**1. Introduction**

The microphysical processes of clouds and precipitation play vital roles in the formation and development of precipitation and the prediction of severe weather. Raindrop size distribution (DSD) is an important feature that characterizes the microphysical process of precipitation [1–3] and is mainly affected by climatic characteristics and precipitation types [3–9]. In recent years, disdrometer DSD measurements have been widely used to study the microphysical characteristics of precipitation [4,10–14]. Many DSD observations and analyses have been carried out in different regions of China. Based on the OTT Particle Size Velocity (PARSIVEL) disdrometer data from Nagqu (4500 m above sea level (ASL)) over the Tibetan Plateau (TP), Chen et al. [15] reported that the discrepancy in DSDs between day and night is nonsignificant in stratiform rainfall but obvious in convective rainfall. The DSDs of different precipitation types (stratiform and convective) between Nagqu over the TP and Yangjiang in southern China were compared and showed that all three gamma parameters for stratiform precipitation over the TP are larger than those in southern China, while the normalized intercept parameter *N*w and the shape parameter

**Citation:** Li, R.; Wang, G.; Zhou, R.; Zhang, J.; Liu, L. Seasonal Variation in Microphysical Characteristics of Precipitation at the Entrance of Water Vapor Channel in Yarlung Zangbo Grand Canyon. *Remote Sens.* **2022**, *14*, 3149. https://doi.org/10.3390/ rs14133149

Academic Editor: Federico Porcù

Received: 8 April 2022 Accepted: 27 June 2022 Published: 30 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

*μ* for convective precipitation are less than those in southern China [16]. DSD statistical analysis was also conducted in Yining, Xinjiang, an arid region of China, and it showed that convective precipitation was neither continental-like nor maritime-like [17]. In addition, the same location will display significant seasonal differences in the microphysical processes of precipitation [18]. The precipitation over the South China Sea (SCS) is dominated by small (midsize) drops during the premonsoon (monsoon) period, while it has the lowest concentration of raindrops in the postmonsoon season [18]. Monsoon precipitation at Thiruvananthapuram, a coastal tropical station in India, has a higher concentration of small drops than in the other three seasons [3]. Krishna et al. [6] found that the mean concentrations of medium and large raindrops in the west monsoon season are higher than those in the east monsoon season in the Palau Islands.

The TP is located in western China, with an average elevation of approximately 4000 m. It is important to the climate and ecosystems of the Asian continent and even the world [19]. The TP is also known as the Water Tower of Asia due to the origination of seven important Asian rivers, including the Yellow River, the Yangtze River, the Yarlung Zangbo River, etc. The westerlies–monsoon synergy zone covers the TP and the surrounding areas. Climate warming has led to anomalies in westerlies–monsoons and an imbalance in the Water Tower of Asia. The Yarlung Zangbo Grand Canyon (YGC), with a total length of 496.3 km and a depth of up to 6009 m [20], is located in the southeast TP and is the largest channel for transporting water vapor to the TP. During the Indian summer monsoon period, warm and wet water vapor is transported northward to the TP along the YGC. The water vapor transport intensity (nearly 2000 g cm<sup>−</sup><sup>1</sup> s<sup>−</sup>1) is equivalent to that from the south bank of the Yangtze River to the north bank in summer [21]. The YGC plays an important role in climate change in the TP and is a typical unit in the TP climate system.

Mêdog, with a mean altitude of 1200 m, is located at the entrance of the YGC. The humid air from the Indian Ocean flows straight into the gorge, giving Mêdog the most annual accumulated precipitation on the TP [22]. Due to inconvenient transportation and frequent debris flows in the rainy season, in situ observation data are lacking along the YGC, especially in Mêdog. To explore the causes and related mechanisms of water resource changes in the Yarlung Zangbo River basin under the synergistic action of westerlies–monsoons in the southeast TP, a comprehensive cloud precipitation observation test base was established at the Mêdog Climate Observatory (95.32◦E, 29.31◦N), supported by "the Second Tibetan Plateau Scientific Expedition" and "the Earth-Atmosphere Interaction in the TP and its Influence on the Weather and Climate in the Lower Reaches" projects. A Ka-band cloud radar, a micro rain radar, an OTT PARSIVEL disdrometer and other instruments were deployed at Mêdog National Climate Observatory to obtain the three-dimensional structure of clouds and precipitation characteristics in the YGC. Based on Ka-band cloud radar measurements, the vertical structure characteristics and diurnal variation in clouds over Mêdog in the southeast TP were analyzed [23]. In addition, precipitation in Mêdog was dominated by small and medium drops, and the convective rain in this region could be classified as maritime-like [24]. However, the seasonal variation characteristics of the raindrop spectrum were not analyzed due to the short observation period. In this study, DSD data collected from an OTT PARSIVEL disdrometer during the period of July 2019 to June 2020 were used to study the seasonal variation in microphysical characteristics for different precipitation intensities and precipitation types. In addition, the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis version 5 (ERA5) data, Fengyun-4A (FY-4A) satellite products and automatic weather station (AWS) observations were used to address the possible reasons for the seasonal differences in DSDs in Mêdog. This study aimed to better understand the seasonal variation in the microphysical characteristics of precipitation processes at the entrance of the water vapor channel in the YGC and its relationship with westerlies–monsoon synergy and water vapor transport, which is beneficial for improving the microphysical parameterization scheme and the precipitation forecast of the models in the TP.

The instruments, data and methods adopted in this study are provided in Section 2. The properties of DSDs and microphysics parameters for different rain rate classes and precipitation types in different seasons are reported in Section 3. Section 4 discusses the possible reasons for the seasonal variations in DSDs. The major conclusions are given in the final section.

### **2. Data and Methods**

The proposed research investigation used different datasets to provide an overall evaluation of seasonal variation in raindrop size distribution in Mêdog. The main steps followed along this study are presented in the flow chart depicted in Figure 1.

**Figure 1.** Methodological workflow diagram adopted in this study.

Measurements from a laser-optical PARSIVEL disdrometer [7] in Mêdog with a time resolution of 1 min were used in this study. The position of the Mêdog National Climate Observatory and a picture of the PARSIVEL disdrometer are shown in Figure 2. A disdrometer can simultaneously measure the size and falling speed of hydrometeors. The size and falling speed ranged from 0.06 mm to 24.5 mm and from 0.05 to 20.8 m s<sup>−</sup>1, respectively, which were divided into 32 nonequidistant bins [25,26].

**Figure 2.** The locations of Mêdog (black solid dot) and Yarlung Zangbo Grand Canyon (YGC), and topography (m) of the Tibetan Plateau (TP) (**a**) and the PARSIVEL disdrometer (**b**). The red arrow indicates water vapor channel in the YGC.

The disdrometer data used in this study were collected from 1 July 2019 to 30 June 2020 and divided into four periods, winter (January–February), premonsoon (March–May), monsoon (June–September) and postmonsoon (October–December) [3], to study the seasonal characteristics of DSDs in Mêdog. During this period, 30 days and 893 min of data were missing due to power outages caused by geological disasters, such as landslides and debris flows.

### *2.1. Quality Control*

In this study, strict quality control was carried out on disdrometer data to eliminate the influence of raindrop classification errors caused by edge landing, strong wind and splashing. Firstly, the falling speed beyond the boundary of ±60% of Beard's empirical speed–diameter relationship was excluded [27]. Given the terrain altitude of Mêdog, the speed–diameter relationship was corrected by multiplying by an air density factor of 1.04 [28]. Figure 3 gives the accumulated raw particle counts by diameter and fall speed observed during the four seasons. Drops beyond a ±60% empirical fall speed–diameter relationship were eliminated. The distribution of fall speed–diameter basically conformed to Beard's empirical fall speed–diameter relationship after quality control.

Secondly, the first and second size bins were removed due to their low signal-to-noise ratio, and the size bins with a number of drops less than 2 or with diameters greater than 6 mm were screened and eliminated [10]. One-minute samples with a total number of raindrops less than 10 or a rainfall rate less than 0.1 mm h−<sup>1</sup> were regarded as instrument noise and eliminated [3,29]. Good agreements between disdrometer observations and gauge measurements in Mêdog have been reported by Wang et al. [24], although disdrometers tend to underestimate gauged rain. This suggests that disdrometer data could be used to explore the seasonal variation in the microphysical characteristics of precipitation in Mêdog.

A total of 73,707 min of precipitation samples after quality control were collected from the disdrometer at the Mêdog National Climate Observatory, with accumulated rainfall of 1237.57 mm. Table 1 shows the total rain duration and accumulated rainfall amount during the four seasons. As seen from Table 1, Mêdog precipitation mainly occurred in the monsoon period, with the rainfall in this period being 699.92 mm, accounting for approximately 57% of the total, followed by the premonsoon period, accounting for approximately 32%. The rainfall in winter was the lowest, accounting for only approximately 4% of the total rainfall. Rainfall exhibited obvious seasonal variation.

**Figure 3.** Accumulated raw particle counts by diameter and fall speed were observed during the four seasons: winter (**a**), premonsoon (**b**), monsoon (**c**) and postmonsoon (**d**). Solid black lines indicate the empirical fall speed–diameter relationship. Dashed lines denote the ±60% empirical fall speed–diameter relationship.



### *2.2. Parameter Calculation*

The raindrop number (*ni*,*j*) of the *i*th size (*Di*) and the *j*th speed (*Vj*) are measured by a disdrometer. The raindrop number concentration, *N*(*Di*) (m−<sup>3</sup> mm<sup>−</sup>1), in the *i*th size can be calculated as follows:

$$N(D\_i) = \sum\_{j=1}^{32} \frac{n\_{i,j}}{V\_j \times S / T \times \Delta D\_i} \tag{1}$$

where *S* m<sup>2</sup> and *T* (s) are the sampling area and sampling time and were set to 54 cm<sup>2</sup> and 60 s in this study, respectively. Δ*Di* indicates the size interval.

The rainfall rate *R* mm <sup>h</sup>−<sup>1</sup>, radar reflectivity factor *Z* mm<sup>6</sup> <sup>m</sup><sup>−</sup><sup>3</sup>, total raindrop concentration *N*T m<sup>−</sup><sup>3</sup> and liquid water content (LWC, g m<sup>−</sup>3) can be obtained from the following equations:

$$R = 6\pi \times 10^{-4} \sum\_{i=3}^{32} \sum\_{j=1}^{32} D\_i^{\;3} \frac{n\_{i,j}}{S \times T} \tag{2}$$

$$Z = \sum\_{i=3}^{32} D\_i^{\;6} N(D\_i) \Delta D\_i \tag{3}$$

$$N\_{\Gamma} = \sum\_{i=3}^{32} N(D\_i) \Delta D\_{i\prime} \tag{4}$$

$$\text{LWC} = \frac{\pi}{6000} \Sigma\_{i=3}^{32} D\_i^{-3} N(D\_i) \Delta D\_{i\prime} \tag{5}$$

In this paper, the gamma distribution model was used to fit the observed DSDs from the disdrometer [30]:

$$N(D) = N\_0 D^\mu \exp(-\Lambda D) \tag{6}$$

where *D* (mm) represents the hydrometeor diameter, *N*0 mm<sup>−</sup>*μ*−<sup>1</sup> m<sup>−</sup><sup>3</sup> is the intercept parameter, and Λ mm<sup>−</sup><sup>1</sup> and *μ* indicate the slope parameter and shape parameter, respectively. Λ and *μ* can be calculated as follows [31]:

$$M\_{\mathbf{x}} = \sum\_{i=3}^{32} N(D\_i) D\_i^{\times} \Delta D\_{i\star} \tag{7}$$

$$\mathbf{G} = \frac{M\_4^3}{M\_3^2 M\_6} \mathbf{'} \tag{8}$$

$$\mu = \frac{11\text{G} - 8 + \sqrt{\text{G}(\text{G} + 8)}}{2(1 - \text{G})},\tag{9}$$

$$
\Lambda = (\mu + 4) \frac{M\_3}{M\_4}.\tag{10}
$$

Testud et al. [32] proposed the normalized gamma distribution:

$$N(D) = N\_{\rm w} f(\mu) (\frac{D}{D\_{\rm m}})^{\mu} \exp\left[ - (4 + \mu) \frac{D}{D\_{\rm m}} \right],\tag{11}$$

$$f(\mu) = \frac{\Gamma(4)}{4^4} \frac{(4+\mu)^{4+\mu}}{\Gamma(4+\mu)},\tag{12}$$

where Γ(*x*) represents a complete gamma function that is defined as follows:

$$
\Gamma(\mathbf{x}) = \sqrt{2\pi} e^{-\mathbf{x}} \mathbf{x}^{\mathbf{x} - \frac{1}{2}} \tag{13}
$$

The mass-weighted mean diameter *D*m (mm) and normalized intercept parameter *N*w (m−<sup>3</sup> mm<sup>−</sup>1) can be used to describe the general DSD characteristics and can be defined as follows [33]:

*D*m = *M*4 *M*3 , (14)

$$N\_{\rm W} = \frac{256}{6} \times \frac{M\_3^5}{M\_4^4}. \tag{15}$$

### *2.3. Different Classes in R, Dm and NT*

The DSD data used in this study were divided into the following six categories according to *R*, *D*m and *N*T, respectively, as shown in Table 2. Percentages of occurrence and relative contributions to total rainfall were calculated for different categories of *R*, *D*m and *N*T. To improve the representativeness of the statistical characteristics, the rainfall rate categories with fewer than 20 samples were excluded.

**Table 2.** Categories of rain rate (*R*), mass-weighted mean diameter (*D*m) and total raindrop concentration (*N*T).


### *2.4. Different Precipitation Types*

To further analyze the DSD characteristics in different seasons, the 1 min DSD samples from the disdrometer were also classified into stratiform rainfall and convective rainfall according to a simple method based on the SD *σR* of rainfall rate *R* [34]. Specifically, for 10 continuous 1 min DSD samples, if *R* ≥ 5 mm h−<sup>1</sup> and *σR* > 1.5 mm <sup>h</sup>−1, convective rainfall was distinguished, while when *σR* ≤ 1.5 mm <sup>h</sup>−1, stratiform rainfall was classified.

In addition to the disdrometer measurements, ECMWF ERA5 reanalysis data, AWS observations and FY-4A products were also used in this study. ERA5 reanalysis data are the fifth-generation ECMWF reanalysis for the global climate and weather for the past 4 to 7 decades recorded by C3S Climate Data Store (CDS, https://cds.climate.copernicus. eu, accessed on 1 September 2021) with a spatial resolution of 0.25◦ × 0.25◦. FY-4A is China's second-generation geostationary meteorological satellite and carries the Advanced Geosynchronous Radiation Imager (AGRI), the Geostationary Interferometric Infrared Sounder (GIIRS) and the Lighting Mapping Imager (LMI). AGRI has 14 channels, and the spatial resolution can reach 0.5–1 km for visible and near-infrared bands and 2–4 km for infrared bands. The AGRI level2 dataset provides the cloud type (CLT), the cloud top height (CTH), the Black Body Temperature (TBB) and other products, which can be obtained at FENGYUN Satellite Data Center (http://satellite.nsmc.org.cn, accessed on 1 September 2021).
