Configuration and verification
We employed version 3.5 of the Weather Research and Forecasting (WRF) model. The model configuration consisted of a two-way interactive domain with two levels of nesting comprising horizontal grid spacings of 9 and 3 km. The simulation adopted terrain-following coordinates, 50 vertical levels, and a top level at 50 hPa. The model physics employed in each domain included the Dudhia (1989) shortwave radiation scheme, Rapid Radiative Transfer Model (RRTM) longwave radiation scheme (Mlawer et al., 1997), WRF single-moment 5-class (WSM5) explicit microphysics scheme (Mielikainen et al., 2012), Yonsei University (YSU) planetary boundary layer scheme (Hong et al., 2006), and Noah land surface parameterization scheme (Chen and Dudhia, 2001). The Kain–Fritsch (KF) cumulus scheme (Kain and Fritsch, 1990) was utilized in the 9-km grid spacing domain, and no implicit cloud scheme was employed in the finest 3-km resolution domain. In addition, intensified, static, surface classification data from the Beijing–Hebei region were applied. The model integration was initialized at 0000 UTC 4 April 2018. The model first guess fields and 3-h interval lateral boundary conditions were obtained from the NCEP 1° × 1° global analysis. The data from automatic weather stations, radar reflectivity factors and radial wind data were assimilated through three-dimensional variational data assimilation (3DVAR) to form the initial fields. The simulation period covered 36 h, and the simulated data were output hourly.
A comparison between the simulated precipitation and the observed precipitation was performed. During the 0000–0600 UTC period on 4 April (Figs. 5a, d), representing the initial stage of snowfall in the Beijing–Hebei region, the distribution pattern and magnitude of the simulated 6-h cumulative precipitation were similar to those of the observed precipitation. The precipitation zone gradually moved from west to east and covered Yangyuan Station in Hebei Province with snowfall of approximately 4 mm. From 0600 to 1200 UTC (Figs. 5b, e), the main stage of snowfall, the simulated precipitation area continued to move eastward and was shaped like a pincer with a maximum magnitude of 16 mm, which was similar to the observed value. However, the simulated snowfall near Yanqing Station in northwest Beijing was smaller than the observed snowfall. Moreover, the significant simulated snowfall belt in northwestern Beijing occurred from 1200 to 1800 UTC (Figs. 5c, f), later than the observations. The simulated snowfall in southeastern Beijing was approximately 2 mm smaller than the observed snowfall. The comparison between the 24-h accumulated snowfall from 0000 UTC 4 to 0000 UTC 5 April (Figs. 5g, h) also indicates agreement between the simulated and observed distributions and magnitudes of precipitation. In addition, the simulated 500- and 850-hPa geopotential height fields, temperature fields, and wind fields were also consistent with the observations. Therefore, simulation data with a resolution of 3 km can be utilized to analyze the causes of RRTS transitions.
Comparisons of the cumulative precipitation (shaded; mm) between the (a, b, c, g) simulation and (d, e, f, h) observations. (a, d) 0000–0600 UTC 4 April, (b, e) 0600–1200 UTC 4 April, (c, f) 1200–1800 UTC 4 April, and (g, h) 0000 UTC 4 to 0000 UTC 5 April.
Characteristics of the RFs in the plain and mountainous areas
Due to the complex topography of North China, including the Bohai Sea, North China Plain, Yanshan Mountains, and Taihang Mountains, the RFs in North China may manifest in complex ways. Therefore, it is necessary to conduct a detailed analysis of the characteristics of RFs according to the different geopotential heights, different heavy snow locations, and different evolution stages.
Cold–moist RF at 1000 hPa and cold–dry RF at 925 hPa in the Beijing–Hebei plain area
The Daxinganling Mountains in western Northeast China are oriented in a north–south direction with an average altitude of approximately 1300 m, and the Changbai Mountains in eastern Northeast China are oriented in a northeast–southwest direction with an average altitude of approximately 1400 m. The Northeast China Plain with an altitude of less than 200 m is located between these two mountain ranges, forming a narrow topographic channel conducive to the southward movement of cold air from the Okhotsk Sea surface high pressure system. At 0300 UTC 4 April (Fig. 6a), due to this topographic forcing, cold air flowed southward along this channel and behaved as northerly winds. After reaching the Bohai Sea, part of this air mass east of 120°E continued to flow southward, while part of the air mass west of 120°E became northeasterly winds that invaded the North China Plain, forming the abovementioned RF. At 0600 UTC (Fig. 6b), accompanying this change in the large-scale background circulation, part of the RF west of 118°E and north of 39°N transformed into southeasterly winds. The time of this change in wind direction corresponded to the beginning of the observed precipitation in the plain area.
Wind fields at 1000 hPa (black streamlines) at (a) 0300 UTC 4 April and (b) 0600 UTC 4 April. The shaded background indicates the terrain, and the light blue infill denotes the ocean.
In late spring, solar radiation is often strong. Due to the different physical properties of the sea and land, the land surface temperature increases abruptly in the afternoon, while the sea surface temperature increases slowly, forming a significant sea–land temperature difference in the near-surface layer. At 0600 UTC (Fig. 7a), a cold pool formed in the Bohai Sea at 1000 hPa, and the boundary of this cold pool was similar to the coastline of the Bohai Sea. The temperature at the cold pool center was 2°C, while the highest temperature on the North China Plain reached 7–8°C, so a strong temperature gradient appeared near the coastline. The Bohai Sea also corresponded to an area with high relative humidity. With the intrusion of dry and cold air at 0600 UTC, the relative humidity in the northern Bohai Sea dropped to 20%, whereas that in the central and southern Bohai Sea was maintained at 40%–60%, forming the cold–moist RF on the leeward side of the Bohai Sea. However, the cooling and humidifying effects of the Bohai Sea existed only in the relatively shallow near-surface layer. At 925 hPa (Fig. 7b), the Bohai Sea area was occupied by cold–dry northeasterly winds, and the temperature and relative humidity gradient belts corresponding to the oceanic boundaries basically disappeared. In addition, due to the low friction of the ocean surface, a low-level jet formed at 925 hPa over the Bohai Sea, facilitating wind speed convergence on the leeward side of the Bohai Sea.
Temperature (red contours; °C), relative humidity (shaded; %), and wind (vectors; m s−1) at 0600 UTC 4 April 2018, at (a) 1000 hPa and (b) 925 hPa.
The abovementioned analyses show that during the period from 0300 to 0600 UTC, the Beijing–Hebei plain area was affected by both a shallow cold–moist RF at 1000 hPa and a relatively deep dry–cold RF at 925 hPa.
Cold–dry RF at 925 hPa in the northwest mountainous area of Beijing
From the simulated relative humidity and wind at 925 hPa, the RF exhibited significant dry characteristics. The main body of the RF corresponded well to the area with a relative humidity of 30% and below. Therefore, the 30% relative humidity contour was selected to track the RF activities. From 0300 to 0600 UTC, the RF advanced from the Bohai Sea to the Beijing plain area, and then gradually intruded the vicinity of Yanqing Station in the mountainous area of northwest Beijing. At 0300 UTC (Fig. 8a), the Beijing plain area was governed mainly by 40%–50% relative humidity. The 30% relative humidity zone invaded the southeast corner of Beijing (approximately 39.7°N, 116.75°E). At 0400 UTC (Fig. 8b), the 30% relative humidity boundary extended approximately 20 km westward and approximately 10 km northward in 1 h, reaching approximately 39.8°N, 116.45°E. At this time, the wind direction also changed from east to southeast. At 0500 UTC (Fig. 8c), most of Beijing was controlled by southeasterly winds, and the 30% relative humidity zone intruded approximately 20 km westward and 40 km northward in 1 h, reaching approximately 40.2°N, 116.25°E, indicating that the southeasterly winds had strengthened. Another notable feature at this time was the intrusion of the RF into the area near the front of the western valley. At 0600 UTC (Fig. 8d), most of the plain area was occupied by the RF. The 30% relative humidity zone intruded 20–30 km northwest in 1 h, reaching approximately 40.4°N, 116.1°E, and the RF intruded the valley near Yanqing Station. These analyses demonstrate that although the mountainous area of northwest Beijing has a higher altitude, owing to the intrusion of the RF along the valley, the snowfall in the mountainous area of northwest Beijing might have been directly affected by the cold–dry RF at 925 hPa.
Simulated relative humidity (shaded; %), temperature (red contours; °C), and wind barbs (full barb is 4 m s−1) at 925 hPa at (a) 0300 UTC, (b) 0400 UTC, (c) 0500 UTC, and (d) 0600 UTC 4 April 2018. The thick black solid line is the contour line of the terrain height of 100 m. The black circle indicates the area around Yanqing Station.
Cold–moist RF at 800 hPa in the northwest mountainous area of Hebei Province
Since the average altitude in the northwestern part of Hebei Province exceeds 1000 m, the northwest Hebei Province is affected predominantly by the circulation system above the 850-hPa layer. Therefore, the 800-hPa level was selected as the representative layer to investigate the characteristics of the near-surface layer and boundary layer in this region. At 0300 UTC (Fig. 9a), an anticyclonic circulation was centered at 800 hPa over the northeastern corner of the Beijing border (approximately 40°N, 117°E) with a radius of 500 km. This anticyclonic circulation was separated from the North China high (at 0000 UTC located at 35°–50°N, 115°–125°E). When cold air flowed southward, the southern part of the North China high strengthened due to the increased surface pressure. Controlled by this anticyclonic circulation, the cold–dry air from the northeast flowed southward, passed over the Bohai Sea, and then flowed westward at 36°–39°N, forming an RF. This RF showed cold and dry characteristics east of 115°E but cold and moist features near Yangyuan Station (39.5°N, 114°E). Note that the anticyclonic circulation affecting Yangyuan was very shallow. At 750 hPa (Fig. 9b), the anticyclonic circulation disappeared, and there were no easterly winds in the entire study area. The region within 112°–116°E was dominated by southerly winds, while that east of 116°E was dominated by northwesterly or southwesterly winds. Therefore, Yangyuan Station was also directly influenced by the RF at 800 hPa. However, this relatively shallow RF was cold and moist, and thus obviously differed from that over the Beijing–Hebei plain area and mountainous area of northwest Beijing.
Simulated (a) relative humidity (shaded; %) and wind (black streamlines) at 800 hPa and the simulated (b) temperature (shaded; °C) and wind at 800 hPa (black streamlines) and wind at 750 hPa (gold streamlines) at 0300 UTC 4 April. The black circle indicates the area around Yangyuan Station.
The characteristics of this RF have been debated in previous studies, in which some researchers considered it cold and moist, while others considered it cold and dry. However, from our abovementioned analyses, the RF features might have manifested differently in different geographical locations, at different terrain heights, and within different periods of snowfall. There was a cold–moist RF at 1000 hPa, a cold–dry RF at 925 hPa, and a cold–moist RF at 800 hPa, forming an interesting characteristic “sandwich” structure.
Characteristics of convergence, vertical motion, and water vapor transport
The location of heavy snow is closely related to the RF. Analyzing the characteristics of boundary layer convergence and vertical movement caused by the RF in different regions, as well as the characteristics of water vapor transport in this heavy snow process, can establish a foundation for further exploring the causes of heavy snow phase transitions.
Plain and Yanqing mountainous areas of Beijing
At 0600 UTC (Fig. 10a), the Beijing plain area was controlled primarily by the easterly winds of the RF at 1000 hPa, and there was no obvious convergence zone in the plain area. At 925 hPa (Fig. 10b), similarly, there were no obvious convergence zones in the plain area or Yanqing mountainous area, which was occupied mainly by the southeasterly winds of RF. At 700 hPa, obvious southwesterly winds were roughly perpendicular to the specific humidity contours and temperature contours, indicating that the warm and humid airflow climbed along the cold–dry air cushion, and its water vapor transport capacity was strong. At 0800 UTC (Fig. 10c), at 1000 hPa, the Beijing plain area changed to be controlled by southeasterly winds, and an obvious convergence zone appeared over southeast Beijing. At 925 hPa (Fig. 10d), the southeasterly winds further strengthened and invaded along the valley to the western mountain area, where they formed an obvious convergence zone near Yanqing Station. The southerly component of the southwesterly winds at 700 hPa further strengthened, and the 3 g kg−1 specific humidity contour moved northward to approximately 40°N, approximately 10–50 km within 3 h, which resulted in significant humidification in the plain area of southeastern Beijing.
Convergence (orange dashed contours with interval 10; 10−4 s−1) and wind (black streamlines) at 1000 hPa (left) and 925 hPa (right) at (a, b) 0600 UTC and (c, d) 0800 UTC 4 April. The plots are superimposed with the specific humidity (green contours; g kg−1) and wind at 700 hPa (blue streamlines). The black circles in (a) and (c) indicate the Beijing plain area, and the black circles in (b) and (d) indicate the area around Yanqing Station.
The appearance of these strong convergence zones in the near-surface layer corresponded to the significant increase in snowfall observed in the Beijing plain area and Yanqing Station. The convergent upward motion of the near-surface layer caused by the RF, the climbing of warm–moist air along the cold–dry air cushion, and the vertical upward motion in front of the surface inverted trough combined to form relatively strong convective motions, which were beneficial to heavy snowfall over Beijing.
The height–time profile at Beijing Station (Fig. 11a) indicates that the easterly winds acted mainly below 850 hPa from 0300 to 0900 UTC. The convergence of the near-surface layer forced upward motions and formed a vertical speed center of 0.2 m s−1 from 0600 to 0700 UTC at 1000–850 hPa. Warm–moist air climbed along cold–dry air in combination with the influence of the upper-level trough, forming a strong ascending center at 700–400 hPa from 0500 to 1100 UTC, which was conducive to the further development and maintenance of snowfall. The 700-hPa layer was very dry at 0300 UTC with a relative humidity of approximately 20%. However, by 0600 UTC, the relative humidity increased to 80%–90%, which corresponded to the transport of warm–moist air at 700 hPa and cloud formation due to convergent upward motion of the boundary level.
Height–time cross-sections of cloud hydrometeor snow particles (shading; 10−1 g kg−1), upward vertical velocity (red solid contours; m s−1), and relative humidity (black dashed line; %) at representative stations on 4 April 2018. (a) Beijing Station and (b) Yanqing Station. The plots are superimposed with the easterly winds (wind barbs, a full barb is 4 m s−1). The gray area represents the terrain.
The height–time profile at Yanqing Station displays a wedge-shaped image (Fig. 11b), indicating that warm–moist air climbed along a cold–dry air cushion. The easterly winds acted mainly at 925–800 hPa above the terrain height and formed a convergence zone in the near-surface layer. With the enhanced convergence, a relatively strong updraft center of 0.6 m s−1 formed near 0800 UTC. Directly above the updraft center of the near-surface layer, there was another updraft center at 700–300 hPa created by the joint action of the climbing motion and upper-level trough, which made the cloud top reach 300 hPa, forcing the cloud body over Yanqing Station to become significantly higher than that over the plain area. This factor might have been one of the favorable factors for Yanqing Station to become the snowfall center in Beijing.
Mountainous area of northwestern Hebei Province
At 0300 UTC (Fig. 12a), 800-hPa southeasterly winds invaded the mountainous area of northwestern Hebei Province and formed an obvious convergence zone near Yangyuan Station, which corresponded to a vertical velocity center with a vertical velocity of approximately 0.2–0.6 m s−1. Concurrently, 700-hPa southwesterly warm–humid airflow existed above the cold–moist air of the RF. At 0600 UTC (Fig. 12b), both the southeasterly winds of the RF and the wind convergence zone were enhanced at 800 hPa, corresponding to a more significant vertical velocity center near Yangyuan. Simultaneously, the 700-hPa warm–humid southwesterly airflow intensified, especially the 3 g kg−1 specific humidity contour, which moved northward approximately 70 km from 39.7°N at 0300 UTC to 40.5°N, indicating that the snowfall area was supplied with sufficient water vapor.
Convergence (shading; 10−4 s−1), upward vertical velocity (contours with 0.1 interval; m s−1), and wind (black streamlines) at 800 hPa at (a) 0300 UTC and (b) 0600 UTC 4 April 2018. The plots are superimposed with the specific humidity (green contours; g kg−1) and wind at 700 hPa (blue streamlines). The black circles in (a) and (b) indicate the area around Yangyuan Station.
The height–time profile at Yangyuan Station (Fig. 13) did not display an obvious wedge-shaped feature of ascending warm air. The southeasterly winds acted at 900–800 hPa from 0300 to 1200 UTC, corresponding to an area of 60%–80% relative humidity and a significant vertical velocity center with a maximum speed of 0.6 m s−1 and forming a cold–humid cushion. Above the updraft center of the near-surface layer, there was a middle–high-level updraft center, which made the cloud top over Yangyuan Station exceed 300 hPa.
As in Fig. 11, but for Yangyuan Station.
Yangyuan Station in Hebei Province is approximately 240 km west of Beijing Station, and the altitude is approximately 0.9 km higher. However, under the action of synoptic-scale circulation, the water vapor transport conditions and thermal dynamical uplift mechanism at Yangyuan were better than those of the Beijing plain area, which was conducive to Yangyuan Station becoming the precipitation center of this heavy snow event over the Beijing–Hebei region.
Diabatic cooling of cloud hydrometeors
In previous studies of precipitation phase transitions, conventional observation data or NCEP data were usually employed, and it was impossible to study diabatic processes such as the sublimation, melting, and evaporative cooling of cloud hydrometeor particles as they fell to the ground. However, our numerical simulation data provide an opportunity to further study the influence of diabatic cooling on the RRTS phase transition.
Cloud hydrometeors and temperature changes in the middle and lower troposphere
The cloud microphysical characteristics and atmospheric temperature from 700 hPa to the ground have a direct effect on the precipitation phase. If clouds are dominated by snow particles, when the surface temperature is 0–1.5°C, a mixture of rain and snow will occur; in contrast, if clouds are dominated by raindrops when the surface temperature is 0–1.5°C, rain will occur (Yuter et al., 2006). The 700-hPa level and the profile along 40°N were selected to investigate the influence of cloud hydrometeors on the RRTS phase transformation.
At 0300 UTC (Fig. 14a), the Beijing area and northwestern Hebei Province were occupied by temperature contours ranging from −2 to −4°C. Snow particles fell in a small area near the western boundary of Hebei Province. As indicated by the longitude–height cross-section (Fig. 14b), near the 700-hPa layer, a shallow warm layer existed at 114°–118°E due to the southwestward warm advection at 700 hPa. Simultaneously, in the plain area (116°–119°E), the temperature at 925 hPa was 0°C, while that at the surface was approximately 4°C; in the mountainous areas (114°–116°E), the surface layer was located at approximately 900 hPa, and the surface temperature was also approximately 4°C. At 0600 UTC, the cloud area at 700 hPa (Fig. 14c) intruded the vicinity of Yangyuan and southwestern Beijing. The area north of 39.1°N was dominated by snow particles with temperatures of approximately −4 to −6°C, and the area south of 39.1°N was dominated by raindrops with temperatures of approximately −2 to −4°C. As revealed by the longitude–height cross-section (Fig. 14d), the cloud area had not yet arrived over the plains by this time. Due to the intensity of solar radiation in late spring, the surface–900-hPa layer temperature at 116°–119°E increased by approximately 2°C, thereby exceeding 6°C. In the mountainous area (114°–116°E), the 900–600-hPa layer changed from a warm layer to a cold layer (with a minimum temperature of −8°C) due to the falling drag of snowfall particles, and the surface temperature dropped to approximately 0°C, which corresponded to the time when snowfall occurred. Unlike the upright cloud body near 114°N, the eastern boundary of the cloud body near 115.5°N sloped eastward, which corresponded to a significant gradient zone of dropping temperatures due to the sublimation-induced cooling of snow particles falling through the dry air layer.
Simulated cloud hydrometeor snow particles (shading; 10−1 g kg−1) and temperature (red contours; °C) at 700 hPa (left) and the longitude–height cross-sections (right) along 40°N at (a, b) 0300 UTC, (c, d) 0600 UTC, (e, f) 0800 UTC, and (g, h) 1000 UTC 4 April. The cloud hydrometeor raindrops (black contours; 10−2 g kg−1) are superimposed onto (a, c, e, g). The thick dashed straight lines in (d, f, h) indicate the sloped boundary of snow particles.
In the subsequent 2-h interval, ending at 0800 UTC (Fig. 14e), at 700 hPa, the cloud area moved northeast to 118°E, and the cloud area north of 39.8°N was dominated by snow particles. The corresponding temperature was −6 to −8°C due to a drop of approximately 2°C compared to the temperature at 0600 UTC. The area south of 39.8°N was dominated by raindrops. In the longitude–height cross-section (Fig. 14f), the sloped boundary of snow particles shifted eastward to approximately 116°E and entered the transition zone between the mountainous area and plain area. The temperatures of the 1000–850-hPa layer near the sloped boundary were −6 to 4°C following a sharp drop of 2°C compared to the temperatures at 0600 UTC due to the sublimation-, melting-, and evaporation-induced cooling of snow particles during snowfall. At 1000 UTC (Fig. 14g) and at 700 hPa, the snow particle cloud area moved eastward to 120°E, while the area characterized by raindrops moved slightly northward. There was a significant cooling zone in the area 39°–41°N, 118°–120°E in front of the eastward-moving cloud area with temperatures approximately 2–4°C lower than those at 0800 UTC. In the longitude–height cross-section (Fig. 14h), the sloped boundary of snow particles continued to move eastward near 117°E, accompanied by a drastic gradient zone of decreasing temperatures. The surface temperature of the plain area (116°–117.5°E) west of the sloped boundary dropped near 0°C, which is consistent with the RRTS transition occurring from 0800 to 1000 UTC.
Influence of the diabatic cooling of snow particles on the phase transition
The abovementioned analyses reveal significant differences in the temperature and humidity characteristics of the RFs that affected the plain area of Beijing, mountainous area of northwestern Beijing, and mountainous area of northwestern Hebei Province. Coupled with the influences of their different terrain altitudes, the diabatic cooling caused by snowfall might vary among these three regions. Therefore, the time–altitude profiles of the snow particles and temperature at the representative stations were utilized to further investigate the influence of diabatic cooling on the RRTS phase transition.
At Beijing Station, representing the plain area (Fig. 15a), the eastern boundary of snow particles sloped eastward, and the temperature near the sloped boundary dropped significantly due to the diabatic cooling caused by the sublimation, melting, and evaporation of snow particles during snowfall. The 1000–900-hPa layer exhibited more melting (≥ 0°C) with a relative humidity of approximately 20%, which was conducive to melting- and evaporation-induced cooling, so there was a drastic temperature decrease of 4°C from 0600 to 0800 UTC in this melting layer. This simulated temperature decrease was consistent with the observed sharp temperature decrease in the stage near the beginning of snowfall at Beijing Station. From 0900 to 1500 UTC, the stage of persistent snowfall, since the surface–500-hPa layer was quasi-saturated and the surface temperature was near 0°C, the diabatic cooling effect basically dissipated, and the temperature changed only marginally.
Height–time cross-sections of cloud hydrometeor snow particles (shading; 10−1 g kg−1) and temperature (contours; °C) at representative stations on 4 April 2018. (a) Beijing Station, (b) Yanqing Station, and (c) Yangyuan Station. A gray profile signifying the terrain and dashed vertical lines indicating the onset of precipitation are superimposed onto the plots.
The temperature drop near the sloped boundary at Yanqing Station (Fig. 15b) was similar to that at Beijing Station. However, Yanqing Station is situated at a higher altitude (approximately 500 m); therefore, the thickness of the surface–900-hPa melting layer was only half that over the plain area, and thus, the diabatic cooling produced by the melting layer was weaker. However, because Yanqing Station sits at a higher altitude and its latitude is approximately 0.5° farther north than Beijing Station, the −8°C isotherm of Yanqing Station extended downward to 850 hPa at 0800 UTC, and the distance from the surface to the −8°C isotherm was approximately 1000 m less than that at Beijing Station. Therefore, when snow particles fell from 850 hPa and passed through a relatively thin melting layer, the snow particles could not melt before reaching the surface, which caused the surface temperature to drop rapidly to nearly 0°C between 0800 and 0900 UTC, which is consistent with the observations.
The RF at Yangyuan Station (Fig. 15c) in the mountainous area of northwestern Hebei Province was shallow with cold and moist characteristics, so the sublimation- and melting-induced cooling of snow particles was relatively weak. An obvious cold zone was located from 850 to 650 hPa from 0400 to 1300 UTC, with the cold center temperature reaching −6°C and the −4°C isotherm extending downward to 850 hPa. This cold zone corresponded to a vertically descending motion (Fig. 13). Simultaneously, the relative humidity of the cold zone was quasi-saturated (> 90%), so the cold zone was attributable primarily to snow particles dragging the cold air from 600 hPa downward to 850 hPa. Due to the relatively high altitude of Yangyuan Station (approximately 900 hPa), the near-surface, temperature change trend resembled that at an altitude of 850 hPa. The surface temperature gradually dropped from approximately 4°C at 0300 UTC to −2°C, which is consistent with the actual situation. Yangyuan Station’s rapid temperature decrease was directly related to the downward entrainment of middle-level, cold air by snow particles, which obviously differed from the rapid cooling mechanism in both the plain and the mountainous areas of Beijing.
In addition, compared to the situation at Beijing Station, the areas with many snow particles at both Yanqing Station and Yangyuan Station were located from 650 to 350 hPa, corresponding to temperature zones of −8 to −18°C. Because −12 to −18°C is considered the most suitable temperature for the formation of dendritic snowflakes (Auer and White, 1982), these temperatures might be an important reason for the larger amount of snowfall and larger snow depth at Yangyuan and Yanqing stations.
Influence of urbanization on the phase transition
Because of the intensive assimilation of automatic station data and the use of static surface classification data in the Beijing–Hebei region in the simulation, the influence of urbanization on the phase transition could be further investigated.
As shown in the height–time profile of the relative humidity at Sanhe Station in Hebei Province (Fig. 16a), from 0300 to 0900 UTC, the surface–850-hPa layer was dry with relative humidity ≤ 30%, accompanied by a 4–6 m s−1 easterly wind. From 0300 to 0700 UTC, the 800–600-hPa layer was the middle–low-level dry area with relative humidity ≤ 30%, and the wedge-shaped feature of the cloud was obvious. A 90%–40% relative humidity gradient belt was detected at 0300–1000 UTC in the surface–500-hPa layer, forming a significant sloped dry–moist boundary. These features are consistent with those observed at Beijing Station. The simulated snowfall at Sanhe Station occurred at approximately 1000 UTC, approximately 2 h later than that at Beijing Station, which reasonably reflects the characteristics of the westward RF intrusion and the gradual eastward movement of the precipitation system in the plain area.
Height–time cross-section at Sanhe Station of the (a) relative humidity (dashed contours; %) and wind (barbs, full barb is 4 m s−1) and (b) temperature (solid contours; °C). Cloud hydrometeor snow particles (shading; 10−1 g kg−1) are superimposed onto the plots, and the dashed vertical lines indicate the onset of precipitation.
As indicated by the temperature height–time profile at Sanhe Station (Fig. 16b), the overall characteristics of the temporal and spatial evolution of temperature were similar to those at Beijing Station. The 800–600-hPa layer behaved as a warm layer (−4 to −6°C) at 0300–0700 UTC but changed to a cold layer (−6 to −8°C) at 0800–1400 UTC. A drastic belt of decreasing temperatures accompanied the sloped dry–moist boundary. Between 1000 and 1500 UTC, the surface temperature remained at approximately 0°C. However, a significant difference in the temperature change rate between Sanhe Station and Beijing Station, especially the drop rate from 2 to 0°C, played a key role in the phase transition from rain to snow. At Sanhe Station, the surface temperature dropped from 6 to 0°C within 0900–1000 UTC; during this period, it took approximately 20 min for the temperature to drop from 2 to 0°C. Comparatively, at Beijing Station, the surface temperature dropped from 4 to 0°C within 0720–0930 UTC, and it took approximately 90 min for the temperature to drop from 2 to 0°C. Therefore, the development of urbanization has extended the duration of mixed rain and snow at Beijing Station, which reasonably reflects the observations. Further comparison shows that it took approximately 50 min for the temperature at Yanqing Station in Beijing to drop from 2 to 0°C and 1 h for the temperature at Yangyuan Station in Hebei Province’s mountainous area to drop from 2 to 0°C. The temperature drops at both of these stations were faster than those at Beijing Station.