**3. Results**

### *3.1. Vertical Distribution Characteristics of Radar Reflectivity Factor*

Figure 3 shows the vertical structure of reflectivity factor. In addition to revealing the vertical structure of precipitation, the maximum frequency profile of the Ze is also a good indicator of its microphysical processes [40]. SP is usually characterized by bright band near FzH, and the bright band is a good indicator of phase change of hydrometeor [41]. The hydrometeors above bright band are mainly ice or snow. In bright band, there are mixedphase hydrometeors including partially melted ice, snow and raindrops, while it occurs at the liquid phase below bright band. As shown in the maximum frequency profile of Ze from 6 to 10 km, Ze of two types of heavy precipitation increases with altitude reducing, indicating that precipitation particles are growing as they are falling. For stratiform heavy precipitation (Figure 3a–c), different to CP, the growth rate of Ze is very large near the freezing layer that the profile tends to be horizontal, showing particles undergoing a rapid transition from ice to liquid phase near the freezing layer. Figure 4 shows the correlation between the elevation and FzH for heavy rainfall. Only Figure 4e,g,h passed the 0.01 significance level, and the r-value is relatively large in Figure 4h, which suggests that the FzH of the two types of heavy precipitation is hardly affected by topography, except for heavy CP over high mountains.

**Figure 3.** Normalized contoured frequency by altitude diagrams (NCFAD) of reflectivity factor of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains (**<sup>c</sup>**,**f**). The color areas and black solid lines indicate occurrence frequency and maximum frequency profile of reflectivity factor, respectively. The black and blue dash lines represent STA and FzH, respectively.

**Figure 4.** Scatter plots of elevation and FzH of two types of heavy precipitation over (**<sup>a</sup>**,**<sup>e</sup>**) three areas, (**b**,**f**) plains, (**<sup>c</sup>**,**g**) mountains, and (**d**,**h**) high mountains. r represents Pearson correlation coefficient, p represents significance level, y and red solid lines are the regression equation and N represents sample size.

Figure 5 exhibits the mean Ze profiles with various RR. Ze increases from STA to the surface with altitude decreasing, and the near-surface Ze increases with near-surface RR increasing. As shown in Figure 5a–e, the bright band characteristics of SP are remarkable at an altitude of about 6 km, and there is a clear turning point near FzH, which is the microphysical property of the SP particles during the falling process, while Ze profiles of CP do not have such characteristic. For heavy precipitation, Ze from FzH to the surface overall increases, indicating that the coalescence of falling particles is more efficient than breakup and evaporation.

**Figure 5.** The vertical profile of Ze with various near-surface RR. (**<sup>a</sup>**–**<sup>e</sup>**) represent SP (top panels), (**f**–**j**) represent CP (bottom panels). The green, black and red solid lines represent plains, mountains and high mountains, respectively. The square and round dots represent STA and FzH, respectively.

### *3.2. Vertical Distribution Characteristics of Mass-Weighted Mean Raindrop Diameter*

DSD is a fundamental property of precipitation, which is very important for understanding the microphysical processes occurring in precipitation systems. DSD contains two parameters: mass-weighted mean raindrop diameter (Dm in mm) and normalized DSD intercepts (Nw in mm<sup>−</sup>1/m3), where Nw denotes the raindrop number concentration as rainwater content and raindrop size are constant. The relationship between dBNw and Nw is dBNw = 10 lg(Nw). For simplicity, hereinafter, dBNw is used rather than Nw. Figure 6 shows the NCFAD of Dm of two types of heavy precipitation over different types of terrain. Small precipitation particles (Dm ≤ 1.2 mm) are mainly concentrated between STA and FzH, and the precipitation particles in this layer mainly exist in the form of ice crystals. With the elevation of terrain, the horizontal distribution domain of Dm increases while the vertical distribution domain decreases. For the same terrain, the horizontal and vertical distribution domain of Dm of heavy SP is smaller than that of heavy CP. In addition, the maximum Dm of the horizontal distribution domain is also smaller than that of CP. For heavy CP, with the elevation increasing, the occurrence probability of large raindrops (Dm ≥ 2.6 mm) increasing, as shown in Figure 6d–f. It can be clearly seen from Figure 6a–c that in heavy SP, precipitation particles with Dm of 1.3–1.6 mm below FzH of mountains and high mountains have a higher probability of occurrence than that of plains. This is mainly caused by the difference of water vapor content near FzH, which will be discussed in Section 4.

**Figure 6.** NCFAD of Dm of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains(**<sup>c</sup>**,**f**). The color areas indicate occurrence frequency of Dm. The black and blue dash lines represent STA and FzH, respectively.

From Dm profiles of different rain intensity levels (Figure 7), it can be seen that the value of CP near the surface is always greater than SP under the same terrain and rain intensity levels. The Dm profiles of light CP (0.5 ≤ RR < 2 mm/h) that are shown in Figure 7f are obviously different from that of other rainfall intensity. The Dm profile of light CP over high mountains decreases as altitude decreases from STA to FzH, while that of other rainfall intensity increases as altitude decreases. Different from light SP (Figure 7a), the Dm profile of plains of CP (Figure 7f) decreases as altitude decreases from FzH to surface because of the less water vapor in low-level atmosphere over plains, which tends to reduce the diameter of small raindrops by evaporation. Above an altitude of 10 km, the Dm of heavy precipitation over high mountains are larger than those over plains and mountain (Figure 7e,j), particularly in heavy SP, which is significantly different from that of other rainfall intensity.

**Figure 7.** The vertical profile of the mean Dm with various near-surface RR. (**<sup>a</sup>**–**<sup>e</sup>**) represent CP, (**f**–**j**) represent SP. The green, black and red solid lines represent plains, mountains and high mountains, respectively. The square and round dots represent STA and FzH, respectively.

There is a close relationship between Dm and Ze. In general, large Dm of precipitation particles correspond to large Ze. As shown in Figure 7e,j, the near-surface Dm of CP is significantly larger than that of heavy SP, and therefore the near-surface Ze is also larger than that of SP (Figure 5e,j).

### *3.3. Vertical Distribution Characteristics of dBNw*

For the same terrain, compared with dBNw NCFAD of two types of heavy precipitation (Figure 8), it shows that the horizontal distribution domain of dBNw of heavy SP over plains and mountains is wider than that of CP while the high frequency areas of dBNw are not as concentrated as those of heavy CP. In contrast, the horizontal distribution domain of dBNw of heavy SP over high mountains is smaller than that of CP while the high frequency areas are more concentrated. For different types of terrain, the vertical structure of dBNw of heavy CP is also different, as shown in Figure 8d–f: The higher elevation is, the wider horizontal distribution domain of dBNw is. Below FzH, the high frequency areas of dBNw over plains are the most concentrated with 34–38 mm<sup>−</sup>1/m3, while those over mountains and high mountains are more scattered, especially over high mountains. The opposite is true for heavy SP. The high frequency areas of dBNw over mountains and high mountains are more concentrated than those over plains (Figure 8a–c).

**Figure 8.** Normalized contoured frequency by altitude diagrams (NCFAD) of dBNw of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains (**<sup>c</sup>**,**f**). The color areas indicate occurrence frequency of dBNw, the black and blue dash lines represent STA and FzH, respectively.

Figure 9 shows the mean dBNw profiles with different surface rainfall intensity. In each panel, dBNw overall increases as altitude decreases. For heavy precipitation (Figure 9e,j), the dBNw above 10 km over high mountains are obviously smaller than those over plains. Combined with the mean Dm profiles of heavy precipitation (Figure 7e,j), compared with plain areas, precipitation particles over high mountains have the characteristics of lower number concentration and larger scale above 10 km.

**Figure 9.** The vertical profile of dBNw with various near-surface RR. (**<sup>a</sup>**–**<sup>e</sup>**) represent SP, (**f**–**j**) represent CP. The green, black and red solid lines represent plains, mountains and high mountains, respectively. The square and round dots represent STA and FzH, respectively.

### *3.4. Distribution Characteristics of Storm Top Altitude and Water Vapor*

In general, for the same type of precipitation, the higher STA is, the heavier precipitation is. STA is usually lower than the cloud top height, and when STA is low, the cloud top height varies widely. The higher STA is, the more similar the cloud top height is to STA [42].

Figure 10 shows the horizontal distribution of STA when heavy precipitation occurred. For the same types of terrain, STA of heavy CP is overall higher than that of heavy SP. For heavy CP, the high value areas of STA over plains (Figure 10d) are mainly located in the central and western of the Sichuan Basin where roughly correspond to the high frequency areas of heavy precipitation. The high value areas of STA over mountains are mainly located in the region where the plain and the mountain meet in the west of the Sichuan Basin (Figure 10e). In general, STA over plains is higher than those over mountains and high mountains, regardless of heavy CP or SP.

**Figure 10.** The horizontal Distribution of storm top altitude of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains (**<sup>c</sup>**,**f**).

In order to have a more intuitive understanding of the distribution of STA, Figure 11 shows the probability distribution functions (PDFs) of STA over plains, mountains and high mountains with different rain intensity levels. It can be seen that when heavy precipitation occurs, STA is mostly above 6 km altitude (cloud system is deep). In the same terrains, STA of heavy CP is generally higher than heavy SP (Figure 11e,j), which also indicates that convective cloud develops more vigorously than stratus in the vertical direction. When the RR is less than 8 mm/h, there is no significant difference in STA of CP and SP over the same terrain. When the surface RR increased to 8–20 mm/h, STA of precipitation over mountains and high mountains was still roughly the same, but STA of precipitation over plains began to show significant differences. STA of CP over plains was significantly higher than that of SP, and its highest frequency was about 4 km higher than that of SP. When CP occurs, with the increase of RR, the height corresponding to the maximum occurrence frequency of STA also increases. For the two types of precipitation with RR < 20 mm/h, STA of precipitation over high mountains is generally higher than those over plains and mountain. When the RR ≥ 20 mm/h for heavy CP, STA over plains is obviously higher than those over mountains and high mountains. STA of heavy CP over plains is mainly concentrated at 10–14 km altitude, while that in mountains and high mountains are mainly concentrated at 9–11 km altitude.

**Figure 11.** Probability distribution functions (PDFs) of storm top altitude with various near-surface RR of SP (**<sup>a</sup>**–**<sup>e</sup>**) and CP (**f**–**j**) over different topographic. The green, black and red solid lines represent plains, mountains and high mountains, respectively.

It is usually considered that convective overshooting is with STA > 10 km, and deep SP is with STA > 8 km. Fu et al. [43] divided STA into three categories, including STA < 5 km, 5–10 km and > 10 km, which represent shallow convective, moderate convective and deep convective (also called convective overshooting) respectively. Likewise, divided STA into below 5 km, 5–8 km and above 8 km, representing shallow stratiform, moderate stratiform and deep stratiform respectively. Statistics on the categories of STA over the Sichuan Basin and its surrounding areas are shown in Table 2. It can be found that with the increase of RR, the proportion of shallow precipitation gradually decreases, while the deep precipitation increases with the increase of RR. For heavy precipitation, the proportion of deep convective over plains is 69.6%, higher than mountains (54.8%) and high mountains (57.2%). For heavy SP, there is little difference in the proportion of deep stratiform between plains and high mountains (64.4% and 65.9%, respectively), both higher than mountains (50.1%).

STA is related to the strength of the updraft and affected by properties of the underlying surface [44]. The updraft not only influences STA, but also plays a significant role in the size of the precipitation particles. Therefore, it is speculated that there may be a correlation between STA and the precipitation particles size, then the correlation between STA and the near-surface Dm was tested for significance (Figure 12). Except for the heavy SP over plains and high mountains that failed with 0.01 significance level, all others passed the significance test, indicating that there is a positive linear correlation between STA and near-surface Dm of heavy precipitation except for the heavy SP over plains and high mountains. For the same types of terrain, the correlation of heavy CP is higher than that of heavy SP, and the correlation of heavy CP over mountains is the most significant, which may be related to different mechanisms of heavy CP and SP. For heavy CP, as shown in Figure 12d–f, the slope of the regression equation of high mountains is the largest, followed by mountains and plains in sequence, which shows that when near-surface Dm grows to the same diameter, STA over high mountains is usually the lowest while that over plain is usually the highest.

**Table 2.** Statistics on the proportion of shallow convective (stratiform), moderate convective (stratiform) and deep convective (stratiform) over different types of terrain in the Sichuan Basin and its surrounding detected by GPM DPR from May to September of 2014–2021.


Numbers in brackets denote SP, and numbers outside brackets denote CP.

**Figure 12.** Scatter plots of storm top altitude and Dm of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains (**<sup>c</sup>**,**f**). r represents Pearson correlation coefficient, p represents significance level, y and red solid lines are the regression equation and N represents sample size.

Water vapor is one of the important physical quantities that affects precipitation. The momentum, water vapor and heat convergence in meteorological boundary layer all contribute to the rainstorm during the heavy rainfall processes. Figure 13 exhibits the horizontal distribution characteristics of total water vapor from the surface to FzH when heavy precipitation occurs. The maximum value of water vapor is mainly located over the plains, and the water vapor gradually decreases as the elevation increases, showing a

stair-like decrease from west to east. The horizontal distribution characteristics of water vapor are roughly similar to STA.

**Figure 13.** The horizontal Distribution of water vapor of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains (**<sup>c</sup>**,**f**).

Figure 14 shows the correlation between the sum of water vapor from the surface to FzH and STA. For the heavy CP and SP over plains and mountains except high mountains passed with 0.01 significance level, which suggests there is a positive linear correlation between the sum of water vapor from surface to FzH and STA. For the same terrain, the correlation of heavy SP is more relevant than that of heavy CP, and it is most significant over plains.

**Figure 14.** Scatter plots of water vapor and storm top altitude of two types of heavy precipitation over plains (**<sup>a</sup>**,**d**), mountains (**b**,**<sup>e</sup>**), and high mountains (**<sup>c</sup>**,**f**). r represents Pearson correlation coefficient, p represents significance level, y and red solid lines are the regression equation and N represents sample size.
