the Z?R relation, which reconfirm previous opinion (Cao et al., 2010; Tang et al., 2014; Zhang et al., 2017).

	Rainfall estimation methods
	R(Z_H, Z_DR)					Z?R
	a	b	c	NB (%)	NES (%)	A	b	NB (%)	NES (%)
PRM	0.009 40	0.730	?1.092	?5.92	34.6	466	1.33	31.4	73.2
MON	0.004 46	0.864	?1.197	?3.16	32.1	346	1.41	12.2	70.5
POM	0.002 39	0.916	?1.245	?3.97	29.9	339	1.46	31.8	67.6

Table4. Results of the Z ?R and R(Z_H, Z_DR) rainfall estimation methods. Z_H denotes the horizontal polarization and Z_DR represents the differential reflectivity. NB is short for the normalized mean bias and NSE is short for the normalized standard error.

2

4.3. GPM retrieval algorithms

-->

4.3. GPM retrieval algorithms

Nowadays, the GPM Core Observatory satellite serves as an optimal platform to estimate precipitation globally. However, more ground validation needs to be performed to improve the rainfall retrieval algorithm.
The DPR on board GPM is the key instrument. It estimates the RSD from the attenuation difference between the Ku-band precipitation radar (13.6 GHz) and the Ka-band precipitation radar (35.5 GHz) (Schwaller and Morris, 2011). The normalized gamma distribution described in Eq. (1) is widely used in the GPM DPR rainfall retrieval algorithms (Grecu et al., 2016). Accordingly, the DSD can be reconstructed from N_w and D_m given a μ value. The radar reflectivity at the Ku band or Ka band can be used to derive the parameters N_w and D_m. The DFR (dB) takes the following form:
The effective radar reflectivity factors at Ku-band (Z_Ku) and Ka-band (Z_Ka) frequencies can be calculated by
where Z_e is the effective radar reflectivity factor at a specific wavelength λ, which can be derived from the onboard disdrometer observations. σ_b(D_i, λ) is the backscattering cross section of a raindrop with diameter D_i, which is directly calculated according to Mie theory. ${{k_{\rm{w}}^{2}}} $

is the dielectric factor, which is related to the complex refractive index of region and is conventionally taken to be 0.93.
As pointed out by Chen et al. (2017), DFR is commonly used for the retrieval of D_m. As shown by the scatterplots of D_m?Z_Ku and D_m?Z_Ka for the three periods in Fig. 7, the values of D_m increase with Z_Ku or Z_Ka increasing. There is a high correlation between D_m and Z_Ku (Z_Ka). As in Chen et al. (2017), we also derived second-degree polynomial relations between D_m and Z_Ku (Z_Ka). The relevant fitting results are presented in Table 5. As shown in Fig. 7, the values of D_m in the monsoon period are the largest for Z_Ku (Z_Ka) below 15 dBZ and those in the post-monsoon period are the largest for Z_Ku (Z_Ka) above 25 dBZ.

Figure7. Scatterplots of (a) D_m (units: mm) and the Ku-band effective radar reflectivity factor Z_Ku (units: dBZ), and (b) D_m (units: mm) and the Ka-band effective radar reflectivity factor Z_Ka (units: dBZ). The fitted curves of the pre-monsoon, monsoon, and post-monsoon periods are represented by solid, dashed, and dash-dotted lines, respectively.

Relation	Data	a	b	c
D_m = ${{aZ_{\rm{Ku}}^{2}}} $ + bZ_Ku + c	PRM	0.000 417 3	0.0201	0.5178
	MON	0.001 232 2	?0.0304	1.1780
	POM	0.001 574 1	?0.0288	0.9831
D_m = ${{aZ_{\rm{Ka}}^{2}}} $ + bZ_Ka + c	PRM	0.000 309 2	0.0316	0.3581
	MON	0.001 161 3	?0.0140	0.9082
	POM	0.001 353 1	?0.0073	0.7637
lg(N_w) =${{aD_{\rm{m}}^{2}}} $ + bD_m + c	PRM	0.0878	?1.1521	4.8647
	MON	0.1178	?1.1023	4.9553
	POM	0.0652	?0.9211	4.5127

Table5. Second-degree polynomial relations of D_m?Z_Ku, D_m?Z_Ka, and lg(N_w)?D_m for the three monsoon periods.

Figure 8 presents scatterplots of the two parameters lg(N_w) and D_m in a normalized gamma distribution. It clearly shows that the lg(N_w) decreases with the increase of D_m. In order to reduce the parameters in the normalized gamma distribution, we further derived an empirical polynomial relation between lg(N_w) and D_m. The fitting results indicate that the lg(N_w) decreases most obviously with D_m in the monsoon period. Comparing the current results with those for the Tibetan Plateau reported by Chen et al. (2017), the coefficients of the second-degree polynomial relation over the ocean are smaller.

Figure8. Scatterplots of lg(N_w) (units: mm^?1 m^?3) and D_m (units: mm) in the three monsoon periods. The fitted curves of the pre-monsoon, monsoon, and post-monsoon periods are represented by solid, dashed, and dash-dotted lines, respectively.

本站小编 Free考研考试/2022-01-02

Corresponding author: Yun ZHANG,zhangyun.1207@163.com;

HTML

2.1. Datasets

2.2. Data-quality issues

2.3. Methods

3.1. Average characteristics of microphysical rainfall parameters

3.2. Composite raindrop spectra

3.3. RSD variations in stratiform and convective precipitation

3.4. Distributions of D_m and N_w

4.1. Z?R relations

4.2. Polarimetric radar applications

4.3. GPM retrieval algorithms

相关话题/Microphysical Characteristics Precipitation

领限时大额优惠券,享本站正版考研考试资料!

Diurnal Variations of Precipitation over North China Regulated by the Mountain-plains Solenoid and B

Microphysical Properties of Rainwater in Typhoon Usagi (2013): A Numerical Modeling Study

Differences between Convective and Stratiform Precipitation Budgets in a Torrential Rainfall Event

Warm-Season Diurnal Variations of Total, Stratiform, Convective, and Extreme Hourly Precipitation ov

Role of the Nocturnal Low-level Jet in the Formation of the Morning Precipitation Peak over the Dabi

Variation in Principal Modes of Midsummer Precipitation over Northeast China and Its Associated Atmo

Estimating the Correlated Observation-Error Characteristics of the Chinese FengYun Microwave Tempera

Determination of Intraseasonal Variation of Precipitation Microphysics in the Southern Indian Ocean

Subseasonal Change in the Seesaw Pattern of Precipitation between the Yangtze River Basin and the Tr

Changes in the Proportion of Precipitation Occurring as Rain in Northern Canada during Spring-Summer