The Contribution of Mesoscale Convective Systems to Intense Hourly Precipitation Events during the Warm Seasons over Central East China

Short-duration intense precipitation events are responsible for serious disasters, including flash flooding, landslides, mudflows, and urban water logging. Hourly precipitation datasets are very suitable for studying the detailed characteristics of short-duration intense precipitation events, given their high spatial and temporal resolution (Chen et al., 2013; Li et al., 2013). In recent years, changes in the frequency and intensity of intense hourly precipitation (IHP) events have received much attention, and the trends in different regions of China have been found to be different (Zhang and Zhai, 2011; Yu and Li, 2012). These different trends might be associated with many factors, of which the different contribution to IHP events by different kinds of convection may be an important one. IHP events might be caused by either isolated convection or organized convection, such as mesoscale convective systems (MCSs). These two kinds of convection are different mainly in terms of their spatial and temporal scales. Studies have shown that convection of different scales have different spatial and temporal distributions and might be caused by different forcing mechanisms and physical processes (Casati et al., 2004; Chen et al., 2014). Generally, well-organized, long-lived MCSs require stronger synoptic forcing than isolated convection, and suitable deep-layer wind shear plays an important role in the upscale evolution from isolated convection to MCSs, as well as the maintenance of long-lived MCSs (Raymond and Jiang, 1990; Lewis and Gray, 2010; Markowski and Richardson, 2010; Wang et al., 2014). Besides, isolated convection and MCSs have different organizational modes and different precipitation efficiencies, and this may be associated with the differences in their precipitation intensity and contributions to intense precipitation (Newton, 1966; Jirak et al., 2003). Due to these differences between isolated convection and MCSs, it is necessary to study them separately. Thus, the possession of good knowledge regarding their respective contribution to IHP events is important.

Central East China is known to be an area where IHP events occur frequently (Zhang and Zhai, 2011), and many efforts have been made to investigate the characteristics of IHP events over this region (Chen et al., 2013; Li et al., 2013). However, the nature of the main contributing factors to IHP events in this region remains unclear. Many studies have shown that central East China has a high frequency of MCSs (Ma et al., 1997; Zheng et al., 2008; Meng et al., 2013), but what percentage of IHP events are contributed by MCSs in central East China? To the best of our knowledge, this question has yet to be addressed. And the answer to this question might be helpful in understanding the mechanisms involved in the temporal and spatial variation of IHP events.

The remainder of this paper is organized as follows: The data and methodology are described in section 2. Section 3 describes the contribution of MCSs to IHP events in terms of spatial and temporal variation, as well as the contributions of MCSs to precipitation events of different intensities. A summary and discussion are given in section 4.

Central East China was defined as the region covering (30^°-37^°N, 110^°-112^°E), as shown in Fig. 1. The radar data used in this study were digital mosaics of composite Doppler radar reflectivity. The data had a spatial resolution of 4× 4 km, with an interval of 20 min (10 min) before (after) 22 September 2008. We chose a 4-yr period from 1 July 2007 to 30 June 2011 to ensure a relatively large and continuous data record. During this period, the average data coverage was about 85%.

Figure 1.  The terrain height (color-shaded; units: m), radar locations (light blue dots), and rain gauge station locations (small black dots) in central East China. The names of relevant provinces are marked in the center of each province. The names of the Yellow River, Huaihe River, and Yangtze River are also marked, along each river. The inner black box is central East China (30$^\circ$-37$^\circ$N, 110$^\circ$-112$^\circ$E).

To incorporate MCSs of various morphologies, the criterion for an MCS in this work, based on radar data, was: a continuous or quasi-continuous band of 40 dBZ reflectivity that extended for at least 100 km in at least one direction and lasted for at least 3 h. This definition is consistent with previous studies (Orlanski, 1975; Parker and Johnson, 2000; Schumacher and Johnson, 2006). Under this criterion, the threshold of 40 dBZ was used to distinguish convective and stratiform echoes (Fowle and Roebber, 2003). The length scale was defined as 100 km so that the Coriolis acceleration was of the same order as other terms in the momentum equations. The appropriate timescale for an MCS was therefore f^-1, which yielded 3 h for a typical midlatitude value of the Coriolis parameter f (Parker and Johnson, 2000; Markowski and Richardson, 2010). The MCSs in this study were identified manually. This was because there were serious data quality problems in terms of ground clutter in the radar images. Although it took a lot of work to identify MCSs on a case-by-case, the results may be more reliable compared with automatically detected MCSs. The track of an MCS was defined as the line that joined the location of its formation and dissipation.

The hourly precipitation dataset used in this work comprised observations from 2420 national rain gauge stations in China during the warm seasons (May to September for most stations) from 1951 to 2012. The data were subjected to strict quality-control procedures by the Chinese National Meteorological Information Center, which is part of the China Meteorological Administration (Yu et al., 2007). In this work, observations from 527 stations (Fig. 1) in central East China during the warm seasons, from 1 July 2007 to 30 June 2011, were used to be consistent with the temporal coverage of the radar data.

In some previous studies on extreme precipitation, the local percentile threshold was used to define extreme precipitation because climatologies in different regions are different. (Frich et al., 2002; Schumacher and Johnson, 2006). In this study, as we focused on the regional aspects of MCSs and intense rainfall, a fixed threshold was used for simplicity. Another issue concerns how to determine an appropriate hourly precipitation threshold for the region. Based on hourly precipitation data from 1961 to 2000, (Zhang and Zhai, 2011) gave the 95th percentile of the extreme hourly precipitation distribution and deemed 20 mm h^-1 as a suitable threshold for eastern China. (Chen et al., 2013) found similar distributional patterns for the thresholds of 10 mm h^-1, 20 mm h^-1, 30 mm h^-1 and 40 mm h^-1. When a threshold of 50 mm h^-1 was used, however, the distributional pattern was significantly different, and events occurred with a much lower frequency. Based on these results, the threshold of 20 mm h^-1 was used in this study, to be consistent with previous research. For any given station, each IHP record exceeding 20 mm h^-1 at that station was termed as an IHP event for simplicity.

For each IHP event, the radar images were examined to determine whether the event was caused by MCSs. During the particular hour with accumulated rainfall exceeding 20 mm at a station, there might be 3-6 radar observations. If the station was affected by convective echoes (>40 dBZ) of an MCS in any of the radar observations, this event was deemed to be caused by an MCS. Otherwise, we deemed it was contributed by isolated convection. The convection that did not reach 100 km before MCS formation or after MCS dissipation was deemed as isolated convection because it was difficult to determine whether the isolated convection was associated with a certain MCS.

According to the definitions of MCSs and IHP events given above, 302 MCSs and 8162 IHP events were identified during the warm seasons of 1 July 2007 to 30 June 2011 over central East China. For each station, the mean number of IHP events was 3.9 per warm season. Among the 8162 IHP events, 3655 were caused by MCSs, accounting for about 45% of all IHP events. The spatial distribution and diurnal variation of MCS contribution, as well as the contributions of MCSs to precipitation events of different intensities, are given below.

Figure 2a shows the distribution of total precipitation over central East China during the warm seasons of June 2007 to July 2011. The total precipitation decreased from the southeast to the northwest. There were several regions where large numbers of IHP events were observed. These regions were distributed along the lower reaches of the Yellow River, the Huaihe River, and the middle and lower reaches of the Yangtze River (Fig. 2b). Most of the regions with large numbers of IHP events were associated with high contributions from MCSs, with the exception of the middle reaches of the Yangtze River where IHP events might mostly have been contributed by scattered or isolated convection rather than organized MCSs.

Figure 2.  The (a) total precipitation (units: mm), (b) numbers of IHP events, and (c) percentages of IHP events caused by MCSs (units: %), for each station over central East China during the warm seasons of June 2007 to July 2011. The inner black box indicates central East China.

The contributions of MCSs to IHP events varied significantly among different stations. Figure 2c shows two local maxima where the contribution was greater than 80%. One local maximum was observed along the lower reaches of the Yellow River near the boundary between Henan and Shandong Province. The other was observed in the Yangtze River-Huaihe River valleys near the boundary between Jiangsu and Anhui Province. As indicated in Fig. 3, the maxima were caused by either locally formed MCSs or upstream MCSs propagating downstream.

There were several regions where MCSs occur very frequently, including the boundary between Anhui and Henan provinces, the boundary between Anhui and Jiangsu provinces, and the southwest of Shandong Province (Fig. 3). These regions were all associated with high contributions of MCSs to IHP events (Fig. 2c). A high contribution of MCSs to IHP events may also have resulted from MCSs propagating from upstream regions. For example, the high contribution of MCSs over the southwest of Shandong Province was partly caused by MCSs propagating northeastward from the source region of MCSs near the boundary between Henan and Shanxi provinces (Fig. 3a).

Figure 3.  The distribution of MCS tracks during the warm seasons of July 2007 to June 2011 over central East China: MCSs moving (a) northeast, (b) southeast, and (c) in other directions. The color shading represents the numbers of MCSs in the corresponding $1^\circ\times 1^\circ$ box. The dots in each figure represent the locations where MCSs formed. The lines in each figure are the tracks of MCSs from formation to dissipation. The inner black boxes in each subplot indicate central East China.

In addition to the spatial distribution of the contribution of MCSs to IHP events, the diurnal variation was also explored. Figure 4 shows the diurnal variation in MCS numbers and the number of IHP events. Peaks in the late afternoon and early morning were observed in the diurnal cycle of MCSs, with a significant peak after midnight in July. In contrast, the diurnal variation in IHP events also exhibited double peaks, but the peak in the early morning was less evident than the peak in the late afternoon, except in July when the early morning peak became nearly comparable with the afternoon peak.

The number of IHP events contributed by MCSs did not show as strong diurnal variation as the total number of IHP events. However, when calculating the percentage of IHP events caused by MCSs for each 3 h interval, it was found that MCSs contributed more to the early morning peak than to the afternoon peak. For example, the percentage of IHP events contributed by MCSs in July was 59% from 0200 LST (Local Standard Time, UTC+8) to 0500 LST, but the percentage was about 47% from 1700 LST to 2000 LST. This result is consistent with (Nesbitt and Zipser, 2003), who found that nocturnal rain is more often caused by MCSs rather than isolated convection. Therefore, more attention should be paid to MCSs when studying IHP events in the early morning.

Figure 4.  Diurnal and monthly variation of (a) MCS numbers and (b) IHP events, at 3 h intervals (indicated by the different color bars), during the warm seasons of July 2007 to June 2011 over central East China. The filled parts of the bars in (b) represent events caused by MCSs, while the hatched parts represent IHP events caused by isolated convection. The time reference is LST (UTC+8).

As reported above, over central East China, the contribution of MCSs to IHP events exceeding 20 mm h^-1 was about 45% on average. But did this percentage vary for IHP events of different intensities? Figure 5 shows the contribution of MCSs to IHP events of different intensities. The number of IHP events decreased as precipitation intensity increased, but the contributions of MCSs to IHP events exhibited no significant difference, which fluctuated near 50% on average over central East China. This suggested that, for IHP events, the contributions of MCSs and isolated convection were similar.

Figure 5.  Number of IHP events for different precipitation intensities. The filled parts of the bars represent events caused by MCSs, while hatched parts are non-MCS events. The percentages of IHP events contributed by MCSs, for IHP events of different intensities, are marked on top of each bar.

Based on radar and hourly precipitation data, the contribution of MCSs to IHP events was evaluated for the warm seasons of July 2007 to June 2011 in central East China. Using the definitions given in section 2, 302 MCSs and 8162 IHP events were identified. Of all the IHP events, 45% of them were caused by MCSs.

The spatial distribution and diurnal variation of the contribution of MCSs to IHP events, as well as the contributions of MCSs to precipitation events of different intensities, were also documented. MCSs contributed the most to IHP events in the lower reaches of the Yellow River and the Yangtze River-Huaihe River valleys, where the contribution could reach more than 80%. These regions were found to be the source regions of MCSs, or situated just along the frequent tracks of MCS. The diurnal variation in the number of IHP events showed two peaks: one in the late afternoon and one in the early morning. The early-morning peak was more pronounced in July than in other months. The contribution of MCSs to IHP events in the early morning was larger than that in the late afternoon. The contributions of MCSs to IHP events of different intensity exhibited no significant difference, which fluctuated around 50% on average over central East China.

Central East China is located in a summer monsoon region, where there is plenty of moisture and a high frequency of IHP events during the warm seasons (Zhai and Eskridge, 1997; Zhang and Zhai, 2011). Intense rainfall events may result from both isolated convection and MCSs. The two kinds of convection should be studied separately due to the different scales involved. Understanding their contribution to intense precipitation could help both forecasting and research communities to focus on key aspects of convection. It was found that organized MCSs and isolated convection may, on average, contribute equally to intense precipitation events in central East China. The contributions depended significantly on spatial and temporal variation. For regions that were frequently affected by MCSs, like the lower reaches of the Yellow River and Yangtze River-Huaihe River valleys, more attention should be paid to MCSs. As MCSs often require strong synoptic forcing compared with isolated convection, studies of convection should concentrate more on synoptic and dynamical forcing. For other regions, such as in the middle reaches of the Yangtze River near the southern boundary of central East China, isolated or scattered convection may play important contrasting roles, and thermal or local effects may be more important for convection. Nocturnal MCSs should also be emphasized when studying intense nocturnal rainfall. These findings are also helpful in understanding the changes in intense precipitation for climate research.

Whilst it is true that the current criteria used to identify isolated convection and MCSs cannot separate convection of different mechanisms and physical processes well, due to the possible transition between isolated convection and MCSs, the present results nonetheless provide us with an indication of the main mechanisms involved in IHP for different regions, considering MCSs generally require stronger large-scale forcing than isolated convection. It would have been useful and interesting to use different criteria in this study. However, due to the huge amount of work involved in testing different criteria when identifying MCSs and isolated convection manually, we decided against carrying out such an analysis on this occasion. Therefore, further research, using radar data that has been subjected to high levels of quality control, is still needed in the future, to better understand this issue.