Comparative Analysis of Summer Deep Convection Systems over the Tibetan Plateau and Sichuan Basin

Abstract

Based on GPM satellite observations during June to September from 2014 to 2023, deep convective systems (DCSs) over the Tibetan Plateau and Sichuan Basin exhibited distinct spatiotemporal and structural characteristics. Over the Plateau, DCSs were primarily concentrated in the central and eastern regions, with echo-top heights typically ranging from 15 to 17 km and 40 dBZ echo tops mostly found between 6 and 8 km. In contrast, the Basin displayed a more spatially uniform distribution of convection, characterized by lower echo-top heights (12–14 km) and higher 40 dBZ echo tops. Although both regions experienced a seasonal peak in DCS frequency in July, their diurnal variations differed significantly. The Plateau exhibited a pronounced unimodal peak between 13:00 and 16:00, which was driven by strong surface heating. In the Basin, a bimodal pattern was observed, with elevated frequencies during 23:00–02:00 and 08:00–11:00. This pattern was likely influenced by local thermodynamic and topographic conditions. The altitude of maximum corrected radar reflectivity (MaxCRF) was predominantly between 4 and 7 km over the Plateau and confined to 2–4 km over the Basin. Over the Plateau, DCS frequency increased significantly with elevation, consistent with the enhancing role of high terrain, whereas no comparable relationship was found in the Basin. Instead, convective activity in the Basin appeared to be modulated primarily by atmospheric instability and moisture availability, highlighting the contrasting environmental controls between the two regions.

1. Introduction

Deep convective systems (DCSs) are intense vertical atmospheric phenomena characterized by vigorous updrafts and the rapid development of cloud clusters. These events are typically triggered by surface heating, which generates strong upward motions that transport water vapor, aerosols, and pollutants from the boundary layer into the upper atmosphere. In some cases, these constituents can penetrate the tropopause and reach the stratosphere, thereby substantially influencing the distribution and chemical composition of stratospheric water vapor [1,2]. Such powerful convective processes are often associated with severe weather events, including thunderstorms, hail, and short-duration heavy rainfall. At the same time, they play a crucial role in the vertical redistribution of heat and moisture. Due to their strong vertical transport capacity, DCSs facilitate the rapid injection of tropospheric constituents into the upper troposphere and lower stratosphere (UTLS) region [3,4], significantly altering the spatial and temporal distribution of atmospheric trace species in this layer. Consequently, deep convection exerts a profound influence on both regional and global atmospheric circulation and ultimately shapes the broader climate system.

In recent years, significant progress has been made in the study of deep convective systems (DCSs), owing to advances in observational technologies and innovative research methods. From an observational perspective, long-term global monitoring has relied primarily on satellite remote sensing, particularly measurements of visible and longwave radiation. The Tropical Rainfall Measuring Mission (TRMM), jointly conducted by the United States and Japan, was equipped with a precipitation radar (PR) capable of capturing the three-dimensional structure of DCSs with high precision. By integrating multi-sensor data from airborne instruments—including the TRMM Microwave Imager (TMI), Lightning Imaging Sensor (LIS), and Visible and Infrared Scanner (VIRS)—researchers have been able to systematically obtain key parameters such as precipitation intensity, lightning frequency, and cloud-top temperature. Deep convective systems exhibit substantial regional contrasts. Xu and Zipser [5] showed that continental deep convection is generally stronger, with higher echo tops, more active mixed-phase processes, and frequent lightning, while oceanic convection is weaker and monsoonal convection lies in between. In tropical regions, sea surface temperature (SST) is a key regulator, and organized convection becomes more frequent once SST exceeds about 28 °C, accompanied by increased cirrus cloud coverage and radiative feedbacks that sustain upward motion [6]. Using TRMM data, Liu and Zipser [7] revealed significant regional differences in the global distribution of DCSs, pointing out that Africa is the region with the highest frequency of deep convective activity. Convective systems penetrating the tropopause commonly occur over Central Africa, the Indonesian archipelago, and South America. Over South America, CloudSat observations have further revealed distinct vertical structures of Amazonian deep convective cores, including a characteristic double-arc reflectivity pattern linked to ice microphysics and updraft strength. This study also highlighted pronounced diurnal variability, with stronger graupel/hail signatures during daytime than nighttime, underscoring the critical role of cloud microphysics and updraft intensity in tropical convection [8]. Although Africa emerges as a major center of deep convective activity globally, its regional and seasonal characteristics are shaped by prevailing climate conditions and large-scale atmospheric phenomena. Hart et al. [9] demonstrated that ENSO significantly influences convective hotspots in eastern southern Africa, with El Niño and La Niña events causing shifts in the location and intensity of deep convection across the region. In addition to large-scale controls, recent studies have provided new insights into mesoscale convective systems (MCSs) over West Africa, highlighting the critical role of low-level humidity, wind shear, and meridional temperature gradients in shaping MCS occurrence and intensity. Nkrumah et al. [10] showed that both dry- and monsoon-season MCS development is closely tied to meridional displacements of the southwesterly monsoon flow and the positioning of the Saharan temperature gradient, with transition-season conditions being particularly favorable for MCS initiation. Such findings underscore the importance of synoptic-scale variability and meridional circulation shifts in modulating the climatology of African deep convection. The analysis by Liu et al. [11] further indicated that intense deep convection is more likely to occur in semi-arid regions, whereas in precipitation-rich areas such as the equatorial oceanic convergence zones, the Southeast Asian monsoon region, and the Indian subcontinent, such phenomena are relatively rare. Similarly, Panasawatwong et al. [12] found that deep convection occurs more frequently over land with stronger precipitation intensity, but due to its smaller horizontal extent, its overall contribution to total rainfall is limited. Moreover, this type of convection tends to be more localized, lacking large-scale moisture transport and strong dynamic support. In North America, recent work has highlighted the spatial and temporal characteristics of deep convection initiation (DCI), showing that favored areas of DCI occur over higher terrain and near the Gulf Coast, with distinct diurnal cycles and interannual variability across the region [13]. Complementing these initiation-focused studies, Li et al. [14] developed a high-resolution (4 km, hourly) climatology of mesoscale convective systems (MCSs) and isolated deep convection (IDC) east of the Rocky Mountains during 2004–2017, demonstrating that while IDC occurs far more frequently than MCSs, both contribute substantially to regional precipitation with distinct spatiotemporal patterns. Such insights into the initiation stage complement research on the broader impacts of deep convection, which include the formation of stratospheric ice clouds (SICs) and the modulation of stratospheric water vapor content [15,16]. Overall, studies conducted in Africa, South America, North America, and the broader Asian monsoon regions have greatly advanced our understanding of deep convective systems. These works highlight the strong regional contrasts, seasonal and diurnal variability, and the influence of environmental factors such as topography, moisture, wind shear, and sea surface temperature on DCS occurrence and development. Insights from these international studies provide a valuable context for examining the characteristics and dynamics of deep convection in other regions.

Based on observations from the Tropical Rainfall Measuring Mission (TRMM) and the Global Precipitation Measurement (GPM) mission, deep convective systems (DCSs) are primarily concentrated in five key regions: the Asian monsoon domain, the western Pacific warm pool, the west coast of Africa, the Amazon Basin, and the Great Plains of North America. Among these regions, the Asian monsoon domain and the western Pacific warm pool exhibit particularly intense convective activity [17,18]. In terms of diurnal variability, deep convection over land shows significantly stronger daily fluctuations than over ocean. The continental diurnal cycle of DCSs typically has a bimodal structure, with a primary peak in the afternoon driven by local diabatic surface heating and a secondary peak around midnight associated with more complex formation mechanisms [19].

As the highest and largest tectonic uplift on Earth, the Tibetan Plateau has attracted extensive attention due to its unique natural environment and pronounced spatial heterogeneity. Numerous studies have demonstrated that the interaction between atmospheric circulation and the Plateau’s complex topography forms a distinctive hydrothermal pattern, characterized by a clear gradient from the warm and humid southeast to the cold and arid northwest [20]. Convective activity over the Plateau is strong in summer but weak in winter, with seasonal transitions in June and October. This variation is mainly controlled by changes in surface sensible heat fluxes driven by solar shortwave and net radiation. The abrupt intensification in early June and weakening in early October align with the “June transition” and “October transition” in Northern Hemisphere circulation identified by Ye et al. [21]. This transition is most pronounced over the Plateau’s main body, implying a possible influence on the hemispheric circulation transitions [22].

Based on 14 years of TRMM observations, Qie et al. [23] found that, compared with oceanic and South Asian regions, deep convective systems (DCSs) over the Tibetan Plateau are generally weaker in intensity and smaller in radar echo area. However, they occur more frequently and more readily reach higher altitudes, with cloud tops often extending into the upper troposphere. During summer, deep convection exhibits a distinct regional pattern, with the central and particularly the eastern Plateau serving as primary centers of convective activity. These systems often propagate eastward and exert significant impacts on downstream precipitation patterns. Gao et al. [24] reported that, compared with the southern slope of the Plateau, convective intensity within the Plateau interior is considerably weaker, overall activity is reduced, and nocturnal convection is notably suppressed. Nevertheless, owing to the Plateau’s mean elevation exceeding 4500 m, strong surface heating can still support the development of DCSs, allowing them to reach the tropopause and, in some cases, generate penetrating convection. Recent studies focusing on the eastern extension of the Plateau, including the Sichuan Basin, have revealed pronounced regional differences in the diurnal cycle of warm-season extreme precipitation events [25]. Long-duration events tend to occur at night, whereas short-duration events typically initiate in the afternoon to evening, features strongly influenced by local topography and mesoscale processes. Zhao et al. [26] noted that convective cores and anvils over the Plateau are relatively narrow and thin. The thin cores weaken radiative cooling at the top of the atmosphere, while denser anvil cloud tops enhance shortwave effects. The width and intensity of convection are positively correlated with vertical wind shear and instability.

The Sichuan Basin, located at the eastern edge of the Tibetan Plateau, is a key geomorphological unit within China’s three-step topographic staircase. Its enclosed terrain leads to significant regional climatic differences and supports diverse ecosystems. As a representative region of the Asian monsoon domain, the basin is simultaneously influenced by both the East Asian and South Asian monsoons and exhibits strong interactions with atmospheric circulation over the Tibetan Plateau [27]. The convective characteristics in this region reflect not only local topographic effects but also meteorological forcings. Variations in atmospheric circulation and topography further modulate the intensity and organization of convection [28,29,30,31]. Lee et al. [32] found that deep convection in the Sichuan Basin can strongly uplift boundary layer pollutants to the tropopause and even higher levels, and transport them westward under the influence of monsoonal circulation, ultimately affecting the atmosphere over southern Nepal. Meanwhile, deep convection in the basin exhibits pronounced seasonal and diurnal variations. The primary convective season occurs from late June to early August, with a peak from late June to early July. In terms of diurnal variation, convection generally shows a bimodal distribution, intensifying in the late morning to early afternoon and weakening in the evening [33,34]. Li et al. [35] found that convection in the Sichuan Basin shows marked nocturnal intensification, closely linked to the diurnal cycle of low-level winds. Stronger mean winds and larger diurnal amplitudes enhance nighttime moisture convergence and instability, favoring deep convection with a peak around midnight.

In addition, the diurnal characteristics of convection differ across topographic types: convection tends to follow a unimodal distribution over plateaus and mountainous areas, whereas in plains and basins it more often exhibits a multimodal pattern [36].

Although deep convective activity over the Tibetan Plateau has been extensively investigated, much less attention has been paid to the adjacent Sichuan Basin. This imbalance limits our understanding of differences in convective intensity, frequency, and spatiotemporal evolution between the two regions, despite their contrasting environments and the Plateau’s role in modulating East Asian atmospheric circulation. Previous studies have mainly examined large-scale patterns—such as the distribution, diurnal cycle, and extreme precipitation of deep convection—using satellite-based observations and basic statistical analyses [18,37], whereas discussions specific to the Sichuan Basin have remained fragmented and lacked direct comparisons with the Plateau. To address this gap, this study applies high-resolution precipitation retrievals from the Global Precipitation Measurement (GPM) mission to compare summer deep convection over the Tibetan Plateau and the Sichuan Basin, focusing on spatial distribution, temporal evolution, and topographic influences. By employing consistent satellite-based methods, this work provides a systematic cross-regional comparison that advances understanding of convection in these regions and offers insights for other plateau–adjacent plain settings, such as the Rocky Mountains and the adjacent Great Plains.

2. Data and Methods

2.1. Study Area and Data Overview

The Sichuan Basin, located on the eastern flank of the Plateau, differs substantially from the Plateau in topography, thermal conditions, and atmospheric circulation. The Plateau is characterized by high elevation, strong surface heating, and dynamic lifting, whereas the Basin is enclosed by mountains with abundant low-level moisture, creating a contrasting environmental background. These contrasting surface and environmental conditions provide context for examining convective systems in the two regions. Figure 1 shows the study area and surrounding major topographic features, including the Tibetan Plateau, Sichuan Basin, Wushan Mountains, Daba Mountains, and the Yunnan–Guizhou Plateau. This map provides the geographical context for the analyses presented in this study.

This study uses the GPM 2BCMB (Version 07) dataset (GPM DPR and GMI Combined Precipitation L2B), provided by NASA’s Precipitation Processing System (PPS), Greenbelt, MD, USA, in collaboration with the Japan Aerospace Exploration Agency (JAXA). The dataset provides precipitation retrievals with a horizontal resolution of 5 km, a vertical resolution of 250 m, and a temporal resolution of 1.5 h. The analysis covers the period from June to September in the years 2014–2023. Given the pronounced seasonality of convective thunderstorm activity over the Tibetan Plateau, and the fact that summer precipitation accounts for approximately 60–90% of the annual total [38], focusing on the summer season allows for a more representative characterization of deep convective features in this region.

Compared with its predecessor, the Tropical Rainfall Measuring Mission (TRMM), the Global Precipitation Measurement (GPM) satellite offers substantially enhanced observational capabilities, including extended coverage into mid- and high-latitude regions. By integrating data from the GPM Microwave Imager (GMI) and the Dual-frequency Precipitation Radar (DPR), the satellite enables more accurate detection of precipitation structure and intensity. These improvements significantly enhance the detection of light precipitation (e.g., <0.5 mm·h⁻¹), solid precipitation, and microphysical properties of hydrometeors. In addition, GPM precipitation products show notable advancements in spatial and temporal resolution, coverage, and retrieval accuracy compared with TRMM [39]. Validation studies across multiple regions in China further confirm that, although occasional discrepancies exist with ground-based observations in certain months, the overall performance of GPM products surpasses that of TRMM in both accuracy and consistency [40,41]. However, it should be noted that uncertainties remain in complex terrain regions such as the Tibetan Plateau and the Sichuan Basin. Mountain-based evaluations indicate that GPM exhibits the highest correlation with ground observations in mid-mountain areas (1250–2800 m, R ≈ 0.71), whereas its correlation declines markedly in high-mountain areas (>2800 m), accompanied by a systematic underestimation of approximately 13–16% [42]. These findings suggest that, although GPM generally performs well, its observation of convective system features may still be subject to certain uncertainties in high-altitude complex terrain. It should also be noted that June 2014 falls within the calibration phase of the GPM mission; therefore, data from that month were excluded from this analysis.

CAPE data for this study come from the ERA5 reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts (ECMWF), Reading, UK. The analysis uses hourly data for the summer months (June–September) from 2014 to 2023. ERA5 is the fifth-generation global atmospheric reanalysis product developed by the European Centre for Medium-Range Weather Forecasts (ECMWF). It provides a horizontal resolution of 0.25° × 0.25°, an hourly update frequency, and covers 37 pressure levels from 1 to 1000 hPa. The dataset includes key meteorological variables such as horizontal wind components, surface pressure, precipitation, and specific humidity. It also provides multi-level atmospheric parameters (e.g., geopotential height, temperature, and relative humidity) as well as surface-related variables including radiation fluxes, cloud properties, and soil temperature and moisture. These features make ERA5 broadly applicable for climate diagnostics and numerical modeling.

This study uses topographic data of the Tibetan Plateau from the ETOPO2v2c global elevation dataset, released by the National Centers for Environmental Information (NCEI) under the National Oceanic and Atmospheric Administration (NOAA), Silver Spring, MD, USA. The dataset provides a spatial resolution of 2 arc-minutes (approximately 3.7 km). The 3000 m elevation contour of the Tibetan Plateau was derived from the Comprehensive Dataset of Tibetan Plateau Boundaries provided by the National Tibetan Plateau Data Center, Beijing, China [43]. The boundary of the Sichuan Basin was extracted from the standard reference map GS(2016)2923, obtained from the official website of the National Administration of Surveying, Mapping and Geoinformation of China. The base map was used without modification.

2.2. Methods

In this study, we follow the approach of Xu et al. [44] to identify deep convective pixels, defining those with Maxht20 (20 dBZ echo-top height) exceeding 12 km (Maxht20 > 12 km) as deep convection. The identification procedure first selects convective precipitation pixels with MaxCRF (Maximum Corrected Reflectivity) above 40 dBZ (MaxCRF > 40 dBZ), and then pixels with Maxht20 exceeding 12 km (Maxht20 > 12 km) are classified as deep convective. Adjacent deep convective pixels detected within the same satellite overpass are grouped into a single deep convective event, with the maximum Maxht20 value representing the event’s intensity and spatial location [24]. For clarity, the identification criteria can be summarized as a simple logical condition: deep convection is identified when MaxCRF > 40 dBZ and Maxht20 > 12 km. Previous studies typically used a 14 km (Maxht20 > 14 km) threshold for Maxht20 to identify deep convective events [7]; Because the Sichuan Basin is low in elevation and convection there rarely reaches 14 km, using this threshold would underestimate its intensity. Based on existing methodology [44], this study adopts 12 km to enable more effective comparison between the Plateau and Basin. Unlike prior studies emphasizing the Plateau and its eastern regions, this work highlights regional contrasts and the role of terrain and surface conditions in shaping deep convection.

2.3. Statistical Methodology

To ensure the robustness of the results, standard statistical procedures were applied to the GPM satellite observations. Deep convective systems were objectively identified based on unified physical thresholds, and event characteristics were aggregated across space and time. Monthly and diurnal variations were derived from compositing and normalization techniques commonly used in climatological analyses. The relationships between deep convection, elevation, and vertical structure were quantified through frequency statistics and distributional metrics, including boxplot summaries and contoured frequency by altitude diagrams (CFADs). These approaches follow established practices in satellite-based convection studies and provide a consistent framework for cross-regional comparison between the Tibetan Plateau and Sichuan Basin.

3. Results

3.1. Spatial Distribution Characteristics

Based on statistical analysis of GPM satellite observations from June–September 2014–2023, 349 deep convective systems (DCS) were identified in the Tibetan Plateau study region (25–40° N, 75–105° E), while 116 DCS were detected over the Sichuan Basin (27–34° N, 102–110° E).

Figure 2 illustrates the spatial distribution and height classification of deep convective systems over the Tibetan Plateau and Sichuan Basin. White areas indicate regions where no deep convective systems (DCSs) were detected. These areas may actually experience occasional deep convection, but due to the GPM satellite’s orbital sampling, which provides intermittent coverage (typically twice daily along its orbit) rather than continuous full-region observations, some events may not be captured. Such gaps reflect an inherent limitation of the GPM dataset. The same explanation applies to the subsequent figures showing DCS spatial distributions. As shown in Figure 2a, deep convective systems are relatively sparse in the western and northern regions of the Tibetan Plateau, while they are more concentrated in the central, eastern, and southern parts. In contrast, deep convection in the Sichuan Basin is distributed fairly uniformly. Over the Tibetan Plateau, echo-top heights generally remain below 18 km and are mainly concentrated in the central and eastern areas. This observation agrees with Qie et al. [23], who reported dense deep convective activity in the central plateau extending eastward and comparatively infrequent occurrences in the west. Wu et al. [45] suggested that strong surface heating makes the lower atmosphere over the plateau a significant heat source. Because of low precipitation and weak latent heating in the western plateau, a climatic pattern develops with ascending motion dominating the east and descending motion prevailing in the west, leading to more frequent and intense convection in the eastern plateau. To aid visualization, the Sichuan Basin is outlined by a black box in Figure 2a, given its relatively small spatial extent compared with the Plateau. This box has no statistical meaning but simply highlights the study region for easier comparison.

Compared to the Tibetan Plateau, the Sichuan Basin experiences fewer deep convective systems. The 20 dBZ echo-top heights of these systems mainly cluster around 13 km, with only a small fraction exceeding 15 km; these more intense events are primarily confined to the southern basin. Zheng et al. [34] demonstrated that deep convective activity in China’s subtropical regions is closely linked to the East Asian summer monsoon. Terrain features play a key role in shaping convection spatially: activity tends to be stronger and more frequent on windward slopes, such as the Tibetan Plateau, and generally weaker on leeward slopes like the Sichuan Basin. Figure 2b further highlights the contrast: echo-top heights over the Plateau mainly range from 14 to 16 km, with the largest proportion (31%) in the 15–16 km category, while in the Sichuan Basin, 68% of events fall within 12–14 km. This clear difference underscores the stronger vertical development of convection over the Plateau and the relatively shallow nature of convection in the Basin.

Figure 3 shows the spatial distribution of 40 dBZ echo-top heights and Convective Available Potential Energy (CAPE) over the Tibetan Plateau and the Sichuan Basin. Over the Tibetan Plateau, the 40 dBZ echo-top heights exhibit a distinct spatial pattern, as shown in Figure 3a, with prominent concentrations in the central and eastern regions. In these areas, echo-top heights generally range from 6 to 7.5 km, suggesting that deep convective activity is primarily confined to the lower troposphere. This observation agrees with Wu et al. [46], who reported that although 20 dBZ echo-top heights may be similar across regions, convective intensity as reflected by 40 dBZ echo-top heights varies considerably. For example, the average 40 dBZ echo-top height on the southern slopes of the plateau exceeds 8 km, while the interior plateau typically shows heights closer to 7 km. In contrast, 40 dBZ echo-top heights in the Sichuan Basin are more uniformly distributed, predominantly ranging from 8 to 9.5 km, with some localized areas reaching up to 10 km. This suggests that deep convection in the Sichuan Basin is both spatially extensive and occurs at relatively higher tropospheric altitudes. Overall, 40 dBZ echo-top heights over the Tibetan Plateau tend to be lower and more spatially concentrated, whereas those in the Sichuan Basin are higher and more evenly distributed. These results indicate that convection over the Tibetan Plateau is weaker in intensity but geographically concentrated, whereas convection in the Sichuan Basin is stronger and spatially more homogeneous.

Figure 3b shows that the spatial distribution of CAPE closely matches that of 40 dBZ echo-top heights in Figure 3a. Over the Tibetan Plateau, CAPE values are generally low, mostly ranging from 200 to 300 J·kg⁻¹, with concentrations in the central-southern and eastern regions. Luo et al. [47] reported that average CAPE over the plateau is about 156 J·kg⁻¹, typically ranging from 0 to 500 J·kg⁻¹. In contrast, the Sichuan Basin exhibits a markedly different CAPE distribution, with values ranging from 200 to 600 J·kg⁻¹. The highest values (500–600 J·kg⁻¹) concentrate mainly in the basin’s central part. This pattern agrees with Riemann-Campe et al. [48], who suggested that CAPE is primarily controlled by near-surface specific humidity and is less sensitive to temperature variations. Geographically, the Tibetan Plateau’s average elevation exceeds 4000 m, resulting in thinner air and low water-vapor content. Even during the monsoon season, these topographic conditions limit near-surface moisture availability and inhibit CAPE development. Conversely, the Sichuan Basin, as a typical inland basin strongly influenced by the monsoon climate, receives abundant moisture during summer. The combination of elevated near-surface specific humidity, warm temperatures, and frequent precipitation creates favorable conditions for higher CAPE values, increasing the likelihood of deep convective activity. This contrast in CAPE further highlights the thermodynamic limitations of the Plateau compared to the Basin, reinforcing the different convective environments in the two regions.

In addition to thermodynamic conditions (CAPE and near-surface moisture), dynamic factors such as vertical wind shear also play critical roles in the development of deep convection, jointly determining its intensity, organization, and lifetime. Figure 3c shows the distribution of vertical wind shear over the Tibetan Plateau and Sichuan Basin. Over the Tibetan Plateau, vertical wind shear generally increases from southwest to northeast, with values reaching up to about 8 m·s in the northeastern region, while remaining much weaker in the southwest. Relatively weak wind shear limits the organizational capacity of convective clouds, making it difficult for convective systems to form stable rotational structures, which often results in localized and short-lived convective cells. In contrast, vertical wind shear across the Sichuan Basin generally exceeds 6 m·s and averages higher than that over the Plateau, providing favorable dynamic conditions for convection. This stronger shear supports the continuous development of vigorous updrafts, thereby sustaining the spatial expansion and longevity of convective systems. Zhao et al. [26] demonstrated that vertical wind shear strongly influences the organization and evolution of deep convective systems. Stronger shear favors larger and more organized convection, promoting horizontal growth and persistence, as seen in the Sichuan Basin. Over the Tibetan Plateau, weaker shear limits convective development. Vertical wind shear not only strengthens updrafts and cold pools but also helps maintain convection by modulating cloud tilting and moisture transport. Overall, although thermodynamic factors also contribute, shear plays a dominant role in convective development and indirectly affects diabatic heating and the regional radiation budget [49].

Figure 3d shows the spatial distribution of near-surface specific humidity. Over the Tibetan Plateau, specific humidity is generally low, with most areas below 10 g·kg. This reflects the thin air and limited water vapor at high elevations, which restricts the accumulation of CAPE. As a result, even under strong local heating, the development of intense deep convection is difficult. In contrast, near-surface specific humidity in the Sichuan Basin is significantly higher, typically between 10 and 15 g·kg. Influenced by the summer monsoon, the basin is characterized by abundant water vapor and high temperatures. This ample moisture not only increases the moist adiabatic potential of the atmosphere but also enhances the persistence of convective updrafts, allowing CAPE to be more fully released and thereby facilitating the formation of high-intensity, spatially extensive deep convection. Morrison et al. [50] demonstrated that environmental relative humidity critically influences the initiation and development of deep convection. Higher humidity reduces buoyancy dilution by entrained air, allowing updrafts to overcome inhibition more easily, while drier environments weaken updrafts and require larger thermal perturbations to trigger convection. Relative humidity thus affects buoyancy loss, cloud-top height, and the spatial scale of initiation. Moreover, both specific and relative humidity are vital for the transition from shallow to deep convection: moist conditions in the free troposphere and boundary layer strengthen updrafts and counteract entrainment, whereas dry conditions suppress buoyancy and favor shallow convection. Therefore, sufficient humidity is a fundamental prerequisite for deep convection, with an influence that may exceed that of vertical wind shear and aerosols [51].

A synthesis of Figure 3a–d indicates that deep convection over the Tibetan Plateau is limited by low CAPE, scarce moisture, and weak vertical wind shear, resulting in weaker, localized, and short-lived activity. In contrast, the Sichuan Basin, with higher CAPE, ample near-surface moisture, and stronger shear, supports more intense, organized, and widespread convection. These contrasts highlight the key role of topography and the combined thermodynamic–dynamic environment in shaping summer deep convection: basin topography and abundant moisture favor strong convection in the Sichuan Basin, whereas high elevation and dryness suppress it over the Plateau.

3.2. Temporal Variation Characteristics

Figure 4 shows the monthly evolution of the average proportion of deep convective systems in the Tibetan Plateau and Sichuan Basin from June–September 2014–2023. Both regions display clear seasonal variations in deep convective activity. On the Tibetan Plateau, deep convection peaks in July, accounting for 34.8% of total events, a share notably higher than in other months. This pattern reflects a marked summer enhancement of deep convection, likely driven by topographic and environmental factors.

In contrast, the temporal variation in deep convective activity in the Sichuan Basin is less pronounced, although the peak also occurs in July. Despite lower amplitude compared to the Tibetan Plateau, both regions share a similar seasonal pattern, with maximum activity in July followed by a marked decline in September. This seasonality likely reflects the advance and retreat of the monsoon system. The intensification of the South Asian and Southwest monsoons during summer supplies abundant moisture and thermal energy to both regions, creating favorable conditions for deep convection, especially in July. As the monsoon weakens in September, moisture transport diminishes, leading to a corresponding decrease in deep convective activity. Seasonal changes in solar radiative heating may also modulate deep convection. It should be noted that due to the GPM satellite’s sampling and limitations in complex terrain, some deep convective events, especially in high-altitude regions, may not be captured. Consequently, the observed frequencies might slightly underestimate the actual occurrence of deep convection.

Figure 5 shows the spatial distribution of deep convective systems from June–September over the Tibetan Plateau and Sichuan Basin, highlighting pronounced seasonal evolution in both regions. On the Tibetan Plateau, deep convection in June is clearly clustered in the central and eastern parts, consistent with limited moisture and lower surface temperatures early in the summer. In July, DCS occurrence increases substantially, covering most of the plateau except its western margins, with activity peaking both in frequency and spatial extent. This expansion is driven primarily by the strengthening southwest monsoon and shifts in large-scale atmospheric circulation [52], which provide favorable dynamic conditions for convective development. In August, convection remains active but shows a slight reduction in density compared to July, while in September activity declines sharply, becoming sporadic across the central and eastern plateau as the monsoon weakens and thermal forcing diminishes.

In contrast to the Tibetan Plateau, the Sichuan Basin displays a distinct spatial evolution of deep convection. In June, convective activity is primarily concentrated in the basin’s northeast, likely driven by localized surface heating and instability. Activity intensifies rapidly in July, with DCSs spreading across nearly the entire basin and reaching their maximum frequency. Compared with the plateau, the basin experiences more uniform and widespread convective coverage during this peak stage. In August, convection weakens and becomes concentrated in the central basin, and by September only scattered systems remain, indicating a rapid seasonal retreat similar to that over the plateau.

Taken together, Figure 5 shows that both regions peak in July, but with notable contrasts: convection over the Tibetan Plateau is strongly shaped by topography and remains geographically clustered, whereas in the Sichuan Basin it is more evenly distributed. These differences underscore how monsoon dynamics, moisture supply, and terrain jointly control the spatiotemporal evolution of deep convection.

Figure 6 illustrates the diurnal variation in the frequency of deep convective systems over the Tibetan Plateau and Sichuan Basin. All times are given in local standard time (LST; UTC+8) unless otherwise stated. On the Tibetan Plateau, deep convection frequency rises sharply around 13:00 and peaks near 16:00. This pronounced afternoon increase is likely driven by strong surface heating, which enhances daytime atmospheric instability and promotes frequent deep convective events. After the peak, frequency gradually declines after 17:00 and remains relatively low from 19:00 until about 12:00 the next day. Qie et al. [23] reported a similar afternoon peak in deep convective activity over the main plateau region, typically around 16:00, with minimal activity from midnight to approximately 10:00. The unimodal diurnal distribution in Figure 6 also aligns with Wu et al. [46], who identified the highest convective frequency near 16:00. Likewise, Na et al. [53] documented a clear diurnal cycle in deep convection initiation, peaking between 14:00 and 15:00 and reaching a minimum near 09:00–10:00. This distinctive afternoon maximum is closely linked to the daytime evolution of the atmospheric boundary layer (ABL) and diurnal moisture transport over the Plateau. Strong surface heating and low air density drive rapid ABL growth (up to ~3 km by afternoon), enhancing vertical transport and convective initiation [54]. At the same time, afternoon moisture transport reaches nearly twice the daily mean, leading to pronounced convergence through thermally driven upslope flows and regional circulation. This “air-pump effect” provides favorable conditions for moisture and energy accumulation, thereby promoting afternoon deep convection [55].

In contrast, the Sichuan Basin displays a distinct diurnal pattern of deep convective activity. Deep convection frequency is relatively high from 23:00 to 02:00, indicating significant nocturnal activity, and a secondary peak occurs from 08:00 to 11:00, likely related to morning radiative heating and local terrain influences. During other periods, convection frequency remains comparatively low, particularly in the afternoon and early evening, yielding a characteristic bimodal distribution. Observational studies support these patterns. Cao et al. [56] found that Convective activity in the Sichuan Basin is more pronounced in the evening to nighttime, with a clear minimum in the morning, in contrast to the Tibetan Plateau where convection is mainly triggered by strong surface heating and orographic lifting during the daytime. Xu et al. [44] also noted that weak local convection often occurs in the afternoon, whereas deep convection and lightning activity peak at night, highlighting the importance of nocturnal mesoscale convective development. The nocturnal convection peak in the Sichuan Basin is mainly driven by low-level moisture transport and the formation of a nocturnal low-level jet (LLJ). Moisture convergence along the southeastern boundary intensifies in the evening (~22:00 LST), several hours before the convective maximum (~02:00 LST), providing favorable conditions for initiation. The LLJ, often developing over the Yunnan–Guizhou Plateau and southern basin, effectively channels moist air into the region, while associated wind shear and turbulence further enhance convective growth [57,58]. A secondary morning peak (08:00–11:00 LST) is linked to residual ascent across the basin, especially along the eastern Plateau slope, whereas afternoon convection is suppressed by widespread subsidence and strong upslope winds that inhibit vertical development [57].

Figure 7 illustrates the diurnal spatial distribution of deep convective systems over the Tibetan Plateau and Sichuan Basin. Daytime deep convection is defined as convective events occurring between 07:00 and 20:00 LST, whereas nocturnal deep convection refers to events occurring during the remaining hours. On the Tibetan Plateau, daytime deep convection is frequent and widespread, covering nearly the entire region, with pronounced concentrations in the central and eastern areas. In contrast, nocturnal deep convection is relatively sparse, confined mainly to the central-western and northern parts, while the eastern region experiences minimal activity at night. Li et al. [59] analyzed deep convective systems across China and adjacent areas, reporting that approximately 95% of deep convection over the Tibetan Plateau occurs during daytime hours (06:00–20:00). This high daytime frequency likely results from daytime temperature increases and enhanced surface heating over the plateau, which favor convective development.

In the Sichuan Basin, nocturnal deep convection is slightly more frequent than daytime activity and is predominantly concentrated in the basin’s central area. This pattern likely reflects the region’s climatic conditions, terrain features, and nighttime temperature variations. Located at the eastern edge of the Tibetan Plateau and surrounded by the Wushan Mountains, Daba Mountains, and Yunnan–Guizhou Plateau, the basin experiences relatively high humidity and localized atmospheric instability at night, creating favorable conditions for nocturnal deep convection in its central area. During the daytime, although deep convective activity is common, its spatial distribution tends to concentrate in the northern and southern parts of the basin. This pattern is closely related to the unique dynamic and thermodynamic circulations characteristic of the Sichuan Basin and its surroundings [59]. Hu et al. [60], investigating regional differences in summer convective systems over the Tibetan Plateau, found that deep convection often clusters in areas with pronounced terrain variation, including mountainous regions, water bodies, and their interfaces. Approximately 25% of deep convection occurs in low-elevation plains, primarily at night.

3.3. Vertical Structural Characteristics

Figure 8 shows percentile box plots of four radar echo intensity parameters for deep convective pixels over the Tibetan Plateau and the Sichuan Basin: the maximum corrected radar reflectivity height (MaxCRF), the maximum height of the 40 dBZ reflectivity echo (Maxht40), the maximum height of the 20 dBZ reflectivity echo (Maxht20), and the storm-top height (TopCRF).

Significant differences in MaxCRF are evident between the Tibetan Plateau and the Sichuan Basin. On the Tibetan Plateau, radar echo intensities are mainly concentrated at higher altitudes, with strong echo systems occurring predominantly between 4 and 7 km. This pattern may result from the plateau’s pronounced thermal instability combined with significant orographic lifting. Gao et al. [24] reported that orographic lifting over the plateau raises MaxCRF values by about 4 km in the main plateau region compared to its southern slopes. In contrast, radar echo intensities in the Sichuan Basin are mostly concentrated closer to the surface, between 2 and 4 km. The median MaxCRF height on the plateau is around 6 km, indicating a clear vertical difference between the two regions.

For Maxht40, echo tops on the Tibetan Plateau mainly range from 6 to 9 km, with a median of around 7 km. In comparison, echo tops in the Sichuan Basin mostly lie between 7 and 10 km, exceeding those on the plateau. This suggests a broader vertical extent of deep convective systems in the basin. The basin’s humid climate and abundant moisture likely enhance updraft strength and promote vertical convective development, resulting in higher echo tops. Conversely, the plateau’s dry climate and thinner atmosphere constrain vertical ascent, producing comparatively lower echo tops. For Maxht20 and TopCRF, echo tops in the Sichuan Basin are generally lower, at about 12 km and 12.5 km, respectively. In contrast, the Tibetan Plateau shows a broader and notably higher distribution of echo tops, with maximum 20 dBZ echo heights reaching about 17 km and storm tops near 18 km. These observations agree with Wu et al. [46].

In terms of echo-top heights, deep convective systems on the Tibetan Plateau exhibit stronger vertical development. Owing to the plateau’s high elevation, deep convection often attains greater altitudes, with echo tops generally ranging from 15 to 17 km. This vertical extent agrees with Zheng et al. [61], who reported an average 20 dBZ echo-top height of about 15.8 km over the Nagqu region in central Tibet. In contrast, deep convection in the Sichuan Basin is more vertically constrained, with echo tops typically between 12 and 12.5 km, significantly lower than those on the plateau. This difference likely results from the basin’s humid climate and lower terrain elevation. The moist environment promotes convection at lower atmospheric levels, while greater atmospheric stability inhibits upward development of deep convective systems.

In summary, pronounced contrasts in echo-top heights exist between the Tibetan Plateau and the Sichuan Basin. The plateau’s high elevation and dry climate facilitate strong vertical development of deep convection, while the Sichuan Basin’s lower altitude and humid conditions favor shallower convection with a comparatively limited vertical extent.

Figure 9 presents the CFAD (Contoured Frequency by Altitude Diagram) composite of deep convective echoes over the Tibetan Plateau and the Sichuan Basin. On the Tibetan Plateau, radar reflectivity varies gradually with altitude. Notably, at higher altitudes near 16 km, reflectivity values cluster predominantly around 20 dBZ, forming a distinct high-frequency reflectivity band. Most deep convective echoes concentrate at relatively high reflectivities (above 15 dBZ) and lower altitudes, roughly between 7.5 and 10 km. Above 15 km, reflectivity decreases markedly, indicating weaker convective activity in the upper troposphere. Specifically, within the 15–17.5 km altitude range, reflectivity generally remains below 20 dBZ, suggesting low intensity and sparse distribution of upper-level convective systems. This vertical reflectivity pattern likely results from the plateau’s unique high-altitude climate, pronounced orographic lifting, and atmospheric stability.

In contrast, deep convective echoes in the Sichuan Basin predominantly occur at lower altitudes. Reflectivity values mainly range from 15 to 25 dBZ, with a higher frequency of strong echoes concentrated between 10 and 13 km. Notably, a pronounced peak in reflectivity frequency appears near 13 km. This feature is closely linked to the basin’s humid climate, abundant moisture supply, and relatively stable convective environment. Overall, significant differences exist in the vertical structure of deep convective systems between the Tibetan Plateau and the Sichuan Basin. The plateau’s reflectivity is concentrated at higher altitudes and exhibits strong vertical shear characteristics, whereas the Sichuan Basin’s convective echoes are centered at lower altitudes but display greater vertical extent and intensity.

Figure 10 presents representative vertical profiles of deep convective systems over the Tibetan Plateau and the Sichuan Basin. Comparing these profiles reveals clear differences in the vertical structure of deep convection between the two regions. On the Tibetan Plateau, deep convective systems are mainly concentrated in the mid- to upper troposphere and exhibit relatively high radar reflectivity at elevated altitudes, indicating strong vertical development. In contrast, deep convection in the Sichuan Basin shows weaker vertical growth, characterized by lower radar reflectivity values in the upper troposphere, reflecting comparatively lower convective intensity.

3.4. Influence of Terrain on Deep Convection Systems

Figure 11a illustrates the frequency distribution of deep convection occurrences across different elevation intervals on the Tibetan Plateau. Results show that deep convection frequency increases with elevation, with a notable rise between 4.5 and 5.0 km, while remaining relatively low between 3 and 4 km. This indicates that high-altitude regions of the plateau are more favorable for convection development, which is consistent with the role of orographic uplift and reduced atmospheric thickness in promoting instability and upward motion. Similar patterns were also noted by Lu et al. [62], who showed that deep convective initiation is positively correlated with elevation, especially when the relief amplitude exceeds 1 km, underscoring the enhancing role of high terrain.

In contrast, Figure 11b depicts the relationship between deep convection frequency and elevation in the Sichuan Basin, which shows a much more gradual variation. The highest frequency occurs in the lowest elevation band (0–50 m), exceeding 20 occurrences, and then declines gradually with elevation. Overall, frequency variations are weak, suggesting that convection in the basin is less directly linked to elevation and more strongly modulated by thermodynamic and geographic factors. This is also consistent with Lu et al. [62], who noted that in regions with small relief amplitude, elevation exerts only a weak control on convection frequency.

In summary, the response of convection frequency to elevation differs markedly between the Tibetan Plateau and the Sichuan Basin. While the plateau’s high terrain enhances convection through stronger orographic lifting and greater atmospheric instability, in the basin elevation plays only a minor role, with convection primarily controlled by atmospheric instability and moisture supply. The use of GPM satellite observations from 2014 to 2023 enables the analysis of elevation–convection relationships over nearly a decade, providing stronger statistical support and complementing earlier studies that relied on shorter periods or different observational datasets. This broader temporal coverage helps refine our understanding of how terrain influences deep convection in contrasting environments.

4. Conclusions and Discussion

4.1. Conclusions

Using GPM satellite data from June to September during 2014–2023, this study investigated deep convective systems over the Tibetan Plateau and Sichuan Basin, focusing on their spatiotemporal distribution, vertical structure, and topographic influences. The main findings are summarized as follows:

• Deep convective systems on the Tibetan Plateau are primarily concentrated in the central and eastern regions, with storm-top heights generally ranging from 15 to 17 km. In contrast, deep convection in the Sichuan Basin is more uniformly distributed, with storm tops mostly between 12 and 14 km; occurrences exceeding 15 km are largely confined to the southern basin. The 40 dBZ storm tops on the Tibetan Plateau are predominantly in the central and eastern areas, typically 6–7.5 km, whereas those in the Sichuan Basin are more evenly distributed, mainly 8–9.5 km.
• Deep convective activity in both regions peaks in July. The diurnal variation over the Tibetan Plateau follows a unimodal pattern, characterized by a rapid increase between 13:00 and 18:00 LST, with a maximum around 16:00 LST. In contrast, the Sichuan Basin exhibits a bimodal diurnal variation, with elevated deep convection frequencies during the nighttime to early morning hours (23:00–02:00 LST) and again in the late morning (08:00–11:00 LST).
• The maximum radar reflectivity height (MaxCRF) over the Tibetan Plateau is mainly concentrated between 4 and 7 km, whereas in the Sichuan Basin, it is primarily distributed between 2 and 4 km. The 20 dBZ storm top heights (Maxht20) on the Tibetan Plateau generally range from 14 to 17 km, while those in the Sichuan Basin mostly fall between 11 and 13 km. Overall, the Tibetan Plateau exhibits a broader and higher distribution of storm top heights compared to the Sichuan Basin.
• The frequency of deep convection over the Tibetan Plateau increases markedly with elevation, indicating that high-altitude regions promote the development of deep convective systems. In contrast, no significant correlation between deep convection frequency and elevation is observed in the Sichuan Basin, where the highest frequencies occur at low elevations and decline only slightly as altitude increases.

4.2. Discussion

The analysis of deep convective systems (DCSs) over the Tibetan Plateau and the Sichuan Basin enhances our understanding of their development and the influence of terrain. Overall, observations from the GPM satellite are consistent with previous studies, confirming the reliability of the dataset. However, satellite observations in complex terrain and high-altitude regions have certain limitations, which may slightly influence the accuracy of spatial–temporal patterns of deep convection derived from GPM, especially in high-altitude areas. Previous studies have shown that although GSMaP V07 generally outperforms IMERG V06B in terms of systematic and random error metrics, uncertainties still exist in high-altitude complex terrain. The newly introduced orographic rainfall classification in GSMaP V07 improves the detection of terrain-induced precipitation and reduces errors, but underestimation or misrepresentation of precipitation may still occur in some high-altitude areas [63]. Similarly, Pan et al. [64] found that GPM products correlate well with ground observations in low-altitude regions, with minor overestimation, but correlations decrease at mid-altitudes (2000–4000 m) due to complex terrain and snow cover and decline further above 4000 m. Although IMERG-F performs slightly better at high altitudes, all products are limited by topography, high-altitude dynamics, and sparse observations. These limitations should be considered when interpreting the spatiotemporal characteristics in this study.

Despite uncertainties in satellite observations, deep convection over the Tibetan Plateau exhibits distinct diurnal variations. Daytime heating drives upslope winds, triggering small-scale convection at mid-to-high elevations, while nighttime cooling induces downslope winds, leading to medium-scale systems along mountain fronts and lowlands that converge with monsoonal moisture to produce a nocturnal precipitation peak [65]. Compared with lower elevations, the plateau favors small- to medium-scale convection due to relatively low TPW and CAPE, lower LNB, and drier conditions, resulting in shallower and less frequent deep convective systems [47]. Xu et al. [66] found that even under low humidity, the plateau can generate considerable low cloud cover (LCC), whereas similar LCC at low elevations requires higher moisture. This is mainly due to strong boundary layer lifting and mixing, lower air density, and reduced lifting condensation levels, which facilitate cloud formation and the initiation of small- to medium-scale convection under relatively dry conditions.

Convection in the Sichuan Basin is strongly influenced by both thermodynamic and dynamic conditions. Dong et al. [67] showed that during a heavy rainfall case in June 2013, high CAPE and low CIN facilitated convective initiation, while strong vertical wind shear enhanced organization and maintenance. In addition, complex terrain and circulation further supported convection: vertical circulation structures guided development, convergence zones and low θse centers near steep slopes provided thermodynamic and dynamic support, and terrain-induced gravity waves strengthened low-level convergence and upward motion. These factors make the Sichuan Basin particularly favorable for strong convection and extreme rainfall [68].

In addition to these local and regional influences, large-scale circulation patterns can also modulate deep convection over the eastern Tibetan Plateau. For example, the negative phase of the Scandinavian (SCA) teleconnection pattern has been linked to enhanced summer precipitation in this region, primarily through intensified vertical moisture advection induced by anomalous ascending motion [69,70]. Such circulation anomalies can modify the frequency and intensity of deep convective systems by altering regional moisture transport and vertical motion fields. Previous research has shown that strongly developed deep convective systems can penetrate the tropopause, facilitating the vertical transport of tropospheric constituents and energy into the stratosphere, with water vapor and ozone as key components in this process [71]. Recent modeling studies over the eastern Tibetan Plateau have projected substantial intensification of precipitation extremes under future climate change, primarily driven by dynamic processes [72]. Given that deep convective systems are key contributors to such extremes, our results on their spatial–temporal characteristics provide a valuable reference for understanding potential future changes in convective activity in this region. Consequently, future studies should focus on elucidating the mechanisms and physical processes by which deep convection contributes to the vertical transport of atmospheric constituents such as water vapor and ozone. Moreover, Xiong et al. [73] reported that ozone levels over the Tibetan Plateau during winter are lower than in other regions at comparable latitudes, suggesting troposphere-to-stratosphere transport in winter, potentially driven by deep convection during this season. Therefore, expanding research to include winter deep convection would provide a more comprehensive understanding of these atmospheric processes.