Effects of anthropogenic heat due to air-conditioning systems on an extreme high temperature event in Hong Kong

The mesoscale meteorological model has been used to study various meteorological phenomena. The advanced research weather research and forecasting (WRF) model (ARW version 3.6.1) is a non-hydrostatic, compressible model with a mass coordinate system (Skamarock et al 2008). The model was set up with four nested domains with 79 × 79, 79 × 79, 118 × 118 and 142 × 136 grid points possessing spatial resolutions of 13.5, 4.5, 1.5 and 0.5 km, respectively (figure 1(a)). The innermost domain covered Hong Kong entirely (figure 1) and provided sufficient information on UHI effects and sea–land breeze circulation (SLBC). The vertical grid contains 51 terrains following full sigma levels from the ground up to 50 hPa. The planetary boundary layer (PBL) scheme of Bougeault and Lacarrere (1989) was used. The other physical schemes used in this study were the single-moment three-class microphysics scheme (Hong et al 2004), Dudhia shortwave radiation (Dudhia 1989), RRTM long-wave radiation (Mlawer et al 1997), the Noah land surface model (Chen and Dudhia 2001), and Kain–Fritsch (Kain and Fritsch 1990) cumulus schemes for the two outer domains. A 120 h simulation (from 0000 UTC June 23 to 0000 UTC June 28, 2016) was conducted with the initial and boundary conditions from the National Centers for Environmental Prediction (NCEP) Global Forecast System Final Analyses data (FNL) at 1° at 6 h intervals and daily updated sea surface temperatures from NCEP Marine Modeling and Analysis Branch data at 0.5°.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Hong Kong has a large spatial and vertical land surface heterogeneity (figure 1(b)). To better represent the dynamics and thermal interacts in the urban canopy layers, multilayer building effect parameterization (BEP, Martilli et al 2002) and the building energy model (BEM, Salamanca et al 2010) were chosen. The multilayer canopy model, BEP/BEM, with a revised drag coefficient, based on plan area fraction, λ_p, has been tested over Kowloon Peninsula in Hong Kong, and has shown reasonably good results for capturing spatial air temperature distribution, diurnal wind rotation, and stagnation wind phenomena in the lower level convergence zone due to the combination of urban heat island circulation and sea-land breeze circulation (Wang et al 2017). The latest gridded building information and land cover and land use data (figure 1(c)) were used for the entire Hong Kong area for better representation of spatial heterogeneity. The default air-conditioning system in BEM is the wall-type scenario (shown in figure 2) in all urban areas. In this study the wall-type air-conditioning system is assumed to be used only in residential areas and central air-cooled air systems in non-domestic areas according to the real situation in Hong Kong (shown in figure 2). To study the effect of air-conditioning systems in non-domestic areas on the urban climate in the city of Hong Kong, several configurations or scenarios of air-conditioning systems (shown in figure 2) in non-domestic areas were simulated, as summarized in table 1.

The new cooling tower scheme (CPCT case) coupled with the building energy model (BEM) is illustrated. In the counter-flow cooling tower system, the gas phase (cool air) flows upwards and the liquid phase (water) downwards (figure 2) and a large interface exists between the two phases (Milosavljevic and Heikkilä 2001). Heat and mass transfer takes place in the cooling tower system, and the water droplets carry out evaporative cooling with the dry and cool air from the atmosphere (Gutiérrez et al 2015). After the process, the water temperature decreases, but the air increases both the temperature and humidity and is released into the atmosphere; further, make-up water is needed for the cooling tower system to maintain the process.

The heat exchange effectiveness, ξ_CPCT, between the refrigerant and the water in the condenser is the heat that is generated inside a building at each floor, H_i, which is calculated via the BEM model, and the heat that is pumped into the cooling tower system, H_T^CPCT. Assume that no heat is lost in the pipe between the condenser and the cooling tower. Then, the heat exchange effectiveness, ξ_CPCT, can be defined as follows:

$$\begin{equation} {{\rm{\xi }}_{{\rm{CPCT}}}} = \frac{{\mathop \sum \limits_{i = 1}^m {H_i}}}{{H_T^{{\rm{CPCT}}}}} \end{equation} \tag{ 1 }$$

where m represents the total number of floors in the building.

If we assume that the water temperature difference between the inlet and the outlet of the condenser system is constant, then the water volume flow rate is as follows:

$$\begin{equation} {q_w} = \frac{{H_T^{{\rm{CPCT}}}}}{{{\rho _w}{C_w}{\rm{\Delta }}{T_w}}} \end{equation} \tag{ 2 }$$

where ρ_w denotes the water density, C_w represents the water heat capacity, and ΔT_w stands for the water temperature difference between the inlet and the outlet of the condenser and the cooling tower.

In the counter-flow cooling tower system, the air volume flow rate, q_a, is as follows:

$$\begin{equation} {q_a} = \frac{{\nu {q_w}{\rho _w}}}{{{\rho _a}}} \end{equation} \tag{ 3 }$$

where v denotes the ratio of the water mass flow rate and the air flow rate in the contour-flow cooling tower system, which is determined by the inter-facial contact area and the heat-transfer coefficient, and ρ_w represents the air density. Here, assume that the ratio v is 0.7 (Chen et al 2014).

Then, the mass flow rate of the air in the cooling tower system is defined as follows:

$$\begin{equation} {m_a} = {\rho _a}{q_a}. \end{equation} \tag{ 4 }$$

The inlet air in the cooling tower is assumed to be from the atmosphere at the appropriate building height level; thus, the enthalpy of the outlet air, h_ao, from the cooling tower system can be defined as follows:

$$\begin{equation} {h_{{\rm{ao}}}} = \frac{{H_T^{{\rm{CPCT}}}}}{{{m_a}}} + {h_{{\rm{ai}}}} \end{equation} \tag{ 5 }$$

where h_ai denotes the enthalpy of the inlet air and can be defined as follows:

$$\begin{equation} {h_{\rm ai}} = {C_p}{T_{{\rm{ai}}}} + {d_{{\rm{ai}}}}\left( {{C_{{\rm{pw}}}} + L} \right) \end{equation} \tag{ 6 }$$

where C_p denotes the specific heat of air at constant pressure, C_pw represents the specific heat of water at constant pressure, L indicates the latent heat of vaporization, and T_ai and d_ai refer to the air temperature and the water vapor mixing ratio of the inlet air, respectively.

If the outlet air is saturated, then the water vapor mixing ratio of the outlet air from the cooling tower is as follows:

$$\begin{equation} \,e = 6.11\, \times \,{10^{\frac{{7.5\quad \times \,{T_{{\rm{ao}}}}}}{{237.3 + {T_{{\rm{ao}}}}}}}} \end{equation} \tag{ 7 }$$

$$\begin{equation} {d_{\rm ao}} = 0.622 \times \frac{e}{{P - e}} \end{equation} \tag{ 8 }$$

where e denotes the water vapor pressure calculated using the Magnus (1844) empirical formula.

Direct cooling systems (DCS) use sea water to cool the condenser in the buildings (Ho et al 2007). If we assume that the sea water volume flow rate, q_sw, is constant, then the temperature difference between the inlet and outlet of the system is as follows:

$$\begin{equation} {\rm{\Delta }}{T_{{\rm{sw}}}} = \frac{{H_B^{{\rm{DCS}}}}}{{{\rho _{{\rm{sw}}}}{C_{{\rm{sw}}}}{q_{{\rm{sw}}}}}} \end{equation} \tag{ 9 }$$

where ρ_sw denotes the sea water density, C_sw represents the sea water heat capacity. For the DCS systems in this study, the estimation of the temperature rises of the sea water during one day is around 1 × 10⁻⁵ °C, which is ignored in this study.

BEM starts from the assumption that the internal target temperature and humidity levels remain within the comfort range (Salamanca et al 2010) and then estimates the cooling load inside the buildings. The default air-conditioning system in BEM is the wall-type scenario (shown in figure 2), and is assumed to be used in residential areas in this study. The BEM model calculates the sensible heat fluxes, H_i^out, and latent heat fluxes, LE_i^out, for each floor, i, into the atmosphere required to maintain standard indoor comfort conditions (Krpo et al 2010). The details of the sensible and latent heat flux calculation can be found in Salamanca et al (2010). Then, the sensible and latent heat fluxes generated at each floor, i, and released into the atmosphere through the air-conditioning systems (Krpo et al 2010) can be expressed as follows:

$$\begin{equation} {H_i} = \frac{{{\rm{COP}} + 1}}{{{\rm{COP}}}}H_i^{{\rm{out}}} \end{equation} \tag{ 10 }$$

$$\begin{equation} {\rm{L}}{{\rm{E}}_i} = {\rm{LE}}_i^{{\rm{out}}} \end{equation} \tag{ 11 }$$

where COP denotes the coefficient of performance for the air-conditioning systems, which is assumed to be constant for all wall-type systems.

The Baseline case (figure 2) assumes that the systems are settled at the rooftop level. Then, the sensible and latent heat fluxes generated at each floor, i, are summed up and released into the atmosphere only at the rooftop level through the air-conditioning systems (followed by Krpo et al 2010); these fluxes can be expressed as follows:

$$\begin{equation} H_T^{{\rm{Base}}} = \mathop \sum \limits_{i = 1}^m {H_i} \end{equation} \tag{ 12 }$$

$$\begin{equation} {\rm{LE}}_T^{{\rm{Base}}} = \mathop \sum \limits_{i = 1}^m {\rm{L}}{{\rm{E}}_i}. \end{equation} \tag{ 13 }$$

The sensible or latent heat fluxes at the rooftop are given by the sum of turbulent fluxes H_tur or LE_tur, from the roof surfaces, which is computed using the Louis formulation (Martilli et al 2002) and H_T^Base or LE_T^Base, followed by the method proposed by Krpo et al (2010).

For the DCS case (figure 2), the sensible heat fluxes generated at each floor, i, are summed up and go back to the sea water through DCS systems, which is assumed to be located in the equipment unit in the building (ASHRAE Handbook 2008); this can be expressed as follows:

$$\begin{equation} H_B^{{\rm{DCS}}} = \mathop \sum \limits_{i = 1}^m {H_i} \end{equation} \tag{ 14 }$$

$$\begin{equation} {\rm{LE}}_B^{{\rm{DCS}}} = \mathop \sum \limits_{i = 1}^m {\rm{L}}{{\rm{E}}_i}. \end{equation} \tag{ 15 }$$

For the CPCT case (figure 2), the sensible, H_T^CPCT, and latent heat fluxes, LE_T^CPCT, to the atmosphere at the rooftop level are shown below:

$$\begin{equation} H_T^{{\rm{CPCT}}} = {m_a}{C_p}\left( {{T_{{\rm{ao}}}} - {T_{{\rm{ai}}}}} \right) \end{equation} \tag{ 16 }$$

$$\begin{equation} {\rm{LE}}_T^{{\rm{CPCT}}} = {m_a}\left( {{h_{{\rm{ao}}}} - {h_{{\rm{ai}}}}} \right) + \mathop \sum \limits_{i = 1}^m {\rm{L}}{{\rm{E}}_i} - H_T^{{\rm{CPCT}}}. \end{equation} \tag{ 17 }$$

Note that for the residential areas, wall-type systems are kept unchanged for all the cases.

To validate the performance of the Baseline model, the simulation results are compared with surface level observations. The model could reasonably reproduce the physical characteristics of the temperature and wind field distribution (shown in figures 3, 4 and S1 available at stacks.iop.org/ERL/13/034015/mmedia). Figure 3 shows that there are some convergence zones (Tong et al 2005) formed at 2:00 pm on 25 June in different areas in Hong Kong. The temperature distribution is basically captured, although the model underestimated the air temperature at some stations, like SEK on 24 (shown in figure S1) and 25 June, which may due to the uncertainty of soil conditions and also the 1° NCEP FNL data chosen as the boundary condition (Takane and Kusaka 2011). The Baseline model could well capture the high temperature area in the convergence zone in Kowloon Peninsula at 3:00 pm on 24 June (figure S1), which is dominated by a westerly sea breeze and easterly wind and 2:00 pm on 25 June (figure 3), which is dominated by a southeastern wind. The observed and simulated 2 m air temperature for all urban and rural sites are compared (shown in figure 4). Note that because the horizontal spatial resolution is limited to 0.5 km, the stations along the coast (CCH, WGL, PEN, and SLW, as shown in figure 1(d)) are excluded because they may represent sea surface characteristics due to the land cover and land use classification. The Baseline model could capture the spatial distribution of 2 m air temperature well. The agreement between the Baseline model and observations was set in accordance with the criteria value of mean bias error (MBE), root-mean-square error (RMSE), and index of agreement (IOA) for 2 m air temperature for 27 available stations, which are 0.224, 1.242 and 0.90 respectively. The MBE value between the Baseline model and observations for the urban areas for 2 m air temperature is −0.09 °C. The Baseline model could capture the wind rotation from southwestern to southeastern on 24 June well, particularly in the central area of the Kowloon Peninsula, as shown in figure 5. The model underestimated the wind field when the extreme high-temperature event ended on 28 June, as shown in figure 5. In the Baseline case, the anthropogenic heat releases at the rooftop level and can affect the street level by vertical diffusion (Martilli et al 2002, Salamanca et al 2010), which is not significant in a high-rise city. In general, the Baseline case could capture the temperature distribution and wind rotation in the urban areas well, which can fulfill the goal of this study.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To evaluate the contribution of air-conditioning systems on temperature and humidity levels, the 2 m air temperature and the water vapor mixing ratio are chosen as the criteria for comparison. Figure S2 shows the 2 m air temperature distribution at 2:00 pm on June 25, 2016, i.e. the time when the maximum air temperature was recorded, and the temperature difference between the Baseline case and different air-conditioning scenarios at the corresponding time. The use of alternative water-cooled air-conditioning systems could decrease the 2 m air temperature during an extreme high temperature event by around 0.4 °C–0.8 °C, particularly along the northern part of Hong Kong Island and the commercial areas in Kowloon Peninsula, where most high-rise buildings and the high-density population are located. The population density data used in this study are downloaded from Columbia University’s Socioeconomic Data and Applications Center. The results also revealed the mean 2 m air temperature differences (figure 6), which were around 0.3 °C–0.8 °C, between the Baseline case and the corresponding air-conditioning scenarios averaged for all the extreme high temperature events from 8:00 am–8:00 pm, when the air-conditioning was switched on. The results are in agreement with the findings of other studies (He et al 2007, Salamanca et al 2014). Further, the large depth evolution of the PBL height during the daytime inhibited the effects of air-conditioning on the urban air temperature as compared to the relatively shallow PBL height and stable atmospheric conditions during the night. If the anthropogenic heat were excluded in the urban areas, the temperature difference could have reached around 4 °C in the commercial areas in Kowloon Peninsula, which was also verified by Wang et al (2017). As the non-domestic buildings are mostly concentrated in Kowloon Peninsula and the northern part of Hong Kong Island, an area, marked by the red rectangle in figure 6, is chosen for further investigation. The averaged 2 m air temperature, water vapor mixing ratio, sensible heat fluxes, and latent heat fluxes are compared among the different scenarios (figure 7). The use of water-cooled air systems can reduce the 2 m air temperature by around 0.5 °C–0.8 °C during the daytime as shown in figure 6(a), and the DCS case presents around 0.1 °C–0.15 °C lower air temperature as compared to the CPCT case. The temperature differences between the Baseline and the alternative systems reach 1.5 °C at 7:00–8:00 pm, as the impact of air-conditioning systems is more significant during the night-time (Fan and Sailor 2005). The use of cooling tower systems in commercial areas can significantly increase the water vapor mixing ratio in the atmosphere by around 0.29 g kg⁻¹ on average. However, an increase in the water mixing ratio in the air will result in an increase in the apparent temperature and the heat index level, which will in turn increase energy consumption to remove the humidity in the supply air (Steadman 1984, Gutiérrez et al 2015). The cooling tower diminishes the sensible heat by around 80%–90% and transforms it into latent heat flux (Milosavljevic and Heikkilä 2001) during the afternoon with the maximum value delayed for around 1–2 as compared to the maximum sensible heat fluxes for the Baseline case, as shown in figure 7(b). As sensible heat fluxes are the sum of H_tur and H_T^Base (Krpo et al 2010), the differences between the Baseline case and the CPCT and DCS cases with respect to the sensible heat fluxes indicate the heat fluxes from the air-conditioning systems (figure 7(b)). The sensible heat fluxes in the CPCT case are slightly higher than those in the DCS case, while in the CPCT case, the latent heat fluxes contribute almost 90% of the energy emitted by the air-conditioning systems (figure 7(b)). Ichinose et al (1999) reported that the anthropogenic heat emission in central Tokyo exceeds 400 W m⁻² in the daytime, and the maximum value could be 1590 W m⁻². The anthropogenic heat flux in Hong Kong using satellite data was studied by Wong et al (2015), and the results show that commercial areas emit the largest anthropogenic heat fluxes around 500–600 W m⁻² compared with other land-use types. The use of different air-conditioning systems affects the surface energy balance, and air-cooled air systems (Baseline case) could significantly increase sensible heat fluxes through the air-conditioning systems and enhanced turbulence fluxes.

The effects of city-scale air-conditioning systems on vertical structures were investigated. The use of different air-conditioning systems could not only modify the surface energy balance but also extended to the areas above the ground level, which affects the development of the PBL height (Fan and Sailor 2005, Chen et al 2009, Sailor 2011, Gutiérrez et al 2015), as shown in figure S3. The PBL height presents diurnal variation, with the peak value of around 1400 m to 1600 m occurring at around 13:00–14:00 pm and around 500–600 m during the night (shown in figure S3). The implementation of the two alternative air-conditioning systems modifies the heat emitted and generation of turbulence, which induces a 100–150 m reduction in the PBL height. A reduction of the PBL height could cause potential air pollution issues (Han et al 2009, Gutiérrez et al 2015).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The Baseline case has a relatively high urban canopy temperature, as shown in figures 8(a1)–(a4) on the vertical potential temperature, which leads to thermal discomfort for pedestrians and high building energy consumption (Nikolopoulou et al 2001, Salamanca et al 2010). The vertical profile of the potential temperature (figure 8) shows a higher value below 300 m in the Baseline case than in the other cases at the same height, which illustrates that anthropogenic heat from the air-conditioning systems enhances vertical mixing (Salamanca et al 2014). Differences in the potential temperature between the different cases are reduced with an increase in height (figures 8(a1)–(a4)), which shows the effect of anthropogenic heat fluxes on the potential temperature distribution at the lower level (Chen et al 2009).

The momentum fluxes for the horizontal surfaces are determined by the bulk Richardson number based on Louis formulation, which is a function of roof roughness length, wind speed and the air temperature (Martilli et al 2002). The differences in the potential temperature affect the wind speed, and the Baseline case has the largest horizontal wind velocity (figures 8(b1)–(b4)). Figures 8(b1)–(b4) further illustrate that the wind direction is not affected by the different schemes of air-conditioning systems.

The selection of an air-conditioning system affects the sensible heat fluxes released into the atmosphere, as discussed in section 2.2, which has a direct effect on the buoyant production of turbulent kinetic energy (TKE) (Martilli et al 2002). The results show that the buoyant production of TKE has the largest magnitude for the Baseline case and the smallest for the no-heat case (figures 8(c1)–(c2)). The largest differences in TKE were found 100 m above the ground level, above the mean building height of the region. Furthermore, figure 9 shows that the differences in the buoyant TKE are extended to a height of around 1.3 km above the ground level in the commercial areas, which has no influence in the non-domestic areas. The decreases in TKE within the urban boundary layer lead to potential air pollution issues, particularly pollutants near the ground level (Chen et al 2009, Salamanca et al 2010, Sharma et al 2016).

The effect of different air-conditioning systems on local wind circulation, particularly UHIC and SLBC, is investigated in the urban areas (vertical East–West cross-section, transect A–B, shown in figure 9(a)) in Kowloon Peninsula. At 06:00 UTC on 24 June 2016, a convergence zone was observed in the urban areas of Kowloon Peninsula (Tong et al 2005), with sea breeze penetration on the left side and a downslope wind on the right side. The urban areas in the convergence zone area are characterized by a low water vapor mixing ratio and enhanced UHIC of up to around 1.5 km (as shown in figures 10(a) and (d)). On the right side of the convergence zone, the UHIC is stronger than on the left side with the assistance of the airflow from across the mountain (as shown in figure 10(a)). Across the mountain, the water vapor mixing ratio is also reduced (figure 10(a)). To further investigate the effects of different air-conditioning scenarios on the local circulation, the difference in vertical wind fields is examined. The CPCT case has a higher water vapor mixing ratio near the ground level in the urban areas (on the left side of the convergence zone, shown in figure 10(e)), which is in agreement with Gutiérrez et al (2015). Figures 10(b) and (c) show that the Baseline case has the largest sensible heat fluxes and can enhance the downslope wind on the mountain side merged into the UHI circulation. The DCS case has less sensible and latent heat fluxes, which leads to a large vertical velocity difference, as shown in figure 10(c). Both cases (figures 10(b) and (c)) show a decrease in the vertical velocity in the urban areas, particularly in the convergence zone areas, which could reduce the strength of the UHIC (Chen et al 2009, Chen et al 2011, Xie et al 2016). Changing to the alternative air-conditioning systems could reduce the near-ground air temperature by around 0.4 °C–0.8 °C (figures 10(b) and (c)) for both cases, which in turn reduces the strength of the sea breeze.

Zoom In Zoom Out Reset image size

At 06:00 UTC on 25 June 2016, the background wind was southeasterly, and the urban areas in the cross-section were dominated by a strong eastern wind and no sea breeze penetration (figure S4(a)). The urban areas were characterized by a low water vapor mixing ratio as well. The largest vertical wind velocity was near the western edge of the cross-section, and the UHIC moved west in comparison to that shown in figure 10(a). The Baseline case, with the largest sensible heat fluxes within the urban canopy layer, had a strong UHIC. The changes in air-conditioning systems could not only affect the vertical distribution of the air temperature and the water vapor mixing ratio, but also the secondary local circulation, particularly the sea breeze circulation and the UHI circulation.