Figure 2 shows a case study of TRMM PR observations over the Tibetan Plateau on 16 June 1998 (Orbit 03166). The black rectangles show the MBRs of the precipitation areas and indicate that, subjectively, although the size and shape of these six precipitation areas are different, the MBR can describe the morphology of the precipitation areas and completely frame them. In addition to these six precipitation areas, there are a large number of precipitation pixels at (32.5°N, 99.5°E) and (32.5°N, 97.5°E). These pixels are not identified as precipitation areas and are not considered in the statistics because they are truncated by the PR swath or consist of fewer than four continuous pixels.
Figure 2. Example of precipitation areas and their corresponding MBR values on 16 June 1998. The dashed line indicates the PR swath.
Table 1 lists the morphological characteristics and precipitation of the six precipitation areas shown in Fig. 2. Area A is the largest precipitation area with a scale of 191.2 km, whereas area D has the smallest scale of 22.9 km. The shape of area F is slender with an α value of 0.327, whereas areas A and C are more square-like with α > 0.6, which is consistent with the subjective cognition in Fig. 2. Because the 3D morphology γ includes the average echo-top height, area A shows characteristics we describe as plump-and-short, whereas area E appears thinner and taller. Although areas D and E show similar scales (22.9 and 24.5 km, respectively), the difference in γ exceeds 0.15, showing a completely different spatial morphology. The five rain areas B–F show very similar values of RRav, all of which are < 0.9 mm h−1.
L (km) α γ RRav (mm h−1) A 191.2 0.636 0.055 1.519 B 30.7 0.456 0.359 0.853 C 40.0 0.622 0.300 0.767 D 22.9 0.541 0.428 0.896 E 24.5 0.563 0.581 0.735 F 37.8 0.327 0.318 0.865
Table 1. Morphological characteristics and precipitation of the six precipitation areas shown in Fig. 2.
The relationship between the morphological characteristics of precipitation areas and the near-surface rain rate is shown by 2D histograms, where the black solid line represents the median RRav within a certain morphological interval. Overall, the precipitation areas over the eastern Tibetan Plateau (Fig. 3c) have a wider distribution of RRav than those of the same scale over the western Tibetan Plateau (Fig. 3a), and there are more large-scale precipitation areas over the eastern Tibetan Plateau. The precipitation areas with RRav > 5 mm h−1 (~100.7) are not large in scale, which is consistent with the well-known isolated convection over the Tibetan Plateau. As the scale of the precipitation areas increases from 10 km (101) to 20 km (~101.3), the median RRav shows a downward trend. As the scale continues to increase, the median RRav increases from 1.12 mm h−1 (100.05) to 1.41 mm h−1 (100.15), although it stabilizes when the scale is ≥ 50 km (101.7) over the western and central Tibetan Plateau (Figs. 3a and b). The median RRav of precipitation areas over the eastern Tibetan Plateau (Fig. 3c) continues to increase to a maximum >2 mm h−1 (100.3), 78% greater than the RRav of small-scale precipitation areas. The most common precipitation areas over the whole of the Tibetan Plateau occur at a scale of 20 km (101.3) and an RRav of 1 mm h−1.
Figure 3. 2D frequency histograms of L and RRav. The black solid line represents the median RRav in a particular L interval.
Overall, the difference in α is not large in the three regions and the distribution of α shows two clear peaks at about 0.6 and close to 1 (Fig. 4). The frequency peak at α close to 1 is consistent with previous studies, with more square-shaped precipitation areas over land and more linear precipitation areas over the oceans (Liu and Zipser, 2013). RRav also shows the widest distribution in these two α peak areas, whereas the most slender precipitation areas (α = 0.2) have the narrowest distribution and smallest value of RRav. Interestingly, the median value of RRav tends to increase with α in all three regions (Fig. 4) and the increased slope is the largest over the eastern Tibetan Plateau (Fig. 4c). When α increases from 0.2 to 0.4, the median value of RRav increases from 1 to 1.12 mm h−1 (100.05) over the western Tibetan Plateau and from 1 to 1.26 mm h−1 (100.1) over the eastern Tibetan Plateau. The median value of RRav increases to 1.58 mm h−1 (100.2) over the eastern Tibetan Plateau as α increases to 0.8, whereas it remains stable over the whole Tibetan Plateau with an increase in α from 0.8.
Figure 4. 2D frequency histograms of α and RRav. The black solid line represents the median value of RRav in a particular α interval.
The echo-top height indicates the depth of the development of precipitation. A higher echo-top height will extend the growth path of precipitation particles and will usually produce stronger precipitation. A large number of studies have focused on the relationship between the echo top and the near-surface rain rate (Hamada et al., 2015; Fu et al., 2018). Our recent study found that the echo-top height is also closely related to the near-surface distribution of droplet size over the Tibetan Plateau (Chen et al., 2020). We can quantitatively analyze whether the precipitation area is thin-and-tall or plump-and-short by dividing the average echo-top height and the average size to obtain the 3D morphology of the precipitation areas. Figure 5 shows that the 2D histograms of γ and RRav have almost the same characteristics over the western and central Tibetan Plateau. The samples are mainly concentrated in the area where γ is 0.3–0.7 and RRav is 10−0.1 to 100.2 mm h−1 (Figs. 5a and b), whereas the distribution of RRav over the eastern Tibetan Plateau is wider, especially in the area where γ < 0.6 (Fig. 5c). At γ < 0.6, the median value of RRav decreases rapidly with the increase in γ and the rate of decrease is greatest over the eastern Tibetan Plateau. After γ exceeds 0.6, the median value of RRav increases with γ. According to the frequency distributions of L–γ and Hav–γ shown in Fig. A1 (in the Appendix), we speculate that when the precipitation areas are small, the vertical development promotes the increase of rain rate; after the precipitation areas reach a certain scale, the increase in the horizontal scale dominates the increase in rain rate. This key turning point occurs when γ is between 0.6 and 0.7. Specifically, the median value of RRav doubles from 1.12 mm h−1 (100.05) to 2.24 mm h−1 (100.35) over the central Tibetan Plateau as γ increases from 0.7 to 0.95 (Fig. 5b). These results indicate that the near-surface rain rate is more sensitive to a thin-and-tall 3D morphology over the central and western Tibetan Plateau, whereas it is more sensitive to a plump-and-short 3D morphology over the eastern Tibetan Plateau.
By setting a threshold for L, the precipitation areas over the Tibetan Plateau can be divided into three categories: large, medium and small (Fig. 6). The CFADs show a similar vertical pattern over the three regions for precipitation areas in the same scale interval, except for slightly different heights overall. There are clear differences in the vertical structure for precipitation areas of different sizes. Taking the eastern Tibetan Plateau as an example, the outermost contour of the small precipitation areas is always below 12 km (Fig. 6c), whereas that of the large precipitation areas can reach 15 km (Fig. 6i). This indicates that although isolated convection is often triggered over the Tibetan Plateau (Fu et al., 2006), the deepest precipitation is embedded in large precipitation areas. From the perspective of the modal contour (> 0.5), the reflectivity of the large precipitation areas increases from 18 dBZ at 8 km to 21 dBZ at 6 km (Fig. 6i). Such a slow growth rate may correspond to the deposition process—that is, the ice particles grow by directly absorbing the surrounding water vapor. From 6 to 5 km, the modal contour increases rapidly from 21 to 27 dBZ and the outermost contour increases to 32 dBZ (Fig. 6i). Such a strong increase in reflectivity corresponds to aggregation and riming processes just above the freezing level (~5.6 km). Melting near the bright band is also important for the increase of reflectivity, because the partially melted ice particles exhibit the complex index of refraction of the water drops (roughly four times greater than that of ice particles) but maintain their sizes. In contrast with the large precipitation areas, the increase in the outermost contour with decreasing height mainly appears above 8 km for small precipitation areas (Fig. 6c) and the increase in the modal contour with decreasing height appears above 6 km.
Figure 6. CFAD of the precipitation-area reflectivity with different values of L. The left-hand, center and right-hand panels represent the western, central and eastern Tibetan Plateau (TP), respectively: (a–c) L < 20 km; (d–f) 20 < L <50 km; and (g–i) L > 50 km.
The response of the precipitation area CFAD to different thresholds of α is less clear (Fig. 7). The outermost contour is almost the same regardless of the reflectivity or height. The only difference is in the modal contour: the square-shaped precipitation areas have an overall modal reflectivity 1–2 dBZ greater than that of the linear precipitation areas and there is a slight bulge at 24 dBZ and 7 km height (Fig. 7g–7i), which is most obvious over the eastern Tibetan Plateau (Fig. 7i). This result indicates that the vertical structure of the extreme precipitation areas (outermost contour) is not sensitive to the shape of the precipitation area. The specific mechanism of the shape of the precipitation area to the increase in precipitation requires further study.
Figure 7. CFAD of the precipitation-area reflectivity with different values of α. The left-hand, center and right-hand panels represent the western, central and eastern Tibetan Plateau (TP), respectively: (a–c) α < 0.4; (d–f) 0.4 < α <0.8; and (g–i) α > 0.8.
Figure 8 shows interesting results in that different 3D morphologies correspond to completely different vertical structures of reflectivity. A value of γ < 0.4 is defined as relatively plump-and-short—that is, the horizontal scale of the precipitation area is much greater than the vertical scale. A value of γ > 0.7 is regarded as a thin-and-tall morphology, where the horizontal and vertical scales are almost the same.
Figure 8. CFAD of the precipitation-area reflectivity with different values of γ. The left-hand, center and right-hand panels represent the western, central and eastern Tibetan Plateau (TP), respectively: (a–c) γ < 0.4; (d–f) 0.4 < γ <0.7; and (g–i) γ > 0.7.
Taking the eastern Tibetan Plateau as an example, the plump-and-short precipitation areas have the highest outermost contour of up to 14 km (Fig. 8c). However, under constant reflectivity, the plump-and-short precipitation areas have a large difference in height between contours. For example, the distance between the 0.05 and 0.1 contours at 18 dBZ is > 2 km (Fig. 8c), whereas this distance is < 1.5 km for thin-and-tall precipitation areas (Fig. 8i). This suggests that although a few pixels of the plump-and-short precipitation areas have high rain tops, more of the other pixels have not been equally developed in terms of depth.
There is a bulge at a height of 5 km (Fig. 8c), which corresponds to the height of the freezing level over the Tibetan Plateau in summer. This indicates that there is a large proportion of stratiform precipitation pixels in the plump-and-short precipitation areas. The reflectivity in the thin-and-tall precipitation areas (Fig. 8i) is enhanced at about 10 km and the 0.5 contour exceeds 27 dBZ at 8 km, whereas the reflectivity of the plump-and-short precipitation areas is only 21 dBZ in the same situation. The reflectivity of the thin-and-tall precipitation areas does not show a sudden increase near the freezing level, indicating that there are more convective pixels with vigorous development in the vertical direction. The precipitation areas with a moderate 3D morphology have the lowest CFAD with a weak bulge at the freezing level (Fig. 8f), meaning that these precipitation areas produce mainly weak convective precipitation and may contain a few stratiform precipitation pixels.
Figure 9 shows the spatial distribution of the precipitation area RRav with different morphological characteristics over the Tibetan Plateau. There are two major centers of precipitation over the Tibetan Plateau: east of the Tibetan Plateau (east of 100°E) and south of the Tanggula Mountains (30°–33°N, 85°–93°E). In both regions, RRav has a clear tendency to increase with the size of the precipitation area. The value of RRav increases from 1.6 to >2.5 mm h−1 over the Tibetan Plateau east of 100°E and to nearly 2.5 mm h−1 over the southern slopes of the Tanggula Mountains. It is worth noting that these two regions are relatively flat: one is 3 to 4.5 km above sea level, and the other is mainly 4.5 to 5 km above sea level (Fig. 1b). By contrast, RRav and the scale of the precipitation area show different relationships in areas with a low RRav. Over the northern part of the Tibetan Plateau, RRav increases from ~1.2 mm h−1 in small precipitation areas to ~1.6 mm h−1 in large precipitation areas. Over the trumpet-shaped region (~30°N, 90°E) and the northwestern Tibetan Plateau (north of 32°N, west of 80°E), RRav is ~1.4 mm h−1 regardless of the scale of the precipitation area. The high altitude and steep terrain in these areas (Fig. 1b) may not be conducive to the development of scale and rain rate.
Figure 9. Spatial distribution of the average precipitation area RRav with different morphological characteristics: (a) L < 20 km; (b) 20 < L < 50 km; (c) L > 50 km; (d) α < 0.4; (e) 0.4 < α < 0.8; (f) α > 0.8; (g) γ < 0.4; (h) 0.4 < γ < 0.7; and (i) γ > 0.7.
The relationship between α and RRav (Figs. 9d–f) shows that RRav increases with increasing values of α, similar to that shown in Figs. 9a–c. As α increases to 0.8, the area with RRav < 1.4 mm h−1 decreases significantly. The RRav values of linear precipitation areas are almost all < 1.8 mm h−1 (Fig. 9d). By contrast, for the square-shaped precipitation areas, part of the central Tibetan Plateau has values of RRav > 2 mm h−1 and almost all of the eastern Tibetan Plateau is > 2 mm h−1 (Fig. 9f). According to our criteria, the typical example of isolated convective rain shown by Fu et al. (2006) is a square-shaped precipitation area. We therefore speculate that square-shaped precipitation systems are produced by the strong solar heating at flat areas over the Tibetan Plateau, and are continuously strengthened by the release of latent heat. However, some precipitation systems are blocked by topography and develop linearly along the mountain slope, which hinders energy transfer and weakens precipitation intensity.
With an increase in γ, RRav shows a tendency to first decrease and then increase. The thin-and-tall precipitation area in the northeast corner of the Tibetan Plateau (Fig. 9i) shows the largest RRav, which can reach >3 mm h−1. In the trumpet-shaped region, the thin-and-tall precipitation areas show a larger RRav than the other 3D morphologies of precipitation areas. We speculate that those thin-and-tall precipitation areas mainly appear in the daytime, corresponding to stronger surface heating and larger convective available potential energy than those in the plump-and-short precipitation areas. This phenomenon has been revealed by using CloudSat observations that the outflow height of deep convection systems is higher but the horizontal range is smaller over the southern slope of the Tibetan Plateau during daytime than those during nighttime (Chen et al., 2019). For precipitation areas with a moderate γ, RRav is < 1.8 mm h−1 in most regions of the Tibetan Plateau.
|L (km)||α||γ||RRav (mm h−1)|