Analysis of the Characteristics of the Low-level Jets in the Middle Reaches of the Yangtze River during the Mei-yu Season

It is important to study how low-level jets (LLJs) contribute to the occurrence and development of heavy rainfall events from both the scientific and social points of view (Chen and Yu, 1988; Cook and Vizy, 2010; Du and Chen, 2019a). LLJs can be broadly divided into synoptic-system-related low-level jets (SLLJs) and boundary layer jets (BLJs) based on their altitude. SLLJs often occur from 600 to 900 hPa and are usually related to synoptic-scale weather systems (Du et al., 2012). BLJs generally occur in the planetary boundary layer, especially in the stable nocturnal boundary layer, with significant vertical shear of the horizontal winds (Whiteman et al., 1997).

The objective criteria of LLJs vary from region to region (Whiteman et al., 1997; Wei et al., 2014; Vanderwende et al., 2015). As a result of limitations in the spatiotemporal resolution of observational data, previous studies of regional LLJs in China were based on the maximum wind speed in a certain layer of the isobaric surface and did not explicitly consider the vertical shear intensity of the wind speed (Qian et al., 2004). As observational technology has developed, more researchers have started to study LLJs in conjunction with vertical shear (Hao et al., 2001; Du et al., 2012; Du and Chen, 2018, 2019a, b). Hao et al. (2001) defined BLJs as events with wind speeds >10 m s⁻¹ at any time, occurring at any altitude below 1500 m in the Zhejiang region, lasting for more than one observation period (2 h) and with an obvious protruding "nose" in the vertical wind profile. BLJs in the Beijing area generally occur against a background of high daytime temperatures or at night when there is locally heavy rainfall, with significant diurnal variations in intensity and an obvious nose-like vertical distribution (Sun, 2005; Li and Shu, 2008).

In addition to the difference in the occurrence height of BLJs and SLLJs, their spatiotemporal characteristics and mechanisms of formation are also different (Blackadar, 1957; Wang and Zhang, 2012; Rajewski et al., 2013; Du and Chen, 2018). Theoretical, observational, and numerical modeling studies of LLJs have been ongoing for many years. The main factors affecting the formation of LLJs include inertial oscillations, orographic blocking, and synoptic weather systems (Blackadar, 1957; Li and Chen, 1998; Zhang et al., 2007; Lin et al., 2011; Du et al., 2012). The earliest theoretical analyses began with Blackadar (1957), who studied the emergence of LLJs in the stable boundary layer. He showed that the inversion layer begins to form in the boundary layer after sunset and that the sudden decrease in frictional restraint excites inertial oscillations in the non-geostrophic wind components, leading to the formation of night-time super-geostrophic wind speed extremes. Some progress has been made in the study of LLJs in eastern and southwestern China, mainly focusing on the effect of LLJs on the transport of water vapor and heavy rainfall (Sun and Zhai, 1980; Zhang et al., 2019).

Nocturnal BLJs have long been of interest in China (Li et al., 1982; Jin et al., 1983; Fu et al., 2019). In Beijing, 30% of nocturnal observations record BLJs. BLJs are well-correlated with local valley wind circulations under stable boundary layer conditions at night (Li and Shu, 2008). The vertical mixing of the boundary layer during the day and the inertial oscillation at night are important processes influencing BLJs and the diurnal variation of precipitation in the Yangtze River Basin (Xue et al., 2018).

LLJs usually occur overnight and in the early morning (from 2200 LST and 0800 LST the next day, LST = UTC + 8 hours) (Bonner, 1968; Astling et al., 1985; Fu et al., 2019). The frequency and intensity of BLJs in both Shanghai and Tianjin are characterized by a greater daily variation at night than during the day. As a result of the local topography and weather conditions, Shanghai is dominated by southwesterly and easterly winds, whereas Tianjin, at a slightly lower altitude, is dominated by northeasterly and southerly winds (Du et al., 2012; Wei et al., 2014). LLJs also affect precipitation by influencing the transport of water vapor, vertical shear, and divergence fields. Du and Chen (2018, 2019a) showed that the outlet zone of BLJs provides low-level convergence, whereas the SLLJ inlet zone provides mid- to upper-level divergence. Both types of jets produce mesoscale uplift near the South China coast, which favors the triggering of convection.

The mei-yu is a specific period of the East Asian summer monsoon and produces most of the summer rainfall from the Yangtze–Huaihe River valleys in central China northeastward to Japan via a quasi-stationary elongated rainband from mid-June to mid-July (Qian et al., 2004). The mei-yu season is one of three periods of heavy rainfall in China, and the frontal-based heavy rainfall of this period contributes to a significant portion of the total amount of precipitation in central China (Wang et al., 2019). BLJs are closely related to the East Asia monsoon and mei-yu fronts. However, few analyses have been conducted to characterize the heavy rainfall related to BLJs in the Yangtze River basin, especially during the mei-yu season. We used intensive sounding and mobile boundary layer wind profiler radar data during the mei-yu season to show the existence and daily variation of BLJs in the middle reaches of the Yangtze River and their subsequent impact on heavy rainfall events. The observation time of conventional sounding data does not match the timing of LLJs because upper-air sounding stations in China usually only observe twice daily (at 0800 LST and 2000 LST).

The remainder of this paper is organized as follows. Section 2 gives a preliminary introduction to our data and methods, and section 3 describes the selection of the jet observation profiles in the middle reaches of the Yangtze River. Section 4 analyzes the statistical characteristics of the BLJs in this region during the mei-yu season, and section 5 gives examples of heavy rainfall events in the Yangtze River Basin accompanied by a BLJ. Our summary and discussion are given in section 6.

We used intensive sounding observations to analyze the vertical structure and daily variations of LLJs. Intensive sounding observations have a higher temporal resolution than conventional sounding observations. Typically, upper-air sounding stations observe only twice daily (at 0800 and 2000 LST). This study used observation times at Wuhan Station of 0200, 0500, 0800, 1100, 1400, 1700, 2000, and 2300 LST with a 3-h temporal resolution from 16 June to 30 July 2010. The data include the wind speed and direction with a 30-m vertical resolution. Figure 1b shows the wind speed observation profile as a solid black line (observed at Wuhan Station at 0500 LST on 2 July 2010).

Figure 1.  (a) Location of Wuhan Sounding Station and the surrounding terrain (units: m; the blue line represents the Yangtze River); (b) Wind speed profile observed at Wuhan Station at 0500 LST on 2 July 2010. The black line represents the wind speed profile using the original data; the cyan line represents the wind speed profile drawn after removing the equal wind speed layer; the red circles represent the noses of the jet-like profile.

We also collected daily precipitation data and used the ERA5 reanalysis dataset (Hoffmann et al., 2019) to make a preliminary analysis of the formation mechanism of LLJs. The mobile boundary layer wind profiler radar data observed at Xianning Station was provided by the Institute of Heavy Rain, China Meteorological Administration. Its observation interval is 4 minutes or 5 minutes (alternately), with the lowest detection height of 43m and the highest detection height of 8142m. We also obtained daily precipitation data from surface rain gauges at weather stations throughout China from the National Meteorological Information Center of the China Meteorological Administration to determine the importance of LLJs for rainfall. These high spatiotemporal resolution data have proven reliable (Wan et al., 2011), and have previously been used to study a short-term severe rainfall process and an extreme precipitation process in the middle reaches of the Yangtze River (Wang et al., 2012, 2019).

The lowest altitude of the vertical range was chosen to be 30 m after considering the influences of the environment and surface features (such as trees) on the observational data. In conjunction with previous studies on LLJs, the highest altitude was chosen to be below 600 hPa (Chen and Yu, 1988; Chen et al., 1994, 2005; ), and the highest vertical range was selected as 4000 m altitude. We conducted validity tests on the observations in this range (Wei et al., 2014) and estimated the missing measurements using linear interpolation.

We used the intensive sounding observations from Wuhan Station to analyze the characteristics of BLJs and SLLJs in the middle reaches of the Yangtze River during the mei-yu season. Wuhan Station (30.6°N, 114.05°E) is located on the Jianghan Plain, a flat region between the Dabie and Jiuling mountains with the relief height of about 24 m. Figure 1a shows the location of Wuhan Station and its surrounding terrain.

The BLJs and SLLJs both exhibit a nose-like feature in the wind speed profile (a wind speed maximum at a certain height). We chose the jet-like profile following the method of Blackadar (1957); however, unlike Blackadar (1957), we attempted to find the wind speed extremes within 4000 m of the ground. It was difficult to extract information about the feature points of the nose-like profile due to the presence of several isotach layers in the original data. Therefore, we eliminated the influence of the multiple isotach layers by extracting them as a single layer and expressing the height of the isotach layers as the average height. The cyan line in Fig. 1b shows the processed wind speed profile. The red circle indicates the nose-like feature of the jet-like profile. The BLJ and LLJ profiles were selected based on the nose-like feature points; the related definitions are detailed in section 3.2.

To make the results more representative, the profiles used to study the LLJs in the middle reaches of the Yangtze River were based on the height–frequency distribution of the nose-shaped features and the frequency distribution of the wind speed. The height of the BLJ or SLLJ was represented by the height of the corresponding nose-shaped feature, whereas the wind speed intensity was represented by the wind speed and direction of the corresponding feature.

Most previous studies of the wind speed of LLJs follow Bonner (1968) (a wind speed maximum ≥12 m s⁻¹ and a falloff ≥6 m s⁻¹). However, the results of Bonner (1968) are based on the characteristics of jets over the Great Plains of the USA, where the topographic and climatic characteristics are very different from those in China. Some researchers also constrain the wind direction of jets, considering them to be southerly or westerly winds that meet certain criteria (Chen and Yu, 1988; Chen et al., 1994, 2005).

The vertical distribution of the wind field in the middle reaches of the Yangtze River from 16 June to 30 July 2010 (Fig. 2) shows that nose-like jet structures in the lower troposphere can also be observed in Wuhan, but the wind speed is slightly lower than the previous standard for LLJs. Therefore, it was necessary to choose an appropriate wind speed standard to study the characteristics of jets in the middle reaches of the Yangtze River. Consequently, we conducted a statistical analysis of the characteristics of the local wind speeds in the middle reaches of the Yangtze River.

Figure 2.  The evolving vertical distribution of the (a) wind speed (units: m s⁻¹) and (b) wind direction (units: °) from intensive sounding observations in Wuhan between 16 June and 30 July 2010.

The wind speeds in the nose-shaped feature below 1500 m in Wuhan were widely distributed, ranging from 4 to 18 m s⁻¹, whereas the wind speeds above 1500 m were more concentrated, ranging from 8 to 12 m s⁻¹ (Fig. 3a). Figure 3b shows that the nose-shaped feature in Wuhan was mainly concentrated below 1400–1600 m, with; a sharp decrease in frequency above this height. Bonner (1968) suggested that the upper limit of the jet should be set as the high-frequency height of the nose-shaped feature of the jet-like profile (a sharp decrease in frequency at higher levels). Referring to previous research (Hao et al., 2001; Sun, 2005), we defined a height of 1500 m to distinguish the BLJs and SLLJs in Wuhan—that is, LLJs occurring below 1500 m were defined as BLJs and those occurring between 1500 and 4000 m were defined as SLLJs. Figure 3b also shows that the distribution of the points in the nose-shaped feature points was positively skewed.

Figure 3.  (a) Frequency distribution of different wind speeds at different heights for the jet-like profile of the nose-shaped feature in Wuhan (horizontal axis indicates wind speed, units: m s⁻¹; shading indicates frequency). (b) Frequency distribution (black bars) and wind speed distribution of the nose-shaped feature at different heights (red dots are mean values and blue dots are median values). (c) Total frequency distribution for different wind speeds. These plots are statistically based on data from the intensive sounding observations.

We next determined the wind speed thresholds of LLJs, which make this research statistically significant and the representative jet samples. We then selected the high-frequency threshold wind speeds. The jet-like profile of the nose-shaped feature in Wuhan showed a single-peak distribution of wind speed frequencies with a maximum frequency interval of 6–8 m s⁻¹, which was observed 70 times (Fig. 3c). Therefore, we defined the wind speed threshold for LLJs in Wuhan as 8 m s⁻¹. In general, the definitions of BLJs and SLLJs were established according to the following criteria: (1) the maximum wind speed for BLJs was >8 m s⁻¹ below 1500 m (between 1500 and 4000 m for SLLJs); and (2) the wind speed must decrease by at least 2.5 m s⁻¹ from the height of the maximum wind speed to the wind speed minimum. A double low-level jet (DLLJ) was defined as when both a BLJ and a SLLJ were present at the same time.

Based on these conditions, there were 184 observed LLJs in the Wuhan area (a detection rate of 55.25%). Du et al. (2012) analyzed the frequency of LLJs before, during, and after the mei-yu season and reported a significant increase in the frequency of SLLJs during the mei-yu season, followed by a decrease after this season. Du and Chen (2019b) also found that the frequency of jets in South China was highest in June, when it was twice that of other months of the warm season. Thus, the specificity of the study period in our work contributes to the high detection results.

We counted the frequency distribution of LLJs at different heights (Fig. 4a). LLJs in the Wuhan area occurred more frequently between 300 and 1200 m and were most frequent (30 occurrences) between 900 and 1200 m. The frequency decreased sharply above 1200 m. SLLJs mainly occurred at altitudes of 1800–3300 m, with a maximum of 20 occurrences.

Figure 4.  Statistical plots based on intensive sounding observational data show (a) the frequency of LLJs at different altitudes and (b) the wind velocity observational profile for jets (red line) and non-jets (pink line).

Figures 4a and 4b both show that the vertical structure of LLJs during the mei-yu season in the middle reaches of the Yangtze River was mainly characterized by the vertical structure of BLJs, with an average height of about 1200 m and an average intensity >8.5 m s⁻¹. Below 4000 m, the non-jet composite wind speed was significantly lower than that of the jet composite. However, the observed non-jet profile in the Wuhan area also had obvious nose-like features at lower levels.

Du et al. (2012) found that SLLJs in Shanghai occurred more frequently and with more intensity during the mei-yu season. In contrast, BLJs occurred more frequently and with greater intensity in Wuhan during the 2010 mei-yu season, and the observations indicated that they differed significantly from the jets in Shanghai. This unique phenomenon of LLJs in the middle reaches of Yangtze River during the mei-yu season further confirms the need for our study.

The low-level wind speed in Wuhan showed significant diurnal variations, and the diurnal variation of the composite jet wind speed profile was opposite to that of the composite non-jet wind speed profile. When jets were observed, the wind speed above 1000 m was greater during the day than at night. The diurnal variation of the wind speed was relatively complex below 1000 m, but, in general, the intensity of nocturnal jets was slightly greater than that of daytime jets. The composite non-jet profiles showed slightly higher wind speeds at night than during the day, which was most pronounced below 600 m.

Previous studies have shown that the diurnal variation of BLJs becomes insignificant when there are mid-altitude jets above them (Zhang et al., 2007); therefore, we only discuss pure BLJ and pure SLLJ conditions to reveal the diurnal variation of the vertical structure of BLJs more clearly. The diurnal variation of the vertical structure of pure BLJs was the opposite of that of pure SLLJs, with the speed of BLJs greater at night and the speed of SLLJs greater during the day. The BLJs had the highest wind speeds at 0200 LST when the composite wind speed exceeded 14 m s⁻¹. The wind speed began to decrease after 0200 LST. The nose of the BLJ varied in altitude from 600 to 1200 m. The SLLJ was strongest at 1100 LST, with the lowest wind speed occurring at 1700 LST.

The structure of DLLJs was more pronounced at night, with stronger BLJ and SLLJ wind speeds at night and greater vertical shear. The BLJs in the DLLJ were significantly stronger than the SLLJs (Fig. 5). The evolutionary features of the double LLJs at different hours (Figs. 5e-f) show that the DLLJs were strongest at 0200 LST when the BLJ and SLLJ wind speeds reached a maximum. The BLJs were less variable in height, and the nose-like features were always evident. In contrast, SLLJs showed a large variation with height, and the nose-like features of the composite profiles were correspondingly weak and highly variable.

Figure 5.  Diurnal variation of the composite observational profile in Wuhan during (a) LLJ observational events, (b) non-LLJ observational events, (c) pure BLJ events, and (d) pure SLLJ events. (e) Diurnal variation of DLLJs and (f) variation in DLLJs at different times in Wuhan. These are statistical plots based on intensive sounding observational data.

Because the number of observations in Wuhan at 0200 LST was significantly less than that at other hours (about 42 for other hours, but only 32 at 0200 LST), a standard frequency was selected to analyze the daily variation of the jets: R_t = NJET_t/N_t × 100, where R_t represents the frequency of jets at hour t, NJET_t represents the occurrence frequency of the jets at hour t and N_t represents the total number of observations at hour t.

In conjunction with the diurnal variation of the LLJs (Fig. 6a), the jet-like profile and the occurrence frequency of LLJs in Wuhan showed a double-peak structure during the mei-yu season. There was only a small difference between the primary (0800 LST) and secondary (2300 LST) peaks in the frequency of LLJs. The primary peak of the jet-like profile was recorded at 1100 LST, three hours after the primary peak of the LLJs. Whiteman et al. (1997) reported that the primary and secondary peaks in the frequency of LLJs in Oklahoma during the warm season were at 2300 CST and 0500 CST, respectively. Therefore, the times of the two peaks in the middle reaches of the Yangtze River were roughly consistent with the results of Whiteman et al. (1997), indicating that the patterns of development of LLJs in Oklahoma and Wuhan were similar. However, the primary and secondary peaks of the LLJ frequency in the two regions were diurnally opposite, with the greatest frequency occurring during the day in the middle reaches of the Yangtze River (with the secondary peak at night) and during the night in Oklahoma in the USA (with the secondary peak in the daytime), suggesting that the influence of local factors on LLJs in these regions differs. These differences between the two regions may be due to the specific weather systems in the middle reaches of the Yangtze River during the mei-yu season, such as the mei-yu front, which is mei-yuusually most vigorous in the early morning. The mesoscale circulation caused by the latent heat of condensation of the mei-yu front affects the wind speed of LLJs (Qian et al., 2004), resulting in the unique diurnal variability of LLJs in the middle reaches of the Yangtze River.

Figure 6.  (a) Occurrence frequencies of jet-like profiles (gray bars) and jet profiles (red bars; including BLJs and SLLJs) in Wuhan at different times. (b) Daily variation of the frequencies of BLJs (red line) and SLLJs (blue line) in Wuhan; 2000−0500 LST represents nighttime, and 0800- 1700 LST represents daytime. These plots are statistically based on intensive sounding observational data.

Figure 6b shows that BLJs mainly occur at night. The frequency of BLJs showed a double-peak structure, with the maximum at 2300 LST and a second peak at 0800 LST. The frequency of BLJs decreased significantly after 0800 LST. The diurnal variation of SLLJs was the opposite of that of BLJs, occurring mainly during the day and most frequently at 0800 LST. The co-occurrence of the peak frequencies of BLJs and SLLJs at 0800 LST made LLJs, as a whole, most frequent at 0800 LST (Fig. 6a).

An interesting phenomenon in Wuhan is that the frequency of BLJs decreased rapidly after 0800 LST, reached a minimum at 1700 LST, but increased again at night. This may be because the mixed layer started to develop after 0800 LST, causing the turbulent mixing in the boundary layer to intensify, which increased the frictional turbulent drag. The vertical distribution of the wind speed in the boundary layer tended to be uniform, leading to a weakening or even the disappearance of the jets. The frequency of BLJs began to increase again at 2000 LST due to the development of a stable nocturnal boundary layer. The stable nocturnal boundary layer is primarily influenced by surface radiative cooling, with atmospheric temperatures decreasing more rapidly near the surface, resulting in a shallow inversion layer in the lower troposphere. Turbulent mixing is weak at night, and any turbulent friction becomes decoupled when the wind speed in the boundary layer is mainly influenced by surface friction drag. As a result, wind speeds are lower near the surface and stronger at the upper levels, leading to the formation of jets (Blackadar, 1957).

Table 1 shows that the frequency rate of BLJs in Wuhan was slightly higher at night than during the day, but their average height was higher during the day (941.2 m) than at night (875.9 m). The diurnal variation of the mean height of SLLJs was opposite to that of BLJs, with a higher frequency rate during the day but a higher mean height at night (2676.3 m) than during the day (2474.3 m). The mean heights of BLJs and SLLJs in Wuhan were both larger than their corresponding median values and showed a positively skewed distribution. The standard deviation of the height of SLLJs was about twice that of BLJs, indicating a high degree of relative dispersion among the SLLJs.

The BLJs in the Wuhan area were generally southwesterly, with wind speeds up to 20 m s⁻¹. The frequencies of BLJs in the SSW and WSW directions were 37 and 36, respectively, accounting for 76.84% of the total (n = 95). The frequency of BLJs in the other three quadrants decreased sharply, although the northeasterly direction was a sub-high-frequency region with a significant decrease in wind speed (Fig. 7a). The wind direction of maximum frequency for the SLLJs was still dominated by southwesterly winds, with wind speeds of up to 23 m s⁻¹. However, the wind direction of SLLJs was more westerly than that of the BLJs, and the frequency of northeasterly jets increased significantly (Fig. 7b). The highest frequency wind direction for the SLLJs was from the WSW, followed by westerly winds and then by winds from an NNE direction.

Figure 7.  Wind roses summarizing all the wind speeds (units: m s⁻¹) and directions of (a) BLJs and (b) SLJs observed by radiosondes in Wuhan. Composite wind stream and geopotential height (red contours, units: dagpm) at 800 hPa from the ERA5 dataset during (c) southwesterly SLLJs and (d) northeasterly SLLJs. The gray shading represents the topography at 800 hPa. The blue triangle represents the location of Wuhan.

When Vanderwende et al. (2015) studied jets below 2 km in the Iowa region of the USA, they found that the jets were almost non-existent in the first quadrant. In contrast, in the middle reaches of the Yangtze River, the first quadrant had a moderate frequency of jets, indicating that these jets differed from those in the USA. This also confirms the need for this study.

Based on these characteristics, we further analyzed the causes of the formation of southwesterly oriented SLLJs and northeasterly oriented SLLJs. At 800 hPa, the formation of southwesterly oriented SLLJs in the middle reaches of the Yangtze River was closely related to the southwest monsoon. The main system of influence was the low-pressure trough at 800 hPa. Wuhan was in front of this trough and was influenced by southwesterly winds. The weather conditions influencing the northeasterly oriented SLLJs were very different from that of the southwesterly oriented SLLJs. There was an anticyclonic and a cyclonic system to the northwest and southeast of Wuhan, respectively, which means that the Wuhan area was controlled by northeasterly winds.

We used the ERA5 reanalysis dataset to make a preliminary analysis of the formation mechanism of LLJs in the middle reaches of Yangtze River during the 2010 mei-yu season. First, we verified the validity of the ERA5 reanalysis data for the simulation of LLJs in the Wuhan area during the mei-yu season. In the following developments, the radiosonde sounding data are grouped according to LLJ events and non-LLJ events. We then used the LLJ events chosen from the sounding data to select the ERA5 data at the corresponding time and to plot the mean vertical profile; the same process was used for non-LLJ events.

Figure 8 shows the wind speed and wind direction profiles from the radiosonde sounding data and ERA5 reanalysis dataset from 975 to 400 hPa during the LLJ (Fig. 8a) and non-LLJ (Fig. 8b) events. In terms of the vertical structure, the ERA5 reanalysis data agree fairly well with both LLJ and non-LLJ events in the Wuhan area, where the ERA5 simulation results for the jet core of LLJs are consistent with the sounding observations (Fig. 8a). This shows that the ERA5 reanalysis dataset can effectively simulate the structure of LLJs in Wuhan. Therefore, we used the ERA5 reanalysis dataset to determine the weather conditions favorable for the formation of LLJs during the mei-yu period.

Figure 8.  Profiles of the mean vertical wind speed (lines; units: m s^-1) and direction (crosses; units: degrees) from the radiosonde sounding observations (red) and the ERA5 dataset (pink) in (a) LLJ and (b) non-LLJ events. The average (c, d) 800- and (e, f) 875-hPa geopotential height (shading; units: dagpm), wind fields (wind barbs; units: m s⁻¹; the black and white wind barbs represent wind speeds greater than and less than 6 m s⁻¹, respectively), equivalent temperature (deep pink contours; units: K) and the 500 hPa geopotential height (cyan contours; units: dagpm) in (c, e) LLJ and (d, f) non-LLJ events during the mei-yu period in 2010. The average (g) daytime and (h) nighttime 875-hPa geopotential height (shading; units: dagpm), wind fields (wind barbs; units: m s⁻¹; the black and white wind barbs represent wind speeds greater than and less than 6 m s⁻¹, respectively), equivalent temperature (deep pink contours; units: K) and the 500-hPa geopotential height (cyan contours; units: dagpm) in LLJ events. The gray shading represents the topography. The red triangle represents the location of Wuhan.

Many scientific studies have been conducted on the formation and development of LLJs. The formation and daily variation of LLJs vary slightly from region to region and are influenced by local diurnal oscillations and weather systems (Blackadar, 1957; Li and Chen, 1998; Lin et al., 2011; Du et al., 2012). Du et al. (2012) showed that the mei-yu front could extend to lower levels during the mei-yu season and affect both SLLJs and BLJs.

We analyzed the formation mechanism of LLJs in the middle reaches of the Yangtze River at 800 hPa and 875 hPa. Figures 8c–f show that the change in the position of the mei-yu front influenced the formation of LLJs in Wuhan. Wuhan was located on the south side of the mei-yu front during LLJ events at both 800 hPa and 875 hPa, whereas it was located on the north side of the mei-yu front in non-LLJ events. The latent heat of condensation on the mei-yu front enhanced the low-level wind field via the mesoscale circulation field (Qian et al., 2004).

The northwest Pacific subtropical high and the low-pressure system over the eastern Tibetan Plateau also significantly affected the formation and development of LLJs in the middle reaches of the Yangtze River. The northwest Pacific subtropical high intensified its westward extension in LLJ events, and the westward ridge point was located at 110°E (115°E in non-LLJ events), which gave the southwesterly airflow from its periphery more influence in the middle reaches of the Yangtze River and favored the formation of LLJs.

The cyclonic circulation near the Sichuan basin was also more pronounced, and the low-pressure system was stronger. Wuhan is located between a strong high-pressure system and a low-pressure system, which causes the pressure gradient to increase in the middle reaches of the Yangtze River, as evidenced by the mid- and low-level geopotential height fields. This large-scale horizontal pressure gradient increased the wind speed in the mid-to-lower levels, promoting the formation and development of LLJs (Xu et al., 2004). Therefore, the influential systems for the formation of LLJs in the middle reaches of the Yangtze River were the mei-yu front, the northwest Pacific subtropical high, and the low-pressure system in the eastern Tibetan Plateau.

We conducted a preliminary analysis of the daily variation mechanism of BLJs, the features of which were more significant in Wuhan. Figures 8g–h show that the intensity of BLJs in the middle reaches of the Yangtze River was significantly stronger at night than during the day and was accompanied by a larger gradient in equivalent temperature. This also implies that the stronger nocturnal mei-yu front promoted the development of the nocturnal BLJs. Regarding geographical location, the BLJs tended to develop stronger when Wuhan was located south of the mei-yu front. The ridge point of the nocturnal northwest Pacific subtropical high was slightly to the west, which was not significantly different from the daytime location (110°E at night, 111°E during the day), but the intensity was significantly stronger at night than during the day. The low-pressure system in the eastern part of the Tibetan Plateau was also stronger at night. Du et al. (2012) identified the inertial oscillation mechanism as the main cause of BLJs in areas with a flat topography.

We investigated the statistical relationship between LLJs and heavy rainfall. A rainstorm day was defined as a day when torrential rain (rainfall >50 mm d^–1) was observed by at least one weather station near Wuhan. During the 45 days of sounding observations, 19 days were rainstorm days; the other days were referred to as non-rainstorm days.

Figure 9 shows the average wind profiles at 0200, 0500, 0800, and 1100 LST on rainstorm and non-rainstorm days. The diurnal variation was completely different between rainstorm and non-rainstorm days. A comparison between Figs. 5a and 9 shows that the height of the jet and the diurnal variation on rainstorm days were consistent with the statistical characteristics averaged on all LLJ days. The wind speed increased rapidly from the ground surface to 1000 m, causing significant jets to form in the boundary layer and the lower troposphere. The vertical profile of the wind speed showed a double-peak pattern, with the peaks in the 500–1000 and 2500–3000 m layers. The wind speed on rainstorm days was 1–2 m s⁻¹ greater than on LLJ days. By contrast, the wind speed on non-rainstorm days was significantly smaller than that on rainstorm days, and there was only a small change in the diurnal variation of the vertical wind speed.

Figure 9.  Wind velocity profile on rainstorm (red) and non-rainstorm (pink) days at (a) 0200, (b) 0500, (c) 0800, and (d) 1100 LST. These plots are statistically based on intensive sounding observational data.

We studied the relationship between LLJs and rainfall for representative cases. We selected four rainfall processes after investigating the heavy precipitation processes that occurred in the middle and lower reaches of the Yangtze River during the mei-yu period: 16 July 2010, 19 July 2010, 1 July 2016, and 6 July 2020. The precipitation during the two 2010 events was mainly in northeastern Hubei. In contrast, the heavy rainfall and extremely heavy rainfall in 2016 and 2020 were concentrated in eastern Hubei and southern Anhui along the Yangtze River and caused significant losses.

Figure 10.  (a) Soundings at Wuhan Station (from 2000 LST 14 July to 2000 LST 16 July 2010). (b) Soundings at Wuhan Station (from 2000 LST 17 July to 2000 LST 19 July 2010). (c) Height- time evolution of the horizontal wind field at Poyang Lake (from 2000 LST 29 June to 2000 LST 1 July 2016). (d) Wind profile at Xianning Station (from 2000 LST 4 July to 2000 LST 6 July 2020).

We used sounding data from Wuhan Station (2010), the ERA5 dataset (2016), and wind profiler radar data from Xianning Station (2020) to show the diurnal variation of SLLJs and BLJs during these four events (Fig. 10). In all four events, the jets appeared below the 3.5 km (650 hPa) layer at night two days before precipitation began and again, with greater intensity, during the night and in the early morning before precipitation began. In all instances, there were large centers below 1.5 km (850 hPa).

The statistical features of the rainstorm days and the characteristics of these four events clearly show the existence of LLJs before the onset of heavy precipitation in the middle reaches of the Yangtze River. The evolution of the wind fields shows the distinctive feature of the jets beginning at night in the boundary layer and moving toward the lower troposphere in the morning. The strong southwesterly winds developed rapidly at night and peaked in the morning before the onset of precipitation, and the BLJ also appeared earlier than the SLLJ. The SLLJs were dominated by southwesterly winds, whereas the BLJs had a more southerly component.

Given China's vast area and complex terrain, the current upper-air sounding stations are relatively sparse and cannot reflect the local characteristics of LLJs. There have been few specific studies of LLJs in the middle reaches of the Yangtze River during the mei-yu season, although there have been more indirect studies. Observations of LLJs in some areas (i.e., meteorological towers, tethered balloons, and radar systems) have achieved relatively good results. To explore the mechanics of the LLJ in more detail, we used a fusion of sounding observations and precipitation data from Wuhan Station during the 2010 mei-yu season to analyze the characteristics of LLJs in the middle reaches of the Yangtze River. Our results can be summarized as follows.

(1) The vertical structure of the LLJs in the middle reaches of the Yangtze River was characterized by a predominance of BLJs, concentrated at heights of 900–1200 m. The differences between the jet and non-jet composite profiles were most pronounced in the boundary layer, where the vertical shear of the composite jet profile was stronger and deeper.

(2) The LLJs in the middle reaches of the Yangtze River had a unique diurnal variation, with the highest frequency occurring at 0800 LST and the second highest at 2300 LST. The difference in frequency between these times was small and could be specifically related to the presence of the mei-yu front in this region. BLJs occurred most frequently at 2300 LST but most strongly at 0200 LST, with composite wind velocities >14 m s⁻¹. SLLJs were most frequent at 0800 LST, but were strongest at 1100 LST, with composite wind velocities >12 m s⁻¹.

(3) Both BLJs and SLLJs were characterized by southwesterly winds. However, the wind direction of SLLJs was more westerly than that of BLJs, and northeasterly SLLJs were significantly more frequent than northeasterly BLJs.

(4) The mei-yu front significantly influenced LLJs in the middle reaches of the Yangtze River. When Wuhan was located to the south of the mei-yu front, the intensification of the westward extension of the northwest Pacific subtropical high and strengthening of the low-pressure system in the eastern Tibetan Plateau favored the formation of LLJs in this region.

The wind speed on rainstorm days was 1–2 m s⁻¹ higher than on LLJ days. Our analysis of typical rainfall events from the reanalysis dataset, wind profiles, and sounding observations showed the existence of LLJs in the middle reaches of the Yangtze River before the onset of heavy precipitation. BLJs developed earlier than SLLJs, wherein the SLLJs were dominated by southwesterly winds while the BLJs had a more southerly wind component.

LLJs are a key factor in the location and intensity of precipitation centers. Many researchers have investigated the characteristics of BLJs and DLLJs in coastal areas and their effects on precipitation, but BLJs in inland areas have received less attention. Our research clearly shows that DLLJs also present themselves in the middle reaches of the Yangtze River. Both the observational and reanalysis data showed the presence of LLJs, especially BLJs, before the onset of heavy precipitation in the middle reaches of the Yangtze River from both a statistical perspective and for individual cases. The generation mechanisms of BLJs and their effects on precipitation, which are not yet well understood, require further study. Considering that our study period was relatively short, we will conduct more observations and related studies in the future to determine in greater detail the characteristics of LLJs in the middle reaches of the Yangtze River.

Acknowledgements. We acknowledge the National Meteorological Information Center of the China Meteorological Administration for providing the daily precipitation data. The Institute of Heavy Rain, China Meteorological Administration, provided the wind profiler radar and radiosonde data. The ERA5 dataset was obtained from https://cds.climate.copernicus.eu/cdsapp#!/home. The topographic data were obtained from https://www.ngdc.noaa.gov/mgg/global/global.html. The authors thank the editor and anonymous reviewers for their constructive comments.

This work was supported by the National Natural Science Foundation of China (Grant Nos. 42230612, 41620104009, 41705019, 42075186, and 41975058) and the Projects of the S&T Development Foundation of the Hubei Meteorological Bureau (Grants No. 2021Q04 and 2020Y04). We give special thanks to Theodore McHardy for reading the manuscript carefully and correcting our English.