HTML
-
The ERA-Interim 6-hourly wind reanalysis data at 500 hPa with a 1° × 1° resolution from May to October during 1981-2016 are used [37]. The gauge-based daily precipitation dataset (version 3.0) from 1200 UTC to 1200 UTC produced by the National Meteorological Information Centre of the China Meteorological Administration are also employed.
-
In this paper, the MSL is identified by a combination of three parameters: the zonal shear of the meridional wind, the relative vorticity and the zero line of meridional wind. The specific equations are as follows.
$$ \left\{\begin{array}{l} \partial v / \partial x>0 \\ \zeta>0 \\ v=0 \end{array}\right. $$ (1) where v represents the meridional wind, $ \zeta$ is the relative vorticity $ \left(\zeta=\frac{\partial v}{\partial x}-\frac{\partial u}{\partial y}\right)$, and x(y) represents the zonal (meridional) coordinate.
Given the spatial resolution and the study area, the identification of an MSL should meet the following conditions. Equation (1) is satisfied at each grid point. Meanwhile, the line connecting these grids crosses over 5 degrees along the y coordinate.
-
New variables are adopted for the statistical analysis.
(i) Definition of an MSL day
If an MSL occurs during any of the 4 observation in a day (i.e., 0000 UTC and 0600 UTC in the current day as well as 1200 UTC and 1800 UTC in the previous day), then this day is considered as an MSL day.
(ii) Definitions of a rainstorm day in the vicinity of the TP
The neighbouring area of the eastern TP is defined as the region of (103° E-110° E, 25° N-40° N) to the east of the TP. The region mainly includes east-central Sichuan Province, Chongqing Municipality, Guizhou Province, Ningxia Autonomous Region, South Shanxi Province, Inner Mongolia Autonomous Region and northeast Yunnan Province.
If the 24-hr accumulated rainfall over 50 mm is observed by more than five stations in the neighbouring area of the TP in one day, then that day is identified as a rainstorm day in the vicinity of the TP.
(iii) Definition of an MSL rainstorm day
If a day is both an MSL day and a rainstorm day in the vicinity of the TP, and the rainstorm is close to the MSL (i.e., less than 5 degrees along the x axis), then the day is defined as an MSL rainstorm day.
-
The altitude of most areas to the west of 85°E in the TP is above 4000 m, and only 4% of the total ground stations locate in this area. Therefore, the statistics of MSL days and MSL rainstorm days refer to those in the area to the east of 85°E.
Statistics show that due to the fast movement of MSLs over the TP, there are much fewer rainstorm days occurring in the TP than those in the neighbouring area of the TP. Over the last 36 years, the MSL-induced rainstorm days in the main body of the TP have been only 30 in total. While the corresponding number in the TP neighbouring area amounts to 885 days in total. This result suggests that most of the MSL rainstorm days take place in the neighbouring area of the TP (the rainstorm discussed in this study refers to the rainstorm in the neighbouring area of the TP).
2.1. Data
2.2. Methods
2.2.1. MSL OBJECTIVE IDENTIFICATION METHOD
2.2.2. DEFINITIONS OF VARIABLES
2.2.3. SCOPE OF STUDY
-
The cumulative frequency distribution of the MSL from 1981 to 2016 is shown in Fig. 2.
Figure 1. Examples of (a) an MSL day without rainstorm (May 5, 2013) and (b) an MSL rainstorm day (May 25, 2013). The blue line indicates the objectively identified MSL; the orange border encircles the main body of the TP; the black wind barbs indicate the 500 hPa wind; the green plus signs indicate daily precipitation < 25 mm; the green dots indicate daily precipitation between 25 and 50 mm; and the red triangles indicate daily precipitation ≥ 50 mm.
Figure 2. Cumulative frequency distribution of the MSL from 1981 to 2016. The thick black line indicates the axis of average high-occurrence frequency (Unit: 103).
It can be seen that there are two high-occurrence frequency centers of MSLs: one is located over the steep-slope area of the eastern TP (100° - 105° E, 27° - 37° N), and the other is over the central TP between northern Tibet and western Qinghai Province (86°-90°E, 30°-37° N). The frequency distribution of MSLs from the central TP to the eastern TP (90° - 103° E) shows a"high-low-high"pattern, which suggests that some of the MSLs generated over the central TP move eastward towards the eastern TP, while the others dissipate locally. There is a high-occurrence frequency value of above 1500 over the border region between the western Sichuan Plateau and Southeast Qinghai (100° E, 33° N), which is one of the source regions of the plateau vortex [19, 38]. This result reveals a close relationship between the plateau vortex and the MSL. The area with the low occurrence frequency of MSLs lies at 85°E to the west of the TP and southern Tibet.
Monthly statistics show (Fig. 3) that the highest occurrence frequency of MSLs occurs in July, followed by June, August, May and October. For the high-occurrence frequency center of MSLs over the central TP, the frequency value increases from June to August and decreases during the three following months. While for the other high value center, its value reaches a similar peak value as that of the one over the central TP, and the meridional span peaks in July and then narrows in May and June. The occurrence frequency value in August is the lowest, but its corresponding meridional span with the frequency above 200 is similar to that in May and July. In September, this center splits into two high-value centers in the vicinity of Chengdu City.
Figure 3. Monthly cumulative frequency distribution of the MSL from 1981 to 2016. (a) May, (b) June, (c) July, (d) August, (e) September, and (f) October.
The occurrence frequency value over 200 (90°-105°E) signifies that the MSL is active in the vast area of the central and northern Tibet and the western Sichuan Province in July and August. The high occurrence frequency centre of MSLs further locates to the north in May, June, September and October, compared with that in the midsummer. Meanwhile, no high occurrence frequency centre is observed over the western Sichuan Plateau during those four months.
-
The sum or difference of a variable's average and its standard deviation is taken as the threshold to determine the high-occurrence and low-occurrence year. We stipulate that when the amount of MSL days exceeds the sum of its average and the standard deviation, this year is considered to be a high-occurrence year for MSL days; when the amount is less than the difference, this year is considered to be a low-occurrence year. This method is also applied for rainstorm days and MSL rainstorm days.
The interannual variation of MSL days from 1981 to 2016 is shown in Fig. 4a. The annual average number of MSL days is 42.2 in the study area (i.e., east of 85°E), which accounts for 23% of the total number of MSL days during the study period. In this study, the number of MSL days in high-occurrence year should exceed 50.3 and that in the low-occurrence year should be less than 34.1. It can be seen that 1998, 2014 and 2016 are three high-occurrence years with the maximum of 62 days observed in 2014. While the low-occurrence years are 1989, 1997, 2000, 2004 and 2006 with the minimum of 22 days in 2006.
Figure 4. Daily variation curves of the number of (a) MSL day, (b) rainstorm day in the vicinity of the TP and (c) MSL rainstorm day during May-October from 1981 to 2016. The dotted line is the average number of days, and the real line is the trend line.
Further analysis shows that the number of MSL days has varied sharply since the 1980s, and the interannual variability has increased since the late 1990s. Both the highest - and lowest-occurrence years occur in the 21st century. It is noted that in the highest-occurrence years of 1998 and 2016, flood disasters happen along the Yangtze River. In addition, in the highest-occurrence year of 2014, the autumn rainfall in west China is abnormally heavy. Also, from September to October in 2014, the number of MSL days is the greatest compared with the corresponding periods in other years. The interannual trend of the vertical shear line occurrence days shows that in the past 36 years, the growth rate of MSL days has been 0.199/ year, showing an increasing trend with time.
As for the number of rainstorm days, the average number is 59.6, featured with obvious interannual variability (Fig. 4b). Similarly, the number of rainstorm days exceeding 67.0 days corresponds to a high-occurrence year and the rainstorm days less than 52.2 days corresponds to a low-occurrence year. The high-occurrence years are 1983 and 1998 with the same maximum number of 78 days. The low-occurrence years are 1994, 1997, 2001, and 2006 and the lowest number of 43 days is seen in 1997. The interannual variation shows an evident decreasing trend with a decline rate of -0.123/year. The decrease is more distinct (also with a smaller amplitude) in the 1980s and 1990s than in the 21st century.
Figure 4c shows that there are 23.8 MSL rainstorm days in boreal summer half-year for the 36-year average. Accordingly, the year with over 29.5 MSL rainstorm days is defined as a high-occurrence year; and the year with MSL rainstorm days less than 18.1 is considered as a low-occurrence year. The high-occurrence years are 1990, 1998, and 2013 and the highest number of 40 days is in 1998. Meanwhile, the low-occurrence years are 1997 and 2006, with the minimum of 10 days in 2006. The interannual variation of MSL rainstorm days
denotes that the amplitude of the MSL rainstorm days changes little from the 1980s to the mid-1990s. In the late 1990s, the amplitude reaches its maximum. After that, it decreases gradually to its average level. The above results indicate that the interannual variation of MSL rainstorm days is not significant (Fig. 4c).
Figure 5 shows the three variables with significant periodic variation based on the Morlet wavelet power spectrum. In the past 36 years, the periodicity for the MSL days has been 2-4 years, and since the mid-1990s, the periodicity of 4-6 years has been also observed. Both of the above periodicities pass the 0.05 confidence test (Fig. 5a).
Figure 5. Morlet wavelet power spectrum of the number of (a) MSL day, (b) rainstorm day, and (c) MSL rainstorm day during May-October from 1981 to 2016.
The rainstorm days also show a 2-4 years periodicity and a 4-6 years periodicity, and both periodicities pass the 0.05 confidence test (Fig. 5b). These periodicities are similar to those of the MSL days, but differ in occurrence time. The 2-4 years periodicity of the rainstorm days is first found in the late 1980s and lasts until 2010, and the 4-6 years periodicity is observed in the early 1980s through the mid-2000s.
As for the MSL rainstorm days, the Morlet wavelet power spectrum is used. Both of the two periodicities of 2-4 and 4-6 years pass the 0.05 confidence test (Fig. 5c). The former periodicity started in the late 1980s and lasted until 2010, while the latter lasts from the mid-1990s until now. These statistical features are similar to those of the MSL days.
-
Figure 6a shows that the total number of MSL days varies in each month. In July, there are 10.2 MSL days/ year, where approximately 1 in every 3 days is an MSL day. The number of MSL days increases gradually from May to July and decreases from July to October. Thereby, October is the month with the lowest number of MSL days (4.2 days/year).
Figure 6. Monthly variation of the number of (a) MSL day, (b) rainstorm day and (c) MSL rainstorm day during May-October from 1981 to 2016.
By analyzing the monthly-mean rainstorm days (Fig. 6b), it is found that the rainstorm day mainly occur during the main flood period from June to August (JJA), with a total of 3 months up to 1599 days, accounting for 75% of the total days in the boreal summer half year. Different from the findings of Zhang et al. [36], the rainstorm days are more concentrated in JJA than the TP rainstorm days. July is the month with the most rainstorm days not only in the neighbouring area of the TP (631 days) but also in the TP. In contrast, October is the month with the least number (61 days), accounting for only 2.8% of the total number in the boreal summer half year.
Like the above two variables, the MSL rainstorm days show distinct monthly variations, with the maximum in July and the minimum in October (Fig. 6c). The number of MSL rainstorm days in JJA amounts to 670, accounting for 78% of the total number in the boreal summer half year.
-
Based on the above statistics, average numbers of MSL days, rainstorm days and MSL rainstorm days are 42.2 days / year, 59.6 days / year and 23.8 days / year, respectively. In terms of the climatology, the number of the three variables in turn accounts for 22.9%, 32.4% and 12.9% of the 184 total days, respectively.
It can be concluded that the 36-year average number of MSL rainstorm day is 23.8, accounting for 56.4% of MSL days (42.2 days) and 39.9% of rainstorm day (59.6 days). The results reveal that about 56% of the MSLs can cause rainstorms, and approximately 40% of the rainstorms are induced by the MSLs.
The correlation coefficient between the MSL days and rainstorm days in August is 0.628, which passes the 0.01 confidence test. The correlation coefficient in May is 0.354 and passes the 0.05 confidence test. In addition, the correlation coefficient in July and JJA is also high (see Table 1). The previous analysis shows that the high-occurrence years of MSL days correspond to those of rainstorm days (e.g., 1998). It is the same case with the low-occurrence years for the two variables (e. g., 1997 and 2006). These results indicate a close relationship between these two variables. Despite the complexity of the mechanisms of rainstorms, the MSL is clearly one of the most important influence systems.
May Jun Jul Aug Sep Oct Main flood season (JJA) Boreal summer half-year Coefficient 0.354 0.076 0.327 0.628 0.098 0.178 0.305 0.194 Table 1. Coefficient of MSL day and rainstorm day.
3.1. Geographic distribution of the MSL
3.2. Interannual variation and periodicity of the three new variables
3.3. Monthly variation of MSL day, rainstorm day and MSL rainstorm day
3.4. Relationship between the MSL and rainstorms
-
Similarities and differences between the climatic characteristics of the TSL and the MSL are investigated using the same dataset (Table 2) as those in the study of TP TSLs by Zhang et al. [36].
MSL TSL Annual average days of shear line 42.2/year 65.3/year The most-active month July June Periodicity 2-4 years and 4-6 years 2-4 years, 4-6years and 4 years Major impact areas the neighbouring area of the TP the main body of the TP Annual growth rate 0.199/year 0.024/year Annual average days of shear line rainstorm 23.8/year 33.1/year Ratio of shear line rainstorm 56% 50.7% Table 2. Characteristic comparison between the MSL and the TSL.
The annual average number of MSL days is 42.2, while that of the TSL is 65.3 and is 1.5 times of the MSL's. This result is consistent with a previous study, in which the number of TSLs is twice the number of MSL. Both of them are most active in midsummer (JJA). The difference is that the MSL mainly appears in July whereas the TSL in June.
The Morlet wavelet analysis shows that they both change mainly with the periodicities of 2-4 years and 4-6 years. However, the change characteristics are different in different periods. For the TSL, a distinct cycle of 4-6 years in the 1980s, a cycle of 2-4 years in the late 1980s to 1990s and a quasi-4-year cycle after the late 1990s exist. While for the MSL, the periodicity of 2-4 years has changed to a cycle of 4-6 years since the mid-1990s. Under the background of climate warming, TSL days do not decrease significantly with time but fluctuate within a certain range, while MSL days increase with time with a growth rate of 0.199/year.
The influence areas of the TSL and the MSL are different. The TSL is generally parallel to the TP topography, and it moves within the area from 30° N to 35° N. The TSL-induced rainstorm often appears over the TP, and only a few are found when the TSL moves eastward or southward to the southeast of the TP. The average number of TSL rainstorm days is 33.1, which is 50.7% of average MSL days (65.3 days). Additionally, over half of TSLs can lead to rainstorms over the TP.
The MSL shows a quasi-north-south direction, and the two high-occurrence frequency centers of it are distributed in the middle of the TP and the steep slope area of the eastern TP. The average number of MSL rainstorm days is 23.8, accounting for 50.7% of the average number of MSL days (42.2 days). In the neighbouring area of the TP, nearly 56% of the MSLs result in rainstorms, indicating a close relationship between the MSL and the rainstorm in this region.