the urban heat island effect apparent. Meanwhile, analyzing
the shadowing effect of buildings on solar radiation is easy.
Sensitivity tests are mainly used to set the building width
and street width in the
BEP
scheme to represent the density of
buildings in the urban underlay area. The details are shown
in Table 3. The Base, Case1, and Case2 experiment settings
analyze the influence of the three types of building density
on the meteorological elements for middle, high, and low.
In particular, the Case1 experiment sets the building density
higher than that of the Base experiment, whereas the Case2
experiment has a lesser building density than that of the Base
experiment.
Table 3. Simulation scheme settings.
Scheme
setting Height distribution (m)
Floor
width (m)
Road
width (m)
Marks
Base 15(30%), 25(40%), 35(30%)
20
20
Medium
density
Case1 15(30%), 25(40%), 35(30%)
10
10
High
density
Case2 15(30%), 25(40%), 35(30%)
30
30
Low
density
Table. 4 Comparison of the simulated and observed
temperature, relative humidity, and wind speed
Elements
OBS SIM MB MAE RMSE R
T2 (°C)
20.89 22.15 1.27 1.30 1.58 0.97
Rh2 (%)
60.92 62.79 1.87 3.60 4.10 0.93
WS10 (m/s)
2.09 3.02 0.93 0.93 0.98 0.76
We selected the observation data from 221 automatic
weather stations in Guangzhou to test the simulation results
and the accuracy of the simulated meteorological field. The
main test elements included 2 m temperature, 2 m relative
humidity, and 10 m wind speed. The results of T2, Rh2, and
WS10, which were simulated using the Base experiment, are
shown in Table 4. The deviations of T2, Rh2, and WS10 are
simulated and the observed values are 1.27°C, 1.87 percent,
and 0.93m/s, respectively (see Table 4). All correlation coef-
ficients are above 0.75 and the correlation of T2 and Rh2 are
over 0.93. The comparison of the average T2, Rh2, and WS10
distribution maps simulated using the Base experiment and
the observed values of the Guangzhou automatic weather
station are shown in Figure 9. The horizontal distribution of
the simulation results of all meteorological elements are also
consistent with the observations. In the downtown area, the
horizontal distribution characteristics of higher temperature,
lower humidity, and lower wind speed were particularly
simulated. In general, the simulation results can substantially
reflect the actual situation of the atmosphere.
In addition, this research focused on the effects of building
density on temperature and analyzed the impact of various
building densities on the average 2 m temperature of Guang-
zhou at daytime (08 to 19 AM) and nighttime (00 to 07 AM,
and 20 to 23 PM). In Figure 10 A-a and Figure 10B-d, the
temperature of the main urban area in Guangzhou is evidently
higher than the other regions, thereby forming a large-scale
urban heat island. Liwan, Yuexiu, Tianhe, Haizhu, central
Panyu, South Baiyun, and Huangpu districts are particularly
warm regions, thereby forming a considerably high-tempera-
ture peak area. After changing the building density, the simu-
lated 2 m temperature of each experiment has an apparent
effect on the urban underlying surface. At daytime, the high
building density (see Figure 10A-b) decreased the tempera-
ture of the urban underlying surface by approximately 0.3°C
compared with the simulation results of the middle building
density (see Figure 10 A-a). The low building density (see Fig-
ure 10 A-c) increased the 2 m temperature by approximately
0.18°C. However, the influence of building density on the 2
m temperature at nighttime is opposite to that of daytime.
The high building density increased the 2 m temperature of
the urban underlying surface by approximately 0.12°C (see
Figure 10 B-e). The low building density decreased the 2
m temperature by approximately 0.08°C (see Figure 10B-f).
The reasons are that the higher building density makes the
street valleys narrower in the same area, the buildings have
a clear shadowing effect on the short-wave radiation during
the daytime, and the amount of heat that reaches the ground
decreases, thereby resulting in a decrease in the 2 m tempera-
ture. However, the increase in the width of the street valley
(low urban density) increases the amount of heat received by
the ground and increases the 2 m temperature. At nighttime,
the urban canopy has a significant trap effect on the long-
wave radiation because of the narrower street and valley. The
heat trapped inside the canopy increases, the 2 m temperature
increases, and the wide street valley decreases the amount of
heat trapped inside the canopy. Thus, the 2 m temperature
also decreases accordingly.
In Figure 10C, the warming effect associated with building
density changes has an apparent diurnal variation in Guang-
zhou. From 08:00 to 19:00, the high building density de-
creases the 2 m temperature (Case1 - Base), whereas the low
building density increases the 2 m temperature (Case2 - Base).
The most apparent effect was observed at noon. The 2 m tem-
perature decreases by 0.41°C (at 12 o’clock) and increases by
0.24°C (at 15 o’clock). From 00:00 to 07:00, the high building
density increases the 2 m temperature from 0.11°C to 0.18°C.
The low building density decreases the 2 m temperature from
0.01°C to 0.07°C. This study shows that the high building
Figure 9. Results of the horizontal distribution of the Base scheme simulation and observations of the Guangzhou automatic
weather station (Polka Dots) (a, T2; b, Rh2; c, WS10)
556
September 2018
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
