Search

Different turbulent entrainment‐mixing mechanisms between clouds and environment are essential to cloud‐related processes; however, accurate representation of entrainment‐mixing in weather/climate models still poses a challenge. This study exploits the use of machine learning (ML) to address this challenge. Four ML (Light Gradient Boosting Machine [LGB], eXtreme Gradient Boosting, Random Forest, and Support Vector Regression) are examined and compared. It is found that LGB performs best, and thus is selected to understand the impact of entrainment‐mixing on microphysics using simulation data from Explicit Mixing Parcel Model. Compared with traditional parameterizations, the trained LGB provides more accurate microphysical properties (number concentration and cloud droplet spectral dispersion). The partial dependences of predicted microphysics on features exhibit a strong alignment with physical mechanisms and expectations, as determined by the interpreting method, thus overcoming the limitations of the "black box" scheme. The underlying mechanisms are that the smaller number concentration and larger spectral dispersion correspond to more inhomogeneous entrainment‐mixing. Specifically, number concentration after entrainment‐mixing is positively correlated with adiabatic number concentration and liquid water content affected by entrainment‐mixing, and inversely correlated with adiabatic volume mean radius. Spectral dispersion after entrainment‐mixing is negatively correlated with liquid water content affected by entrainment‐mixing, turbulent dissipation rate and relative humidity of entrained air. Sensitivity analysis further suggests that number concentration is mainly determined by cloud microphysical properties whereas spectral dispersion is influenced by both cloud microphysical properties and environmental variables. The results indicate that the LGB scheme has the potential to enhance the representation of entrainment‐mixing in weather/climate models. Plain Language Summary: Entrainment‐mixing processes occurring between clouds and environmental air have significant effects on cloud‐climate feedback, precipitation, and radiative transfer. Accurately representing these processes has been challenging with previously proposed parameterizations. Machine learning (ML) excels at identifying complex nonlinear relationships and avoids the inherent limitations of conventional parameterizations. Thus, we explore the use of ML to acquire the number concentration and cloud droplet spectral dispersion affected by entrainment‐mixing processes. Simulation data from Explicit Mixing Parcel Model are employed to train, validate, and test the four MLs, including Light Gradient Boosting Machine (LGB), eXtreme Gradient Boosting, Random Forest and Support Vector Regression. After evaluation, the LGB is shown to obtain the most accurate microphysics among the four ML schemes and traditional parameterizations. Additionally, the interpreting method for peeking inside the "black box" reveals that a smaller number concentration and a larger spectral dispersion are indicative of the more inhomogeneous entrainment‐mixing. The relative correlations between predicted microphysics and various features align with our expectations and the underlying physical principles. Sensitivity tests further confirm that incorporating more features produces a more robust and efficient prediction. Overall, this study affirms the reliability and applicability of ML to develop/replace subgrid parameterizations in actual weather/climate models. Key Points: Machine learning schemes are trained to predict cloud microphysical properties affected by turbulent entrainment‐mixing processesThe proposed Light Gradient Boosting Machine provides more accurate microphysical properties compared with traditional parameterization schemesThe partial dependencies of microphysics on features prove a robust alignment with physical mechanisms through the interpreting method [ABSTRACT FROM AUTHOR]

The concave mountainous area of Pearl River Delta (PRD) is a summer rainfall hotspot of South China due to the presence of warm‐moist monsoon flow and complex orography. This study evaluated the performance of a convection‐permitting Weather Research and Forecasting (WRF) model in forecasting nocturnal rainfall in this area, focusing on days with low level southwesterly winds during the summers of 2013–2015. Results showed that the nocturnal rainfall exhibited two centers, one located along the large‐scale northern mountains and the other along the small‐scale Huadu Hill. WRF demonstrated superior performance in predicting rainfall over the northern mountainous region. In contrast, WRF significantly underestimated nocturnal rainfall both near local Huadu Hill and in the foothill area of northern mountains, which were strongly influenced by local forcings. Using high‐resolution analyses from Variational Doppler Radar Analysis System (VDRAS) and Automatic Weather Stations observations, we firstly identified the sources of forecast errors in triggering a typical localized nocturnal convection. Results revealed that WRF severely underestimated thermal contrast between the Guangdong‐Hong Kong‐Macao Greater Bay Area urban agglomeration and the concave mountains, leading to the absence of the northeastern/northern inflows toward the cities. Consequently, low level convergence and updrafts near the convection initiation position were too weak to lift air parcels above the severely overestimated level of free convection, thereby failing to trigger the convection. VDRAS‐based sensitivity experiments, with a specific focus on assimilating surface temperature, validated the crucial role of urban‐mountain thermal contrast on local winds that triggered the nocturnal convection. Plain Language Summary: The concave mountainous region of Pearl River Delta serves as a summer rainfall hotspot in South China, and heavy rainfall at night, especially those influenced by local forcings, pose significant challenges for forecasters. This study evaluated the performance of convection‐permitting Weather Research and Forecasting (WRF) model in forecasting nocturnal rainfall on days with low level southwesterly winds during the 2013–2015 summers. Results showed that WRF exhibited limitations in forecasting nighttime precipitation that developed locally within the concave mountainous area. Based on a localized heavy rainfall event, we identified sources of errors in predicting nocturnal convection initiation (CI). There was a significant cold bias over the Guangdong‐Hong Kong‐Macao Greater Bay Area urban agglomeration, which led to a seriously underestimated temperature difference between urban area and concave mountains. Consequently, WRF failed to predict urban inflows, and the predicted low level convergence and updrafts near the CI location were too weak to lift air parcels above the seriously overestimated level of free convection, resulting in a failure to trigger the convection. This study emphasizes that, over a complex geographical environment characterized by concave mountains and urban agglomeration, accurately predicting the urban‐mountain thermal contrast and local circulations is crucial for nighttime precipitation forecast. Key Points: Nocturnal rainfall that developed locally within the concave mountainous area of South China was severely underestimatedForecast bias in localized convection initiation was related to underestimated thermal contrast between urban area and concave mountains [ABSTRACT FROM AUTHOR]

Classical bow echoes (CBEs) normally develop due to rear‐inflow jets, while merger‐formation bow echoes (MFBEs) evolve from the merger between linear systems and preline convection. MFBEs are rarely documented before, and the key processes remain unclear because of the small‐scale and fast‐evolving merging processes. Using radar observations during 2015–2019, this study examines the characteristics of these two types of bow echoes over South China. Eighteen MFBEs and 11 CBEs are identified, accounting for 62% and 38% of the total 29 bow echoes, respectively. MFBEs commonly occur over coastal regions in the afternoon, where southerly sea breezes provide favorable thermodynamic conditions. About half CBEs develop over mountainous areas in the morning, and northerly winds enhance the local baroclinity and kinematic convergence. A new Merger‐Classical index is proposed to distinguish the environments for the development of two types of bow echoes, disclosing the more favorable near‐surface thermal conditions for MFBEs. In addition, more severe weather occurs in metropolitan area near the Pearl River Delta region during MFBEs, in which the accumulated precipitation is about 2.5 times higher than that during CBEs. This study reveals the discrepancies between MFBEs and CBEs over South China, emphasizing the MFBEs account for more than half of the total bow echoes and cause more high‐impact weather in the metropolitan area. This study proposes a new perspective on the environment in which bow echo forms by merger process over South China, urging for more research to explore the underlying processes. Plain Language Summary: Bow echoes produce damaging surface winds and heavy rainfall. Previous studies normally explained bow echoes using the classical theory that emphasized the importance of rear‐inflow jet, and such echoes are termed as the classical bow echoes (CBEs) in this study. Additionally, the warm moist environmental conditions over South China facilitate bow echo formation through merger process between linear systems and convective cells. Bow echoes formed in this manner are termed as the merger‐formation bow echoes (MFBEs). Due to the small‐scale and fast‐evolving process of merging, MFBEs are rarely documented in previous literature. This study is the first to investigate the spatiotemporal distributions of MFBEs and CBEs over South China using 5‐year radar observations, pointing out the MFBEs account for more than half of the total bow echoes and cause more high‐impact weather in metropolitan area than CBEs. In addition, we propose a new Merger‐Classical index to distinguish the environments for MFBEs and CBEs. The differences between MFBEs and CBEs are statistically significant at the 95% confidence level. This paper emphasizes a new perspective on the environment in which bow echo forms by merger process, underscoring the need for more in‐depth research in the future. Key Points: Merger‐formation bow echoes (MFBEs) and classical bow echoes (CBEs) account for 62% and 38% over South China during 2015–2019, respectivelyMFBEs mainly occur over the coastal regions in the afternoon, while about half CBEs develop over the mountainous areas in the morningThe proposed Merger‐Classical index distinguishes developing environments for MFBEs and CBEs, and MFBEs produce heavier rainfall near metropolitan area [ABSTRACT FROM AUTHOR]

The first rainy season (April–May–June) extreme precipitation events over South China tend to frequently occur in specific hotspots, resulting in a great threat to local society and economy. Abundant moisture supply is essential to precipitation formation. In this study, we employ the FLEXPART model to investigate the geographical moisture sources for extreme precipitation events in three hotspots of Guangdong from west to east (termed as Area‐A, B, and C). During both the periods before and after the onset of the South China Sea Summer Monsoon (hereafter SCSSM), the moisture contributions of ocean regions are larger than that of land regions for all the hotspots, but the main moisture source regions are different. Before the onset of SCSSM, the main source regions are South China Sea and Southeastern Asia, which contribute 36.50 and 31.79% to Area‐A, 35.85 and 27.83% to Area‐B, while 42.27 and 23.81% to Area‐C, respectively. After the onset of SCSSM, the contribution of moisture from Indian Ocean increases a lot. Thus, the main source regions are Indian Ocean, Southeastern Asia and South China Sea, which contribute 31.52, 24.58 and 18.13% to Area‐A, 29.21, 21.82 and 14.99% to Area‐B, 26.62, 20.70%, and 28.19% to Area‐C, respectively. Among the three hotspots, the contribution of ocean regions for Area‐C is the largest. Southeastern Asia is the main land contributor before and after the onset of SCSSM for all the hotspots. This study implies the important role of both ocean and land moisture sources on extreme precipitation variations over South China. [ABSTRACT FROM AUTHOR]

This work compares the characteristics of the first echoes of thunderstorms and nonthunderstorms retrieved from S‐band polarimetric radar observations. Observations of 57 (39) isolated thunderstorm (nonthunderstorm) cells with roughly equivalent aerosol and water vapor conditions but different convective available potential energy were obtained with a S‐band polarimetric radar and three independent lightning location systems during 2016/2017 in southern China. Storms with the first echoes were divided into three types based on echo top heights, namely, type 1 (below 0°C layer), type 2 (0°C to −10°C), and type 3 (above −10°C layer). Our observations show median values of radar reflectivity (ZH) and differential reflectivity (ZDR) of type 1 and type 2 in warm phase layer (below 0°C layer) are obviously greater in nonthunderstorms than in thunderstorms, but this feature is not significant in type 3 storms. In the mixed 1 phase layer (0°C to −10°C), median ZH in type 2 is greater in nonthunderstorms while median ZDR in type 3 is slightly smaller. In the mixed 2 phase layer (−10°C to −38°C), median ZH is greater in thunderstorms while median ZDR is smaller, and ZDR values in nonthunderstorms are closer to zero. Although results of ZDR comparisons in the mixed phase are likely affected by random errors and/or residual bias errors, these different signatures suggest different characteristics of liquid or ice particles between thunderstorms and nonthunderstorms. This study is expected to advance our understanding of physical processes responsible for the generation of the first flash. Plain Language Summary: Thunderstorms generally show stronger convective activity than nonthunderstorms do. What are thunderstorms and nonthunderstorms like at the beginning of precipitation radar detection? Observations of 57 (39) isolated thunderstorm (nonthunderstorm) cells based on a S‐band polarimetric radar and three independent lightning location systems during 2016/2017 in southern China are investigated. We found that polarimetric radar parameters of the first echoes are different between thunderstorms and nonthunderstorms, especially, the larger echo intensity in nonthunderstorms below −10°C layer. Comparative results in different height layers suggest different characteristics of liquid/ice particles of the first echoes between thunderstorms and nonthunderstorms. This study would help us to understand the characteristics of microphysics in the formation of thunderstorms. Key Points: The first echoes of thunderstorms and nonthunderstorms determined using radar are manually retrievedThe parameters of polarimetric radar of the first echoes are different between thunderstorms and nonthunderstormsDifferent properties suggest different characteristics of liquid/ice particles of the first echoes between thunderstorms and nonthunderstorms [ABSTRACT FROM AUTHOR]

Sedimentological and chronological analyses of coastal dunes can determine their sedimentation processes and yield information that can reflect changes in local marine environments (e.g., information on climate, biology, and sea‐level changes in tectonically stable areas). In this study, Late Quaternary deposits on Shidao Island, one of the Xisha Islands, South China Sea, were analysed using techniques such as U/Th dating to obtain the following results: (1) Four lithofacies associations and two unconformities were identified from the deposits on Shidao Island. On this basis, the Shidao Island sediments can be divided into three stages of aeolian sediments and one stage of beach rock sediments. (2) U/Th dating revealed that the aeolian deposits on Shidao Island formed during three periods, namely, marine isotope stages 5.2, 5.1, and 4, and that the beachrocks were formed in the Holocene when the sea level rose to its present height. (3) The aeolianites on Shidao Island all formed in an environment in which the sea level was relatively stable or slowly falling. This study determined for the first time the specific sedimentary ages of the aeolianites on Shidao Island, and reconstructed the sedimentation process of the deposits on Shidao Island based on palaeo‐sea‐level changes. These results can provide more reliable geochronology data for the future study of Late Quaternary marine environments across the East Asian region. [ABSTRACT FROM AUTHOR]

The northern mountainous area of the Pearl River Delta (PRD) is a center of convection and heavy rainfall of coastal southern China during the summer monsoon season. Although there are frequent nighttime convective activities and rainfall over the region, the key dynamical and physical processes responsible for the nocturnal convection initiation (CI) are complex and not yet well documented. In this study, high‐resolution reanalyses from the Variational Doppler Radar Assimilation System (VDRAS) were used to explore the triggering mechanisms of a representative nocturnal convection event that occurred over the northern PRD on August 14, 2014. Results showed that the enhanced prevailing low‐level southerly winds, deflected easterly winds caused by the blocking of multi‐scale orography, and northern downslope winds associated with an orographic cooling effect were crucial for generating low‐level convergence and strong updrafts before CI. The low‐level convergence and moisture flux in the boundary layer started to increase at 00:20 local solar time (LST), and by the CI time (01:00 LST), the convergence intensity more than doubled. The intensification of low‐level convergence and moisture transport was strongest before 00:40 LST and was dominated by a zonal component that was closely related to the deflected easterly winds. Using Doppler radar data assimilation, this is the first study to reveal how the deflected flows and the thermally induced downslope winds combine with prevailing monsoonal winds to affect a nocturnal CI over the complex mountainous region of the South China coast. Key Points: The combination of complex orography and monsoonal flows is crucial for the nocturnal convection initiation along the South China coastEnhanced low‐level monsoonal flows, terrain‐induced easterly winds, and downslope winds converge and generate strong updraftsThe terrain‐deflected easterly winds play an important role in low‐level convergence and moistening [ABSTRACT FROM AUTHOR]

Extreme precipitation is an issue of worldwide concern, but its microphysics remain elusive. The convective and microphysical characteristics of extreme precipitation features (EPFs) in a monsoon coastal area (South China) are investigated mainly using 2‐year observations from a dual‐polarization radar and distrometers. The EPFs are accompanied by a broad range of convective intensity, and categorized into the "intense", "moderate", and "weak" convection accounting for 17.3%, 48.6% and 34.1% of the total population, respectively. The EPFs with weaker convection show weakened size sorting and less breakup of large raindrops, but a larger ratio of liquid water path to ice water path and more prominent coalescence warm‐rain process. All the three categories are dominated by the coalescence in the liquid‐phase processes, and have much more populous raindrops than the "continental" with a mean size larger than the "maritime" regime. These results improve our understanding of extreme precipitation from the microphysical perspective. Plain Language Summary: Extreme precipitation contributes significantly to the global precipitation and could induce flooding and severe hazards. Such a scenario may become more often in the warming climate. However, we do not know well whether extreme precipitation is mainly produced by intense convection with presence of large solid precipitating particles such as hail and graupel, or by rapid growth of many raindrops below the 0°C level with a lack of large solid precipitating particles aloft. To document the convective and microphysical characteristics of extreme precipitation, we use 2‐year, high‐resolution observations from a dual‐polarization radar, six 2D‐Video‐Distrometers (which measure raindrop size distribution), and Low‐Frequency E‐field Detection Array (which provides the three‐dimensional position and time of flashes) on a typical monsoon coast (South China). Results show that the extreme precipitation events have convective intensities ranging from relatively weak to very intense. The raindrops are larger‐sized than the "tropical maritime type" and much more populous than the "continental type" of convective precipitation. Key Points: Only 17% of the extreme precipitation features (EPFs) overlap with intense convection, while over 30% EPFs show limited mixed‐phase processCoalescence dominates (>75%) the liquid‐phase processes of EPFs over the monsoon coast, even for those with intense convectionMonsoonal EPFs show a mean raindrop size larger than the "maritime" and a raindrop concentration much higher than the "continental" regime [ABSTRACT FROM AUTHOR]

An extreme rainfall event with maximum 12 hr accumulated rainfall of 464.8 mm associated with a quasi‐stationary mesoscale convective system occurred over the western coastal region of South China in June 2017. An observational analysis shows that early convective storms were initiated near the mountains and moved northeastwards. Moreover, a convergence line between cold northerly winds and warm southerly winds in the lower levels was formed, which favored the development of the quasi‐stationary system. Cold northerly winds were associated with land breeze, downslope winds as well as previous rainfall during earlier period, and cold pool outflows during later period. Cloud‐resolving simulations were performed to examine the roles of terrain, land‐sea surface roughness contrast, and cold pool outflows in the formation of heavy rainfall. Results demonstrate that land‐sea surface roughness contrast and mountains in the middle of Yangjiang and Jiangmen facilitated the formation of the convergence line during earlier period, and cold pool outflows sustained it during later period. Besides orographic lifting, coastal hills helped reduce the stability. Mountains in the middle of Yangjiang hindered the movement of the convective system, without which the associated heavy rainfall would shift farther north. Mt. Tianlu with westward concave morphology played a vital role in the formation of local convergence and the concentration of heavy rainfall. Without it, the total rainfall in the region of interest was reduced. This study suggests the importance of representing processes associated with the complex underlying surface in models for the prediction of heavy rainfall in coastal regions. Key Points: The relative roles of terrain, surface roughness, and cold pool outflows in the formation of heavy rainfall were investigatedA convergence line associated with winds affected by underlying surface, cold pool outflows, and persistent southerly flows was favorableLand‐sea surface roughness contrast and terrain played an important role in the initial formation of the convergence line [ABSTRACT FROM AUTHOR]

A double‐moment version of the SBU‐YLIN cloud microphysical scheme in WRF is introduced. It predicts the mass and number mixing ratios of cloud droplet, rain, cloud ice, and precipitating ice. In addition, a number of physical processes, like rain evaporation, collection between rain and snow are also optimized in the new scheme. The scheme is evaluated and compared with the original one‐moment scheme for a squall line case. We found that the double‐moment approach gives a better representation of rain evaporation, which is critical for the development, morphology, and evolution of the simulated squall line, especially for the enhanced trailing stratiform cloud and leading convective line. The relationship between key microphysical processes and squall line dynamics is investigated to identify the driving mechanisms of the descending rear inflow, cold pool, and slantwise updraft. Furthermore, formation of the transition zone in the simulated squall line strongly depends on the flexible description of ice particle properties, such as size, degree of riming and fall speed. Plain Language Summary: A double‐moment version of the SBU‐YLIN cloud microphysical scheme in WRF is introduced and evaluated. The scheme improves a squall line simulation in terms of the shape and structure of the squall line with enhanced stratiform precipitation. The improvement is mainly related to more realistic rain evaporation and flexible description of ice particle properties. Close interactions between cloud and precipitation physics and dynamics is key for the development, maintenance and movement of squall lines. Key Points: A double‐moment version of the SBU‐YLIN scheme is introducedThe scheme improves the simulation of a squall line in terms of the transition zone and stratiform precipitationRain evaporation and flexible description of ice particles are critical for a realistic simulation [ABSTRACT FROM AUTHOR]

This study evaluates how clouds evolve into thunderstorms in terms of the turbulence characteristics producing the first flash. Observations of 57 (39) isolated thunderstorm (non‐thunderstorm) cells during 2016/2017 in Southern China are provided by an S‐band polarimetric radar and three independent lightning location systems. The vertical turbulence characteristics of clouds associated with thunderstorms are obviously different from non‐thunderstorms. For thunderstorms, the maximum of the eddy dissipation rate (mean value in each height layer) in the entire height is 0.19 m2 s−3, which occurs at the first flash stage, and the achievable height of turbulence exceeds the −30°C layer. For non‐thunderstorms, however, the maximum is 0.12 m2 s−3, and the achievable turbulence height hardly exceeds the −10°C layer. Additionally, the turbulence intensities of the locations where the initial discharge pulse events of the first flashes occur are weak. These turbulence characteristics are useful for lightning nowcasting. Plain Language Summary: The first flash is a significant indicator of the development of a thunderstorm. Hence, studying how clouds evolve into thunderstorms producing the first flash is an interesting topic. Understanding the occurrence of the first flash is valuable for developing nowcasting tools and lightning parameterizations in numerical weather prediction models. However, the role of turbulence in producing the first flashes in the development of thunderstorms is poorly understood. Turbulence indicating the disorder in clouds provides the relative speed differences for hydrometeors with different masses and inertias. Consequently, it is essential for electrification process and the occurrence of the first flash. The vertical turbulence characteristics are both demonstrated in thunderstorms from the initial generation to the occurrence of the first flashes and in non‐thunderstorms. The resulting turbulence characteristics may provide useful information for lightning nowcasting. Key Points: This study evaluates how clouds evolve into thunderstorms in terms of the turbulence characteristics producing the first flashThe vertical turbulence characteristics of clouds show obvious differences between thunderstorms and non‐thunderstormsThe turbulence intensities of the locations where the initial discharge pulse events of the first flashes occur are weak [ABSTRACT FROM AUTHOR]