Characteristics of Atmospheric Rivers and the Impact of Urban Roof Roughness on Precipitation during the “23.7” Extreme Rainstorm against the Background of Climate Warming

Abstract

In July 2023, Baoding in Hebei Province experienced unprecedented torrential rainfall, breaking historical records and causing severe flooding. However, our understanding of the multi-scale circulation systems and physical mechanisms driving this extreme precipitation event remains incomplete. This study utilizes multi-source observational data and the Weather Research and Forecasting (WRF) numerical model to conduct a weather diagnosis and numerical simulation of this extreme rainfall event, focusing on the impact of atmospheric rivers (ARS) and urban rooftop roughness on the precipitation process against the background of climate warming. The study found that this extremely heavy rainstorm occurred in the circulation background formed by the factors of subtropical high ectopics, typhoon residual vortex retention, double typhoon water-vapor transmission, and stable high-level divergence. The ARS provided abundant moisture, with its vapor pathway significantly altered following the landfall of Typhoon Doksuri. The interaction between the ARS and the Taihang Mountains was crucial in triggering and intensifying the rainstorm in the foothills. Urbanization significantly affected the distribution of precipitation, with moderate urban roughness enhancing rainfall in and around the city, whereas excessive roughness suppressed it. These results contribute to a deeper understanding of the mechanisms behind extreme precipitation under climate change and provide a scientific basis for improving the forecasting and mitigation of such events.

1. Introduction

Extreme precipitation events have profound impacts on human society and the ecological environment. They can cause severe weather-related disasters and are closely linked to the distribution and utilization of regional water resources. With the ongoing intensification of global warming and rapid urbanization, the frequency and severity of extreme precipitation events are projected to escalate significantly [1,2]. This trend is primarily attributed to the thermodynamic Clausi-us-Clapeyron relationship, which dictates that atmospheric water-vapor content—the primary source of precipitation—increases by approximately 6–7% for each degree Celsius rise in temperature [3,4]. This relationship underscores the intricate connection between rising global temperatures and the heightened potential for more intense precipitation events. According to the “2023 Global Climate Status” report, multiple climate indicators reached new records in 2023, with the near-surface average temperature being 1.45 ± 0.12 °C higher than the average level from 1850 to 1900 [5]. Against this backdrop of climate warming, the densely populated and topographically complex region of Baoding experienced an unprecedented, sustained heavy rainfall event from 29 July to 2 August 2023, influenced by the low-pressure system left in the wake of Typhoon Doksuri. This event, referred to as the “23.7 event”, led to severe secondary disasters such as flash floods, mudslides, landslides, and urban flooding.

During this extreme precipitation event, moisture conditions played a crucial role. ARs, as key pathways for water-vapor transport in atmospheric circulation, are of significant importance to the global water-vapor cycle. The standardized definition of an atmospheric river in the Glossary of Meteorology (GoM) describes it as a long, narrow corridor of intense water-vapor transport in the lower atmosphere. These atmospheric rivers typically originate from tropical oceans and are embedded in the warm conveyor belt (WCB) ahead of the cold front in mid-latitude cyclones. They are responsible for transporting over 90% of the water vapor in the mid-latitudes [6,7,8]. When ARs encounter mountainous terrain, the water vapor is forced to ascend, leading to substantial orographic precipitation at specific altitudes, significantly contributing to regional precipitation [9,10,11]. ARs are closely associated with extreme precipitation events [12,13]. They directly influence the transport of water vapor and the dynamic uplift during rainstorms, providing important indications for forecasting extreme precipitation and even flood events [14,15]. However, existing studies exploring the impacts of large-scale circulation systems on atmospheric rivers often focus on relatively singular factors such as blocking highs, subtropical highs, low-pressure troughs, cyclonic activities, and El Ni-ño. This approach fails to fully capture the comprehensive effects of complex weather systems on atmospheric rivers. Therefore, this study examines a more intricate circulation background, investigating the combined effects of subtropical high displacement, the persistence of typhoon remnants, dual typhoon water-vapor transport, and upper-level divergence stability on atmospheric rivers [16,17,18,19]. This research aims to provide a more thorough and in-depth understanding of the behavior and role of atmospheric rivers within complex weather systems, thereby enhancing the forecasting capabilities for extreme precipitation events.

During this rainstorm, Baoding’s urban area was also at the center of the precipitation. The impact of urbanization on precipitation has long been a focus of atmospheric science research. Existing studies have shown that urbanization can alter local thermodynamic and dynamic conditions, thereby affecting the distribution and intensity of precipitation [20]. Urban dynamics can lead to increased rainfall downwind of cities [21,22] or increased rainfall upwind [23], or even cause the “splitting” of rainstorm cloud movement paths due to building obstructions [24]. Urban canopy parameterization is crucial for simulating the urban boundary layer, and the wind field near urban areas is highly sensitive to canopy roughness [25]. Larger surface roughness can influence boundary layer secondary circulation by enhancing local convergence, thereby affecting storm intensity [26]. However, while urban canopy roughness can characterize different stages of urbanization, existing studies often focus on the overall urban roughness’s impact on wind fields [27,28,29], with less attention paid to the relationship between roof roughness length (Z0R) and precipitation. This limits our understanding of the microscopic mechanism of urbanization affecting extreme precipitation. Therefore, this study uses the mesoscale numerical weather prediction model WRF (Weather Research and Forecasting Model) to investigate the impact of different Z0R values on boundary layer dynamic conditions and rainstorm cloud movement during this precipitation event. This research will contribute to a comprehensive understanding of the mechanisms behind the “23.7 extreme rainstorm”, offering significant theoretical and practical value.

2. Model Description and Experimental Design

2.1. Description of Study Data

This study utilizes hourly precipitation data from surface and automatic weather stations provided by the China National Meteorological Information Center, as well as multi-source merged precipitation products. Using precipitation benchmark data observed from automatic stations, we precisely corrected biases in radar and satellite precipitation estimates through the probability density function matching method. Subsequently, the Bayesian Model Averaging (BMA) fusion method was employed to successfully integrate radar and satellite precipitation information. This method involves collecting data from both radar and satellite precipitation sources within a specific time and spatial range that matches ground observations. Through training, the weights and errors for each type of data are determined, allowing for the calculation of analysis values and variances at each grid point. This process ultimately generates a satellite-radar dual-source data fusion background field that closely approximates ground observations. The result is a nationally unified precipitation background field, producing an hourly updated multi-source precipitation fusion product [30,31]. Due to the impact of heavy rainfall, several stations in Baoding were damaged. We evaluated the multi-source merged precipitation data for 2021–2022, finding that the root-mean-square error (RMSE) for each month was below 3 mm (Figure 1), indicating high accuracy. Therefore, we used the merged precipitation data to analyze precipitation conditions. Additionally, we incorporated ERA5 reanalysis data published by the European Centre for Medium-Range Weather Forecasts (ECMWF), which has a temporal resolution of one hour and a spatial resolution of 0.25° × 0.25°. All times mentioned in this article are in Beijing time.

The identification and definition of ARs are crucial for understanding their role in global hydrometeorological processes. Zhu and Newell [4] proposed a method for calculating integrated horizontal water-vapor transport (IVT), finding that regions with IVT greater than 250 kg/(m·s) exhibited distinct elongated structures globally, with a uniform distribution in both hemispheres. Consequently, IVT ≥ 250 kg/(m·s) has been widely adopted as the standard for identifying ARs [32,33]. However, this standard is not universally applicable. Ralph et al. [34] found a close relationship between IVT and integrated water vapor (IWV), suggesting that IWV ≥ 2 cm could also be used to identify ARs. Mahoney et al. [35] directly used IVT = 500 kg/(m·s) as the threshold for AR boundaries in the southeastern United States.

Considering the unique characteristics of East Asian summer ARs, Pan and Lu [36,37] developed an AR identification and tracking algorithm suitable for this region and established an East Asian AR database based on ERA5 reanalysis data. The main objective of their AR pathway detection is to find contiguous grids exceeding dual IVT thresholds: a local threshold and a regional threshold. For each grid (0.75° × 0.75°) inside the AR detection region (40° E to 120° W, 20° S to 60° N), they calculated the 85th percentile of IVT intensity over all time steps during the boreal summer (JJA) of 1985–2016 as the local threshold. They innovatively set the 80th percentile of IVT intensity (334 kg/(m·s)) within the detection region during the Northern Hemisphere summer from 1985 to 2016 as the regional intensity threshold. This work provides a solid foundation for understanding the physical processes of East Asian summer ARs. In this study, we also use 334 kg/(m·s) as the standard for ARs.

The calculation formulas are shown in Equations (1) and (2), respectively. q, g, p₀, u, v, $\mathsf{\alpha }$, $\overline{{\mathrm{I}\mathrm{V}\mathrm{T}}_{\mathrm{u}}}$ and $\overline{{\mathrm{I}\mathrm{V}\mathrm{T}}_{\mathrm{v}}}$ are specific humidity, gravitational acceleration, surface pressure, zonal wind, meridional wind, average direction angle of ARs, regional average zonal and meridional components of IVT, respectively. The flow direction is 90° towards due north and 0° towards due east.

$$\mathrm{I}\mathrm{V}\mathrm{T}=\sqrt{{\left(\frac{1}{g}{\int }_{p_0}^0 qu\mathrm{d}p\right)}^2+{\left(\frac{1}{g}{\int }_{p_0}^0 qv\mathrm{d}p\right)}^2}$$

(1)

$$\alpha ={\mathrm{t}\mathrm{a}\mathrm{n}}^{-1}\frac{\overline{{\mathrm{I}\mathrm{V}\mathrm{T}}_{\mathrm{v}}}}{\overline{{\mathrm{I}\mathrm{V}\mathrm{T}}_{\mathrm{u}}}}$$

(2)

2.2. Experimental Design

The Weather Research and Forecasting (WRF) model is a state-of-the-art atmospheric modeling system designed for both meteorological research and operational forecasting. It features high-resolution capabilities, a wide range of physical parameterization options, and the ability to assimilate observational data to improve initial conditions. WRF supports nested grids for detailed local simulations within larger-scale contexts and is utilized across various applications, including weather prediction, climate studies, air quality assessments, and hydrological forecasting. This study employs the WRF 4.3.3 [38] model for simulations, covering the period from 20:00 on 28 July to 08:00 on 1 August 2023, with the simulation center located at 38.85° N, 115.5° E. The mode is set to a 12 h start time. The simulations use a three-level nested grid configuration with grid sizes of 81 × 81, 121 × 121, and 271 × 271, corresponding to grid spacings of 9 km, 3 km, and 1 km, respectively. The model utilizes the WSM-3 cloud microphysics scheme [39], the RRTM longwave radiation scheme [40], the Dudhia shortwave radiation scheme [41], the Noah LSM land surface scheme [42], the YSU boundary layer scheme [43], the MM5 surface layer scheme [44], and the BEP urban canopy scheme [45]. The above configuration can be referred to in Table 1. Due to the outdated urban data in WRF, we incorporated 1 km resolution urban canopy parameter (UCP) data for 60 Chinese cities around 2018 [46]. The numerical experiments use ERA5 as the driving field. To enhance the accuracy of the simulation results, STRM1 30 m terrain data, 27 m land use data based on secondary classification, and CLDAS (0.0625° × 0.0625°) soil moisture data provided by the China Meteorological Administration were also introduced. Four sensitivity experiments were designed: a control scenario without urban canopy scheme (CTR), the default urban roof roughness scheme (URBAN), a 30% increase in urban roof roughness (URBAN + 30%), and a 50% increase in urban roof roughness (URBAN + 50%).

3. Case Overview

3.1. Observed Precipitation

Analysis of Figure 2 and Table 2 indicates that from 29 July to 2 August 2023, Baoding experienced an unprecedented rainstorm. The rainfall was concentrated between 31 July and 1 August, with prolonged duration, high total precipitation, uneven distribution, and strong convection, leading to severe flooding. Statistics show that the average rainfall across Baoding was 357.2 mm, with the highest precipitation recorded at Zijingguan in Yixian, reaching 752.7 mm. A total of 36 stations (10.3%) recorded process rainfall exceeding 500 mm, with eight counties (cities/districts), including Shunping, exceeding historical maximums. Fu Ping Liao Dao Bei recorded a single-day rainfall of 491.7 mm, also breaking historical records, highlighting the unprecedented intensity of this rainstorm. Additionally, the rainfall persisted for a long duration and was unevenly distributed. Central areas experienced rainfall for over 60 h, with Zijingguan in Yixian receiving rain for up to 70 h, and Fugang station for 84 h. The centers of heavy rainfall were scattered, with ten stations recording over 600 mm, primarily located in Yixian (six stations), Fuping (one station), Laiyuan (one station), the main urban area (one station), and Xushui district (one station). The prolonged and uneven distribution of rainfall increased the difficulty of flood control and drainage.

3.2. Circulation Background

Typhoon Doksuri made landfall on the southern coast of Fujian province on the morning of 28 July. By 20:00 on the same day, the typhoon had moved to northwest Fujian, while the western Pacific subtropical high (WPSH) showed a block-like distribution with its core region located over the Sea of Japan (Figure 3a). Doksuri continued to move northwestward along the southeastern flow on the western side of the WPSH, gradually weakening in intensity. By 08:00 on 29 July (Figure 3b), the WPSH still exhibited a block-like distribution, extending significantly northward and westward. As Doksuri moved northward along the southwestern edge of the WPSH into Anhui province, it began to weaken, while a southeast jet stream with wind speeds exceeding 20 m/s on its eastern side transported warm, moist air from the coast to Hebei province. A shear line appeared at 850 hPa at the border between Hebei and Shandong. As the residual circulation of the typhoon moved northwestward, this spiral rainband advanced towards the mid-section of the Taihang Mountains, intensifying on the windward slopes. By 08:00 on the 30th (Figure 3c), the subtropical high had connected with the Mongolian high ridge, with the subtropical high positioned anomalously west and north. The 500 hPa geopotential height deviated from the climatological state by more than 2–3 standard deviations, blocking the further northward progression of the typhoon’s low-pressure system. The southeast low-level jet on the eastern side of the typhoon continued to supply warm, moist air to Baoding, with specific humidity in the area exceeding 15 g/kg. By 08:00 on 31 July (Figure 3d), the WPSH continued to move westward, the typhoon’s low-pressure system weakened and dissipated, and the strength and extent of the southeast jet stream diminished. However, Typhoon Khanun near East China continued to transport moisture to Baoding. Between 29 and 31 July, Baoding was situated in the strong divergence zone on the right side of the entrance region of the 200 hPa upper-level jet stream, providing stable and robust upper-level divergence conducive to extreme precipitation in the area (figure omitted). By 08:00 on 1 August, the subtropical high had taken control over the Hebei region, leading to a rapid decrease in precipitation in Baoding. Thus, the interaction of the typhoon’s residual circulation, the anomalous subtropical high, the dual typhoon moisture transport, and the strong upper-level divergence collectively established the atmospheric environment for the extreme precipitation in Baoding.

4. The Relationship between ARs and Rainstorms

Figure 4 demonstrates that the hourly regional average integrated water-vapor transport (IVT) and the moisture flux divergence (VIMD) in the Baoding area both exhibit strong correlations with the hourly regional average precipitation. This relationship is particularly evident in the intensity of water-vapor flux and its convergence, which corresponds to the precipitation intensity, especially during the peak and weakening periods. The development of atmospheric rivers (ARs) is directly related to the location and intensity of heavy precipitation. During the landfall of the typhoon, a significant amount of water vapor absorbed by the tropical ocean is transported to higher latitudes, providing ample moisture for the ARs. The South China Sea, East China Sea, and Yellow Sea are the primary sources of water vapor in this region (Figure 5a). This moisture is conveyed by the atmospheric river to southeastern China and further advances into northern China. As the low-pressure system of Typhoon Doksuri moves northward, the water-vapor transport path of the ARs shifts, with the sources of moisture gradually transitioning to the East China Sea, Yellow Sea, and Bohai Sea (Figure 5b). By 31 July (Figure 5c,d), after ARs crossed the mountains, the main moisture sources for the vapor center were local moisture near Hebei and the Bohai Sea, resulting in a significant reduction in overall water-vapor flux. Additionally, although Typhoon Khanun was somewhat distant from the water-vapor channel of ARs, its low-pressure system and rotational wind field facilitated the convergence of water vapor into ARs, enhancing the intensity and extent of water-vapor transport.

ARs extended from the southeast to the northwest, reaching the Taihang and Yanshan mountainous and hilly regions in Hebei Province. During this process, the terrain’s blocking and orographic lifting effects significantly influenced the movement and precipitation of ARs. By 08:00 on the 29th (Figure 6a), ARs reached the southern mountains of central Hebei, where it was forced to ascend due to the mountain barrier, triggering intense precipitation at a certain altitude. In addition to orographic lift, radiative heating also played a role. By 20:00 on the 29th (Figure 6b), ARs further intensified, encountering obstruction on the western side of the Taihang Mountains, enhancing vertical motion on the windward slopes. This obstruction facilitated the continued development of ARs in the Baoding region, corresponding to the first phase of heavy rainfall there. By 13:00 on the 30th (Figure 6c), the core area of inland water-vapor flux had expanded to cover the entire Taihang and Yanshan mountainous regions. On the windward slopes, the dense isopleths of integrated water-vapor flux indicated strengthened horizontal convergence and enhanced vertical motion. Under the combined influence of ARs and the terrain, heavy rainfall reached its second peak. By 20:00 on the 31st (Figure 6d), ARs gradually weakened, with its main body moving northeast around the mountains. At this point, the orographic lifting effect diminished, leading to a decrease in precipitation intensity.

In the initial phase of the rainfall event (Figure 7a), ARs developed vertically from the lower to the upper layers. The maximum pseudo-equivalent potential temperature (θse) reached 350 K, and the maximum water-vapor flux exceeded 16 kg/(m·s), with values over 6 kg/(m·s) extending up to the 700 hPa level. The core area was located at a relatively low altitude, indicating the development phase of ARs. By 02:00 on the 30th (Figure 7b), both horizontal and vertical wind speeds significantly increased, and the extent of the horizontal and vertical ranges expanded. At this time, the maximum water-vapor flux exceeded 28 kg/(m·s). By 13:00 on the 30th (Figure 7c), a strong and stable low-level jet had formed, with wind speeds over 12 m/s extending from the 500 hPa level to the surface above the heavy rainfall region. The core area of ARs was rich in humidity, with the maximum center water-vapor flux reaching 32 kg/(m·s), and regions with values over 6 kg/(m·s) extended upward to the 500 hPa level. The dense distribution of θse isopleths at lower levels, with a core maximum of 350 K, indicated the presence of unstable energy and energy fronts, fostering strong convective development. At this point, ARs were in a vigorous development phase. However, by 20:00 on the 31st (Figure 7d), the low-level jet weakened, the humidity in the core area of ARs decreased, and the maximum water-vapor flux diminished. The intensity of the θse isopleths at lower levels decreased, with the maximum value dropping to 345 K. The weakening of the water-vapor field, low-level jet, and energy field were the main factors contributing to the reduction in precipitation intensity.

The analysis of divergence and vertical velocity vertical cross-sections along the transect reveals the development and dynamical characteristics of the rainfall event. From 08:00 on the 29th to 02:00 on the 30th (Figure 7a,b), the low-level divergence field transitioned from weak convergence to strong convergence, with the height of the convergence center gradually increasing. In the region around 115.5° E, the 900–750 hPa levels showed negative divergence values, while the 700 hPa level exhibited weak positive divergence. This combination of low-level convergence and mid-level divergence collectively enhanced vertical motion, with the vertical velocity center reaching 1.3 Pa/s and its maximum located around 700 hPa. The strengthening of vertical motion contributed to an increase in precipitation intensity. The combination of low-level convergence and mid-level divergence enhanced vertical motion. By 13:00 on the 30th (Figure 8c), the divergence values ranged up to −22 × 10⁻⁵/s from the eastern plains of Baoding to the windward mountains, with the thickness of the convergence layer further increasing. Meanwhile, a divergence center with a value of 18 × 10⁻⁵/s was present at the 600 hPa level, and the corresponding vertical velocity peaked at 2.8 Pa/s. The strong low-level convergence and mid-level divergence further intensified the low-level upward motion, leading to a significant increase in precipitation intensity. The strong development of vertical motion and divergence field near ARs promoted the occurrence and development of heavy rainfall. By 20:00 on the 31st (Figure 8d), the divergence and vertical velocity in the precipitation area had significantly weakened, and the previously well-organized vertical structure from the lower to the upper layers had dissipated. The reduction in low-level vertical velocity corresponded with the main body of ARs moving northeastward around the mountains, resulting in a marked decrease in precipitation.

5. Relationship between Urban Roof Roughness and Rainstorms

Figure 9 shows the impact of different levels of urban roof roughness on precipitation distribution in two different stages. From 08:00 on the 29th to 08:00 on the 30th (Figure 9a–c), considering urban effects, the rainfall amount in the southern urban area increased compared to the no-urban scenario, while rainfall within the urban center decreased. When urban roof roughness increased by 30%, rainfall in the southern, western, and northern areas of the city increased, but the internal urban rainfall continued to decrease. Further increasing urban roof roughness to 50% resulted in an overall decrease in rainfall compared to the no-urban scenario. From 08:00 on the 30th to 08:00 on 1 August (Figure 9d–f), considering urban effects, rainfall within the urban area increased compared to the no-urban scenario. With a 30% increase in urban roof roughness, there was a significant increase in urban rainfall. However, when urban roof roughness increased to 50%, urban rainfall decreased while rainfall in the eastern urban area increased. Comprehensive analysis of the two periods indicates that moderate urbanization enhances rainfall in urban and surrounding areas, but excessive urban roof roughness inhibits urban rainfall development and may even shift precipitation to peripheral areas. This influence demonstrates consistent patterns across different periods.

To investigate the impact of urban roof roughness on vertical wind speed and specific humidity profiles, numerical simulation results for the vertical profiles of wind speed and specific humidity along 115.4° E from 02:00 to 03:00 on 30 July are presented (Figure 10). These results exhibit the following characteristics:

During the first phase of precipitation, under the no-urban scenario (CTR), the convective system translated with the wind direction with little change in intensity, but the height of the upward motion was relatively low, and a secondary circulation center near the surface up to 700 hPa appeared at 38.9° N. When urban influences were considered (URBAN scenario), the 18 g/kg isohume height in urban areas was generally higher than in the no-urban scenario, indicating more favorable humidity conditions. Although convection intensity increased in urban areas, the main upward motion was concentrated on the southern boundary of the city, resulting in significant rainfall in that region, while rainfall within the urban area itself was weaker. With a 30% increase in urban roughness (URBAN + 30% scenario), the northern side of the city maintained a larger upward motion area, the southern secondary circulation structure was more complete, and the humidity field changed little, leading to increased rainfall on both the northern and southern sides of the city. Further increasing urban roughness by 50% (URBAN + 50% scenario) reduced the specific humidity field and diminished the urban area’s influence on vertical wind speed. At this point, convection on both the northern and southern sides of the city significantly weakened, and convective activity did not extend into the urban area or downwind, potentially altering the movement direction and resulting in overall reduced rainfall in the city.

During the second phase of precipitation (vertical profiles of wind speed and specific humidity along 38.9° N from 12:00 to 13:00 on 31 July), all urban-influenced scenarios showed higher specific humidity conditions compared to the no-urban scenario (CTR). In the URBAN scenario, with easterly winds, convection near the city edge began to intensify. Compared to the CTR scenario, urban convection strength increased and persisted longer, with the impact extending downwind with the prevailing wind. The URBAN + 30% scenario exhibited similar patterns of enhanced convection. When urban roughness increased by 50% (URBAN + 50% scenario), wind speeds decreased, the movement of convection centers slowed, and convection intensity was relatively weaker. Thus, with further increases in urban roughness, the urban area’s influence on the vertical wind speed field gradually diminished, exhibiting the same pattern as during the first phase of precipitation.

In both precipitation processes, the vertical motion exhibited a pattern of initially increasing and then decreasing as the urban roof roughness increased. However, the impact on specific humidity differed between the two processes. During the first phase, specific humidity also showed a pattern of initially increasing and then decreasing with increasing urban roof roughness. In the second phase, due to the high intensity of moisture transport and overall stronger upward motion, the changes in specific humidity caused by urban roof roughness were less noticeable.

By analyzing the impact of varying levels of urban roof roughness on the surface wind field during two different stages, we observed that during the first phase of precipitation, the CTR scenario exhibited an enhanced northeast wind effect on the ground, significantly intensifying wind field convergence in the eastern urban area and influencing precipitation distribution. In the URBAN scenario (Figure 11e,f), although wind field convergence was more pronounced in the southwestern part of the city, the overall decrease in wind speed resulted in reduced convergence in the central and eastern areas, leading to increased precipitation in the southwestern part of the city. In the URBAN + 30% scenario (Figure 11i,j), wind field convergence persisted in the northern and southwestern parts of the city, with an increased westerly component in the southwestern wind direction, further enhancing wind field convergence and increasing precipitation in these regions. Conversely, the URBAN + 50% scenario revealed that significant increases in urban surface roughness led to notable flow-around effects, creating alternating wind speed convergence and divergence zones within the urban area, thus reducing overall precipitation (Figure 11m,n).

During the second phase of precipitation, the CTR scenario mainly observed wind field convergence within the city at 13:00 on the 31st. In the URBAN scenario, urban wind field convergence persisted and intensified between 12:00 and 13:00 (Figure 11g,h). In the URBAN + 30% scenario, increased urban roughness resulted in more pronounced wind speed and direction changes, further enlarging the intensity and extent of wind field convergence (Figure 11k,l). Although the URBAN + 50% scenario saw a general decrease in wind speed to the east of the city, the intensity of wind field convergence within the city diminished (Figure 11o,p). In this phase of precipitation, all urban-influenced scenarios showed that urban wind field convergence led to increased precipitation within the city. However, with further increases in urban roughness, the trend of increasing precipitation did not continue, and precipitation began to decrease once a certain roughness threshold was reached.

The analysis shows that with increasing urban roof roughness, wind speed decreases and wind direction changes, affecting convergence and divergence patterns. During the first phase of precipitation, characterized by predominant northeasterly winds, moderate increases in urban roof roughness, combined with the north-south orientation of the urban layout in Baoding, enhance convergence on the northern and south-western sides of the city. This strengthens upward motion, leading to increased precipitation in these areas. In the second phase, with primarily easterly surface winds, moderate increases in urban roof roughness and the east-west urban layout further enhance internal wind field convergence, resulting in more precipitation within the city. In both phases, excessive roughness inhibits the development of urban precipitation.

6. Conclusions and Discussion

Through meteorological diagnostic methods and sensitivity experiments using the Weather Research and Forecasting (WRF) model, this study reveals the principles underlying the influence of atmospheric rivers and urban roof roughness on the “23.7” extreme precipitation event in Baoding under the context of climate warming for precision and clarity. The analysis indicates that this extraordinary rainfall event occurred within a complex circulation context characterized by the factors of subtropical high ectopics, typhoon residual vortex retention, double typhoon water-vapor transmission and stable high-level divergence. The moisture source of ARs exhibited dynamic shifts, initially originating from the South China Sea and the Yellow Sea coast, then transitioning along the northward path of the typhoon, and finally localizing around Hebei and the Bohai Sea. The convergence of ARs flux facilitated moisture accumulation in the precipitation region, providing crucial conditions for this extreme rainfall. Additionally, the moisture channel formed by Typhoon Khanun and the associated low-pressure circulation contributed significantly, highlighting the synergistic impact of multi-scale systems on extreme precipitation.

Within ARs region, potential instability in the lower layers favored convective development. The orographic lift on the windward slopes of mountains forced ARs to ascend, enhancing vertical motion and sustaining heavy rainfall. Moderate increases in urban roof roughness promoted precipitation in urban and surrounding areas due to changes in local circulation induced by altered roughness and thermal conditions. However, further increases in roughness suppressed urban precipitation and potentially shifted it to peripheral regions.

These conclusions elucidate the characteristics of the ARs system driving the precipitation event, the impact of urbanization, and the associated dynamic mechanisms, providing valuable insights into the causes of extreme precipitation events in North China. The study’s findings align with previous research in several aspects, such as the meridional transport of large amounts of moisture by ARs and the associated intensification of precipitation [7,47], the potential instability in the lower layers of ARs [48], and the significant enhancement of precipitation when encountering topography [49,50]. Urbanization has been shown to increase surface roughness, leading to heightened sensible heat fluxes and an intensified urban heat island effect, thereby increasing urban precipitation [51]. The increased surface roughness within urban areas is linked to the deceleration of moving air masses, resulting in enhanced surface moisture convergence and increased rainfall intensity [52]. The momentum transport caused by urban friction plays a crucial role in strengthening low-level convergence and moist convection, contributing to heavy rainfall over urban areas [53]. Furthermore, the increased surface roughness in urban areas raises instability in the lower atmosphere, causing earlier and higher peak rainfall rates in these regions [29]. While previous research has considered the effects of increased roughness leading to reduced wind speeds, increased instability, and enhanced moisture convergence, which in turn augments precipitation, our study extends these findings by exploring the implications of increased roughness beyond certain thresholds. Specifically, our study investigates the micro-scale impacts of rooftop roughness on extreme precipitation—a factor not thoroughly examined in the existing literature. We delineate how different stages of precipitation exhibit distinct characteristics under varying degrees of rooftop roughness. Additionally, this study introduces new scientific insights into the impact of atmospheric rivers and urban roof roughness on extreme precipitation under multi-scale systems. These insights enrich and refine existing theories, providing valuable guidance for flood control and urban planning in regions prone to extreme rainstorms. The study also underscores the importance of incorporating urbanization effects into weather forecasts and climate models to enhance accuracy and inform policy-making.

It is noteworthy that this study only investigates a single extreme precipitation event, making it challenging to generalize the findings to other types of extreme weather over a broader region. Future research should focus on the in-depth examination of interactions between ARs, urban effects, and topographical influences to enhance the prediction and mitigation of extreme weather. Additionally, more customized analyses are necessary for different regions and types of extreme events to fully understand and enhance scientific knowledge of regional extreme weather phenomena. Future research should also focus on the high-resolution modeling of urban environments, such as large-eddy simulations, and long-term observational studies to better understand the interactions between urban structures and atmospheric processes.