1 Introduction

Northeast China (NEC), with its vast area of arable land, is an important commodity grain production base. Under the background of global warming, both precipitation and heavy precipitation have increased during recent years (Xu et al. 2011; Sun and Ao 2013; IPCC 2021). Anomalous variations in precipitation in NEC might lead to increased instability in grain production (Du et al. 2013). Moreover, heavy precipitation events can cause erosion, landslides, debris flows (Ban et al. 2015), and floods (Cao et al. 2018; Breugem et al. 2020). In comparison with general heavy precipitation, clustered heavy precipitation (CHP) can result in disasters of greater magnitude (Li et al. 2016). Therefore, it is essential to understand the rules that govern CHP events in NEC during summer.

Summer is the main season of precipitation in NEC (Gong et al. 2006; Liang et al. 2011; Gao and Gao 2018). Previous studies have revealed the spatial characteristics of summer precipitation over NEC, i.e., reduction in total precipitation amount from the southeast to the northwest (Wu et al. 2021; Xu and Li 2021; Yu and Ma 2022). Additionally, summer precipitation over NEC exhibits spatial heterogeneity and regional diversity (Gong et al. 2006; Yao et al. 2010; Song et al. 2015). Some studies have suggested that the systems that affect precipitation over NEC are different during the summer months (Zhao et al. 2018; Breugem et al. 2020). Precipitation over NEC during early summer (June) is affected by the activities of the Northeast China cold vortex (Lin et al. 2021), whereas midsummer precipitation (July–August) is mainly influenced by the western North Pacific subtropical high (WNPSH) and by the mid-latitude westerly belt system (Shen et al. 2011). Meng et al. (2022) indicated that eastern and central Pacific El Niño events are followed by different precipitation anomaly patterns in the following summers. Also, summer precipitation over NEC shows interdecadal variation (Gao et al. 2014; Li 2019). Specifically, precipitation over NEC has reduced since the mid-1960s, increased after the early 1980s, and diminished remarkably since the late 1990s (Ding and Chen 2015; Han et al. 2015). During 1983–1997, the regimes associate with NEC precipitation comprised a zonal tripole pattern in the mid–high latitudes of the North Asia–Pacific region and an anomalous anticyclone centered over Japan (Han et al. 2017). After the late 1990s, summer precipitation over NEC is influenced mainly by a zonal dipole pattern (Huang et al. 2013) and an anomalous anticyclone elongated from NEC to Mongolia (Xu et al. 2015).

With increasing occurrences of extreme weather events during recent years, some research efforts have been devoted to investigating the changes in heavy precipitation and in associated atmospheric circulations (e.g., Du et al. 2013; Cao et al. 2018). Previous studies reported that heavy precipitation is the greatest contributor to total precipitation in NEC (Song et al. 2015; Wang et al. 2013), and that the contribution rate exhibits insignificant trend during 1961–2013 (Fang et al. 2017). Wu et al. (2019) documented that short-term heavy precipitation events are limited in spatial scale, and that long-term events cover a wide area. Heavy precipitation events can be influenced by the East Asian summer monsoon (Wang et al. 2017), meridional shift of the WNPSH (Gao et al. 2014), meridional position and intensity of the upper-level westerlies over East Asia (Cao et al. 2018), cut-off lows (Zhao and Sun 2006; Hu et al. 2010), and East Asian trough (Wang and He 2015). In addition, the topographic uplift affects heavy precipitation by the modulation of vertical movement (Zhang et al. 2022). The eastern and southern NEC are vulnerable to heavy precipitation events (Du et al. 2013; Wang et al. 2019; Xu and Li 2021). From the above discussion, previous studies mainly focus on spatiotemporal changes of heavy precipitation over NEC, and atmospheric circulation anomalies related to one certain event, whereas few refer to the characteristics of multiple clustered heavy precipitation cases during the past six decades and associated atmospheric regimes, which is the central issue of this study.

It is well acknowledged that water vapor transport (WVT) is of great importance for precipitation (Chen et al. 2012a; Gimeno et al. 2012; Sun et al. 2011; Li et al. 2016; Hu et al. 2018). Additionally, precipitation anomalies are highly connected with anomalous WVT (Zhou and Yu 2005; Huang et al. 2015; Gimeno et al. 2016). The WVT budget is obviously reduced in association with the diminished precipitation over NEC after the late 1990s (Han et al. 2015). The intensity of WVT related to heavy precipitation events is stronger than that associated with general precipitation (Li et al. 2016). Li et al. (2019) identified that the difference in WVT between heavy precipitation and ordinary precipitation over NEC is related to meridional WVT. However, the WVT anomalies across each boundary associated with the CHP events, especially the WVT anomalies leading the occurrence of CHP, remains unclear, which is an issue of this study.

The remainder of this paper is structured as follows. Section 2 introduces the datasets and methods used in the study. Section 3 describes the interannual and interdecadal characteristics of CHP events during summer over NEC and discusses the associated atmospheric circulations and moisture transport. Finally, a brief summary of the main findings is presented in Sect. 4.

2 Data and methods

This study uses an advanced high-resolution (0.25° × 0.25°) daily rainfall observation dataset (i.e., CN05.1) (Wu and Gao 2013), which is constructed based on interpolation of data from more than 2400 meteorological stations in China. Daily and 6-hourly mean reanalysis data for 1961–2020, with horizontal resolution of 2.5° × 2.5°, are obtained from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) (Kalnay et al. 1996). The study area comprised NEC, which is defined as the region within 38°–54° N, 115°–135° E.

Composite and correlation analyses are used to analyze the circulation and WVT related to CHP events over NEC during summer. In this study, a CHP event is determined on the basis of the following two conditions: (1) daily precipitation is greater than the threshold value of heavy precipitation, and (2) more than 40% of the grid points in NEC satisfied condition (1) at the same time. The threshold value of heavy precipitation is defined as the 90th percentile value of precipitation > 0.1 mm in the daily precipitation data during 1961–2020. Summer is defined as the months of June, July, and August. Additionally, to analyze atmospheric stability, we calculate the \(\mathrm{K}\) index as follows (Chen et al. 2012b):

$$K=\left({T}_{850}-{T}_{500}\right)+{T}_{d850}-{\left(T-{T}_{d}\right)}_{700}$$

and we calculate the moisture flux vector \(\mathrm{Q}\) as follows (Trenberth et al. 2007):

$$Q={Q}_{\lambda }\overrightarrow{i}+{Q}_{\phi }\dot{j}$$
$${Q}_{\lambda }=-\frac{1}{g}{\int }_{{p}_{s}}^{{p}_{t}} \left(qu\right)dp, {Q}_{\phi }=-\frac{1}{g}{\int }_{{p}_{s}}^{{p}_{t}} (qv)dp$$

where \({\mathrm{Q}}_{\uplambda }\) is the integral of vertical WVT longitudinally, \({\mathrm{Q}}_{\upphi }\) is the integral of vertical WVT latitudinally, \({\mathrm{p}}_{\mathrm{s}}\) is the surface pressure, \({\mathrm{p}}_{\mathrm{t}}\) is 300 hPa and \(\mathrm{g}\) is 9.80665 (m s−2). The vertical integration of water vapor flux is calculated approximately according to the trapezoidal rule, and because complex terrain will affect the vertical integration of water vapor flux, this study used the method of integrating from the surface pressure to eliminate the impact of complex terrain (Li et al. 2016). In addition, we used the method in Sun et al. (2011) to calculate the moisture budget and horizontal moisture flux divergence.

3 Results

3.1 Characteristics of clustered heavy precipitation events in NEC

According to the definition in Sect. 2, 75 CHP events occur in summer over NEC during 1961–2020 (Table 1). Overall, 5 events occur during June and 70 cases appear in July–August, with more than half of those 70 cases observed in July (Fig. 1), i.e., CHP in NEC mainly occurs during midsummer. Moreover, the numbers of CHP events during the latter 10 days of July significantly exceed those during both the first and middle 10 days of the month. In August, CHP events appear most frequently during the middle 10 days of the month, follow by the latter 10 days; the numbers of CHP events during the first 10 days of August are the lowest. Figure 2 shows the decadal characteristics of the monthly and summer-mean CHP events for the previous 60 years. Higher occurrence of CHP during summer is observed during the 1980s and 1990s than in other decades, which is consistent with the decadal variation in CHP during July. Additionally, 59 of the 75 CHP events (approximately 79%) occur before 2000, and the remaining 16 events (approximately 21%) occur after 2000. The reduction in CHP events during the 2000s is mainly attributed to reduced CHP in July, and the reduction in CHP events during 2011–2020 is attributed to reduction in CHP during June and August (Fig. 2a).

Table 1 The 75 CHP events over NEC during 1961–2020
Full size table
Fig. 1
figure 1

Ten-day characteristics of CHP frequency (day) during summer over NEC during 1961–2020. The first, middle and latter 10 days of June refer to 1st June to 10th June, 11th June to 20th June, 21st June to 30th June, respectively. The first, middle and latter 10 days of July (August) refer to 1st July (August) to 10th July (August), 11th July (August) to 20th July (August), 21st July (August) to 31st July (August), respectively

Full size image
Fig. 2
figure 2

Decadal variations of CHP frequency (day) over NEC during 1961–2020: a summer months and b summer mean

Full size image

Figure 3 illustrates the temporal evolutions of the frequency/intensity of summer CHP events, and the CHP amounts over NEC during 1961–2020. In-phase variations are observed in the time series of frequency/intensity/amounts of CHP events during the past six decades. Specifically, they all have decreased around the mid-1970s, increased after the 1980s, and then reduced since the late 1990s (Fig. 3d). Moreover, the averages of CHP intensity and amounts are 11.30 mm per day and 20.27 mm per year, separately. The maximum of the CHP intensity is 24.66 mm per day, occurring on 22th July, 2012. The maximum of the CHP amounts is 66.15 mm, occurring in 1969. The CHP events are highly linked with atmospheric circulations and moisture transportation anomalies, which will be explored in Sect. 3.2.

Fig. 3
figure 3

Temporal evolution of a CHP frequency (day, gray line), b CHP intensity (mm day−1, gray line) and c CHP amounts (mm year−1, gray line) over NEC in summer during 1961–2020. The red line and black line represent the 9-year moving mean and the average for the past six decades, respectively

Full size image

3.2 Atmospheric circulation associated with CHP

Atmospheric circulation anomaly is the most direct contributor to the occurrence of heavy precipitation events. Different precipitation modes are attributed to different circulation anomalies (Han et al. 2019). Therefore, atmospheric circulations associated with CHP events over NEC are explored in this section. Figure 4 shows the spatial distributions of sea level pressure anomalies, 850-hPa geopotential height/horizontal wind anomalies, 500-hPa geopotential height anomalies, and 200-hPa zonal wind anomalies associated with the studied CHP events over NEC. As shown in Fig. 4a, the CHP events occur with marked negative height anomalies over East China, with an anomalous center over NEC, and positive height anomalies over the Northwest Pacific. In the middle troposphere, anomalous positive geopotential height can be seen over the Japanese archipelago and southern China, together with negative anomalies centered over NEC (Fig. 4b, c). Previous study has suggested that an anomalous low over NEC is favorable for enhanced local precipitation (Shen et al. 2011). Consistently, remarkable cyclonic wind anomalies are evident centered over NEC, together with anticyclonic wind anomalies centered over the Northwest Pacific. Anomalous southwesterly or southerly winds on the western flank of the anticyclone transport warm humid air from the western Pacific toward NEC across the southern boundary (Fig. 4b). When CHP events occur, the WNPSH ridge is located at approximately 33° N, which is approximately 8° further north of its summer mean position (Fig. 4c). Generally, the rain belt is often located 6°–10° north of the WNPSH ridge (Zhu 2007), which covers part of NEC. Additionally, the western point of the WNPSH ridge is located at approximately 123°E when CHP events occur over NEC, which is 10° further west of the summer mean position (Fig. 4c). Previous studies showed that northwestward shift of the WNPSH is conducive to intensified transport of moisture toward NEC (Gao and Gao 2018; Cao et al. 2018). Therefore, the westward and northward shifts of the WNPSH are favorable for the occurrence of CHP over NEC. Additionally, in the upper levels, a significant westerly anomaly is elongated over northern China and an easterly anomaly is extended from southern China to the Northwest Pacific Ocean (Fig. 4d). The position of the westerly jet axis is abnormally northward when CHP events occur over NEC. Specifically, it is generally located to the south part of NEC, whereas it is located over the middle part of NEC when CHP occurs. Cao et al. (2018) showed that northward shift of the upper-level westerlies over NEC is beneficial for the occurrence of heavy precipitation events.

Fig. 4
figure 4

Circulation anomalies associated with CHP events during summer: a sea level pressure (SLP, mb), b 850-hPa geopotential height (hPa, shadings) and horizontal wind (m s−1, vectors), c 500-hPa geopotential height (hPa, shadings) with a 588-gpm (red lines) and a 586-gpm (blue lines) mean position in summer at 500 hPa (dashed lines are for CHP events and solid lines are the summer mean position during 1961–2020), and d 200-hPa zonal wind (m s−1, shadings) and the jet axis (black lines). Stippling areas indicate values that significantly exceed the 95% confidence level, estimated using the Student’s t-test. The red rectangle represents the NEC region

Full size image

Atmospheric stability is of major importance regarding the occurrence of heavy precipitation events (Li et al. 2016). The K index is a meteorological index that describes atmospheric stability (Chen et al. 2012b); the larger the K value, the more unstable the atmospheric circulation. Anomalies of wind divergence at lower and upper levels are shown in Fig. 5a, b, respectively, and K value anomalies at 850 and 200 hPa, as well as the vertical velocity anomalies, are shown in Fig. 5c, d, respectively. Characteristically, CHP events are associated with significant lower-layer convergence anomalies and upper-layer divergence anomalies over NEC, which excite dominant local upward movement anomalies (Fig. 5c, d). Moreover, an anomalous anticyclone is centered over southeastern parts of NEC, together with easterly anomalies over southern China and southwesterly anomalies over NEC. This is consistent with the zonal wind anomalies related to CHP (Fig. 4d). Accordingly, the K values centered on NEC increase significantly when CHP affects NEC (Fig. 5c, d). It suggests that increased unstable energy over NEC is beneficial for the occurrence of CHP.

Fig. 5
figure 5

Spatial distributions of a 850-hPa horizontal wind divergence anomalies (10−6 s−1, shadings), b 200-hPa horizontal wind divergence anomalies (10−5 s−1, shadings) and horizontal wind anomalies (m s−1, vectors) between CHP events and the summer mean during 1961–2020, c same as a but for the K index (°C, contours) and 850-hPa vertical velocity (10−2 m s−1, shadings), d same as c but for 200-hPa vertical velocity (10−2 m s−1, shadings). Stippling areas indicate values that significantly exceed the 95% confidence level, estimated using the Student’s t-test. The red rectangle represents the NEC region

Full size image

The formation of precipitation is closely related to WVT, and abnormal WVT has important indicative significance regarding change in regional precipitation. The vertically integrated WVT during summer and the moisture anomalies for CHP events over NEC are presented in Fig. 6a, b, respectively. The moisture during summer over NEC mainly originates from the subtropical western Pacific, South China Sea, and South Asian monsoon areas (Fig. 6a), which is consistent with the findings of Sun et al. (2007). Compared with the climatology, when CHP events occur over NEC, the southerly flow over East China strengthens markedly and transports moisture northward to NEC across the southern boundary. Accordingly, anomalous moisture convergence dominates NEC and a cyclonic moisture anomaly is centered over NEC. Additionally, an anticyclonic moisture anomaly is located over the Sea of Japan. The southwesterly flow on the western flank of the anticyclonic moisture anomaly transports water vapor from the Northwest Pacific to NEC across the southern boundary. Moreover, the anomalous westerly flow crossing the western boundary is intensified and transports water vapor from inland areas to NEC.

Fig. 6
figure 6

a Climatology of vertically integrated moisture flux from the surface to 300 hPa (kg m−1 s−1, vectors) and its divergence (10−4 kg m−2 s−1, shadings) in summer during 1961–2020. b Spatial distributions of integrated water vapor transport (kg m−1 s−1, vectors) and divergence of water vapor flux (10−4 kg m−2 s−1, shadings) anomalies between the CHP events and the climatology during 1961–2020. Stippling areas indicate values that significantly exceed the 95% confidence level, estimated using the Student’s t-test

Full size image

Figure 7 illustrates the WVT anomalies and the associated moisture divergence at different levels associated with CHP over NEC. Because atmospheric moisture lies primarily in the lower layers of the atmosphere (Sun et al. 2011; Piao et al. 2021), the WVT anomaly diminishes in the upper levels (Fig. 7c, d). The anomalous WVT associated with CHP over NEC can be tracked southward to the subtropical Northwest Pacific, and northward to inland areas near the surface (Fig. 7a). Anomalous moisture divergence appears over the Northwest Pacific and both southern and western China when CHP events occur. The intensified southerly flow over eastern China carries warm moist air from the Northwest Pacific and southern China northward across the southern boundary of NEC. Additionally, near the surface, the westerly flow from inland areas across the western boundary and the easterly flow from the mid-latitude western Pacific across the eastern boundary also contribute to the moisture convergence anomalies over NEC. Hence, notable moisture convergence anomalies dominate NEC. At 850 hPa, the transport of moisture across the eastern boundary plays a negative role in the net moisture budget over NEC (Fig. 7b), which is different from the case of the near-surface transport (Fig. 7a). It can also be observed that anomalous southwesterly flow transports water vapor from the subtropical Northwest Pacific and southern China northward to NEC across the southern boundary, which is consistent with the near-surface transport. The northwesterly flow is enhanced in comparison with the climatology and it transports water vapor from inland areas eastward to NEC across the western boundary. Thus, remarkable moisture convergence anomalies affect NEC, although their magnitude is less than that near the surface. However, different anomalies are observed at middle and upper levels (Fig. 7c, d). The WVT at the northern and eastern boundaries plays a dominant but negative role in the net moisture budget over NEC in comparison with that in the lower troposphere. The WVT across the southern and western boundaries contributes only minimally to water vapor accumulation over NEC. Hence, anomalous moisture divergence is dominant over NEC at middle and upper levels.

Fig. 7
figure 7

Spatial distributions of water vapor transport (m s−1, vectors) and divergence of water vapor flux (10−7 s−1, shadings) anomalies between the CHP events and the climatology during 1961–2020: a 1000 hPa, b 850 hPa, c 500 hPa, and d 300 hPa. Stippling areas indicate values that significantly exceed the 95% confidence level, estimated using the Student’s t-test

Full size image

We calculate the averaged WVT budget across each boundary during summer and for the CHP events during 1961–2020 (Table 2). Generally, the WVT across both southern and western boundaries of NEC contributes positively to water vapor accumulation over NEC, while the WVT across both northern and eastern boundaries makes a negative contribution. Climatologically, the net WVT budget is 12.00 × 107 kg s−1 across the southern boundary, − 2.21 × 107 kg s−1 across the northern boundary, 13.58 × 107 kg s−1 across the western boundary, and − 20.47 × 107 kg s−1 across the eastern boundary. Comparatively, when CHP events occur, the net influx is 32.6 × 107 kg s−1 across the southern boundary and 20.36 × 107 kg s−1 across the western boundary. The net outflux is 6.84 × 107 kg s−1 across the northern boundary and 31.02 × 107 kg s−1 across the eastern boundary. Hence, water vapor influx and outflux are both intensified remarkably during CHP events relative to the climatology. Specifically, the increment of net WVT influx across the southern (western) boundary of NEC for CHP events accounts approximately 171.67% (49.93%) to summer climatology. The summer net moisture budget is 2.90 × 107 kg s−1 at NEC for the climatology, and increases to 15.10 × 107 kg s−1 for CHP events. The increment of net moisture budget for CHP events accounts approximately 420.69% to the climatology. Such an abundant supply of moisture is beneficial to the occurrence of CHP events over NEC.

Table 2 Water vapor transport budget across each boundary of NEC for summer mean and CHP events during 1961–2020 (WVT budget units: 107 kg s−1)
Full size table

Figure 8 shows the evolution of the areal mean precipitation, net WVT budget, and WVT budget across each boundary for 3 days before and 1 day after the occurrence of CHP events over NEC. The CHP is represented as time step 0, and the time steps of − 1 and 1 indicate the day before and the day after a CHP event, respectively. The net moisture budget generally begins to increase 2 days before the CHP, peaks 1 day before the CHP (time step: − 1), and then declines rapidly (Fig. 8a). Clearly, results from the moisture budget across the southern boundary are nearly the same as the net moisture budget (Fig. 8b). The net moisture influx across the southern boundary plays an important role on the day of occurrence of the CHP. It is notable that the net moisture budget of the southern boundary is mostly > 0 (i.e., net moisture influx) for the 3 days before the occurrence of the CHP, which implies that WVT across the southern boundary always contributes to the net moisture budget over NEC. Although weaker than that across the southern boundary, the moisture budget across the western boundary begins to accumulate 3 days before the CHP (Fig. 8d), and it contributes positively to the net moisture influx during the CHP event (Table 2). Moreover, the moisture budget across the northern boundary changes little from 3 days before the CHP to 1 day after the CHP (Fig. 8c). The net moisture outflux across the eastern boundary increases after time step − 1 (Fig. 8e). The above results show that WVT 1 day before a CHP event plays a key role in the water vapor accumulation over NEC.

Fig. 8
figure 8

Evolution of a the net budget (Q_net), the WVT budget across, b the southern (Q_s), c the northern (Q_n), d the western (Q_w), and e the eastern (Q_e) boundaries of NEC and the regionally averaged daily precipitation (P) in the 3 days before and 1 day after the CHP over NEC, f same as a, but showing the average evolution of WVT budget and precipitation of multiple cases. Time step 0 represents the start of the CHP. The gray lines represent the WVT budget of the 75 CHP cases, the blue solid lines represent their averages, the blue dashed lines represent zero values of the WVT budget, and red lines represent precipitation of multiple cases averages (WVT budget units: 109 kg s−1, precipitation units: mm)

Full size image

Generally, the net moisture budget that begins to increase 2 days before the CHP and peaks 1 day before the CHP is predominantly attributed to the pre-accumulated net WVT influx across the southern boundary. The CHP is closely connected to the WVT budget at the southern boundary at time steps − 1 and 0 with correlation coefficients of 0.41 and 0.40, respectively (significant at the 99% confidence level). This verifies that CHP depends not only on coinstantaneous WVT but also on pre-accumulated moisture influx across the southern boundary, which is in accord with the findings of Sun et al. (2007) and Sun and Wang (2013). Thus, the details of the WVT budget across the southern boundary for 1 day before and 1 day after CHP events is discussed in the following.

The evolution of the net moisture budget and the moisture influx across the southern boundary for 24 h before and 18 h after CHP are shown in Fig. 9a, b, respectively. Further analyses show that both the net moisture budget and the moisture influx across the southern boundary increases consistently approximately 24 h before the CHP, peaks 6 h before the CHP, and then diminishes gradually, indicating that the phase of precipitation lags the phase of the net moisture budget by 6 h. This implies that the accumulation of water vapor 1 day earlier, especially the net moisture budget 6 h earlier, is of great importance for prediction of the occurrence of CHP events.

Fig. 9
figure 9

Evolution of a the net WVT budget and b the WVT budget at the southern boundary in the 24 h before and 18 h after CHP in NEC. Time step 0 represents the time of the CHP, the gray lines represent the WVT budget of the 75 CHP cases, the blue solid lines represent their averages, and the blue dotted lines represent the zero values of the WVT budget (WVT budget units: 109 kg s−1)

Full size image

4 Conclusions

Based on the threshold method, this study defined CHP events over NEC and then explored the characteristics of CHP events, and the associated atmospheric circulation and moisture transportation anomalies. The main conclusions derived are summarized as follows.

The CHP events mainly occur during midsummer, especially in July. The numbers of CHP events during the latter 10 days of July significantly exceed those in other 10-day periods during the summer months. The frequency and the intensity of CHP events over NEC display significant interdecadal variation during the previous six decades, i.e., both decreased/weakened after the mid-1970s, increased/strengthened after the early 1980s, and decreased/weakened after the late 1990s. The occurrence of CHP events diminished during the 2000s and 2010s. In comparison with the climatology, CHP events are usually accompanied by marked negative height anomalies centered over NEC in the lower troposphere, westward and northward shifts of the WNPSH in the middle troposphere, and northward shift and intensification of the westerly jet in the upper troposphere. Additionally, significant lower-layer convergence anomalies and upper-layer divergence anomalies over NEC lead to increased atmospheric instability and upward movement in the vertical direction that play important roles in the occurrence of local CHP. The CHP events co-occur with strengthened water vapor transport, especially that in the lower troposphere. The net moisture budget begins to increase 2 days before the CHP and it peaks 1 day before the CHP. Further results indicate that the phase of the net moisture budget, especially the moisture influx across the southern boundary, leads the CHP by 6 h. Furthermore, to reduce the uncertainties from the different criterion to define the CHP, another two classifications (including 30% and 50% of the grid points over NEC) are calculated (Tables S3 and S5). Similar results can be obtained from these different cases (Tables S4 and S6, Figures S7–S24). It suggests that the results in this study are robust.

It should be mentioned that consistent atmospheric circulation anomalies and moisture transportation are observed when the strongest five CHP events appear, but with larger magnitude (Tables S1–S2, Figures S1–S6). The net moisture budget begins to increase 2 days before the occurrence of the strongest five CHP cases and peaks 6 h earlier. These results suggest that the moisture conditions 1 day earlier, especially 6 h earlier, are of predictive significance for the occurrence of CHP events over NEC.