Experimental study of the urban microclimate mitigation potential of green roofs and green walls in street canyons

This paper presents the design and the results of a reduced-scale experimental bench to study the effect of green roofs and green facades on local urban microclimate. An analysis of experimental data was carried out on three street canyons: one reference street, one street between two buildings with green roofs and one with a green wall to the west. The results show a hygrothermal effect of green envelopes on buildings on urban heat island mitigation. In the detailed analysis of these green coating techniques, we highlighted that green facades modify strongly the radiative balance of the street and improve the hygrothermal comfort in urban canyons through reducing the overheating in the street in hot summer days.


INTRODUCTION
The urban heat island (UHI), observed in city temperature peaks compared with rural areas, is caused mainly by the storage of solar irradiance by city structures and materials, the release of anthropogenic heat, a reduced turbulent transfer of heat from within streets and the lack of water and green spaces, which reduce evapotranspiration [1].The increase in temperature in cities leads to a decline in thermal comfort of city dwellers, who account for more than 70% of the population in Europe.
Several published studies [2][3][4][5][6] have linked UHI intensity with physical parameters and urban characteristics.These parameters take into account the urban morphology, the urban surfaces, building materials and green areas.These studies have demonstrated a correlation between UHI intensity and the narrowness of the streets, the surface absorptivities and the lack of green areas.The use of cool paints and revegetation is likely to reduce the peak temperature in most cases.The benefits of cool roofs were shown in several studies; the average of the peak ambient temperature indeed decreases between 0.1 and 0.33 K per an increase in roofs albedo [7][8][9] of 0.1K.For green roofs, in addition to their positive effects on the energy performance highlighted by several simulation studies, the results show that the average ambient temperature can decrease by 0.3-3 K [10][11][12][13] when this roof technology is applied on a large scale.
In urban environment, the use of these coating techniques for facades is as important as for roofs, especially in dense cities.When the urban structure is characterized by narrow street canyons, the radiation trapping increases the surface temperature and the reduced airflow recirculation leads to higher air temperatures.Doya et al. [14] found, through the same reduced-scale bench presented in this paper, that the use of a cool selective paint on street facades reduces the air temperature within the street by about 1.58C.Alexandri & Jones [15] performed simulations for different canyon aspect ratios and climates.They studied a canyon with two green roofs and two green walls and found that the air temperature diminution within the canyon could reach about 108C for a hot and arid climate.Actually, there is a lack of experimental data on these effects, so our study focuses on the experimental verification of such results and gives further verification data for developed models.
The urban canyon air pollution has to be considered also for the outdoor air quality.Microscale dispersion models with different levels of complexity have been used to assess the urban air quality [16].Besides, the analysis of experimental data performed in Oklahoma City in 2003 showed the relationships between contaminant residence time in urban street canyons, urban boundary layer winds and urban topography.An experimental study [17] shows that vegetation improves air quality in urban canyons by particle removal.This purification effect is correlated with the canopy density.Baik et al. modeled green roofs [18] through specific cooling flux intensity; the results of the computational fluid dynamics (CFD) simulation demonstrated the improvements on air quality obtained by the combined effects of air temperature decrease within the street canyon and reduced pollutant concentration.
In this paper, we present an experimental approach to determining the hygrothermal impact of green roofs and green walls on the built environment in street canyons.The study was conducted on a reduced-scale model similar to a typical urban scene characterized by five rows of street canyons.The main objective was to assess the impact of vegetated building envelopes through relative comparisons with a reference street without vegetation and in real climatic conditions.The study also aimed to provide a database of experimental measurements; this will allow further comparisons with numerical models and their experimental validation.

Description of the bench ClimaBat
The experimental bench ClimaBat (see Figure 1) is a platform that was set up at the University of La Rochelle in France (46810N, 01809W).This platform had already been used in several studies that aimed to determine the impact of different facade and roof coatings on buildings' thermal performance and on local microclimate [14, 19 -21].It consists of five rows of concrete empty tanks, which stand for reduced-scale buildings.Each row is made up of three tanks, which form a block 5 m long, 1.3 m high and 1.12 m wide.The aspect ratio defined as the height to width of the street canyons is equal to 1.2 and the scale reduction is approximately 1:10.The experimental bench is built on a 10 Â 20 m terrace of concrete tiles.The canyon facades, oriented to east and west, and the roofs are painted in white.
Two adjacent reduced-scale buildings (index 4 and 5 on the east side of the experimental platform) have green roofs composed mainly of sedum.A green wall system was set up on the west facade of building 2. The green wall growing medium consists of Chilean sphagnum 15 cm thick and is fixed on a metal grid that forms a weakly ventilated air layer of 5 cm.There are six different species of vegetation planted on the green wall.This latter is watered twice a day by an automatic drip watering system.One row of concrete tanks and one street canyon were kept as a reference for further comparisons.
The experiment aimed to assess the impact of the revegetated envelopes on diurnal and seasonal variations in indoor and outdoor environments.The conducted work consisted of † Experimental monitoring of a green wall.Thermal and hydric variations are monitored at different depths and heights in the growing medium and at different heights in the air layer behind the green wall, and thus, the thermal and hydrological behaviors of the green wall are assessed and the drip is optimized.† The assessment of the impact of green roofs on building energy performance through measurements and comparisons of the air temperatures inside reduced-scale buildings with different vegetated and paint coatings.† The study of the impact of green walls and green roofs on the microclimate within typical street canyons.In this paper, we focus on the last point and the study of the impact of vegetated roofs and facades on the air temperature, air humidity and on the radiative trapping within the reduced-scale street canyons.
For further analysis and discussions on the presented results in this paper, the dimensions of the experimental setup and the physical properties of materials that form it are summarized in Table 1.Some of these properties have an important effect on the extent to which the green wall and the green roofs affect the air temperature and humidity within the canyon.

Similarity of the bench measurement results with real scale
This experimental bench is designed in order to study the different heat transfer fluxes through the canyon surfaces: the shortwave radiation, the longwave radiation, the sensible and the latent convective heat transfer and the conducted heat flux.The order of magnitudes of the reduced-scale mock-up and its similarity to a real-scale street have been studied to design the mock-up first.
It is important here to note that beyond the absolute results for green walls or green roofs, the simultaneous measurements allow us to highlight the temperature and flux variations, and the potential benefits, of different configurations.
Considering all the heat fluxes and the surface temperatures of the buildings, we have compared the experimental bench with a real-scale building and found very similar values, through dynamic simulations (TRNSYS) [14].Indeed, solar shortwave irradiance is the main energy flux that influences the surface temperatures in summer.The absorbed solar radiation depends only on the orientation, the radiative properties of materials and on the aspect ratio, which determines the solar masks.All these parameters are conserved on the reduced scale.The longwave radiation fluxes depend not only on the geometric aspect but also on the surface temperatures, and these two factors are almost the same on the reduced and the real scales.In addition, this is due to a similar convective heat transfer coefficients and air temperatures.
Considering the canyon microclimate, the convective heat transfer leads to increase in the air temperature within the canyon.Considering the prevailing winds, there are three regimes of across-canyon flows (according to Oke [22]), depending only on the aspect ratio of height to width (H/W) of the street: (a) isolated roughness flow for wide canyons (H/W , 0.3), (b) wake interference flow for more closely spaced buildings (0.3 , H/W , 0.5) and (c) skimming flow for regular canyons (H/W .0.5).So, in our experiment, the model has direct similarities to a real-scale street with skimming flows.
The heat flux from the canyon to the urban canopy is correlated with the airflow _ V and the temperature difference (T C -T A ; 8C).This convective heat flux is linked to the sensible heat transferred by the canyon surfaces S i (m 2 ) to the air.Considering a scale reduced by a factor n and similar wind magnitudes, the airflow from the street canyon to the canopy is approximately reduced by a factor n 2 as the surface is reduced by n 2 .Simultaneously, the convective heat transfer is reduced by a factor n 2 due to the reduced heat exchange surface.So the temperatures within the canyon should be in the same order as in the real scale one.The main differences could be found in the convective heat transfer coefficient variations, yet they will evolve with the wind speeds simultaneously as the airflow _ V.

Experimental measurement devices
The experimental instrumentation work on ClimaBat has been designed for three main objectives on various scales.First, special envelope coating properties and behavior were studied.Second, we studied the impact of these envelope coatings on the energy performance of buildings, and finally their effects on urban microclimate at local level were studied.
The sensors installed on ClimaBat provide † Annual and daily monitoring of temperature variations inside the building blocks and on their external and internal surfaces at different locations.This allows the assessment of the impact of various external coating techniques ( paint cool, green roofs, green walls) on the thermal behavior of the reduced-scale buildings.These measurements can be used for experimental validation of real building models.† Monitoring of hydric and thermal behaviors of the vegetated components (green roofs and facades).This allows a better understanding of the physics of both heat and moisture transfer through their special materials.5 cm, weakly ventilated † Diurnal and seasonal monitoring of the microclimatic variations in temperature, humidity and radiation trapping within street canyons at local level.In this article, we focus on this part of the experimental work.There are three street canyons that were instrumented with several sensors.The first street canyon (noted GRoofC) is located between buildings 4 and 5, which have green roofs; the second one (noted GWallC) is located between buildings 1 and 2, and it has the west-oriented facade covered with a green wall system; the third one (RefC) is the reference located between buildings 2 and 3 (see Figure 3a -c).
A weather station that is installed on the site permits the acquisition of meteorological data, including air temperature and humidity, wind speed and direction, short-and longwave horizontal irradiance and rainfall.The sensor distribution on the platform for the canyon study is outlined in Figure 2.
The thermocouples used in this study were type K (chromel -alumel) with an accuracy of about 0.38C.Hygrometers (Campbell Scientific CS215 temperature and HR sensor, which uses CMOSensw technology and incorporates SHT75 digital element) were placed in the center of the streets for simultaneous measurement of relative humidity (+2% between 10 and 90% RH) and air temperature (+0.48C between 5 and 408C).To compare the reflected solar radiation and the longwave emittance of the streets, a pyranometer (Kipp & Zonen CMP3, spectrum waveband 310-2800 nm, field of view 1808) and a pyrgeometer (Kipp & Zonen CGR3, spectrum waveband 4500 -42 000 nm, field of view 1508) were placed slightly above the roof ' only in the middle of the reference and the green wall streets (see Figure 2).In fact, the GRoofC and the RefC streets have identical walls and soil radiative properties; there are no radiative sensors on the GRoofC canyon.Moreover, a 2D sonic anemometer (WindSonic TM , wind speed 0-60 m s 21 (+2% at 12 m s 21 ), wind direction 0-3598(+38 at 12 m s 21 )) placed at 28 cm above the roof of the reference building records the local intensity and the direction of wind.These sensors are connected to a datalogger (CR1000 by Campbell Scientificw), which records all the measurements in less than half a second at 5-min time intervals.
Some sensor tuning was performed during the set up to reduce errors.At first, the pyranometer and pyrgeometer mounts should not be too large.In fact, the orientation of the facades implies that the mount shadow will cover the various sensors located at the middle of the facades.The solution was to set the radiative sensors on a metal plate that has been fixed on tensioned wire cables (see Figure 3d) and minimize errors due to the shadow of the devices.Second, the hygrometers CS215 should be protected against sunlight and rain.Indeed, shelters marketed for these types of sensors are very large to be placed in the streets.They may alter the airflow and produce also an important shadow.To remedy this problem, small shelters were made from flexible PVC elbows (see Figure 3e).These shelters protect the sensors from rain and solar radiation at least within the limits of exposure angles in the street.

Weather conditions
The analyzed measurements were performed during August 2012.The main meteorological data recorded during this month (see Figure 4) are characterized by moderate summer temperatures T a (8C) with some peaks reaching 358C.The specific climate of La Rochelle on the Atlantic coast implies moderate or high relative humidity RH(%) throughout the year The rise of wind close to the experimental bench (see Figure 4) shows that the airflow at local level is strongly influenced by the neighboring buildings and the experimental bench orientation.Hence, the regional wind intensity is reduced and its direction is modified.Therefore, the main direction of airflow over ClimaBat is from the south.The airflow is less channeled in city canyons than in a separate urban scene as ClimaBat.The reduced confinement of the street canyons plays an important role in the measured differences in temperature and relative humidity within the three studied streets.Hence, we can expect larger hygrothermal effects on a real scale in large cities.

Radiative impact of green wall
The heterogeneity of the vegetation on the green wall in terms of species and fractional coverage provides it with a solar reflectivity with a high spatial variability.Spectrometric  measurements of the local reflectivity averaged over a 20 cm diameter disc showed that the green wall reflectance varies between 0.15 and 0.30 depending on the location [23].Because the white paint, with which the buildings facades are painted, has a solar reflectance of about 0.64 (see Table 1), the equivalent albedo of the street canyon is reduced when the green wall is set up.
The overall street canyon solar-reflected radiation varies depending on the day time.We assume that the measurements performed by the pyranometers and pyrgeometers downwardly directed and positioned slightly above the middle of two roofs represent the mean-reflected solar radiation by the entire street.This is justified by the field view of such radiative sensors, which is about 1808.Indeed, the reflected solar radiation is composed of the reflected diffuse radiation and the reflected direct radiation.In the case of the GWallC street, the intensity of the reflected direct radiation depends especially on which facade, white paint or green wall, is directly exposed to the direct radiation.This is why the albedo of the reference street (RefC) is symmetric when the albedo of the green wall canyon (GWallC) is asymmetric and is lower in the afternoon (see Figure 5).The mean equivalent albedo is 0.272 for the reference street and is 0.153 for GWallC.
In order to assess the impact of vegetation on the comfort in the street, an analysis of the longwave radiation flux is necessary.The two pyrgeometers positioned in the middle of the streets at the roofs' height allow measuring the longwave emittance to the sky of the entire streets.Figure 6a compares the measured variations by these two pyrgeometers.We note that the RefC longwave emittance is greater than the GwallC longwave emittance during the day.This is due to higher temperatures in the reference street compared with the green wall street especially the west facade temperature where the green wall is set up.This difference in temperature recorded during the day is lower during the night.Hence, the longwave radiation loss seems the same at night. Figure 6b compares the net longwave radiation exchange between the sky and the street.The net longwave radiation exchange is given by the difference between the sky longwave irradiance, which is measured by the weather station, and the emitted longwave radiation measured by each pyrgeometer (net longwave radiation exchange ¼ longwave horizontal irradiance 2 longwave emittance of the canyon).The reduction in the longwave radiation flux emitted by the vegetated streets is accompanied by an attenuation of the longwave radiation loss of about 20 -25 W m 22 in the afternoon.We conclude in terms of radiative impact that vegetation increases the absorption of solar radiation and reduces longwave radiation loss in summer.The radiative balance of the street is then more important when there are vegetated facades.Although the vegetated street absorbs more radiation, it remains at a lower temperature, thanks to the evapotranspiration.
In terms of thermal comfort in the street canyon, this longwave heat flux from the canyon surfaces can be modeled by a theoretical mean radiant temperature, assuming black body surfaces.Therefore, the mean radiant temperature at the sensor location is given by: where T MR,C is the mean radiant temperature of the canyon surfaces (C), f LW,C is the measured longwave (LW) radiation flux emitted by the canyon surfaces (C), and s is the Stefan-Boltzmann constant.
Using equation ( 1), the evolution of the mean radiant temperature (T MR,C ) of the street facets is plotted in Figure 7.This figure shows that the difference between the RefC and the GWallC streets reaches 48C for a day with clear sky and at an air temperature of about 378C.This radiant temperature difference reflects the difference in the various facet temperatures for both streets.The west facade is obviously colder or hotter when it is vegetated or not.Because the west facade is exposed to direct sunlight in the afternoon, the maximum difference is recorded around 18:00 h with a lag of 4 h compared with the peak of solar horizontal irradiance which occurs at 14:00 h.

Thermal impact of vegetated envelopes
The thermal impact of buildings' revegetation can be identified through the analysis of the measured changes in temperature  within the studied street canyons.The comparison of these temperatures shows significant differences especially in daytime.All the streets reach temperatures higher than the local temperature (see Figure 8a).These recorded temperature differences are more important when the urban air temperature is lower.This is mainly due to radiative trapping effect where the streets floor and walls heat up; the big difference in temperature between these surfaces and the air intensifies the convective heat transfer and leads to overheating of the air in the streets.
The mean hourly temperature difference between the vegetated streets and the reference street is illustrated in Figure 8b.We note that GwallC and GRoofC streets remain cooler during the day.This effect is more pronounced in the GwallC street where the temperature difference with respect to the RefC street reached 1.58C.The results show that during the night the green wall located on the west side restores the solar radiation absorbed in the afternoon in sensible heat which keeps the street slightly warmer by about 0.58C.
The overheating in the RefC street, which reflects the potential of UHI created by the experimental platform, is illustrated in Figure 8c.This plot shows the difference between the air temperature within the RefC street and the air temperature measured by the weather station.The temperature difference is more important when it is hot and reaches up to 58C at noon.During the night and very early morning before sunrise, a flow of fresh air near the ground that comes from around the experimental platform leads to cooling of the air within the studied streets.Indeed, the experimental platform is surrounded by built and grassy areas, which are cooler than the roof meteorological station during the night due to longwave radiation loss.
To assess the mitigation potential of revegetation, we quantify the overheating reduction potential (ORP) by the decrease in the temperature in vegetated canyons versus the overheating in the reference canyon: The temperature difference between the GwallC and the RefC streets and between the GRoofC and the RefC streets are plotted against overheating in the reference street.We find that the highest overheating and the larger temperature reductions due to vegetation are both related to the intensity of solar irradiance on urban canopy.Negative temperature differences correspond to the night period and are previously explained.Thus, the scatter plots shown in Figure 9 allow estimation of the extent to which the vegetation reduces the overheating in the streets.The ORP is deducted directly from the slope of the scatter plots.A least-squares correlation is established for each scatter, which reveals a strong correlation between these two variables.For the green wall, the linear correlation (R 2 ¼ 0.864) indicates that the living wall mitigates summer temperature by about 1 per 38C of overheat.Similarly, the two green roofs mitigate the temperature within the street canyon by about 1 per 88C of overheat.The scatter plots are colored following the solar irradiance magnitude (see Figure 9).As previously stated, the observed temperature differences are negative or low when the solar irradiance is low or null, i.e. during night or cloudy periods.The same analysis for the measurements during high solar irradiance periods points out the interest on the higher overheating reduction during the highest overheating period for the reference street.There are local dispersions of measured points due to other coupled phenomena such as evapotranspiration and inertia of materials.Yet, the overall trends are consistent especially for the studied green facade, which has a more correlated effect on the street.

Hygrometric impact of vegetated envelopes
In addition to the radiative and the thermal effects of green roofs and green facades used on building coatings, the vegetated envelope may modify the air humidity in the confined street canyons.The vegetated building surfaces do not reach high s due to leaves and substrate latent cooling.The intensity of this phenomenon depends strongly on the water content of the growth medium [19] and leads to a humidity increase in the canyon.Figure 10 shows (a) the changes in the relative humidity RH% measured at the center of the three streets, (b) the difference in RH% found between the two vegetated streets and the reference street and finally (c) the vapor content difference between these streets expressed as partial vapor pressure.
According to the graphs shown in Figure 10a and b, the relative humidity RH% is lower in the reference street at daytime.The maximum of RH% difference reaches 6-8% at 17:00 h in the GwallC street.At this hour, the living wall is watered again and receives a maximum of direct solar irradiance.Hence, the evapotranspiration is accentuated on the living wall and led to an increase in relative humidity in the street canyon.We note, however, that the minimum value of the RH% reduction remains well above zero in this street at night.Although the same trends are observed in the GRoofC street, the trends are more uncertain, since the differences remain lower than sensors accuracy.
To quantify the real change over time in the vapor content of the air independently of changes in the temperature within the streets, the partial vapor pressure is calculated for each time in the three streets and is plotted in Figure 10c.It is found, in average, that the vapor pressure is greater in the GwallC street relative to that in the RefC street by about 100 Pa.This explains the difference in relative humidity, which remains above zero during the night when the temperature is almost the same in all the streets.Indeed, although the plant transpiration is near zero during the night, the evaporation of water from the green wall substrate keeps humidity higher in the GwallC street.For green roofs, the situation is less significant, since the two green roofs lack water in summer and can, therefore, even fix the moisture through buffering capacity.
In terms of hygrothermal comfort in the outdoor environment, the increase in relative humidity has an effect on the thermal sensation depending on the climate.For hot-dry climates, a little rise in relative humidity does not compromise the positive effect, but by cons, it may lead to better thermal sensation, as for direct evaporative cooling systems [24].For the current study, a comparison of the heat index between the three canyons indicates no change in this index with vegetation presence in daytimes ever since there is simultaneous decrease in temperature and increase in relative humidity.In spite of this observation, the green wall reduces not only the air temperature, but also the facade surface temperature, the reflected solar radiation from the concerned facade.All these positive effects contribute obviously to UHI mitigation and consequently to the improvement of the urban air quality.Further study may be accomplished to assess greening impact on thermal comfort in detail using more precise outdoor comfort indicators such as physiologically equivalent temperature (PET).

CONCLUSION
This paper presents a part of an experimental work performed on a reduced-scale model buildings and street canyons.The results highlight the radiative, thermal and hygrometric effects of the use of vegetated roofs and facades on urban microclimate.The results make evident local effects on outdoor thermal comfort from the air temperature evolutions and the mean radiant temperatures within the streets.It was found that the green wall may reduce overheating by one-third in street canyons through maintaining moderate temperatures on the green facade by evapotranspiration.Thus, when vegetated envelopes are used, the air temperature increases less during the day and decreases less at night.Green roofs have the same but less noticeable effects.Similarly, we found that the use of vegetated facades can reduce by 48C the mean radiant temperature, which directly affects the thermal comfort in the street canyon.
The overall conclusions of this experimental study are consistent with the bibliography on the subject given in the introduction, namely the effect of introducing vegetation in cities and its potential to mitigate the UHI.In addition, the measured thermal impact, of few degrees, on this reduced-scale experiment within the vegetated street canyons could be greater at city level.In this work, a lot of experimental data have been collected and the results will be valuable for both further numerical studies and the design of larger scale (in situ) experiments.
The research work conducted at the University of La Rochelle on the impact of use of green roofs and green facades in dense cities has the objective of studying the hygrothermal interaction of the building-green envelope-microclimate system.This work is part of a French research project (VegDUD) [25] that aims to study the multiple effects of vegetation in cities.In this paper, the presented results are focused on microclimate effects.To determine the direct and indirect effects (thought microclimate change) of vegetated envelopes on buildings energy performance, the incorporation of our green envelope model [19] to a dynamic thermal buildings simulation code is in development.Measurements carried out on the experimental platform and further analysis will be processed to obtain a reference dataset in order to verify the accuracy of such modeling approach.

Figure 2 .
Figure 2. Location of sensors on ClimaBat, measuring the microclimate impact of vegetated building envelopes.

Figure 4 .
Figure 4. Main meteorological data recorded at the experimental site and the local wind rise on ClimaBat bench.

Figure 5 .
Figure 5. Measured, reflected solar radiation from the streets and the equivalent albedo.

Figure 6 .
Figure 6.(a) Measured streets' longwave emittance to the sky; (b) the net longwave radiation exchange between the sky and the streets.

Figure 7 .
Figure 7. Evolution of the mean radiant canyon surfaces temperature for the RefC and the GwallC streets.

Figure 8 .
Figure 8.(a) Mean hourly temperature variations in the streets.(b) Relative differences in temperature within the vegetated streets compared with the reference street.(c) Overheating in the reference streets compared with local weather.

Figure 9 .
Figure 9. Microclimate mitigation potential of green wall (a) and green roof (b).

Figure 10 .
Figure 10.(a) Mean hourly relative humidity variations in the streets.(b) Relative differences in RH(%) within the vegetated streets compared with the reference street.(c) Calculated differences in vapor pressure p v (Pa) within the vegetated streets compared with the reference street.

Table 1 .
Main dimensions and properties of the experimental bench ClimaBat.