Overshooting convection that penetrates the lapse rate tropopause is defined globally using 3 years of Global Precipitation Measurement (GPM) observations and ERA-Interim data. Overshooting convection in the subtropics is mainly found over a few hot spot regions, including central North America and Argentina. A relatively high density of events with overshooting convection is also found over northeast China in the summer months, where 203 events are identified during 2014–16. These convective events extending above the tropopause occur under various synoptic conditions. The synoptic conditions during these events are categorized into three different types, namely, trough, cutoff low, and ridge types, with a subjective analysis based on the wind and pressure fields at 500 hPa. The precipitation systems with overshooting convection ahead of a deep trough have larger sizes than other types. Those in the cutoff low environment are mostly embedded within a large precipitation system. The ridge-type systems have a stable midtroposphere and a high moist instability at low levels and are mostly isolated convective systems, characterized by smaller sizes, higher radar echo top, and larger convective area and precipitation fraction than the other two types.
Convective clouds play an important role in the global energy and water budgets (Braham 1952; Houghton 1968; Gamache and Houze 1983; Young et al. 1992). Deep convection that penetrates the tropopause plays an important role in the stratosphere–troposphere exchange (STE) (Holton et al. 1995). Recently, Anderson et al. (2012) have proposed that deep convection at midlatitudes could transport a considerable amount of water vapor into the stratosphere and have an impact on the stratospheric ozone concentration. Therefore, it is important to understand the global distribution of deep convection that reaches above the tropopause.
The global hot spots of deep convection have been demonstrated for more than a few decades with the help of satellite observations. As early as the 1980s, Spencer and Santek (1985) identified intense convection using the scattering signature in the polarization corrected temperature (PCT) at 85.5 GHz from Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR). In the last decades, the Tropical Rainfall Measuring Mission (TRMM; Kummerow et al. 1998) Precipitation Radar (PR) reflectivity and Microwave Imager (TMI) brightness temperatures at 85 and 37 GHz have been used to examine intense convection and their vertical structures throughout the tropics and subtropics (Cecil et al. 2005; Zipser et al. 2006; Romatschke et al. 2010; Houze et al. 2015). Based on the radar reflectivity top, convection reaching the tropopause in the tropics and subtropics has been identified (Alcala and Dessler 2002; Liu and Zipser 2005). After the launch of the Global Precipitation Measurement (GPM) Core Observatory satellite in February 2014 (Hou et al. 2014), with a global coverage from 65°S to 65°N, both the dual-frequency radar and passive microwave radiometer on board the GPM have enabled studies of precipitation and convection globally, such as enhanced global precipitation detection by radar with higher sensitivity than the TRMM PR (Hamada and Takayabu 2016) and global intense convection, including those at the mid- and high latitudes (Liu and Zipser 2015).
Using 1 year of GPM observations (March 2014–February 2015), Liu and Liu (2016) have identified tropopause-reaching deep convection globally by GPM Ku-band radar echo top reaching above the tropopause. Using a similar approach, Fig. 1 shows the geographical locations of deep convection reaching above the lapse rate tropopause based on the World Meteorological Organization (WMO) definition (WMO 1957) in different seasons using 3 years of GPM observations (April 2014–March 2017). In agreement with previous studies, tropopause-reaching convection is mainly found over central Africa, the Amazon, the ITCZ, and the South Pacific convergence zone (SPCZ) in the tropics, and over central North America, Argentina, Europe, and Russia at the mid- and high latitudes. The convection over many of these regions has been extensively studied, including south of the Himalayas (Houze et al. 2007; Romatschke et al. 2010; Romatschke and Houze 2011), South America (Romatschke and Houze 2010; Rasmussen and Houze 2011; Rasmussen et al. 2014), Africa, and near-equatorial oceans (Zuluaga and Houze 2015). The high occurrences of convection over these regions have been related to favorable large-scale flow, topographic influences, the diurnal cycle, and the low-level airflow patterns (e.g., Liebmann et al. 1999; Garreaud 2000; Yang and Slingo 2001; Vera et al. 2006; Romatschke et al. 2010; Zuluaga and Houze 2015; Rasmussen and Houze 2016).
In addition to these well-studied regions, there are also many tropopause-reaching events over northeast China in summer (enclosed by the black box in Fig. 1). Previous studies of convection over this region have mainly emphasized the importance of certain large-scale flow patterns (e.g., Wang et al. 2012; Shi et al. 2013; Ren et al. 2014), especially the northeast China cold vortex (NCCV), which is prevailing in summer (Sun et al. 1994). Considerable attention has been given to describing and understanding the NCCV that is supportive of intense convection. Wang et al. (2007) stated that the NCCV is one of the main large-scale processes that are responsible for the convection over northeast China. However, radar observations have not been used to understand the relationship between NCCV and the convection.
A better understanding of the nature of the development of deep convection and its relationship to the large-scale environment would facilitate an estimate of the contribution of tropopause-reaching convection to stratosphere–troposphere exchange. Therefore, the purpose of this investigation is to supplement prior studies on large-scale environments that are related to deep convection with the help of GPM observations. On seeking insight into the mechanisms leading to the observed deep convection over northeast China, this study uses GPM observations and the European Centre for Medium-Range Weather Forecast interim reanalysis (ERA-Interim; Dee et al. 2011) to explore the factors that favor the initiation of deep convection over this region. To provide the synoptic basis for improving understanding and predicting the warm-season deep convection over northeast China, we try to answer the following questions:
Does tropopause-reaching convection over northeast China tend to occur in any particular synoptic environment?
What are the properties of deep convection under various synoptic conditions over this region?
To answer these two questions, GPM-identified overshooting convection cases over northeast China are first categorized into three synoptic types based on the large-scale environments. Next, the thermodynamic conditions and the characteristics of overshooting convection over this region under different kinds of synoptic conditions are explored. The paper is arranged as follows: section 2 describes the data and methodologies, section 3 gives the results, and section 4 includes the summary and discussion.
2. Data and methods
a. Definition of overshooting precipitation features
As the next-generation space-based precipitation measuring system, GPM provides frequent and global precipitation estimates (Hou et al. 2014). The core satellite carries the first dual-frequency radar in space, operating at Ku and Ka bands, and advanced passive microwave radiometers. The datasets used in this study are from the Ku-band precipitation radar (KuPR) and GPM Microwave Imager (GMI) on board the Core Observatory. The GPM KuPR provides a three-dimensional radar reflectivity structure of extreme convective systems. As this is written, 3 years of observations have been collected by GPM, beginning in March 2014. These observations provide an opportunity to study deep convection with an adequate number of samples over the domain of interest.
The GPM precipitation feature (PF) dataset is used to identify deep convection reaching above the tropopause. The PFs are defined by grouping the contiguous areas with nonzero near-surface precipitation derived from the Ku-band radar (Seto et al. 2013). The properties of each PF are summarized using various parameters. The maximum height of 20-dBZ radar echo (Maxht20) is an indicator of PF depth. The maximum heights of 30 and 40 dBZ are the indicators of large ice particles being lofted to upper levels and are used as proxies of convective intensity. Other proxies of convective intensity include minimum 37- and 89-GHz PCTs that are directly related to the ice-scattering signature in the column (Spencer et al. 1989; Cecil 2011). The convective area and precipitation fractions in each PF are calculated by dividing the convective area and precipitation by the total precipitation area and precipitation.
We choose the lapse rate tropopause as the reference level to define overshooting convection in this study, which is one of the methods used by Liu and Liu (2016). Similar previous studies have extensively demonstrated that ERA-Interim lapse rate tropopause heights are suitable for such efforts (Solomon et al. 2016). First, to calculate the lapse rate tropopause, ERA-Interim temperature and geopotential height profile, with 6-hourly intervals at 37 pressure levels (Dee et al. 2011), are temporally and spatially interpolated to the time and location of each PF. Second, the vertical profile for each PF is interpolated into 0.1-km intervals to calculate the lapse rate tropopause heights according to the WMO definition (WMO 1957). Lapse rate tropopause is defined as the lowest level of the temperature lapse rate less than 2 K km−1 for a depth at least 2 km. Then, the overshooting PFs (OPFs) are defined by PFs with 20-dBZ Ku-radar echo tops above the lapse rate tropopause heights. To control the quality of the data, the overshooting PFs must have significant ice-scattering signals from passive microwave radiance. Here, minimum 89-GHz PCT less than 220 K is used to remove some of the noise seen in radar signals in PFs. In total, 203 OPFs are found over the region (40°–55°N, 115°–133°E) in three summers during 2014–16. The large-scale environmental conditions for each OPF are obtained by interpolating the closest preceding and succeeding 6-hourly ERA-Interim dataset at 0.75° × 0.75° horizontal resolution (Dee et al. 2011).
b. Categorizing synoptic environments for overshooting convection cases
The large-scale synoptic condition is considered to play an important role in the developing of deep convection, as demonstrated by many earlier studies (Rockwood and Maddox 1988; Tuttle and Carbone 2004; Trier et al. 2006). Therefore, it is useful to diagnose large-scale environments where deep convective systems develop. We start the analysis by examining the synoptic conditions when these overshooting events occur. Since only 203 cases are found in 3 years of GPM observations, we are able to examine their large-scale environments individually. The synoptic conditions over a large region (25°–60°N, 105°–150°E) are examined when OPFs occur. For each of these OPFs, fields of geopotential height, temperature, and winds at 850, 700, 500, and 300 hPa, as well as the wind field and temperature at 2 m above the surface from the ERA-Interim dataset, are created and examined subjectively.
Three typical flow patterns at 500 hPa are subjectively distinguished when OPFs occur over northeast China: trough, cutoff low, and ridge. It is common that disturbances of an upper-level trough may enhance the instability and lead to intense convection. Cutoff lows have been extensively studied because of their important role in the weather system (García-Herrera et al. 2001) and stratosphere–troposphere exchange (e.g., Bamber et al. 1984; Holton et al. 1995). In cutoff low systems, the tropopause is anomalously low, which is an important mechanism to STE by gas tracer and radiative erosion of the tropopause (Nieto et al. 2005). Oltmans et al. (1996) and Gimeno et al. (1999) also suggested that STE associated with cutoff low systems is essential to explain anomalous values of tropospheric ozone in the northern midlatitude areas. Occasionally, convection may occur under ridge conditions as well. For example, geopotential heights (red contour), relative humidity (color fill), and wind vectors at 500 hPa under these three flow patterns are shown in Fig. 2. The red cross denotes the center location of the OPF. Figure 2a shows that a trough at 500 hPa is to the west of the OPF over the area of interest (black box). The second case is characterized by a cutoff low at the 500-hPa level (Fig. 2b). A ridge characterizes the synoptic pattern at 500 hPa over this region in the third case.
Figure 3 shows three example cases of OPFs observed by the GPM Ku-band radar under the synoptic conditions shown in Fig. 2. Cross sections through the convective cores are made to illustrate the vertical structures of OPFs. The black dots denote the locations of lightning flashes from the World Wide Lightning Location Network (WWLLN) dataset (Rodger et al. 2005; Abarca et al. 2010; Hutchins et al. 2013) within 10 min of the GPM overpass times. The three cases (Figs. 3a,c,e) show that these OPFs have radar echo tops above the estimated lapse rate tropopause (LRT) using ERA-Interim (black dashed line). The maximum echo-top height is approximately 15 km, with the tropopause height around 13 km in both trough and ridge cases (Figs. 3b,f). The collocated ERA-Interim wind profiles are shown with wind barbs on the right of the cross-section panels. Wind direction veers from south at the ground to southwest in the mid- and upper troposphere under trough and cutoff low conditions. Wind speed does not increase rapidly with height under the cutoff low condition. For the ridge case, even though it is still characterized by southerly wind near the ground, the wind speed is relatively low through almost the whole troposphere.
Examination of upper-level flow patterns for all OPFs (not shown) reveals that they share some resemblance to the three cases shown in Fig. 2. Therefore, the synoptic conditions prevailing over northeast China when OPFs occur have been subjectively classified into trough, cutoff low, and ridge types, according to the flow pattern at 500 hPa and the relative locations where OPFs are observed by GPM. The flow pattern at 500 hPa, similar to Fig. 2, is generated for all 203 cases. The trough condition is defined as an OPF ahead of a clear trough within the study region (black box in Fig. 2). The cutoff low condition is defined as a close contour of geopotential height within the study region. A ridge condition is defined as a ridge dominating the region while the OPF is not to the west of the ridge. Based on these criteria, we have subjectively categorized these cases independently and then discussed and reached agreements on the categorization. As summarized in Table 1, of the 203 cases, 81 are identified as trough type, 58 are cutoff low type, and 14 are ridge type. In the next section, the typical synoptic condition fields are composited to examine the similarities and differences among the categorizations. The properties of OPFs in different categorizations are also examined.
a. Major synoptic-scale environments for OPFs over northeast China
As shown in Fig. 1, most of the OPFs in northeast China are found in June, July, and August (JJA). The 203 cases in three summer seasons provide a decent sample size to investigate overshooting convection under different synoptic conditions over this region. To characterize the meteorological environment favorable to support deep convection, the composite large-scale environments at different levels are analyzed for each category.
It is well known that there is usually synoptic-scale ascent ahead of a trough and, therefore, a favored area for convective forcing. Figure 4 shows the composite geopotential heights, wind vectors, and relative (specific) humidity at different pressure levels for 81 trough-type OPFs. These synoptic conditions are composited within the fixed region of interest, regardless of the characteristics and locations of OPFs. The composite fields show a clear trough from the upper to the lower troposphere. A weak trough accompanying a westerly jet at 300 hPa is found in the domain region (Fig. 4a). At 500 hPa, the trough is found farther to the east, and the midlevel troposphere is relatively dry (Fig. 4b). At 700 hPa, the trough is found farther to the east and deeper, which might be related to the orography. Under certain conditions, orography (see Fig. 7) could affect the orientation and intensity of the trough (Carlson 1961; Steenburgh and Mass 1994). The moist air at low levels is an important factor for deep convection to develop. At 850 hPa, moisture transport from the south of China and the Yellow Sea to the studied region is evident, with a southwesterly low-level jet at 850 hPa (Fig. 4d). Favorable conditions for deep convection are present when warm, moist air from the Yellow Sea is in association with the ascent east of a trough. A stable and relatively dry air layer from the west (Fig. 4b) capping the warm, moist air at low levels could provide a favorable condition for deep convection.
Characterized by a local cold core and manifested as a closed cyclonic circulation in the mid- and upper troposphere developed from a deep trough in the westerlies (Palmén 1949; Palmén and Nagler 1949), a cutoff low is also an important midlatitude synoptic condition over northeast China (Zhang et al. 2008; Nieto et al. 2005). Cutoff lows are often associated with increased vertical wind shear, which in turn may affect the intensity of local afternoon convection (Wu 1976; Chen and Chi 1990). Sun (1997) proposed that the occurrence of the days affected by cutoff lows could be 42% in JJA. Figure 5 shows the composite synoptic condition for 58 OPFs under the cutoff low. By definition, a cutoff low over the examined region at 500 hPa is the dominant feature (Fig. 5b). These features, commonly known as upper- and midtropospheric features, may sometimes extend to the lower troposphere (Figs. 5c,d) (Nieto et al. 2005). Conditional instability occurs when warm, moist air at low levels comes beneath the cold low from the south or southwest, sometimes in the form of a low-level jet. The combined effects lead to an environment that is favorable for deep convection.
The third type of synoptic condition, in which 14 OPFs occur, is composited and shown in Fig. 6. In contrast with the trough, a weak ridge is found at the mid- and upper troposphere over the region (Figs. 6a–c). The westerly wind at the upper level shifts to southwesterly at 850 hPa (Fig. 6d). The moisture-laden low-level jet west of the ridge from the south of China and Bohai Sea is present at 850 hPa. At the midtroposphere, the air is relatively dry (Figs. 6b,c). In spite of the absence of large-scale lifting, a sufficient amount of humidity in the lower troposphere may still lead to the initiation of deep convection with a local forcing, such as topographic lifting. Unlike OPFs occurring under the trough and cutoff low conditions, OPFs under the ridge conditions are mostly found in June and July (Fig. 7c), coinciding with the mei-yu rainy season (e.g., Tao and Chen 1987; Ninomiya 1999; Sampe and Xie 2010). This indicates that the surface moisture from the mei-yu zone may play a role in the development of deep convection over this region.
All three types of OPFs share the characteristic of warm, moist air from the south capped by midlevel warm and dry air flowing over the high terrain from the west. This bears some similarities to other regions favoring the deep convection in the vicinity of mountains. These regions include, for example, the southern Great Plains in the United States, where moist air from the Gulf of Mexico is typically capped by hot dry air from the Mexican plateau (Carlson et al. 1983); the low plain of Argentina, where the low-level jet from the Amazon transports moist air capped by warm, dry air off the Andes (Romatschke and Houze 2010; Rasmussen and Houze 2011, 2016); and north India and Pakistan, where warm, dry air from Afghanistan or the Tibetan Plateau often caps the moist air from the Bay of Bengal or the Arabian Sea (Houze et al. 2007). This capping inversion inhibits the release of instability. Then, the forced ascent of impinging flow, topographical lift, or strong sensible heating could initiate deep convection.
Figure 7 shows the composite of near-surface wind fields. The red dots mark the locations of the OPFs in each category. The low-level wind field exhibits an abrupt change while crossing the southern mountains, which is consistent with the results shown in previous studies that show the topography could modify the baroclinic waves (Davis 1997; Schultz and Doswell 2000). Most of the OPFs under a deep trough scenario are associated with southerly flow near the surface and are found in the valley surrounded by high terrain (Fig. 7a). Examination of individual large-scale environments for each OPF shows that overshooting convection is mostly ahead of a trough, which may provide sufficient lifting to trigger the convection. OPFs associated with cutoff lows are relatively uniformly distributed around the center of the cutoff low (Fig. 7b). This is because the troposphere below the cutoff lows is usually unstable and is favorable for convection when the low-level southerly moist flow transports moisture to this region. Nieto et al. (2008) have shown that convective rainfall could occur within a radius of about 300 km of the cutoff low center, with a peak close to the cutoff low’s center. The locations of OPFs occurring under the ridge conditions are mostly near mountain slopes, as shown in Fig. 7c, with a strong southerly flow.
b. Thermodynamic conditions
Deep convection requires substantial convective available potential energy (CAPE) in the temperature and moisture stratification. However, the presence of CAPE is a necessary but insufficient condition for convection initiation. To initiate vigorous deep convection, air parcels need to reach a level of free convection to release CAPE. The convective inhibition (CIN) also plays a role in setting the stage for building up CAPE for convective storms. In addition, the vertical wind shear could profoundly affect the structure and evolution of convection (Chisholm and Renick 1972; Rotunno et al. 1988). All of these parameters together with several other thermodynamic parameters, such as the level of neutral buoyancy (LNB), lifting condensation level (LCL), and equivalent potential temperature Θe, are summarized in Table 1. With the lowest CAPE and LRT, OPFs under the cutoff low conditions are also characterized by the lowest CIN and low-level wind shear among all three conditions. OPFs under the ridge condition have the largest mean CAPE, CIN, LNB, LCL, and Θe, compared to those under the cutoff low and trough conditions.
Composite soundings are shown for each type of OPF in Fig. 8. The mean maximum heights of 20-dBZ radar echo tops of OPFs and lapse rate tropopause heights for each category are also shown in each composite sounding. There is no obvious difference in soundings between the trough and cutoff low conditions, as shown in Figs. 8a and 8b. However, cutoff low cases have relatively colder temperatures than trough cases (Fig. 8d). Usually, cutoff lows have longer life times than trough conditions locally and, thus, a colder column of troposphere due to longer periods of cold advection. With a colder column of troposphere, it is easier to develop a deep convection if warm, moist air is introduced at low levels for the cutoff low condition. Therefore, cutoff low cases have the smallest CAPE and CIN among the three groups (Table 1). The cutoff low–type sounding is also characterized by the lowest LNB, which is consistent with the lowest mean tropopause height. The large-scale ascent ahead of a trough is favorable for the development of deep convection. Compared to more vertically aligned flow in a cutoff low, trough cases have a stronger vertical wind shear and stronger southerly flow at low levels (Fig. 8a). It is known that the strength of the low-level vertical wind shear could modulate the strength of convective systems (e.g., Rotunno et al. 1988). Compared to the other two categories, the ridge-type sounding presents a drier midlevel troposphere (300–600 hPa), a larger CAPE, a higher LNB, and a higher tropopause. Under ridge conditions, due to a large-scale descent, the midtroposphere is relatively drier and warmer (Figs. 8c,d). Therefore, it is relatively difficult to develop deep convection as a result of strong convective inhibition with the highest CIN value as shown in Table 1. Whenever deep convection occurs under this condition, a strong instability with high surface equivalent potential temperature and large CAPE is required. This would also lead to a higher LNB (Table 1).
c. Characteristics of OPFs under different types of large-scale environments
As shown above, troughs, cutoff lows, and ridges may provide favorable, but slightly different, conditions for deep convection. The properties of deep convection, such as storm heights, intensity, and size (see examples in Fig. 3) may also vary in response to these three synoptic types. The proxies of convective properties under different kinds of large-scale environments, including the radar echo tops, system size, coldest brightness temperatures at 37 and 89 GHz, flash rate, and convective precipitation fraction, are listed in Table 1. With lower echo tops and warmer passive microwave brightness temperatures, overshooting convection under the cutoff low condition is relatively weaker than under the other two types. This is consistent with the lower LNB, smaller CAPE, and CIN in a colder and lower wind shear environment. Characterized by the smallest size and the largest 20-dBZ echo-top heights, the ridge-type overshooting convective systems are more isolated and deeper, compared to the other two categories. The highest 30- and 40-dBZ echo tops, the lowest brightness temperatures at 37 GHz, and the largest convective area and precipitation fraction among the three categories suggest that the overshooting systems developed under this condition are the most intense. This is consistent with the largest CAPE and CIN and the highest LNB for ridge type in Table 1 and Fig. 8. Larger CIN allows for the storage of more moist potential energy. Therefore, it is more likely for an intense convection to occur once the large CIN barrier is overcome and releases a large amount of energy. The histograms of the sizes of the OPFs under different upper-level patterns are shown in Fig. 9. Most of the cutoff low- and trough-type OPFs have a large horizontal extent. There is no ridge-type OPF greater than 5000 km2.
To demonstrate the vertical structure of overshooting convection under different kinds of large-scale environments, the top 10%, the median, and the top 90% of maximum radar reflectivity values inside the OPFs are shown in Fig. 10. The radar echo top of the ridge-type OPFs reaches the highest, followed by the trough type and the cutoff low type. This is consistent with the finding that the ridge-type overshooting convection is the deepest, as listed in Table 1. The median and top 90% maximum radar reflectivity profiles show that ridge-type OPFs are characterized by weaker low-level maximum Ku-radar reflectivity than the other two types.
Figure 11 shows the mean area of 20, 30, and 40 dBZ at different altitudes for different types of precipitation systems with tropopause-reaching convection. Similar to the maximum radar reflectivity, the mean area of Ku-radar reflectivity decreases with height. The trough type has the largest area of 20, 30, and 40 dBZ at mid- and low levels, followed by the cutoff low and ridge. The modest peaks at around 4 km indicate the radar bright band near the freezing level. The radar reflectivity of 40 dBZ, indicating large ice particles, mostly reaches above 8 km for all three types of OPFs.
The vertical outlines of the three types of systems, using the median-length scale, are drawn in Fig. 12. This schematically demonstrates the vertical size of each overshooting convective system, assuming that they are in conical shapes. The trough-type overshooting convective systems are the largest. The median size of ridge-type overshooting convective systems is the smallest. With the highest median tropopause height, the ridge-type overshooting convection is the deepest among the three categories. This is consistent with the composite soundings shown in Fig. 7. The differences in CAPE, CIN, and LNB of trough, cutoff low, and ridge types may help explain the differences in the convective intensity, but the differences in the sizes of convective systems of these three types need alternative interpretation. It is clear that ridge cases have much smaller sizes than the other two types. Most ridge-type OPFs occur over the mountain slopes (Fig. 7c). We speculate that under ridge conditions, the CIN is too large to break without the help of orographic lifting. Once convection initiates, it is stronger and may reach higher altitudes than the other two types. It takes multiple convective cells to build a large convective system. However, over complex terrain, it is relatively difficult to keep a continuous feeding of moisture to sustain the multicell convective system. Most trough and cutoff low OPFs are located in the basin (Figs. 7a,b). Large-scale circulations of these two types (Figs. 4, 5) may provide a more sustained supply of low-level moisture that is critical to build a large convective system. Because of relatively smaller CIN, CAPE, and LNB, the convective intensity is relatively weaker than that of the ridge type. Compared to the cutoff low, the trough type has a stronger low-level wind shear that would enhance the strength and organization of convection. The above speculations require validation through further studies, either with more observations or more model simulations.
Using three summers of observations from the GPM Core Observatory satellite, we have related convective systems overshooting the tropopause over northeast China to their synoptic-scale environments from ERA-Interim. The favorable synoptic conditions for overshooting convection have been categorized into three major types (trough, cutoff low, and ridge) according to the upper-level flow. Major findings can be summarized as follows:
There is a hot spot of convection reaching above the lapse rate tropopause over northeast China in summer. It is known that the overshooting convection may transport a significant amount of water vapor into the stratosphere and may lead to ozone loss over North America (Anderson et al. 2012). Therefore, it is important to pay attention to the overshooting convection hot spot over northeast China.
Characterized by relatively larger size than in other conditions, and mostly found in the valley between the mountains, the overshooting convection under the trough condition is frequently found in a precipitation system ahead of a deep trough.
Characterized by the lowest tropopause heights and the lowest maximum height of 20 dBZ, overshooting convection in the cutoff low environment is mostly embedded within a large precipitation system (>1000 km2).
Though relatively rare, the overshooting convection under the ridge-type condition features a relatively stable midtroposphere and a high moist instability at low levels. The precipitation systems with overshooting convection under this condition are more isolated convective systems and have relatively smaller sizes. Ridge conditions also exhibit larger convective area and precipitation fraction, and they are deeper and stronger than those in the other two conditions.
Overshooting convection in all three environments shares a general feature of the low-level moisture transport from the south and dry midtropospheric air from the west, though the convection initiation mechanisms are likely different.
The above conclusions about the favorable synoptic conditions and the properties of the overshooting convection also have some caveats. First, only 203 cases and 3 years of observations are used in this study. More observations are needed to help confirm and improve the results in the future. Second, when and where convection may develop involves favorable synoptic-scale circulation, mesoscale temperature, and moisture inhomogeneities, and in many instances, topographical forcing. Here, we arbitrarily categorize the favorable synoptic conditions of overshooting convection into three types using a subjective analysis. The subjective method has its pros and cons, compared to mathematical methods. Another uncertainty in the analysis is from ERA-Interim. Since we have used 6-hourly products, the temporal and spatial interpolations of the data to the cases may lead to an inaccurate description of the true synoptic conditions. Composite fields may help remove some uncertainties, but they also may weaken strong forcing factors near convective initiation points. Last, the lapse rate tropopause heights are used to find overshooting convection cases in this study; however, it is clear that different tropopause definitions could result in significantly different quantitative statistics for the overshooting convection (Liu and Zipser 2005; Liu and Liu 2016). Nevertheless, this study points out the important hot spot of overshooting convection over northeast China and demonstrates its commonly favorable synoptic conditions. It is important to understand the impact of these events on the troposphere–stratosphere exchange over the region and their influences globally.
Thanks to Dr. Edward Zipser and two anonymous reviewers for their valuable suggestions. This research was supported by NASA Global Precipitation Measurement Mission Grants NNX16AD76G and NNX16AH74G under the direction of Dr. Ramesh Kakar. Thanks also go to Dr. Erich Stocker and Patty McCaughey and the rest of the Precipitation Processing System (PPS) team at NASA Goddard Space Flight Center (Greenbelt, Maryland) for data processing assistance. ERA-Interim data are obtained from http://apps.ecmwf.int/datasets/data/interim-full-daily.
This article is included in the Global Precipitation Measurement (GPM) special collection.