Atypical Occlusion Process Caused by the Merger of a Sea-breeze Front and Gust Front

A sea breeze is a type of circulation driven directly by the land-sea thermal contrast, which flows inland at the coastline on fine days. Its leading edge is sometimes in the form of a sudden squall, resembling a minor cold front, which is called a sea-breeze front (SBF) (Simpson, 1994).

(Miller et al., 2003) reviewed and summarized some general features related to sea breezes, including their forcing, structure, life cycle, and forecasting. In particular, sea breezes can interact with external meteorological phenomena, which has long been studied. For example, (Byers and Rodebush, 1948) found that the convergence of sea breezes from the two coasts of the Florida peninsula are important in producing thunderstorms over central Florida, i.e., convection initiation. (Pielke, 1974) further concluded, via numerical simulation, that sea breeze interactions are the dominant form of control over summertime convection in Florida. (Fovell, 2005) studied the interaction between SBFs and horizontal convective rolls. (Lu et al., 2012) investigated 50 SBF cases in Bohai Bay (in China) from 2004 to 2009. Their results showed that ∼44% of SBFs merged with external meteorological phenomena. The interactions between SBFs and gust fronts (GFs) (i.e., thunderstorm outflow boundaries) has also been studied extensively (Nicholls et al., 1991; Fankhauser et al., 1995; Kingsmill, 1995; Carbone et al., 2000; Kingsmill and Crook, 2003; Lu et al., 2012; Liang et al., 2013; Wang et al., 2014). As we show in the present study, the merger of SBFs with GFs can initiate local convection and form an occlusion-like structure, which usually occurs within extratropical cyclones. However, no previous study has reported this interesting structure.

Figure 1.  Left: The area where the severe convective weather occurred. The black dot indicates the location of the sounding derived from the NCEP GFS data; the black triangle indicates the location of Tianjin radar station (TJR). Right: Geographic location of the domain in the left panel.

The occlusion process is often observed in association with midlatitude cyclones. In the early 20th century, Norwegian scientists drew upon previous research and a mesoscale observational network to create a conceptual "ideal cyclone model" (also known as the "classical Norwegian cyclone model") for the structure and evolution of midlatitude cyclones (e.g., Bjerknes, 1919; Bjerknes and Solberg, 1921, 1922). In this model, the cold front that rotates around the center of the low pressure moves fast and finally catches up with a slow-moving warm front. As a result, the intervening warm-sector air is detached from the surface, leaving a wedge of secluded warm air aloft. An occluded front is formed at the interface between the two cold air masses that were earlier behind the cold front and ahead of the warm front. This kind of occlusion is called a "catch-up" process. However, (Schultz and Vaughan, 2011) argued that the catch-up process does not provide any explanation for occlusion; rather, it is simply the result of occlusion. Instead, occluded fronts form through a "wrap-up" process. In this process, the warm-sector air is separated from the low center through the wrapping up of the thermal wave (or baroclinic zone around the cyclone). It then lengthens due to the flow deformation and rotation around the cyclone.

While past studies have provided many insights into synoptic-scale occluded fronts and the associated occlusion process, there is little understanding of the occlusion-like structure (called an "atypical occlusion" in this paper) associated with an SBF. In light of this, an atypical occlusion due to the merger of an SBF with a GF is investigated in this study. The main purpose of the paper is to propose and describe the basic features of the atypical occlusion process. The convection initiation caused by the merger of the SBF and GF will be reported in a companion paper.

The remainder of the present paper is organized as follows: Section 2 briefly introduces the dataset and provides an overview of the environmental conditions and system evolution of the atypical occlusion process. Section 3 described the setup and verification of the numerical experiment. The simulated occlusion process and its influence on the development of local convective weather are reported in section 4 and, finally, a summary and further discussion are provided in section 5.

A severe convective weather event (e.g., thunderstorm, short-term heavy rainfall, high wind and hail) occurred in Tianjin and northern Hebei Province in North China (Fig. 1) from 0600-1200 UTC 14 July 2011. This convective weather event produced high winds of up to 26.8 m s^-1, heavy rainfall of up to 33.6 mm h^-1, and hail as large as 15 mm in diameter. In addition, more than 20 acres of greenhouses and 3500 acres of corn were damaged (Wang and Fan, 2012).

Figure 2.  Geopotential height (blue solid lines; units: gpm), temperature (red dashed lines; units: ^°C), and wind fields (half barbs, full barbs and flags represent 2, 4 and 20 m s^-1, respectively) at (a) 500 hPa and (b) 850 hPa. Panel (c) is a skew T-logp diagram at (39.5^°N, 117^°E) at 0600 UTC 14 July 2011 derived from NCEP GFS data. The shaded area in (b) represents the equivalent potential temperature (units: K) in excess of 340 K. The small rectangle areas in (a, b) denote the domain shown in the left-hand panel of Fig. 1. The black dotted lines in (a, b) represent troughs. The location of the sounding is marked with a dot in (a, b).

Figure 3.  (a-c) Composite reflectivity (color scale; units: dBZ) observed at Tianjin Radar Station (black triangle) and surface wind (full and half barbs represent 4 and 2 m s^-1, respectively); (d-f) MTSAT visible satellite images (shading denotes albedo; units: %); (g-i) sea level pressure (color scale; units: hPa); and (j-l) surface temperature (color scale; units: ^°C). Panels (a, d, g, j) are at 0600 UTC; (b, e, h, k) are at 0700 UTC; and (c, f, i, l) are at 0800 UTC 14 July 2011. The wind, sea level pressure and surface temperature are from ground-based automatic stations. Cressman interpolation is used to obtain the sea level pressure and surface temperature fields. The locations of the GF (indicated by the northwestern black dashed line) and SBF (southeastern black dashed line) are also shown.

Figure 4.  (Continued.)

Figure 5.  (a) The locations of four ground-based automatic stations (S1, S2, S3, and S4) in the severe convective weather area. (b-e) Observed temperature (black solid lines) and wind (full and half barbs represent 4 and 2 m s^-1, respectively) from 0000 UTC to 1000 UTC 14 July 2011 at the four stations. The black and grey arrows indicate the wind shift and temperature drop induced by the passage of the SBF (at S1 and S2) and GF (at S3 and S4), respectively.

The dataset utilized in this work includes conventional observational data from ground-based automatic stations, Doppler radar, visible satellite images from the geostationary satellite MTSAT, and the gridded 0.5^°× 0.5^° Global Forecast System (GFS) analyses at 6 h intervals produced by the National Centers for Environmental Prediction (NCEP).

At 0600 UTC 14 July 2011, a moderately intense cold low was present over the border of eastern Mongolia and northeastern China at 500 hPa, in association with a trough (i.e., the black dotted line in Fig. 2a). The severe convective weather event occurred in an area just ahead of the southern tip of the trough, where the ambient winds were very weak (≤6 m s^-1). At 850 hPa (Fig. 2b) there were two pressure lows accompanied by two troughs. The severe weather area lay between these two lows, with relatively warm and moist air collocated at this position (i.e., high equivalent potential temperature, θ _e). The environmental flow was also weak at this low level.

A representative sounding derived from the GFS analysis at 0600 UTC exhibited a moderate CAPE of 2080 J kg^-1 (Fig. 2c). The atmosphere was rather dry near the surface, showing a dewpoint depression (T-T _d) of ∼10 K. However, the humidity increased quickly with height, with the atmosphere nearly saturated at around 850 hPa. In general, the lower troposphere was quite moist between ∼900 hPa and 600 hPa, whereas a dry layer was present in the middle troposphere. In addition, due to the weak winds in both the middle and lower troposphere, the environmental vertical wind shear was fairly weak, which was not conducive to the development of intense convection (e.g., Rotunno et al., 1988).

Figure 3 shows the evolution of a GF and an SBF related to the severe convective weather. The slow-moving SBF came from the Bohai Sea, while the GF was generated by the decaying convective storm over north central Beijing (Figs. 3a and d). An area of high pressure and low temperature (Figs. 3g and j) in association with the convective storm was identifiable in the northwestern part of the plotted area. Several small, weak and isolated convective cells were found (Fig. 3a) near the GF, with a weak pressure low ahead of the GF (Fig. 3g). The air mass between the GF and SBF was relatively warm (Fig. 3j).

Over time, the GF and SBF moved towards each other, almost encountering one another at 0700 UTC (Figs. 3b, e, h and k). At this time, several intense convective cells developed between, and in the vicinity of, the GF and SBF (Figs. 3b and e); the relatively warm air mass between the two fronts was narrowed (Fig. 3k). By 0800 UTC, the GF had merged with the SBF and the convective cells between them developed significantly in both size and intensity (Figs. 3c and f), attendant with several cold pools (Fig. 3l). The thunderstorm high only appeared at the location of the largest and most intense convective cell near the triple point of the merger of the GF and SBF (Fig. 3i).

The temperature gradients near the GF and SBF were quite weak (Figs. 3g-i), probably owing to the use of Cressman interpolation. To better reveal the features of the GF and SBF, the temperature and wind observations at four ground-based automatic stations (S1, S2, S3, and S4; see Fig. 4a) are examined. As the SBF passed station S1 (Fig. 4b), cyclonic wind shift (from southwesterly to southeasterly) was observed between 0400 and 0500 UTC, whereas temperature dropped in the next hour by about 1.5 K. This kind of time lag was commonly observed and studied in depth in (Schultz, 2005) by reviewing a number of different mechanisms for prefrontal troughs and wind shifts. A similar cyclonic wind shift and temperature drop were observed at station S2 between 0600 and 0700 UTC (Fig. 4c). At stations S3 and S4 (Figs. 4d and e), there was also a dramatic anticyclonic wind shift during the passage of the GF, between 0500 and 0600 UTC and between 0600 and 0700 UTC, respectively. Furthermore, a similar time lag in temperature drop was found at station S3. In general, both the GF and SBF possessed some characteristics that were similar to a cold front.

At about 0800 UTC, the GF, moving from inland towards the Bohai Sea, merged with an SBF towards the inland region of Tianjin, where severe convective weather occurred. However, the observations of this merger process and accompanying severe convective weather are insufficient for a detailed analysis and exploration of their features. Thus, we resorted to numerical simulation, the results of which we report in the following part of the paper.

The simulation was conducted using the Weather Research and Forecasting (WRF) model, version 3.5, with the Advanced Research dynamics core (Skamarock et al., 2008). Quadruple-level, two-way nested domains were used, with horizontal resolutions of 36 km, 12 km, 4 km, and 1.333 km, respectively. The time step for the outermost domain was 180 s. The innermost domain (D4) focused on the convective weather of interest over Tianjin and northern Hebei Province. There were 35 vertical levels from the surface to the model top at 50 hPa, with 15 levels below 2 km. For all four model domains, the model physics adopted the YSU (Yonsei University) planetary boundary layer scheme (Hong et al., 2006), the WDM6 (WRF Double Moment 6-class) microphysics scheme (Lim et al., 2010), the Rapid Radiative Transfer Model (RRTM) longwave radiation scheme (Mlawer et al., 1997), the Dudhia shortwave radiation scheme (Dudhia, 1989), the Unified Noah land surface model (Tewari et al., 2004), and the MM5 (Fifth-Generation Penn State/National Center for Atmospheric Research Mesoscale Model) similarity surface layer scheme (Paulson, 1970). The Kain-Fritsch scheme (Krain, 2004) was used in domains D1 and D2 for cumulus parameterization, but was turned off in domains D3 and D4.

Figure 6.  Simulated wind field (full and half barbs represent 4 and 2 m s^-1, respectively) at 10 m height and composite reflectivity (color scale; units: dBZ) at (a) 0730 and (b) 1000 UTC 14 July 2011. (c, d) As in (a, b) but for temperature at 2 m height (color scale; units: ^°C). Black dashed lines represent the GF and SBF.

Figure 7.  (a) Modeled 6 h accumulated precipitation (units: mm) from 0600 to 1200 UTC 14 July 2011. (b) Observed 6 h accumulated precipitation (units: mm) from 0400 to 1000 UTC 14 July 2011.

Figure 8.  Simulated equivalent potential temperature (color scale; units: K) at the modeled first level and wind barbs (full and half barbs represent 4 and 2 m s^-1, respectively) at 10 m height at (a) 0800, (b) 0900, and (c) 0950 UTC 14 July 2011. (d-f) Equivalent potential temperature (color scale; units: K) and 10 m height wind in the vertical cross section along line AB in the left panel of Fig. 1; the 305 K potential temperature isolines are overlaid (black solid lines). The purple line in (f) represents the occluded front.

Figure 9.  Conceptual model for the "merger" occlusion process. Panels (a-c) show the plane view of the three major periods of the process: (a) formation of the cold GF and SBF at an appropriate distance; (b) the two fronts move towards each other, with the warm air between them squeezed; and (c) the warm air is separated from the surface and forced aloft, on top of the SBF. Panels (d-f) are the same as (a-c) but in the vertical cross sections along the line AB in the left panel of Fig. 8.

Figure 10.  Simulated composite reflectivity (color scale; units: dBZ), 305 K potential temperature isolines (black solid lines), and wind barbs (full and half barbs represent 4 and 2 m s^-1, respectively) at 10 m height at (a) 0800, (b) 0850, (c) 0910, and (d) 1000 UTC 14 July 2011. The black dashed ellipses indicate the parent convective system of the GF; the two red rectangles in (a) indicate the isolated convective cells near and/or behind the GF; red ellipses in (b-d) indicate relatively intense convective system developed in the merger area; the small red rectangle in (d) indicates convective cells generated behind major convection.

Figure 11.  Simulated vertical velocity (color scale; units: m s^-1) in the vertical cross section along line AB in the left panel of Fig. 1 at (a) 0800, (b) 0850 (c) 0910, and (d) 1000 UTC 14 July 2011; the 305 K potential temperature isolines are overlaid (black solid lines). Panels (e-h) are the same as (a-d) but for the simulated reflectivity (color scale; units: dBZ). The wind at 10 m height (full and half barbs represent 4 and 2 m s^-1, respectively) along line AB in the left panel of Fig. 1 is also shown in (a-d).

All domains were initialized at 0000 UTC 14 July 2011, and integrated for 18 h. The model's initial and lateral boundary conditions were created using the six-hourly 0.5^°× 0.5^° GFS data.

In the following, the composite reflectivity (i.e., column maximum), temperature at 2 m height, wind field at 10 m height and accumulated precipitation obtained from the WRF simulation are compared with observations, with the goal of establishing the credibility of the numerical simulation.

Figures 5a and b show the simulated composite reflectivity and wind at 10 m height. Before the merger of the GF and SBF, it can be seen that the simulated northwestern convective system (Fig. 5a) developed about 1.5 h later than its observational counterpart, and was biased by about 0.8^° (i.e., about 90 km) to the south of the observation (Figs. 3a and d). The convection within the simulated convective system was stronger than observed, with the cold GF closer to the system. The simulated SBF showed a farther inland penetration than observed, with its northern half orientated in an almost north-south direction. The two simulated fronts (i.e., the GF and SBF) also moved towards each other and merged together at around 1000 UTC (Fig. 5b), i.e., about 2 h later than observed (Figs. 3c and f). Once the two fronts had merged, relatively intense convection developed along the interface of the two fronts. Seen from the simulated 10 m height temperature (Fig. 5c), the air mass between the two fronts was relatively warm. After the merger of the two fronts (Fig. 5d), a cold pool formed as a result of the intense convection, which was in agreement with observation.

Figure 6 compares the simulated 6 h accumulated precipitation with observations. Due to the time delay in the development of the convective system, the precipitation in the simulation also occurred about 2 h later than observed. The overall pattern of the 6 h accumulated precipitation was also biased by about 0.8^° to the south. However, the intensity of the main heavy precipitation (indicated by the red circle) was captured well, albeit it covered a smaller area. Another heavy precipitation event (indicated by the red rectangle), located to the northeast of the main precipitation, was also simulated fairly well, in spite of the lack of precipitation on its eastern part. In contrast, the numerical simulation showed an overestimation of precipitation to the northwest of the main precipitation (indicated by the red ellipse), and produced some weak precipitation to its west that was not observed.

In general, in spite of some timing and positioning biases, the overall pattern of the two cold-type fronts and their movements (i.e., moving towards each other) was reproduced well by the WRF simulation. The numerical simulation also captured the merger process of the two fronts, with the intense convection that developed in the merger area and the attendant heavy precipitation simulated well.

In the following, the simulated surface wind fields and thermal environments are studied to gain more insight into the features of the two fronts and their interaction.

Figure 7 shows the simulated equivalent potential temperature and horizontal wind at 10 m height. At 0800 UTC (Figs. 7a and d), the northwestern GF (i.e., the cold outflow boundary of a convective system) exhibited a greater horizontal gradient of equivalent potential temperature and stronger horizontal wind shear, as compared to that of the SBF. The mean height of the SBF was about 1.2 km, higher than the 0.9 km mean depth of the GF. [The 305 K potential temperature isoline was chosen to outline the two fronts according to the potential temperature and wind fields in vertical cross sections (not shown).] The two fronts moved towards each other, with the GF moving at a faster speed. At 0900 UTC (Figs. 7b and e), the two fronts almost merged, with the warm and moist (i.e., high-θ _e) air between them squeezed and narrowed. Meanwhile, a cold (i.e., low-θ _e) downdraft occurred on top of the GF's head and separated the warm air mass into two parts. This cold downdraft was generated by a convective cell, which is analyzed later in section 4.2.

By 0950 UTC (Figs. 7c and f) the cold downdraft had intensified further, producing divergent outflow near the surface, as evidenced by the 10 m height wind (Fig. 7f). At this time, the high-θ _e air was squeezed further and detached from the surface, exhibiting a structure very similar to the occlusion structure in the classical Norwegian cyclone model. However, this kind of occlusion process occurs in the absence of an extratropical cyclone and there is only one type of front; namely, a cold front. Moreover, the spatiotemporal scales of the occluded front are much smaller in comparison with traditional occluded fronts (i.e., mesoscale versus synoptic scale). It is thus called an atypical occlusion process. Unlike the well-known "catch-up" and "wrap-up" occlusion processes, this atypical occluded front is generated due to the merger of two fronts.

Figure 8 illustrates a conceptual model for the "merger" occlusion process. In the early stage (Figs. 8a and d), there is an SBF moving inland. Meanwhile, a GF located inland moves towards the SBF at a relatively high speed, with warm air between them. Later, as the two fronts become increasingly close, the warm air between them tends to be squeezed and detached from the surface (Figs. 8b and e). In the final stage (Figs. 8c and f), the two fronts merge, and the warm air is completely separated from the surface and forced aloft, above the SBF. An occluded front is then formed at the interface of the two fronts.

As shown in previous sections, intense convection developed in the region where the two fronts merged, accompanied by heavy precipitation. Figure 9 shows the temporal evolution of the simulated composite reflectivity, which is indicative of the development of convection. At 0800 UTC (Fig. 9a), the GF moved about 40 km away from the parent convective system (see the black dashed ellipse), which decayed and disappeared in the next two and a half hours. Meanwhile, a number of isolated convective cells (indicated by the two red rectangles) were found near and/or behind the GF. However, there was no evident reflectivity in association with the SBF. Later, as the northern SBF merged with the GF (Figs. 9b and c), an intense quasi-linear convective system (QLCS) developed in the merger area (indicated by the red ellipses). At 1000 UTC (Fig. 9d), the convective cells at the northern and southern ends of the QLCS (see the two red ellipses) were enhanced further, with several convective cells generated behind (indicated by the small red rectangle).

Figure 10 displays the temporal evolution of the simulated vertical velocity and reflectivity in the vertical plane approximately normal to the occluded front (see line AB in Figs. 7 and 9). At 0800 UTC, the two fronts were far away from one another. Ascending motion was apparent mainly above the GF, with a localized updraft (about 2-3 m s^-1) found in the low level, on top of the GF leading edge (Fig. 10a). In the meantime, weak reflectivity of less than 25 dBZ was found above the GF at a height of about 4-8 km (Fig. 10e). As the two fronts moved towards each other with time, the initially weak and shallow convective updraft on the leading edge of the GF developed into deep convection, accompanied by a compensating downdraft (Figs. 10b and c), which separated the warm air between the SBF and GF into two parts (Figs. 7e and f). Low-θ _e air was found with this downdraft, likely due to evaporative cooling and/or melting of frozen hydrometeors. Moreover, the downdraft produced a divergent outflow near the surface (see the blue arrow in Fig. 10c). The reflectivity also increased. For example, at 0850 UTC, the peak reflectively was greater than 45 dBZ and occurred at 4-5 km height, with reflectively >35 dBZ extending from the surface to above 9 km height (Fig. 10f). By 1000 UTC when the two fronts had already merged, the previous leading-edge convective cell almost died out (Figs. 10d and h). However, a new cell was generated, with intense updraft (>12 m s^-1) and reflectivity (>50 dBZ) at about 7 km height. This convective cell was initiated by the convergence of the westward-moving cold outflow with the eastward-moving GF flow (see the red arrow in Fig. 10c and red ellipse in Fig. 10d).

As noted in section 2.1, the thermodynamic and kinetic environmental conditions were not particularly favorable for the development of strong convection. However, severe convective storms did occur in the merger area near Tianjin. Therefore, the merger of two fronts can help enhance local convection.

The present reported study investigated an atypical occlusion process that occurred in North China on 14 July 2011. Observations of Doppler radar and ground-based station observations showed that this process was caused by the merger of an SBF with a GF. The SBF was generated near the coastline of the Bohai Sea and then moved inland. In contrast, the GF was generated by a mesoscale convective storm located over north central Beijing and moved southeastward towards the Bohai Sea. As the two fronts encountered each other, an occlusion-like structure was formed, with severe convective weather occurring in the merger area near Tianjin.

This atypical occlusion process was reproduced reasonably well by a real-data, high-resolution WRF model simulation. The model domain had four levels that were two-way nested, with the grid spacing of the innermost domain up to 1.333 km. The evolution and structure of the atypical occlusion process were examined according to the numerical results. Unlike the traditional occlusion process, i.e., the "catch-up" or "wrap-up" process (Schultz and Vaughan, 2011), the present case featured much smaller temporal and spatial scales, and occurred beyond (i.e., in the absence of) an extratropical cyclone. Instead, it was caused by the merger of two cold-type fronts, one of which was an SBF. During their movement towards each other, the warm air between the two cold-type fronts was squeezed and detached from the surface. As the two fronts merged together, the warm air was eventually forced aloft, above the sea breeze front, forming an occlusion-like structure. Consequently, this kind of occlusion process was named a "merger occlusion".

Moreover, the results also showed that the merger of the two fronts was able to trigger or help enhance the local convection in the merger area, even in an environment of moderate thermodynamic instability and weak vertical wind shear. Therefore, this atypical occlusion process deserves consideration in nowcasting and short-range weather forecasts.

The atypical occlusion process and conceptual model of the merger occlusion process proposed in this work were based on analysis of one case. Whether or not an occluded frontal structure forms for every such merger case of an SBF is not clear. Moreover, the mechanism responsible for the development of strong convection during the merger occlusion process is not well understood. For example, is it the thermodynamic or kinetic conditions that are significantly changed as the two fronts merge? Lastly, besides SBFs, can the merger occlusion process occur for other kinds of front? As such, further studies are needed to help improve/generalize the findings of this work.