The interaction of an urban heat island with a sea breeze front during moist convection over <scp>Tianjin, China</scp>

Abstract

Graphical abstract(a) Full map of China including the location of Tianjin; (b) the terrain height (m) of Tianjin with the urban boundary (red line); (c) photograph of Tianjin city. This study analysed a moist convection process which took place during the daytime on 22 July 2018 over Tianjin (China), which was caused by an interaction between an urban heat island (UHI) and the sea breeze front (SBF). The Weather Research and Forecasting (WRF) model was used for simulation and two sensitivity experiments (SEs) were designed to investigate the different effects of the UHI and the SBF. The results showed that the simulation can represent the convection well. During the event, the easterly wind behind the SBF transported a cool, moist air mass from the sea (which provided the moisture supply) over the dry, warm urban centre, which provided triggering mechanism for the moist convection. The UHI was enhanced at the beginning of the process, providing turbulence mixing through the sensible heat flux from the urban surface but weakened after the precipitation occurred. The UHI and urban surface obstructed the westward movement of the SBF and maintained it near the urban centre, providing moisture and energy for the moist convection. Tianjin is a modern industrial city in North China. In recent years, with the development of the economy, urbanisation has been expanding. Temperatures over urban areas have typically been found to be higher than that over surrounding rural areas, a phenomenon which is commonly known as the urban heat island (UHI) (Oke, 1995; Rizwan et al., 2008; Leroyer et al., 2010; Li et al., 2013; Yang and Li, 2015). Many studies (Rozoff et al., 2003; Miao et al., 2011; Wang et al., 2013) have found that a greater sensible heat flux from urban surfaces produces a stronger upward motion and makes atmospheric instability stronger, increasing the chance of convective precipitation. As Tianjin lies beside the Bohai Sea (Figure 1), the sea breeze front (SBF) is another important factor influencing the weather in Tianjin. The land–sea thermal contrast induces a local-scale pressure gradient force, which consequently drives a shallow layer of marine air inland, and creates a SBF (Miller et al., 2003; Ezber et al., 2015). The low-level atmospheric convergence of the SBF can generate updrafts, which are responsible for the generation of turbulence, convective systems and related precipitation (Atkins and Wakimoto, 1997; Gahmberg et al., 2010; Comin et al., 2015). Studies have also been conducted to investigate the relationship between UHIs and SBFs. The ‘chain flow’, generated by the combination of the sea-breeze–heat-island circulations over coastal urban areas, contributes to some of the summertime rainstorm processes (Wan et al., 2013; Li et al., 2016). The urban environment can have an impact on the evolution of the sea-breeze mesoscale boundary layer (Lemonsu et al., 2006; Lin et al., 2008; Carter et al., 2012). More specifically, sea-breeze penetration speeds decline in proportion to the size and the degree of irregularity of cityscapes and due to intensification from UHIs, which enhance vertical air circulation systems (Khan and Simpson, 2001; Martilli, 2003; Thompson et al., 2007; Dandou et al., 2009). The Weather Research and Forecasting (WRF) model has been commonly used to study the UHI and SBF circulations in the diurnal variations in Tianjin (Miao et al., 2015; Li et al., 2017; Wang et al., 2020). A few studies have shown WRF to successfully simulate the effects of the UHI, which could strongly influence the local circulation of sea–land breeze in low ambient wind speed conditions, including the development, thermal intensity, movement, as well as the convection ahead of a sea breeze circulation (Fovell, 2005; Khan, 2010; Chen et al., 2011). Most previous studies have focused on how UHIs affect the circulation and boundary layer structure of the sea–breeze. This study focuses on the effects of the urban thermal circulations and a sea breeze during a moist convection process in Tianjin. During the daytime from 0300 utc to 0600 utc on 22 July 2018, convection caused by the UHI and the SBF occurred over the urban area of Tianjin. This study analyses the interaction between the UHI and the SBF. The WRF 4.0 mesoscale model was used to simulate the moist convection process during the daytime of 22 July 2018. A bi-directional nesting scheme consisting of five domains was used for simulation (Figure 1(a)). The horizontal resolution of the innermost domain in the model was set to 333m. The parameter settings used in simulation are listed in Table 1. The model was run from 1800 utc on 21 July to 1200 utc on 22 July 2018, the first 9 hours (from 1800 utc on 21 July to 0200 utc on 22 July) were used for the spin-up time, and the analysis was from 0300 utc to 0600 utc. The multi-layer, building environment parameterisation (BEP) of the urban canopy model (UCM) scheme was used in the outer four domains (d01–d04), and the single-layer UCM was used in the innermost domain (d05). In the boundary layer scheme, the outer four domains used the Bougeault and Lacarrere (BouLac) planetary boundary-layer (PBL) scheme and the innermost domain used the large-eddy-simulation (LES) boundary layer scheme. For cumulus convection, the Kain-Fritsch scheme was used in the outer two domains and was closed in the innermost three domains. The forcing wind data came from the National Center for Environmental Prediction (NCEP) 6-hour Final (FNL) reanalysis data with a spatial resolution of 0.25°, whilst the sea surface temperature data came from National Oceanic and Atmospheric Administration (NOAA) daily datasets. To distinguish between the effects of the UHI and the SBF, two sensitivity experiments (SEs) were undertaken for the purposes of comparison. In a control experiment (CE), the model used the IGBP-Modified MODIS 20-category land use categories from the unified 'Noah' land surface model (Figure 1b). In sensitivity experiment 1 (SE1), we completely changed the land use categories of the urban and built-up areas of Tianjin into croplands, to eliminate the effects of the UHI (Figure 1(c)). In sensitivity experiment 2 (SE2), we changed the sea surface of Bohai Sea into croplands to eliminate the SBF effects (Figure 1d). Hourly wind data observed from automatic weather stations, presented in traditional wind barb format, together with the composite reflectivity product calculated from S-band doppler radar at Tanggu station, Tianjin (39.04°N, 117.72°E), are shown in Figure 2. The composite reflectivity product was created by calculating the maximum basic reflectivity factor on a horizontal plane across the whole area. The SBF was determined from the automatic wind stations as the wind shear line between the southeasterly breeze from the sea and northeasterly breeze over the land along the coastline in Tianjin, where there was also a weak composite reflectivity line. The spatial intensity variation of Q 1 can reflect changes in the diabatic heating components during the day (Yanai, 1973). Positive Q 1 values represent the release of diabatic heating into the air, including the sensible heat flux from the surface and latent heat flux from condensation. Negative Q 1 values represent the diabatic heat loss from the atmosphere. Figures 2(a–d) show the composite reflectivity of the radar at Tanggu station, Tianjin (39.04°N, 117.72°E) from 0300 utc to 0600 utc on 22 July, supplemented with the hourly wind observed data from the automatic station. At 0300 utc (Figure 2a), the SBF appeared in the east area of Tianjin along the Bohai Sea coastline. The surface wind at 10m behind the SBF (on the east side) was southeasterly, but it was northeasterly in front of the SBF (on the west side). The southeasterly airflow encouraged moisture transport from the sea into the urban area. In the urban area (39.1°N, 117.25°E), a convergence centre formed in front of the SBF, which provided the conditions for the initiation of the moist convection process. At 0400 utc (Figure 2b), the SBF pushed slightly inland. The intensity of the SBF became enhanced, which is clearly seen in the composite reflectivity images. The convection then developed rapidly over the urban area (39.1°N, 117.25°E), where the maximum composite reflectivity exceeded 50dBZ at 0500 utc (Figure 2c), when the moist convection was at its strongest. The SBF reached the urban centre, and the convection plume stretched from the northeast to the south of Tianjin. At 0600 utc (Figure 2d), the SBF pushed further westward, but the southern part of the SBF broke-up and began to disappear. As a result, the moist convective over in the urban centre began to disappear. Compared with the maximum reflectivity and wind field in the CE (Figures 3a–d), the SBF appeared in the southeast area of Tianjin at 0300 utc (Figure 3a). The wind behind the SBF was southeasterly, and the wind ahead of the SBF was northeasterly. The position of the SBF in this control run was simulated to be more southerly than actually observed at 0300 utc, but the intensity and the wind field were simulated well. At 0400 utc in the CE (Figure 3b), the position of the SBF pushed further inland. An obvious convective centre occurred over the urban area (around 39.25°N, 117.25°E) and the max reflectivity in this area reached 45dBZ. Compared with the observed data at 0400 utc, the position of the moist convective centre was a little further to the north and of weaker intensity. However, the occurrence time of the moist convection and the position of the SBF were consistent with the observed data. At 0500 utc (Figure 3c), the moist convection in the urban centre further developed and the SBF pushed further inland but the intensity of the convection weakened at 0600 utc (Figure 3d) and began to disappear, which was consistent with the observed data. From the comparison between the CE and the observed data, WRF simulated the moist convection process and the westward movement of the SBF well, even though the intensity of the moist convective centre was weaker and located slightly more to the north than in the real case, as indicated by the observed data. From the maximum reflectivity and the wind field in SE1 (Figures 3e–h), the SBF was very evident from 0300 utc to 0600 utc. At 0300 utc in SE1 (Figure 3e), the location of the SBF was more westward compared with that in the CE. This indicated that without the 'resistance effect' of the UHI, the SBF could push further west into the inland area before the moist convection started. At 0400 utc in SE1 (Figure 3f), the convection developed rapidly in the area (39.25°N, 117.1°E) where it used to be the urban, which was an hour earlier than in the CE. The location of the SBF in SE1 was further west and inland than in the CE at 0400 utc. At 0500 utc in SE1 (Figure 3g), the moist convective centre declined rapidly, and the SBF was pushed further west compared with that in the CE in the same time. At 0600 utc in SE1 (Figure 3h), the intensity of the SBF weakened until it disappeared, which only occurred in the south of Tianjin. From the maximum reflectivity and 10m wind field in SE2 (Figures 3i–l), the SBF and the southeasterly wind behind the SBF no longer existed, and as a result the moist convective centre was more dispersed. From 0300 utc to 0400 utc (Figures 3i,j), there was dispersed convective reflectivity, but it was not as organised as it was in the CE and SE1. At 0500 utc in SE2 (Figure 3k), the moist convection strengthened in the urban area from the west to the east, but the structure was also loosely organised. At 0600 utc in SE2 (Figure 3l) the convection decayed away. To analyse the horizontal thermal change, the horizontal spatial distribution of 2 m temperature in the CE and SEs are shown in Figure 4. At 0300 utc in the CE (Figure 4a), the strongest temperature gradient was clearly to be seen in the east of Tianjin along the coastline, where the SBF was located. The temperature over the centre of Tianjin was very high at this time (306–308K). At 0400 utc in the CE (Figure 4b), the high temperature in the centre of Tianjin intensified and the strongest temperature gradients extended westward. This indicated that the UHI effect was enhanced and the SBF stretched westward and was closer to the urban centre. At 0500 utc in the CE (Figure 4c) the temperature in the centre of Tianjin decreased because precipitation occurred. The zone of strongest temperature gradients extended further westward. At 0600 utc in the CE (Figure 3d), the temperature in the urban area rose slightly after the end of the rainfall, and the strong temperature gradient zone had stretched further inland. Compared with the 2m temperature change in SE1 (Figures 4e–h), the high temperature region in the urban centre was cooling down, which indicated that the UHI was eliminated in SE1. Otherwise, the boundary of the SBF had stretched further westward inland without the resistance from the urban area. At 0300 utc in SE1 (Figure 4e), the temperature in the east of the urban area was lower than it was in the CE, which indicated the westward pushing of the SBF without the urban obstruction effect. At 0400 utc in SE1 (Figure 4f), the zone of strongest temperature gradients expanded further westward, and the temperature in the urban area cooled down compared with that in the CE. Negative values of the differences between SE1 and the CE were present from 0500 utc to 0600 utc (Figures 4g,h), but their intensity was weakened. Without the obstruction of the UHI, the SBF could transport more moisture inland; thus, the precipitation intensity in SE1 was stronger than in the CE. From 0300 utc to 0600 utc, the total rainfall in the urban area was 4.7mm in the CE, while it was 7.0mm in SE1, and the rainfall period was mainly concentrated from 0400 utc to 0500 utc. From the 2 m temperature distribution in SE2 (Figures 4i–l), the strongest temperature gradients in the east of Tianjin no longer existed, which was used to represent the SBF in the CE and SE1. The area that used to be Bohai Sea changed to a high temperature area with an obvious positive difference value compared to the CE. The temperature in the urban area in SE2 was also higher than that in the CE, and strengthened from 0300 utc to 0600 utc, which meant that the UHI was enhanced without the SBF effects. Water vapour transport was also an important factor in the convection. Figure 5 shows the relative humidity at 2m changing from 0300 utc to 0600 utc in the CE and SEs. The urban centre had a low relative humidity in the CE, and it was drier from 0300 utc to 0400 utc (Figures 5(a, b)) with a relative humidity of less than 45%. From the urban centre to the sea at the coast, the relative humidity increased sharply giving rise to strong gradient along the coastline. Between 0500 utc and 0600 utc in the CE (Figures 5c,d), during the rainfall process, the relative humidity in the urban area increased slightly. The zone of strong gradients of relative humidity also expanded westward from 0300 utc to 0600 utc, which was consistent with the SBF moving west. When compared with the 2m relative humidity distribution in SE1 (Figures 5e–h), the value of the difference between SE1 and the CE in the urban area was positive, which meant that the area that used to be the urban became moister when changed into croplands. The zone of strongest relative humidity gradients expanded further west in SE1 than in the CE. This further showed that when the urban area was changed into cropland, the moist easterly wind could transport more water vapour inland, as the SBF pushed westward. The analysis above showed that the urban area could not only induce the UHI effect in this process, but could also obstruct the transport of water vapour inland. Compared with the 2m humidity change in SE2 (Figures 5i–l), there was no strong relative humidity gradient in the east of Tianjin and the SBF was no longer recognisable. The dry feature of the urban area in SE2 was more distinct, and became enhanced from 0300 utc to 0600 utc. The air became drier in the urban area due to the loss of the water vapour supply from the SBF. To further analyse the structure of vertical circulations of this event, the vertical cross-sections of the potential temperature from the sea to land, through the moist convection centre, are shown in Figure 6. As can be seen, there was an easterly wind from sea to land at a lower altitudes (below 2km) and a westerly wind from land to sea at a higher altitudes (from 2–4km), which followed the principle of land and sea wind circulation during the daytime. The 305K potential temperature line represented the approximate boundary of the SBF. At 0300 utc in the CE (Figure 6a), the SBF was located near 39.23°N, 116.93°E and there was upward airflow before the front due to the uplift effect. The high potential temperature and the turbulence above the urban area were due to the UHI effects. At 0400 utc in the CE (Figure 6b), the SBF (305K potential temperature line) had a little westward movement. The warm region indicates that the UHI effects and turbulence above the urban centre were enhanced. From 0500 utc to 0600 utc in the CE (Figures 6(c, d)), the SBF stretched more to the west and the warm region in the urban area weakened due to the rainfall. The turbulence above the city area was also weakened and mainly controlled by a compensating downward flow. We further compared the vertical cross-section of potential temperature in SEs. Compared with Figures 6(e–h) in SE1 to the CE, there was no high potential temperature region above the urban area caused by the UHI. The SBF had a distinct westward movement without the obstruction of the urban area from 0300 utc to 0600 utc in SE1. The uplift before the SBF was enhanced at 0400 utc (Figure 6f). The turbulence before the SBF was weakened in SE1. Compared with Figures 6(i–l) in SE2, there was no SBF in SE2. The irregular turbulent motion was enhanced, and the warm region was enhanced above the urban area in SE2 from 0300 utc to 0600 utc. To further analyse the effects of the diabatic heating on the vertical circulation structure, the apparent heat source (Q 1 ) was calculated during the moist convection. From 0300 utc to 0600 utc in the CE (Figures 7a–d), there was a positive value of Q 1 ahead of the SBF because the frontal uplift effects made the wet air condense as it rose, releasing latent heat of condensation. There was also a positive value of Q 1 from the surface to 1km in altitude, which was induced by the surface sensible heating. Some strong positive and negative values of Q 1 existed at higher altitudes (above 1.5km) as a result of the turbulent mixing. This indicated there was not only upward vertical airflow, which induced a positive Q 1 value through the condensation latent heat release, but also downward compensatory airflow, which induced a negative Q 1 value as the precipitation absorbed heat by evaporation. The positive Q 1 was stronger than the negative Q 1 at 0300 utc (Figure 7a), at the beginning of moist convection development, which indicated that the upward movements were dominant at 0300 utc. From 0400 utc to 0600 utc (Figures 7b–d), the value of Q 1 was gradually reduced, which meant that the development of the moist convection was gradually weakening, with less latent heat of condensation released after 0400 utc. For the comparison of the Q 1 between SE1 (Figures 7e–h) and the CE (Figures 7a–d), the intensity of Q 1 was weaker in SE1 at 0300 utc due to the reduction of the surface sensible heat flux and the turbulence above the area which used to be urban. However, at 0400 utc in SE1 (Figure 7f), the turbulence developed in front of the SBF which induced a stronger upward vertical airflow. Between 0500 utc and 0600 utc in SE1 (Figures 7g,h), both the positive and the negative Q 1 were weakened quickly before the SBF, which indicated that without the UHI effects, the moist convection developed and then decayed more rapidly. Compared with SE2 (Figures 7i–l), the SBF no longer existed. The Q 1 values in SE2 were larger for the whole region throughout the entire time period from 0300 utc to 0600 utc, for both the positive and negative values, which indicates the enhancement of turbulent mixing over the whole region, especially over the urban area. A moist convection process caused by the interaction of the urban heat island and the sea-breeze front that occurred over Tianjin, China, from 0300 utc to 0600 utc on 22 July 2018 was analysed. The WRF mesoscale model was used to simulate the process and two Sensitivity Experiements were designed to ­eliminate, in turn, the effects of the urban heat island and the sea-breeze front, for comparative purposes. The results were compared with the observed data from the radar and automatic weather stations, which showed that the location of the moist convective centre and the development of the SBF matched well with the observed data, but the intensity of the moist convection was weaker in the Control Experiement compared with the observations. In conclusion, the sea-breeze had two main effects on the development of the moist convection: the first was providing the moisture supply for the convection, and the second was providing a focus for the uplift of air ahead of the sea-breeze front. The urban heat island obstructed the westward movement of the sea-breeze. In addition, the strong turbulent mixing induced by the UHI was another forcing mechanism for the moist convection. This work was supported by the National Natural Science Foundation of China (No. National Research and of China Tianjin Natural Science Foundation of the China Administration Science and of and Tianjin Science and

Topics & Concepts