Dynamic Diagnosis of an Extreme Precipitation Event over the Southern Slope of Tianshan Mountains Using Multi-Source Observations

Abstract

The southern slope of the Tianshan Mountains features complex terrain and an arid climate, yet paradoxically experiences frequent extreme precipitation events (EPEs), which pose significant challenges for weather forecasting. This study investigates an EPE that occurred from 20 to 21 August 2019 using multi-source data to examine circulation patterns, mesoscale characteristics, moisture dynamics, and energy-instability mechanisms. The results reveal distinct spatiotemporal variability in precipitation, prompting a two-stage analytical framework: stage 1 (western plains), dominated by localized convective cells, and stage 2 (northeastern mountains), characterized by orographically enhanced precipitation clusters. The event was associated with a “two ridges and one trough” circulation pattern at 500 hPa and a dual-core structure of the South Asian high at 200 hPa. Dynamic forcing stemmed from cyclonic convergence, vertical wind shear, low-level convergence lines, water vapor (WV) transport, and jet-induced upper-level divergence. A stronger vorticity, divergence, and vertical velocity in stage 1 resulted in more intense precipitation. The thermodynamic analysis showed enhanced low-level cold advection in the plains before the event. Sounding data revealed increases in precipitable water and convective available potential energy (CAPE) in both stages. WV tracing showed vertical differences in moisture sources: at 3000 m, ~70% originated from Central Asia via the Caspian and Black Seas; at 5000 m, source and path differences emerged between stages. In stage 1, specific humidity along each vapor track was higher than in stage 2 during the EPE, with a 12 h pre-event enhancement. Both stages featured rapid convective cloud growth, with decreases in total black body temperature (TBB) associated with precipitation intensification. During stage 1, the EPE center aligned with a large TBB gradient at the edge of a cold cloud zone, where vigorous convection occurred. In contrast to typical northern events, which are linked to colder cloud tops and vigorous convection, the afternoon EPE in stage 2 formed near cloud edges with lesser negative TBB values. These findings advance the understanding of multi-scale extreme precipitation mechanisms in arid mountains, aiding improved forecasting in complex terrains.

1. Introduction

Extreme precipitation events (EPEs) have become increasingly frequent and intense due to global climate change [1,2,3]. Although classified as low-probability weather phenomena, these events often lead to severe socioeconomic losses through mountain floods, urban waterlogging, and other disasters [4,5]. In arid regions, large-scale short-duration EPEs have become a focal point in severe convective weather research due to their abrupt onset and high destructive potential [6,7,8]. Considering the above, it is of great importance to conduct more in-depth research on the formation mechanisms of extreme precipitation events over complex surface types.

In recent years, the academic community has conducted systematic research on the spatiotemporal distribution characteristics and future evolution trends of global EPEs [9,10]. Most studies confirm that EPEs exhibit significant increasing trends in three dimensions: spatiotemporal intensity, duration, and frequency, with this growth trajectory projected to persist over the coming decades [2,11]. Focusing on the arid Northwest China, analyses based on meteorological station observations reveal that from 1961 to 2018, 92% of monitoring stations recorded increasing annual precipitation [12]. Notably, summer precipitation demonstrated the most pronounced enhancement, with the Ili Valley and western Tarim Basin emerging as core areas of significant growth, where summer rainfall contributes over 40% on average to annual precipitation totals [13]. Quantitative indicators specifically show that in this arid region, the annual proportion of heavy rain days, total precipitation, and annual maximum daily precipitation have been increasing at rates of 0.067%, 0.49 mm, and 0.42 mm per year, respectively [14]. The Tarim Basin, the second-largest basin in China, has a northern region characterized by an arid to semi-arid climate [15]. However, in recent decades, it has experienced unprecedented meteorological changes, particularly the occurrence of record-breaking EPEs [16,17]. The characteristics of EPEs near the study area differ significantly from those in the eastern monsoon regions of China [18,19]. EP in this region is primarily concentrated in the summer and is characterized by intense precipitation, short duration, strong extremes, and highly uneven spatial distribution [20].

However, current observation networks present notable limitations: insufficient station density restricts the ability to capture spatial gradient variations, a deficiency particularly exacerbated in vast uninhabited areas, short-duration precipitation events, and arid climates [21,22]. This constraint makes it challenging to accurately characterize high-resolution EPE features relying solely on station data [23]. Consequently, integrating novel remote sensing technologies has become imperative. Precipitation retrieval techniques utilizing geostationary satellites and multispectral satellite data, leveraging their advantages in high spatiotemporal resolution and extensive coverage, provide new technological pathways for enhancing the analysis of extreme precipitation characteristics in arid regions.

The formation mechanisms of EPEs have been extensively investigated, with large-scale atmospheric circulation identified as a critical modulator [24,25,26]. Case studies across global arid regions reveal distinct spatial heterogeneity in driving processes. In Chile’s Atacama Desert, northern Altiplano precipitation stems from Amazonian moisture advection driven by Bolivian high-induced circulation adjustments, while southern events result from Pacific air masses interacting with mid-latitude frontal systems and mid-tropospheric cut-off lows through baroclinic disturbances [27]. The seasonal differences in EPE in arid regions are also highly pronounced: U.S. Midwest spring extremes link to northward-shifted westerly jets and western Atlantic high-pressure anomalies, while summer droughts stem from the sustained high-pressure blocking of moisture transport [28]. In arid Northwest China, sea surface temperature anomalies across the Atlantic, Indian, and Pacific Oceans induce planetary-scale circulation adjustments through teleconnections, enhancing regional moisture transport [14,29,30]. Particularly in the Tarim Basin, a representative intermontane basin, torrential rainfall events demonstrate remarkable topographic forcing characteristics [31]. The bimodal pattern of the South Asian high in the upper troposphere (responsible for 70–90% of summer precipitation in Xinjiang) vertically couples with synoptic systems (e.g., Central Asian vortex and Siberian trough) [32], with the Central Asian vortex contributing to over 60% of torrential rainfall initiation events [33], unveiling the unique multi-scale interaction mechanisms governing EPEs in arid environments [34].

At the same time, it is crucial to identify the source of WV during EPEs in arid areas [35,36]. The Lagrange model, serving as a principal methodological framework in hydrometeorological studies [33,37], effectively delineates hydrometeorological transport pathways and associated thermodynamic evolution. The application of HYSPLIT trajectory diagnostics demonstrates that the Atlantic–Europe–Africa and Indian Ocean moisture source regions dominate the seasonal moisture transport patterns for precipitation over the Pamirs Plateau, collectively contributing over 70% of the total precipitable water during spring and winter seasons [38]. This research focuses on the northern side of the Qinghai–Tibet Plateau, which is bordered by high mountain ranges such as the Kunlun Mountains, Tianshan Mountains, and the Pamir Plateau, forming a unique “horseshoe” shaped terrain. Under the influence of atmospheric circulation, this topographic feature creates complex and dynamic WV transport channels and water circulation patterns, resulting in highly unstable precipitation behavior in this region [39,40]. Such thermodynamic–dynamic couplings pose fundamental challenges for short-term heavy precipitation (STHP) prediction in arid environments, where the rapid convective initiation processes, poorly constrained terrain–atmosphere feedbacks, and insufficient observational networks collectively limit the mechanistic understanding of EPEs [41].

This research takes EPEs on the southern slope of the Tianshan Mountains as its focal point. By employing multi-source data fusion, it systematically investigates three interconnected aspects: the spatiotemporal evolution characteristics of three-dimensional water vapor flux, the coupling relationship between mesoscale convective triggering mechanisms and topographic forcing, and the sensitivity of thermodynamic–dynamic parameters to precipitation extremes. This research seeks to enhance the understanding of the characteristics, potential causes, and underlying mechanisms associated with sudden EPEs in arid regions. The findings may contribute to ongoing efforts in improving early warning systems for extreme weather phenomena, particularly in key regions along the Belt and Road Initiative. The remainder of this paper is organized as follows: Section 2 provides a brief overview of the data sources and methods. Section 3 details the characteristics, causes, and mechanisms of STHP. Section 4 and Section 5 present the conclusions and discussion.

2. Data and Methods

2.1. Overview of the Study Area

The Aksu region is located in Xinjiang, northwestern China. Situated in the northwestern Tarim Basin on the southern slopes of the midwestern Tianshan Mountains, it exhibits elevations that gradually decrease from the northwest to the southeast. Bordered by the Pamir Plateau to the west and the Tianshan Mountains to the north, the region encompasses diverse landforms—including deserts, plains, and mountains—which contribute to its complex ecosystem and climatic patterns [42]. The Kuqa meteorological station in eastern Aksu has an elevation of 1081.3 m (Figure 1b). The analysis of historical climatological data indicates that the study area frequently experiences intense convective weather events, such as EPEs during the summer months (June to September), which often trigger flash floods, posing significant threats to tourism and traffic safety [32,43]. Figure 1c depicts a large canyon (>50 m depth, 5 km length) in the northern mountainous area of Kuqa, which served as the study site for stage 2 of the EPE analyzed in this study.

2.2. Data

This study utilizes multi-source observational data from the China Meteorological Administration, which includes hourly meteorological station records (wind direction/speed, temperature, relative humidity, pressure, and precipitation) from 290 locations (Figure 1b), Doppler weather radar data for the analysis of three-dimensional convective systems, and FY-4A satellite infrared black body brightness temperature (TBB) data [17,44], which has a spatial resolution of 4 km and an hourly temporal resolution sourced from the National Satellite Meteorological Center. The TBB data effectively identify cloud presence and capture key features of cloud evolution, enabling the observation of large-scale cloud distributions as well as the tracking of the complete lifecycle of small- to medium-scale cloud systems, including their formation, development, maturation, dissipation, and evolution processes [21,45].

Additionally, this study utilizes the Level-3 IMERG (Integrated Multi-satellite Retrievals for Global Precipitation Measurement) product from the GPM (Global Precipitation Measurement) satellite [46]. The latest IMERG V07 version provides a spatial resolution of 0.1° and a temporal resolution of 30 min. It is calibrated using station data from the Global Precipitation Climatology Centre (GPCC) and optimizes the Kalman filtering method along with quality metrics, significantly improving accuracy and making it more suitable for monitoring and assessing EPEs [23,47]. The precipitation in different regions of China varies greatly, and there is no unified standard (

Table 1. Comparison of precipitation thresholds and classification standards between Xinjiang and other regions of China.

Region	Heavy Precipitation		Short-Term Heavy Precipitation
	(24 h)	(12 h)	(1 h)
Xinjiang [48]	≥24.1 mm	≥20.1 mm	≥10 mm
Other parts of China [19,49]	≥50 mm	≥30 mm	≥20 mm

).

The ERA5 reanalysis dataset is a global high-resolution atmospheric reanalysis product with a spatial resolution of 0.25° × 0.25° and a temporal resolution of 1 h. It is developed and maintained by the European Centre for Medium-Range Weather Forecasts (ECMWF). This dataset provides continuous, consistent, and high-quality data on various meteorological and climatic variables, including temperature, humidity, wind speed, and surface temperature [50]. Additionally, the Global Data Assimilation System (GDAS) dataset features a spatial resolution of 1° × 1° and a temporal resolution of 3 h, incorporating 23 vertical layers. This dataset is integrated into the hybrid single-particle Lagrangian integrated trajectory (HYSPLIT) model by the Air Resources Laboratory of NOAA and serves as the primary input data for the HYSPLIT model [51].

2.3. Methods

The methodology used in this study is shown in a flowchart in Figure 2. ERA5 reanalysis data were utilized to extract geopotential height, wind, temperature, and vertical velocity fields across multiple levels [31]. The mid- to high-latitude circulation pattern at 500 hPa was identified, including key features such as the low vortex position and westerly trough structure. Concurrently, vorticity, divergence, and vertical velocity were calculated in conjunction with the spatiotemporal evolution of the South Asian High and upper-level jet streams at 200 hPa, enabling the analysis of the coupling mechanism between mid-to-low-level cyclonic convergence and upper-level jet-induced divergence. Sounding data were utilized to compute CAPE values preceding and following the EPE. The temperature advection equation was applied to quantify low-level cold/warm advection intensity, with its precipitation-triggering role validated through spatial correlation with frontal zone positions. Starting from the thermodynamic equation and neglecting diabatic heating and frictional terms, the horizontal advection term corresponds to temperature advection (the first term on the right-hand side).

$${\displaystyle \frac{𝜕T}{𝜕t}}=-\mathrm{V}\cdot \nabla T-\omega {\displaystyle \frac{𝜕T}{𝜕\mathrm{z}}}+{\displaystyle \frac{\alpha }{c_p}}\omega$$

(1)

$$-\mathrm{V}\cdot \nabla T=-(u{\displaystyle \frac{𝜕T}{𝜕x}}+v{\displaystyle \frac{𝜕T}{𝜕y}})$$

(2)

For the horizontal divergence of the water vapor flux, the corresponding formula is given below:

$$A=\nabla \cdot ({\displaystyle \frac{1}{\mathrm{g}}}\mathrm{V}q)={\displaystyle \frac{𝜕}{𝜕x}}({\displaystyle \frac{1}{\mathrm{g}}}uq)+{\displaystyle \frac{𝜕}{𝜕y}}({\displaystyle \frac{1}{\mathrm{g}}}vq)$$

(3)

where T denotes temperature; V = (u,v) denotes horizontal wind vector; and q represents specific humidity.

The HYSPLIT model is a tool used for computing backward trajectories, while cluster analysis is frequently utilized to investigate the origins and movement paths of WV [52,53]. Trajectories are classified into distinct clusters based on similarities in spatiotemporal transport characteristics. Each cluster is defined by its representative mean trajectory, which minimizes intra-cluster variability while maximizing inter-cluster differences [51]. The clustering algorithm utilizes the Total Spatial Variance (TSV) method, a metric that optimizes trajectory grouping by minimizing the sum of squared Euclidean distances between individual trajectories and their cluster mean [54]. The corresponding formula is provided below:

$$P^{\prime }(t+\mathrm{\Delta }t)=P(t)+V(P,t)\mathrm{\Delta }t$$

(4)

$$P(t+\mathrm{\Delta }t)=P(t)+0.5[V(P,t)+V(P^{\prime },t+\mathrm{\Delta }t) ]\mathrm{\Delta }t$$

(5)

where P(t) denotes the initial position; P′(t + Δt) represents the position determined by the first interpolation; V(P, t) is the initial average velocity vector; and Δt indicates the integration time step, which may vary throughout the simulation. P(t + Δt) refers to the final estimated position, while V(P′t + Δt) signifies the mean velocity vector at the first estimated position.

3. Result

3.1. Extreme Precipitation Overview

Based on the temporal and spatial distribution characteristics of this EPE, the precipitation can be categorized into two distinct stages. Stage 1 occurred between 20:00 on 20 August 2019, and 08:00 on 21 August. During this period, EP was concentrated along the edge of the oasis in the western plain of the study area, primarily during the evening of 20 August. In the western part of the study area, 19 stations recorded EP. The maximum precipitation, 59.9 mm, was recorded at the Wenshu Y5906 station (Figure 3a). Additionally, precipitation measurements at Y5990, Y8659, and Y8653 stations were 40.2 mm, 36.5 mm, and 30.8 mm, respectively. The maximum precipitation intensities at these stations were 38.1 mm h⁻¹, 29.3 mm h⁻¹, 15.7 mm h⁻¹, and 26.1 mm h⁻¹, respectively, with the heaviest precipitation occurring between 21:00 and 23:00 on 20 August.

During stage 2 (21 August 2019, 08:00–20:00; Figure 3b), localized heavy rainfall occurred in Keping and the northeastern mountainous areas near the Tianshan Grand Canyon. Specifically, Keping’s Y5956 station recorded a total of 46.3 mm of torrential rainfall, peaking at 29.6 mm h⁻¹ at 14:00. Subsequently, four stations near the Tianshan Grand Canyon (Y8625, Y8626, Y8636, and Y8627) experienced sudden downpours, with precipitation concentrated over approximately 3 h (16:00–19:00). The cumulative rainfall amounts at these stations were 33.6 mm, 32.3 mm, 31.3 mm, and 25.0 mm, respectively, while the maximum precipitation intensities reached 19.0 mm h⁻¹, 20.6 mm h⁻¹, 15.2 mm h⁻¹, and 8.4 mm h⁻¹, respectively.

Considering the site distribution within the study area, the number of observation stations is limited. By integrating these data with GPM satellite observations to analyze the timing of the strongest precipitation, it was found that stage 1 exhibited higher cumulative precipitation, greater hourly precipitation, and stronger precipitation intensity compared to stage 2. During the EPE, stations Y5906, Y5956, Y8625, and Y5990 recorded daily precipitation and maximum hourly precipitation values that were the highest since their establishment. Additionally, many stations reached or exceeded the threshold for hourly EP levels in Xinjiang, where the 97th percentile hourly precipitation is approximately 20.0 mm in northern Xinjiang and 10.0 mm in southern Xinjiang [48,55]. This precipitation event was the strongest EPE of 2019, characterized by a short duration (concentrated within 3 h), intense hourly rainfall, significant extremes, and relatively concentrated STHP. Compared to the Ili region on the northern slope of the Tianshan Mountains, although the daily precipitation was lower, the precipitation intensity was significantly greater [56,57]. This EPE caused mountain flood disasters in the western part of the study area and along the Duku Highway, resulting in damage to transportation, water management systems, tourism infrastructure, and other facilities. It also caused substantial economic losses in downstream agricultural sectors, including facility agriculture, forestry, and the fruit industry.

To accurately depict the diurnal variation characteristics of meteorological elements during this EPE, three stations were selected based on the location of the precipitation center and topographic characteristics. At 12:00 on 20 August, the wind direction at station Y5956 shifted from southeast to northeast. Around 20:00, atmospheric pressure began to rise, and wind speed increased significantly from approximately 3 to 10 m s⁻¹ by 21:00. Concurrently, the temperature dropped sharply and relative humidity rose, indicating the intrusion of cold advection. During this period, the station recorded a maximum precipitation intensity exceeding 36 mm h⁻¹, with relative humidity nearing 100% during precipitation. On 21 August, around 13:00, the wind direction at Y5956 shifted again, from north to south, indicating wind shear. During stage 2, the temperature decreased sharply, atmospheric pressure increased, and wind speed strengthened. By 14:00, the station recorded a maximum precipitation intensity of nearly 30 mm h⁻¹, marking the onset of stage 2. Similarly, in the northwest of the study area, station Y8625 experienced increased pressure, a drop in temperature, a shift in wind direction from north to south, and a rise in wind speed. The highest precipitation intensity, approximately 18 mm h⁻¹, occurred at 17:00. These observations highlight the distinct meteorological conditions driving precipitation formation in the two stages of the EPE. Stage 1 was primarily influenced by a strong northerly cold air mass, which created favorable conditions for EP. In contrast, stage 2 was influenced by a southerly air mass, with smaller changes in temperature and wind speed compared to stage 1. Consequently, the precipitation intensity and cumulative precipitation during stage 2 were also lower than those observed in stage 1.

3.2. Atmospheric Circulation Pattern

A dual-core structure of the South Asian High was observed in the 200 hPa, with two high-pressure centers identified over the Iranian Plateau and the Tibetan Plateau, respectively, during the period from 08:00 on 19 August to 20:00 on 21 August 2019. This configuration is recognized as a characteristic pattern of the South Asian High associated with EPEs in Southern Xinjiang [17,35]. Meanwhile, under the influence of the Balkhash Lake trough, the wind speed in the jet core ahead of the trough reached 62 m s⁻¹. The study area was situated on the left side of the entrance region of the upper-level jet, where the jet exhibited divergence and a suction effect, facilitating the development of vertical upward movement in the mid-to-lower atmosphere [31]. The Tashkent trough moved eastward toward the western part of southern Xinjiang and remained stable at 500 hPa on August 19. The southwest airflow ahead of the trough weakened to 8–10 m s^–1. The distribution of temperature-dew point difference indicated drier conditions in the south and wetter conditions in the north of the Tarim Basin (Figure omitted).

The cold trough in the west of southern Xinjiang transformed into a cold vortex form during stage 1, with the temperature remaining around −12 °C. That was accompanied by significant cyclonic wind field convergence in the western study area. These mesoscale conditions provided the necessary triggering mechanisms for the EP [32]. Combined with favorable moisture conditions and mesoscale cyclones, the low vortex contributed to STHP in the western portion of the study area (Figure 4a). At 850 hPa, a cyclonic wind field was observed over the Hotan region in southern Xinjiang, with northwesterly winds and easterly wind shear between Kashgar and Aksu. A low-level jet, with a speed of 10 m s⁻¹, extended from Qiemo to Korla, coinciding with a steep temperature gradient in the precipitation center during stage 1 (Figure 4b). By 21 August, at 500 hPa, cold shear was evident between Kashgar and Aksu, characterized by northwesterly and southwesterly winds. Warm shear occurred between Aksu and Kuqa, marked by southwesterly winds. At both 700 hPa and 850 hPa on the same day, cyclonic wind fields were present in the Tarim Basin. The low trough west of Aksu interacted with shear and convergence lines at 700 and 850 hPa, triggering a STHP in the southwestern Aksu region between 13:00 and 14:00 (Figure omitted).

With the eastward movement of the 500 hPa trough, the center of EP moved eastward to the east of Aksu, causing a sudden EPE in the Tianshan Grand Canyon from 15:00 to 18:00 on 21 August. The wind speed on the right side of the 200 hPa jet entrance area as well as the wind speed at 500 hPa further increased, and the temperature difference near the precipitation center increased (Figure 4c). In stage 2, a cutting edge between the southwest and west wind directions was observed near the precipitation center. Additionally, there was a clockwise wind shear observed south of the study area, where temperatures were warm in the south and cold in the north, resulting in a temperature difference of nearly 6 °C (Figure 4d).

The center of this EPE was located far from the sounding observation station, and the timing of precipitation was significantly offset from the actual sounding observation time. Previous studies have demonstrated that temperature profiles, humidity profiles, wind profiles, and warm and cold advection derived from ERA5 reanalysis data below 400 hPa closely correspond with the observations recorded at the sounding station [58]. In this study, ERA5 reanalysis data were utilized to generate sounding maps for analyzing the atmospheric junction characteristics associated with this EPE.

Sounding data from the Y5906 station (Figure 5a) indicated that vertical wind shear was observed between 700 and 500 hPa. The convective available potential energy (CAPE) was 314 J kg⁻¹, while the convective inhibition (CIN) was significant, and the Showalter index (Shox) was 1 °C. Combined with the high lifting condensation level (Plcl) at 870 hPa, these conditions suggest limited convective precipitation during periods of weak convection. By 21:00, shortly before and after precipitation (Figure 5b), counterclockwise wind shear was evident near 700 hPa, indicating cold advection at low levels. Below 300 hPa, CAPE increased substantially to 2312 J kg⁻¹, and the precipitable water (Pwat) value rose to 5 cm. These changes indicate an unstable atmospheric stratification conducive to the initiation and development of precipitation during stage 1.

At the Y8625 station (Figure 5c), significant vertical wind shear was observed near 750 hPa prior to stage 2 of precipitation. During this period, the temperature-dew point difference between the upper and lower layers was large. However, the CAPE value at this point exceeded 1060 J kg⁻¹. By 17:00, a “dry above and wet below” stratification appeared near 500 hPa, accompanied by vertical wind shear in the lower levels. This configuration supported the maintenance and enhancement of vertical velocity. During this time, the Plcl decreased, CAPE rose to 2460 J kg⁻¹ (Figure 5d), and the Shox dropped to −2.0, indicating highly unstable atmospheric stratification. In this EPE, there was significant vertical wind shear in the middle and lower layers. CAPE increased and Plcl decreased before and after the precipitation; however, the cold advection was obvious in the lower layers in stage 1. The variations in Plcl and CAPE values in stage 1 were larger than those observed in stage 2, and the corresponding precipitation intensity was also larger than that in stage 2.

3.3. Water Vapor Transport and Budget Analysis

To investigate the impact of long-distance WV transport on this EPE, the HYSPLIT model, developed collaboratively by the NOAA, was utilized for analysis [59,60]. Using cluster analysis, the model can identify the source and propagation path of WV during EPE by analyzing the backward trajectory of each WV particle [54]. For this study, two layers—3000 m and 5000 m above ground level (corresponding to average isobaric surfaces of 700 hPa and 500 hPa, respectively)—were selected as the initial simulation heights for the first and second phases. The time of the strongest precipitation in each phase served used as the initial point, and the backward three-dimensional trajectory of WV particles was tracked for 168 h, with track point positions outputted hourly. To visualize each track path more intuitively, a cluster analysis method was applied to group multiple track paths. By analyzing the spatial variance growth rate (Figure omitted), it was found that at an elevation of 3000 m (5000 m), the variance growth rate of the precipitation tracks in the clustering process increased rapidly when the number of clusters was less than four (six) for stage 1 and four (two) for stage 2. Based on this analysis, the final number of clusters for the simulated WV particle tracks was determined.

During stage 1, the primary WV transport channels above 3000 m from the EPE center originated from the Black Sea (channel 3) and the area west of Balkhash Lake (channel 1), contributing to 77% of the total WV. From 168 to 96 h before precipitation, WV was transported to Central Asia along a westerly flow, which was redirected southward along a northwesterly flow associated with the Central Asian Trough from 96 to 48 h before precipitation. Subsequently, a southwesterly flow carried the WV over the Pamir Plateau and into the area of EP. As shown in Figure 6, channel 1 exhibited the highest specific humidity (SH), followed by channel 3. Twelve hours before precipitation, the SH in all channels rapidly increased to ~4 g kg⁻¹.

The primary WV transport channels at 5000 m are the Aral Sea (channel 6), the southwestern Mediterranean (channel 3), and the Mediterranean (channel 4), which collectively transport WV along westerly pathways, accounting for 68% of the total transport. The WV transport paths and the evolution of weather systems at this altitude are similar to those observed at 3000 m. The height variation analysis shows that the altitude of channel 5 is approximately 8000 m up to 48 h before precipitation, while channel 6 maintains a lower altitude of around 4000 m. This pattern indicates that more distant WV transport channels tend to occur at higher altitudes. The SH analysis reveals that channel 6 consistently maintains values above 2 g kg⁻¹, making it the most significant source of WV among the six channels.

During stage 2, the primary 3000 m WV transport channels for the EP center originated from the Mediterranean Sea (channel 2) and eastern Central Asia (channel 1), contributing 73% of the total WV transport (Figure 7). These channels delivered moisture to the EP area under the influence of the low trough system. The height variation analysis of the WV channels showed that the altitude of channel 3 ranged between 4000 m and 7000 m up to 48 h before precipitation, while the altitude of channel 1 remained below 3000 m. The SH analysis indicated that channel 1 had the highest value at approximately 3.5 g kg⁻¹, followed by channel 2 at around 2 g kg⁻¹. The primary WV transport channel originated at 5000 m from the Aral Sea (channel 1), accounting for 51% of the total WV transport. The transport path at this altitude was similar to that of stage 1. Additionally, channel 2, which originated from the north of the Black Sea, accounted for 49% of the WV transport. The height variation analysis showed that the altitude of channel 2 ranged between 6000 m and 7000 m up to 48 h before precipitation, while channel 1 remained at a lower altitude of 4000–5000 m. The SH analysis revealed that channel 1 and channel 2 exhibited values of approximately 2 g kg⁻¹ and 1 g kg⁻¹, respectively.

The analysis reveals that the mid-level WV during both phases of the EPE primarily originated from the Caspian Sea and the Aral Sea, being transported to the EP area along a westerly path. The weather system over Central Asia contributed more than 30% of the total WV. However, the sources and transport paths at the 5000 m altitude layer differed significantly. The SH of each WV track in stage 1 was greater than in stage 2. Additionally, the SH in stage 1 increased significantly 12 h before precipitation, providing more WV both before and during the EPE. This corresponds to the observation that precipitation in stage 1 was stronger than in stage 2.

The analysis above demonstrates that the WV sources and transport paths in the mid and upper layers of the atmosphere differ between the two stages of precipitation. This raises an important question: what are the characteristics of the WV sources in the lower layers of the study area, and how do they contribute to the development of precipitation?

By analyzing the 700 hPa WV flux field before the onset of the two precipitation stages, a WV convergence center with a magnitude of approximately −16 × 10⁻⁵ g hPa⁻¹ cm⁻² s⁻¹ was identified northeast of the Y5906 station at 20:00 on 20 August (Figure 8). This indicates that WV flux converged from the northeast to the southwest near the station, supplying moisture for the EPE in stage 1. Conversely, at the Y8625 station, the divergence of WV flux was not conducive to precipitation during this period. At 15:00 on 21 August, divergence was observed at lower levels near the Y5906 station, while a southwest wind prevailed near the Y8625 station, accompanied by a WV flux of approximately −6 × 10⁻⁵ g hPa⁻¹ cm⁻² s⁻¹. This formed a foremountain WV convergence zone along the southern slope of the Tianshan Mountains, corresponding to WV accumulation in the northeastern mountainous region of Aksu. This convergence was favorable for the occurrence and intensification of EP near the Tianshan Grand Canyon during this period.

By comparing the relative humidity, wind field, and temperature in the EP centers during the two stages, distinct differences can be observed [34]. At the onset of stage 1 of precipitation at 20:00 on 20 August, as shown in the temperature variations in Figure 3, the near-surface temperature dropped, and the relative humidity increased to over 90% within about one hour. In the evening, radiative cooling along the hillsides caused cold air to flow into the basin, further lowering temperatures. This led to an increase in relative humidity, with near-surface WV reaching saturation. The wind field analysis revealed a sudden change in wind direction between 850 and 650 hPa before and after precipitation, accompanied by an increase in wind speed. The wind direction shifted from easterly to northerly, corresponding to low-altitude mountain inflow in the evening. The atmospheric vertical structure during the precipitation phase exhibited a distinct “dry-moist-relatively dry” stratification, with a dry upper troposphere, a moist mid-level layer (700–500 hPa), and a comparatively drier boundary layer. Following the initial phase, moist air progressively infiltrated the mid-troposphere (~3 h post-precipitation), while the lower atmosphere retained limited humidity due to insufficient moisture replenishment. This pronounced vertical dichotomy—mid-level moistening juxtaposed with persistent low-level dryness—effectively suppressed buoyancy-driven updrafts, thereby inhibiting the sustenance of organized convective systems. During stage 2 of precipitation at 15:00 on 21 August, an eastward-moving cold trough passed over the study area. At this time, the relative humidity gradient increased significantly, with low-level relative humidity rising from 65% to 90%. A distinct wet layer formed between 850 and 550 hPa, creating a “dry at the top and wet at the bottom” distribution. Additionally, from 800 to 750 hPa at 14:00, the wind direction shifted from northwest to south, accompanied by wind shear.

3.4. The Dynamic and Thermodynamic Mechanism of Extreme Precipitation

At 850 hPa at 20:00 on 20 August, a positive vorticity distribution was observed near the Y5906 station, accompanied by dense negative divergence contour lines and negative vertical velocity contour lines. This indicates strong low-level convergence and uplift. As shown in the weather circulation pattern in Figure 4, a “low-altitude convergence—high-altitude divergence” structure was present, which supported convective triggering and precipitation formation [32]. In contrast, at the same time, the vorticity value at the Y8625 station was negative, divergence was positive, and vertical velocity was at a minimum. By 15:00 on 21 August, the area of positive vorticity near the Y8625 station was marked by dense negative divergence isolines and negative vertical velocity isolines. This configuration was favorable for the initiation and development of a convective system.

The study area is situated between a mountain and a basin, where differences in cooling rates between the hillside and the basin create distinct airflows. In the evening, the air along the hillside cools faster than the air in the basin, causing cold air to flow downslope [61]. The relationship between warm and cold advection and precipitation is influenced by several factors, including terrain, wind direction, temperature, and humidity. In mountain and valley environments, warm and cold advection not only shapes the local climate but can also generate specific precipitation patterns [62,63,64]. Around 21:00 on 20 August, cold air moved southward from the higher slope of the terrain, intersecting near the surface of the Y5906 station with warm advection transported by the southeast wind from the western edge of the anticyclone (Figure 9). This interaction between cold and warm air masses likely produced localized convection. When sufficient WV was present, the uplifted air condensed into clouds, leading to a STHP at the Y5906 station, located in the western part of Aksu. During this period, moderate-intensity warm advection (~4 × 10⁻⁵ °C s⁻¹) was observed in the northeastern mountainous region of Aksu. However, the convergence line formed by northerly and easterly winds in the southeastern area remained distant from the Y8625 station at this time, thus lacking localized convergent lifting support and creating unfavorable conditions for precipitation in the eastern regions. By 15:00 on 21 August, the southwest wind (4–6 m s⁻¹) intersected with the east–west oriented Tianshan Mountain Range in eastern Aksu. The warm, moist air carried by the southwest wind was forced to ascend along the mountain slopes. This topographic uplift induced vertical upward movement, cooling and condensing the air, which triggered convection and increased precipitation intensity in the Tianshan Grand Canyon.

Following the divergence and vertical velocity profile at 41°N over the EP center during stage 1, multiple negative areas of divergence were observed below 500 hPa at 20:00 on 20 August (Figure 10a). The maximum negative divergence reached −8 × 10⁻⁴ s⁻¹ at the 850 hPa and 750 hPa layers above the Y5906 station. At 300 hPa, positive divergence values reached 10⁻³ s⁻¹, and densely packed vertical velocity contours were observed, with the center reaching up to −10 Pa s⁻¹. These features indicate strong low-level convergence, high-level divergence, and significant vertical upward motion near the EP area. Similar characteristics are evident in the divergence and vertical velocity profile along 80.5°E (Figure 10b). Cold air flowing into the lower levels of the basin at 850 hPa from the northern mountain area, combined with upper-level flow at 400 hPa, created a small area of divergence near the surface. At the EP center, negative divergence reached −8 × 10⁻⁴ s⁻¹ at 750 hPa, while a larger area of divergence of 8 × 10⁻⁴ s⁻¹ was observed at 300 hPa. This vertical structure of high-altitude divergence and low-altitude convergence, accompanied by intense vertical upward motion, facilitated the formation and development of the convective system during stage 1, ultimately resulting in EP.

An analysis of the profile at 42°N, which shows divergence and vertical velocity during stage 2 of the EPE (Figure 10c), reveals a negative divergence distribution ranging from −2 to −4 × 10⁻⁴ s⁻¹ between 850 and 550 hPa at the Y8625 station, located in the northeastern mountainous region of the study area. In contrast, a positive divergence distribution ranging from 2 to 10 × 10⁻⁴ s⁻¹ is observed at 350–200 hPa. The divergence and vertical velocity profile along 83°E further indicates a central vertical rising velocity value of −10 Pa s⁻¹ in the eastern mountainous region of the study area (Figure 10d). This strong convergence and uplift at lower levels, coupled with divergence at higher altitudes, creates favorable conditions for the development of EP [33].

3.5. Characteristics of Mesoscale Weather Systems

Radar echo and vertical cross-section analysis are crucial for monitoring and predicting STHP events. Particularly in the hour preceding precipitation, detailed observations of radar echo morphology, intensity variations, and vertical structure can effectively identify the developmental stages, thermodynamic characteristics, and moisture transport patterns of convective systems. The Aksu radar observations revealed multiple convective cells exceeding 45 dBZ in the Tianshan mountainous area, northwest of the radar station, as well as to the east and above the station around 20:00 on the 20th (Figure 11a). The vertical cross-section analysis along the AB profile, located 20–60 km northeast of the radar station, showed a meso-γ scale (35 km) [65] convective cluster at an altitude of 2 km (Figure 11c), which persisted until after 00:00 on the 21st before gradually weakening and moving eastward.

From 15:03 to 16:31 on the 21st, a meso-β scale, comma-shaped cloud system was observed above the Shaya radar station, lasting approximately 1.5 h (Figure 11b). At 15:03, convective development began approximately 40 km north–northeast of the radar station. Subsequently, the tail of the cloud system, influenced by southwesterly flow, transported moisture in a vortical path toward the heavy precipitation area, promoting the rapid development of the cloud system. Between 15:30 and 14:01, strong echoes exceeding 40 dBZ appeared in and near the heavy precipitation area, with echo tops reaching 7 km, resulting in intense precipitation in the northern mountainous region (Figure 11d).

By 16:11, the comma-shaped cloud system persisted from the southwest to the northeast of the radar station; however, the echo intensity near the area of heavy precipitation had weakened. At 16:31, strong convective cells of 40–45 dBZ remained in the southern and southwestern portions of the comma-shaped cloud system, moving northeastward with the southwesterly flow, while echoes in the northern part of the cloud system decreased to 30 dBZ. Monitoring data indicated that, although the intensity of precipitation in the heavy precipitation area had weakened, it remained relatively steady. As the echoes continued to weaken, the precipitation eventually ceased.

Due to the limited detection range of the Doppler weather radar in the study area, the two-stage EP centers, particularly those located in mountainous regions, were not adequately captured by the radar observations [43]. In contrast, the hourly black body brightness temperature (TBB) product provided by the FY-4A satellite offers a more direct representation of the development and evolution of mesoscale convective systems [21]. The regions of EP correspond well with areas of low TBB values. A lower TBB value indicates stronger convection and a higher convective cloud top height, making it a valuable tool for analyzing convective processes [17].

Before the onset of stage 1 of the EP at 19:00 on 20 August, precipitation clouds were primarily distributed along the southern slopes of the Tianshan Mountains, with a TBB below −40 °C. At this time, the TBB near the Y5906 station was above −20 °C (Figure 12). As the upper cold trough moved eastward and deepened, the cloud system over the study area rapidly developed. By this stage, the TBB near the Y5906 station, located directly above the center of the EP, dropped below −40 °C. Strong convective clouds were observed extending from the northwest of the study area to the Tianshan Mountain region, characterized by neat and smooth boundaries, indicating intense convection. The central TBB reached below −45 °C, approximately 3 °C higher than the mesoscale convective cloud cluster in the southern part of the study area. This pattern corresponded to an hourly precipitation intensity of 38.1 mm at the Y5906 station, located at the maximum TBB gradient in the southwest portion of the convective cloud cluster, at 21:00 on 20 August. Such conditions, with intense precipitation and strong convection, are rare.

At 17:00 on 21 August, a convective cloud cluster was situated over the Tianshan Mountain area, extending from the southern part of Ili to the northeastern part of Aksu. Additionally, a mosaic-like strong convective cloud cluster was present east of Alar, with a TBB center of −40 °C. At this time, the TBB near the Y8625 station exceeded −20 °C, but the TBB gradient was larger than 15 °C, corresponding to a maximum precipitation intensity of 19.0 mm h⁻¹ Y8625 station. By 19:00, the clouds near Alar had moved eastward and shifted northward toward Shaya, while the convective clouds south of Ili expanded eastward and merged with those in the northern mountains of Kuqa. The area with TBB values of −40 °C, both north and south of the Y8625 station in the Grand Canyon, increased rapidly. After the STHP weakened at the Y8625 station, stratiform clouds dominated the nearby area. The TBB dropped below −20 °C, but the gradient decreased, resulting in significantly reduced precipitation, with hourly precipitation ranging between 0.1 and 2.4 mm from 18:00 to 19:00. Stage 1 of precipitation was characterized by cold cloud processes, where a low TBB value indicated high cloud tops, strong convection, and a strong correlation between EP and low TBB areas. In contrast, stage 2 involved warm cloud precipitation, where relatively higher TBB values indicated lower cloud tops and weaker convection compared to stage 1. These findings indicate that EP can occur not only at the edge of cold cloud regions with strong convective development but also in warm cloud areas with moderately low TBB values. Furthermore, the center of EP does not always coincide with the center of low TBB values but is often located near areas with larger TBB gradients. A rapid decrease in TBB values is often a precursor to the occurrence of EP [66].

An analysis of the changes in precipitation and TBB during each period reveals a strong correlation between decreasing TBB values and increasing rain intensity. At the Y5906 station, the TBB value dropped by nearly 30 °C between 19:00 and 21:00 on 20 August, preceding the EPE in stage 1. The hourly precipitation intensity at the Y5906 station reached a maximum of 38.1 mm when the TBB value reached its lowest point at 21:00. Similarly, prior to stage 2 of precipitation, the TBB value near Y5956 station decreased by approximately 40 °C from 12:00 to 14:00 on 21 August. At 14:00, when the TBB value reached its lowest point, the hourly precipitation intensity at the Y5956 station peaked at 29.6 mm. Both Y5906 and Y5956 stations are located in the western piedmont plain. The consistent pattern of decreasing TBB values and increasing precipitation intensity observed in the plain area aligns with previous studies, although slight differences in TBB values were noted. These discrepancies may be attributed to variations in environmental conditions across different regions.

The TBB value over the Y8625 station fluctuated between 13:00 and 20:00 on 21 August during stage 2. Between 15:00 and 16:00, the TBB value decreased by 15 °C, corresponding to a precipitation rate of 10.8 mm h at 16:00. At 17:00, the TBB value increased by 13 °C, coinciding with an hourly precipitation rate of 19.0 mm. From 18:00 to 21:00, the TBB value dropped from −15 °C to its lowest point of −42 °C, representing a decrease of 17 °C. This decline may indicate that after the occurrence of EP in the mountainous areas, the convective cloud transitioned to an altostratus cloud, leading to a reduction in WV. While the TBB reached its lowest value, the exhaustion of WV likely resulted in a cessation of significant precipitation.

4. Discussions

The rapid formation and development of convective clouds were observed during both stages of EPE [45]; the observational analysis revealed distinct convective–precipitation coupling mechanisms between these stages (

Table 2. Comparison of key metrics between stage 1 and stage 2.

Metric	Stage 1 (20:00 on 20 August–08:00 on 21 August 2019)	Stage 2 (08:00–20:00, 21 August 2019)	Unit
Precipitation intensity	38.1	20.6	mm h−1
Shox	−1	−2	-
Pwat	5	5	cm
CAPE	2312	2460	J kg−1
Main water vapor sources	channel 1 and 3 channel 3, 4, and 6	channel 1 and 2 channel 1	3000 m AGL 5000 m AGL
WV flux divergence	−3 (850 hPa) −1 (700 hPa)	0 (850 hPa) −7 (700 hPa)	10−5 g hPa−1 cm−2 s−1
Horizontal wind speed	8 (Y5906)	5 (Y8625)	m s−1
Radar reflectivity	30 (0.2−3 km)	40 (1−6 km)	dBZ (Height)
TBB	−39.8 (21:00)	−38.9 (16:00)	°C (Time)

). In stage 1, abrupt TBB decreases in cold cloud regions preceded EP initiation in the western Aksu Plain [67], with EP zones located southwest of convective cores exhibiting steep TBB gradients [17,44]. Contrastingly, stage 2 demonstrated spatiotemporal decoupling between mountain precipitation maxima and TBB minima. Peak mountain rainfall occurred during the convective energy release phase (16:00–17:00), coinciding with slight TBB increases, with EPs localized in warm, low-cloud regions featuring moderate negative TBB values [68]. This warm-cloud precipitation mechanism, characterized by efficient moisture conversion via southwest flow lifting without deep convective organization, diverges from conventional EP paradigms in northern China that associate extremes with cold cloud convection [21]. While challenging the assumed TBB–precipitation intensity correlation [45], these findings necessitate further investigation into (1) terrain-mediated warm rain processes in Tianshan’s northeastern ranges [31,32], and (2) generalizability across synoptic regimes through expanded case analyses [69].

While HYSPLIT modeling elucidated synoptic-scale moisture transport pathways, three fundamental limitations require explicit acknowledgment. First, the 1° × 1° input data resolution inadequately resolved mountain–valley circulations, potentially underestimating local moisture recycling [70]. Second, the Lagrangian assumptions neglect precipitation-driven moisture depletion critical to warm-cloud processes [71]. Third, the model’s inability to parameterize sub-grid topographic effects obscures key cloud formation mechanisms, including slope-induced updrafts and lee-wave cloud development. Sensitivity analyses demonstrate that these combined constraints introduce non-negligible uncertainties in short-range moisture source attribution, necessitating future studies employing convection-permitting mesoscale simulations (horizontal resolution ≤ 3 km) to better resolve terrain–atmosphere interactions [72,73].

Building on these findings, three critical research priorities emerge to advance the understanding of arid-region precipitation extremes. First, implementing WRF simulations with advanced microphysical parameterizations could elucidate ice-phase processes in northern Tarim Basin EPEs [74], where preliminary evidence suggests riming mechanisms may substantially enhance precipitation yields [75]. Second, developing dynamically downscaled climate projections is essential to evaluate how synoptic-scale moisture increases—consistent with thermodynamic theory under warming scenarios—interact with Tianshan’s orography to modulate extreme event recurrence [76,77]. Third, expanding the temporal scope to include winter precipitation mechanisms remains imperative [38], particularly given reanalysis indications of enhanced moisture flux variability during specific large-scale teleconnection patterns like ENSO Modoki events [14]. While the single-case study design limits a comprehensive assessment of interannual variability—such as ENSO phase influences on transregional moisture transport—our ongoing work addresses this through multi-decadal CMIP6 ensemble analysis [78]. Complementary field campaigns employing mobile radar networks and airborne sensors will establish an integrated framework combining high-resolution modeling, process diagnostics, and observational validation.

5. Conclusions

ERA5 reanalysis data, along with observational data from the GPM satellite, FY-4A satellite, and ground meteorological stations, were used to analyze an EPE in the northern Tarim Basin from 20 to 21 August 2019. The diagnostic methods of weather dynamics and the HYSPLIT model were applied to investigate the circulation background, high- and low-altitude configurations, water vapor (WV) and dynamic thermal conditions, instability conditions, and the triggering mechanisms of the STHP induced by the low vortex of Balkhash Lake in the central northwest and northeast regions of the study area. The main findings are as follows:

The high-pressure system at 200 hPa over Iran and the Qinghai–Tibet Plateau exhibited a “dual-center” distribution before the EPE. A strong west–southwest jet stream, with a velocity of 60 m s⁻¹, diverged over the study area, creating a “suction” effect that enhanced upper-level divergence. At 500 hPa, the Balkhash Lake in eastern Central Asia, Tashkent, has a low trough and a cold center with temperatures ranging from −12 to −16 °C. Ahead of the trough, warm and moist air from the southwest was transported northeastward, converging with the cold center and resulting in increased atmospheric instability, which favored the development of EP. Additionally, the southwest wind interacted with the east–west-oriented Tianshan Mountains at an altitude of approximately 5 km, leading to topographic uplift and wind convergence, further supporting vertical air movement. The combined influence of the 200 hPa and 500 hPa circulations, along with the configuration of the wind field, played a critical role in the initiation and intensification of the EP during both stages. Significant vertical wind shear and stratification instability are present in the middle and lower atmospheric levels. Additionally, a low-level convergence line enhances dynamic and thermal instability, providing favorable triggering conditions for the development of the EPE. Once precipitation begins, the lifting condensation level decreases. However, during stage 1, pronounced low-level cold advection is evident. Changes in the lifting condensation height and CAPE values are more pronounced one hour before and during precipitation in stage 1 compared to stage 2, resulting in heavier precipitation during stage 1.

The WV tracking analysis using the HYSPLIT model reveals that during both stages of the EPE, the majority of WV in the 3000 m layer originates from the Mediterranean Sea and the Aral Sea, accounting for more than 70%. This moisture is transported to the EP area via a westerly path, with the weather system in eastern Central Asia contributing over 30% of the WV. However, significant differences are observed in the WV sources and transport paths at the 5000 m layer between the two stages. During stage 1, WV is transported through four channels, with the southeast Iranian Plateau and the southern Aral Sea contributing the most significant amounts (64%). In contrast, during stage 2, only two transport channels are active, and WV primarily originates from the Aral Sea and the Mediterranean Sea, contributing 51% and 49%, respectively. SH along each WV transport path is higher in stage 1 than in stage 2. Additionally, SH increases 12 h prior to precipitation in stage 1, providing a greater supply of WV before and during the EPE. This corresponds to the heavier precipitation observed in stage 1 compared to the second stage 2.

The conceptual model diagram indicates the presence of a cyclone field in the western part of the study area during stage 1 of the EPE. An easterly wind of approximately 8 m/s interacts with the distinctive “horseshoe” topography in southern Xinjiang, creating a cold cushion effect (Figure 13). This interaction, combined with westerly airflow over the mountains, generates a mesoscale convergence line. The easterly jet and the Pamir Plateau further contribute to topographic uplift, enhancing vertical upward motion. Alongside the northern mountain wind observed on the evening of 20 August, these factors triggered a short-duration rainstorm. At various atmospheric levels, the interplay of three air streams—easterly, westerly, and mountain winds—supports lower-level convergence and upper-level divergence, sustaining vertical upward motion, facilitating WV transport, and intensifying EP. The mesoscale convergence line formed between the low-level easterly wind and the westerly flow over the mountains was crucial in initiating the short-duration rainstorm in the western plains during stage 1. In stage 2, during the afternoon of 21 August, the overturning wind in the northwest of the study area weakened. Simultaneously, cold air from the eastern part of the study area moved southward along the north–south-oriented Tianshan Grand Canyon and converged with weak easterly and southwest winds near the EP area, creating favorable triggering conditions. Additionally, uneven cloud cover and solar heating caused variations in surface radiation, generating thermal contrasts that formed a vertical circulation cell. This circulation influenced convective cloud clusters in the northeastern mountainous area of Aksu during the afternoon, ultimately leading to stage 2 of EP.