4.1 Corresponding relationship between cyclones and different levels of precipitation
Because of the irregular shapes of ECs, especially in inland areas, traditional auto-detecting methods mainly focus on the cyclone center identification or considering a given value as its radius (Raible et al. 2007). Meanwhile, the above results show that both vqmax and Rmax are mainly distributed in the central region. To simplify the definition of intensity index and improve the operational monitoring efficiency, the vqmax in the central region (vqmax_300) is taken as the cyclone overall comprehensive intensity index. The correlation coefficient between vqmax and vqmax_300 is 0.87, showing a good consistency. As shown in Fig. 3a, vqmax_300 still maintains an evident positive correlation with Rmax (correlation coefficient: 0.7). Accordingly, as the value of vqmax_300 increases, its corresponding precipitation notably intensifies (Fig. 3b). Specifically, when vqmax_300 < − 0.58, the probability of light rain is 26.9%; when − 0.17 > vqmax_300 ≥ − 0.58, the probability of moderate rain is 88.3%; when 0.51 > vqmax_300 ≥ − 0.17, the probability of heavy rain is 50.2%; when vqmax_300≥0.51, the probability of rainstorm is 55.4%. Therefore, according to the value of vqmax_300, the ECs are grouped into four levels as follows. I: vqmax_300 < − 0.58; II: −0.17 > vqmax_300 ≥ − 0.58; III: 0.51 > vqmax_300 ≥ − 0.17; IV: vqmax_300≥0.51.
To further show the relationship between different levels of ECs and the associated maximum precipitation, Fig. 3c demonstrates the probability of different levels of ECswhen different levels of precipitionoccurs. It can be seen that the EC level with the largest probability increases with the increase of precipitation level. More importantly, 32.8% of summer heavy rain events in East Asia are related to level-4 EC. Booth et al. (2018) also reported that there is no time lag between the peaks of precipitation and EC intensity, and there is also no time lag between the increase of precipition and the dynamic intensity caused by precipitation.Based on the above analysis, it is reasonable to divide the summer East Asia ECs into four intensity levels according to the value of vqmax_300, which can well indicate the associated maximum precipitation.
4.2 Cases and evolution of level-4 ECs with different horizontal scales
As mentioned earlier, heavy precipitation events are often associated with level-4 ECs, but the horizontal scales of these strong cyclones vary widely. In the past 40 years, there have been 3,435 level-4 ECs accompanied by heavy rain, among which the frequencies of mesoscale ECs (radius < 150 km), sub-synoptic scale ECs (150 km < radius < 500 km), and synoptic scale ECs (radius > 500 km) are 0.44%, 48.47% and 51.09%, respectively. This indicates that the EC that breeds heavy precipitation does not necessarily have a large horizontal scale, and some heavy precipitation may come from a sub-synoptic scale or even mesoscale EC (low vortex). To further reveal the differences of horizontal scales of level-4 ECs, Fig. 4 shows three level-4 ECs with different horizontal scales, which are all associated with heavy precipitation. The first case (Fig. 4a) shows that at 1800 UTC on July 4, 2013, a long and narrow rain belt appeared in the Huanghuai River Basin, and the largest precipitation occurred in southern Shandong Peninsula and the Yellow Sea. This event was accompanied with a mesoscale vortex on the Meiyu front, and the southwesterly air flow continuously transported water vapor to the rain belt. The shear line at the lower troposphere dynamically facilitates the upward motion and water vapor convergence. The analysis of the 850 hPa geopotential height suggests that the mesoscale cyclonic circulation is embedded from northeast to southwest in the low trough over southern Shandong Peninsula. Although the horizontal scale of this cyclone is relatively small (radius = 144 km), the 24-hour accumulated precipitation associated with it can reach the rainstorm level (Rmax=96.25 mm). During this period, the maximum wind speed in the cyclone reached 26.7 m/s, accompanied with a favorable water vapor condition (the specific humidity value at the maximum wind speed point is 15.2 g/kg). The vqmax_300 index is up to 2.92, so it is considered as a level-4 EC with heavy precipitation.
The second case (Fig. 4b) is an EC at sub-synoptic scale (radius = 458.8 km) which caused a heavy precipition event in East China at 0000 UTC on June 26, 2005. The EC was a typical Huanghuai cyclone. It was generated in the Sichuan Basin with a cold core. The cold air carried by the EC from the middle and high latitudes converged with the southwesterly warm-humid airflow carried by the East Asian monsoon, inducing pronounced precipitation in East China and northern South China. The maximum precipitation was located in the northern part of the cyclone in North China, with a value of 105.5 mm. The EC maximum wind speed was 25.1 m/s, and the corresponding specific humidity value was 15.8 g/kg, indicating that the EC carried large kinetic energy and water vapor. The vqmax_300 index value is 2.82, which satisfies the criteria of level-4 EC.
The third case (Fig. 4c) occurred at 0600 UTC on July 25, 2014, and it was a heavy rain event caused by a synoptic-scale cyclone. The 24-h accumulated maximum precipitation reached the rainstorm level. The EC was also formed in Southwest China, and it moved northeastward and continuously deepened. When it reached Huanghuai region, it connected with the western Pacific subtropical high. As a result, the zonal land-sea pressure gradient between them was intensified (figure not shown), which was favorable for the strengthening of the southwesterly warm-humid airflow, and provided abundant water vapor for the rainstorm. The maximum precipitation area was located in Northeast China, and the 24-h accumulated maximum precipitation reached 114.1 mm. The maximum wind speed reached 25.1 m/s, and the corresponding specific humidity value was 15.8 g/kg. The vqmax_300 index is 2.87, which is a level-4 EC.
To analyze the evolution and differences of the vqmax_300 between the sub-synoptic scale level-4 EC and synoptic scale level-4 EC, Fig. 5 shows the evolution of the 50 strongest vqmax_300 cyclones during their life cycles. The proportions of these two scales level-4 ECs are both 50%. As shown in Fig. 5a, most of the sub-synoptic scale cyclones develop rapidly during their early stages. In particular, for 52% of the cases the vqmax_300 reaches the peak in the first 24 hours. Then, the intensity gradually decreases from 24 to 72 hours, and a small part (24%) of the sub-synoptic scale ECs can live over 72 hours. Compared with the sub-synoptic scale EC, the synoptic scale EC develops more slowly, with an average vqmax_300 value of 3.81, slightly larger than that of the sub-synoptic scale ECs (3.51). However, only 5 ECs reach the strongest in the first 24 hours, and 48% of ECs reach their strongest intensity at around 24–36 hour. Then, they gradually weaken from 36 to 90 hours. Some of them could even last for more than 90 hours (Fig. 5b). During the evolution of these two scales of ECs, the maximum intensity mostly appears in the early stage (72% of sub-synoptic scale ECs and 52% of synoptic scale ECs), which is consistent with the result revealed by Martina and Simmonds (2021) that the extreme wind speed and precipitation are more likely to appear before the cyclone half-life time. In general, compared with sub-synoptic scale ECs, synoptic scale ECs have a relatively higher maximum vqmax_300 and a longer duration.