Geoscience Reference
In-Depth Information
In short, PET was computed for different measuring points (1-10 on Figure 5.4c)
using i) parameters measured in Telheiras during two 10-day measurement
campaigns in winter and in the summer (temperature, humidity, short and long-wave
radiation needed to compute MRT); and ii) parameters measured at the airport (solar
radiation); iii) modeled parameters (MRT, wind speed). PET was modeled
separately for daytime and night-time situations and in function of weather type
[AND 03; AND 08]. Multiple regressions were calculated using PET as the
independent variable and both geographical parameters and meteorological
predictors at the airport as independent ones. One of the most important
geographical parameters is the SVF. It is a very good descriptor of urban geometry
and better than H:W ratio [LAN 81], particularly in irregular canyons, courtyards,
and open spaces. Each independent variable was depicted in a different layer.
Different equations were selected for night-time and daytime, and within the latter
shade and sun situations, and winter and summer periods were separated.
Independent variables were selected by stepwise multiple regression.
Two night-time situations will be described here. The best fit equation to night-
time estimated PET values was the following:
PET (t) = 1.46184 + 1.07735 Ta Airport - 3.07473 SVF - 2.77031 ln V(t)+1,
where Ta is the air temperature and ln V(t) stands for natural logarithm of wind
speed. The determination coefficient (r
2
) attained 0.98, only 10% of the residuals are
greater than 1ºC and only 0.1% greater than 2ºC. The ȕ coefficients showed that the
Ta measured at the airport is the main independent variable, followed by wind speed
and by SVF. Therefore, one can consider that Ta at the airport represents the
mesoscale thermal conditions to which variations in wind speed and SVF introduce
microclimatic modifications. This model allows for the simulation of PET for any
location in Telheiras and to other city districts in a similar geographic context.
Subsequently, spatial interpolation of PET among the measurement points was
carried out using the regression equation determined in the previous stage. In this
process a GIS with a 5×5 m pixel grid was used. Each independent variable
corresponded to a separate layer. The estimated thermal pattern for two night-time
summer conditions will be given as examples (Figure 5.13a-b). PET values between
18 and 23 were predominant and are considered comfortable according to
Matzarakis
et al.
[MAT 99]. On a warm summer night, PET varies between 16 and
more than 22ºC (Figure 5.13a), i.e. most of the area is “comfortable”. PET ranges
from 12-18ºC in a windy summer night (Figure 5.13b): conditions are “cool”
(between 12 and 14ºC) or “moderately cool” (between 14 and 16ºC [MAT 99]
[MAT 07]), and require an increase in the level of thermal insulation of clothing
(close to 1.8 Clo) in order to maintain thermal balance of individuals. This indicates
that even in summer, these areas are not always suitable for long stays outside
without adequate protection when the N wind is blowing. The cooler areas lie in
prevailing wind paths, which was confirmed by wind tunnel experiments (for
example, the area marked (x) in Figure 5.13a and b, see next section and Figures
5.15 and 5.16). In both cases, the highest PET(t) values were found in areas
sheltered from the wind and less exposed to the sky. The thermal behavior of the
