T

LD

a

ARAA

KUCEHS

1

eaaeGvim

awhbBDeooc

(

0d

Landscape and Urban Planning 92 (2009) 179–186

Contents lists available at ScienceDirect

Landscape and Urban Planning

journa l homepage: www.e lsev ier .com/ locate / landurbplan

he cooling efficiency of urban landscape strategies in a hot dry climate

imor Shashua-Bar, David Pearlmutter, Evyatar Erell ∗

esert Architecture and Urban Planning, The Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sde Boqer Campus, Midreshet Ben Gurion 84990, Israel

r t i c l e i n f o

rticle history:eceived 11 February 2009ccepted 27 April 2009vailable online 27 May 2009

eywords:rban microclimateoolingfficient water useot-arid regionshading

a b s t r a c t

This paper describes a climatic analysis of landscape strategies for outdoor cooling in a hot-arid region,considering the efficiency of water use. Six landscape strategies were studied, using different combina-tions of trees, lawn, and an overhead shade mesh. The effects of these treatments were tested duringthe summer season in two semi-enclosed courtyards located at an urban settlement in the arid NegevHighlands of southern Israel. Compared to a non-vegetated exposed courtyard, which on average reacheda maximum air temperature of 34 ◦C in mid-afternoon, a similar courtyard treated with shade trees andgrass yielded a daytime temperature depression of up to 2.5 K, while shading the courtyard with a fabricshading mesh, counter-intuitively, caused a relative increase of nearly 1 K. Unshaded grass was found tocause only a small air temperature depression and had the highest water requirement. However whenthe grass was shaded, either by the trees or by the shade mesh, a synergic effect produced greater coolingas well as a reduction of more than 50% in total water use. The “cooling efficiency” of these strategies was

calculated as the ratio between the sensible heat removed from the space and the latent heat of evapo-ration, with the latter representing the amount of water required for landscape irrigation. This measure

for e

is proposed as a criterionscarce.

. Introduction

Many studies have discussed the importance of climatic knowl-dge in the process of urban design and planning (e.g., Paparelli etl., 1996; Eliasson, 2000; Svensson and Eliasson, 2002). Appropri-te measures may moderate the urban heat island, reduce buildingnergy demand and improve pedestrian comfort (Steemers, 2003;rimmond, 2007). Measures such as shading and judicious use ofegetation are of special importance in hot-arid regions, wherentense solar radiation and high air temperatures may have detri-

ental impacts on even the most basic human activities.Vegetation in parks and streets may generate localized cooling,

phenomenon sometimes referred to as the “park cool island”,hich stands in contrast to the more commonly discussed “urbaneat island” effect. Temperature reductions of up to 3–4 K haveeen observed in urban parks at mid-day during summer (e.g.,ernatzky, 1982; Oke, 1989; Shashua-Bar and Hoffman, 2000;imoudi and Nikolopoulou, 2003; Chen and Wong, 2006). How-

ver, as Spronken-Smith and Oke (1998) showed, the actual extentf this cooling may vary significantly, depending on the attributesf the park—such as the vegetation and irrigation regime, the adja-ent urban fabric, and the aridity of the location. This finding was

∗ Corresponding author. Tel.: +972 8 6596878; fax: +972 8 6596881.E-mail addresses: shashual@bgu.ac.il (L. Shashua-Bar), davidp@bgu.ac.il

D. Pearlmutter), erell@bgu.ac.il (E. Erell).

169-2046/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.landurbplan.2009.04.005

valuating landscape strategies in arid regions, where water resources are

© 2009 Elsevier B.V. All rights reserved.

enhanced in other comparative studies on parks located in the samevicinity (i.e. Potchter et al., 2006; Chang et al., 2007).

Evaluating the cooling effects of vegetation within an urbancontext is further complicated because they are interrelated withother building effects (Stabler et al., 2005; Shashua-Bar et al., 2006).The microclimate in an urban space is influenced by the adjacentbuildings and landscape elements, and by the complex interactionsbetween them (Erell and Williamson, 2006). Thus, conditions atdifferent points within the urban canopy layer may differ signifi-cantly even in the same overall climatic context, and they can beaffected by a variety of factors relating to geometry and surfaceproperties (Pearlmutter et al., 2006) as well as by anthropogenicheat release. It is therefore important to investigate the impact ofvegetation within the context of planning strategies appropriate tothe climatic region and to the related urban environment.

Vegetation has the potential to moderate air temperature notonly through shading and the reduction of surface temperatures,but also through evaporative cooling (McPherson et al., 1994). Therelative importance of each of these mechanisms may vary with cli-mate, the characteristics of the plants involved and their responseto environmental conditions. Conversely, elevated external air tem-peratures increase the irrigation requirements of urban vegetation

(Guhathakurta and Gober, 2007).

The presence of trees in the urban matrix may affect air tem-perature at a variety of spatial scales, from individual streets to thelarger urban boundary layer. However, the magnitude of this effectmay depend on a variety of factors, due to the complex interaction

1 nd Ur

bTsftasads(mmieeda

taMbidss(

gIn(aunlodIrsn2

dhS

Ffb

80 L. Shashua-Bar et al. / Landscape a

etween trees and other constituents of the urban environment.rees intercept not only incoming solar radiation, but also reflecthort wave radiation from their surroundings, long wave radiationrom the ground, building surfaces and the sky—and in some caseshere is significant sensible heat exchange between the warm urbanir and the cooler leaves. The dissipation of this heat load by tran-piration cooling depends on the water balance, the wind climatend the tree species (Oke, 1989). Species from hot dry habitats canissipate heat and conserve water by regulating the opening of leaftomata, or in small leaves by minimizing the density of stomataHemsely and Poole, 2004). The transpiration rates from such trees

ay accordingly be lower than those of broadleaf trees, and thusay have a smaller effect on the air temperature in their surround-

ngs. Recent studies of the microclimate of vegetated areas in desertnvironments report substantially different cooling effects (Saaronit al., 2004; Potchter et al., 2008). The inconsistent results might beue to differences in the behavior of the plant species of the studyreas or in the variability of the surrounding locales.

Unlike trees, grass reduces temperatures mainly through evapo-ranspiration at ground level, in urban lawns (Bonan, 2000) as wells in green walls and roofs (Onmura et al., 2001; Takebayashi andoriyama, 2007; Alexandri and Jones, 2008). The ultimate contri-

ution of lawns to thermal comfort may in fact be limited sincet does not affect the direct incoming radiation, which has such aominant impact on the daytime thermal stress in hot dry urbanpaces (Pearlmutter et al., 2006). The maximum cool island inten-ity found in a large irrigated lawn park in Mexico City was 2 KSpronken-Smith et al., 2000).

Concerns over scarcity of water have focused attention on irri-ation, which is the primary use of water world-wide (Perry, 2007).n arid regions, where the mean annual precipitation (P) is sig-ificantly less than the characteristic potential evapotranspirationPET) (Bruins and Berliner, 1998), appropriate selection of plantsnd efficient watering systems can conserve a large amount ofrban irrigation water (Ferguson, 2007). Differences in irrigationeeds may be substantial. For example, in Israel a well-irrigated

awn in the Mediterranean coastal region consumes just over 3 l/m2

f water on a summer day, while in the arid Negev region theaily requirement is as much as 6 l/m2 (Kremmer and Galon, 1996).

n addition to such objectively high water requirements in aridegions, waste of water in many sectors of agriculture and land-caping, is further aggravating water scarcity and emphasizing theeed for developing ways to improve irrigation efficiency (Lankford,

006).

Shading can be achieved not only by trees but also by shadingevices such as a lightweight mesh. Recently, shade mesh fabricsave been used extensively for agricultural crops in greenhouses.uch a mesh has been found to act as an effective means for regulat-

ig. 1. Plan showing location of measurement points (in both courtyards), of trees in westor convenience only in west courtyard shows extent of grass cover in both courtyards, onare soil in both courtyards; the rest of the ground surface was paved.

ban Planning 92 (2009) 179–186

ing the solar radiation as well as the evapotranspiration rate of thecrops in its shade, thus leading to significant water savings (Molleret al., 2004). While the use of fabric shading is also fairly commonin open spaces of some arid zone cities, its effect on urban microcli-mate has not yet been explored in relation to other environmentalstrategies.

As this brief overview indicates, the cooling effect of vegeta-tion in urban open spaces is well-documented. However, the extentof this effect has not been analyzed in a systematic manner withrespect to the water resources required to achieve it. In hot-aridregions, water availability is a limiting factor and must be consid-ered. This study addresses this shortcoming by focusing on thewater consumption of several combinations of shade and vege-tation in relation to the cooling effect they produce in an urbancontext.

2. Methodology

2.1. Sites and observations

As mentioned above, the microclimatic conditions within urbanopen spaces are affected by adjacent buildings as well as by vege-tation. Due to the complex interactions between these elements, itis difficult to identify comparable urban sites in which the effectsof individual parameters such as landscape treatments may beanalyzed empirically. The methodology used in the present studyaddresses this problem by establishing a controlled experiment intwo adjacent courtyard spaces, which are similar in their geome-try and material attributes but differ in their landscape treatments(Fig. 1). Both are oriented along an approximately N–S axis and havean H/W ratio of about 0.5 (Figs. 1 and 2), giving similar exposure tothe environment (Meir et al., 1995).

The courtyards are located at the Sde-Boqer campus in the aridNegev Highlands region of southern Israel (30.8◦N latitude, 475 maltitude). The region is characterized by hot dry summers and coolwinters. Diurnal temperatures and relative humidity fluctuationsare wide, with summer daily maxima and minima well above andbelow the thermal comfort zone, respectively (Bitan and Rubin,1994).

Six different landscape strategies were studied in the courtyards,using different combinations of trees, grass, and shade mesh. The sixstudy cases are summarized in Table 1, and two cases, “Mesh-bare”and “Trees-grass”, are illustrated in Fig. 2. The ground surface in

the two courtyards initially consisted of light gray concrete pavingtiles (covering about 70% of the area) and exposed soil (occupy-ing the remaining 30%). One of the courts had three trees plantedalong its center line, two of which were Prosopis juliflora and thethird Tipuana typu. Both species are common in hot-arid regions

courtyard (left) and of shading mesh in east courtyard (right). The grey area drawnce it was applied. The grey areas shown in the east courtyard indicate the extent of

L. Shashua-Bar et al. / Landscape and Urban Planning 92 (2009) 179–186 181

nd sh

aGttjvs

mpsat

DacDwpctiaapppt

2

coeicmw

TT

O

ETS

Fig. 2. Case study courtyards with bare pavement a

nd are considered economical water consumers (Kremmer andalon, 1996): the water consumption coefficient (defined here as

he ratio of evapotranspiration per unit horizontal area to evapora-ion from a Class A pan) is 0.3 for Tipuana typu and 0.2 for Prosopisuliflora. Both tree species have a medium leaf density that allowsentilation and sufficient solar penetration for grass to grow in theirhade.

The fabric mesh installed to shade the exposed courtyard wasade from black polyester netting with a nominal 70% opacity, thus

roviding about the same density of shade as the trees (whose mea-ured spatially averaged solar transmissivity was 30%). The extentnd geometry of the shaded areas were made as similar as possibleo those provided by the trees in the second courtyard.

The grass subsequently planted in the two courtyards wasurban grass with a measured pan evaporation coefficient ofpproximately 0.8, which is typical of values for short-cut grassited in previous studies (Brutsaert, 1982; Pearlmutter et al., 2009).urban grass was selected mainly for its ability to grow in the shade,ith a minimum requirement of only three hours of direct sunlighter day. The grass sod units were placed on a polyethylene sheetovering about 80% of the ground area of each court. The trees andhe grass were irrigated separately: a drip irrigation system wasnstalled around each tree trunk, providing water for several hourst a time. Water sprinklers for the grass were located in each courtnd activated each morning at 6:00, for approximately 12 min, afterreliminary experiments showed that this was sufficient to com-ensate for the daily water loss. The two irrigation systems wererogrammed to provide each type of vegetation with enough watero allow unrestricted evapotranspiration.

.2. Measurement setup

Dry- and wet-bulb temperatures were measured using copper-onstantan thermocouples in aspirated psychrometers at fivebservation points set up on instrument masts (with the maximum

rror estimated at 0.1 ◦C) in each of the two courtyards. For the hor-zontal profile, three points were situated along the long axis of theourtyard at a height of 1.5 m. On the main “inner” mast locatedidway between the two Prosopis juliflora trees, two more pointsere situated at different heights, creating a vertical profile from

able 1he six landscape strategies analyzed.

verhead treatment Ground surface

Bare soil and concrete pavers Irrigated Durban grass

xposed “Exposed-Bare” “Exposed-Grass”rees “Trees-Bare” “Trees-Grass”hade mesh “Mesh-Bare” “Mesh-Grass”

ading mesh (right), and with grass and trees (left).

0.5 to 2.5 m. In addition, one aspirated psychrometer was locatedat roof level. Wind velocity was measured using a Campbell 014Acup anemometer (with a sensor accuracy of ±0.11 m s−1) in the barecourt, and with a Young 81000 ultrasonic anemometer in the courtwith trees. Radiant temperatures of the various built and vegetatedsurfaces were measured in the two courtyards using shielded ultra-fine thermocouples (attached to wall, paving and soil surfaces) andan IR thermometer (for the grass surface). Incoming solar radiationwas measured with a Kipp and Zonen CM5 pyranometer and netall wave radiation was recorded with an REBS Q7.1 net radiometer,both located at roof level.

All the readings were taken at a 1-s sampling rate with the result-ing data averaged and stored every 10 min using Campbell CR21Xand CR23X data loggers. Along with the in situ measurements, ref-erence climatic data corresponding to the given measurement dayswere obtained from the nearby meteorological station.

Evaporation from the lawn was estimated using custom-mademini-lysimeters, whose dimensions and material were optimizedto ensure representative measurement of evapotranspiration (ET)from the grass-soil volume (Grimmond et al., 1992). The instru-ments consisted of rectangular (0.05 m × 0.10 m) sheet-metal panswith a vertical depth of 0.03 m, embedded in the grass-soil layerof sod of similar thickness. The evapotranspiration rate was deter-mined from the periodic change in lysimeter weight, measuredhourly with a high-resolution electronic scale starting immedi-ately following the daily irrigation. Two mini-lysimeters wereplaced in each of the courtyards: in the court with trees, oneinstrument was located at the “inner” point shaded by the twoProsopis juliflora trees and the second in the gap between theseand the Tipuana typu tree, which was exposed to sun most of theday. Two other mini-lysimeters were placed in the second courtat corresponding locations, and were intermittently exposed tosun or under the mesh shade, depending on the studied strategy(see Fig. 1).

Transpiration from the trees was measured by the sap flowmethod, which relates the transpiration rate to the speed of sapflow in the tree trunk (Gash and Granier, 2007). The method uses apair of cylindrical probes inserted into the sapwood, with the upperprobe heated by the Joule effect at a constant rate and the lower(reference) probe unheated. Sap flow probes were built accordingto a method developed by Oren, Lab & Phillips (Duke University)and S. Cohen (Volcani Institute, Israel). When there is no sap flow(during the night, when the vapor pressure deficit is equal to orclose to zero), all of the energy in the heated probe is dissipated by

conduction in the wood, provoking a maximum temperature differ-ence with respect to the unheated probe. When the sap circulatesthrough the xylem, the temperature difference decreases in pro-portion to the fraction of heat that is dissipated by convection fromthe heated probe.

1 nd Urban Planning 92 (2009) 179–186

tpe

3

Jmfip

wnndb

lermfmtpsas

T

wstt

pstaer

3

lnt

aa

TNp

C

EEMMTT

82 L. Shashua-Bar et al. / Landscape a

To account for variations in sap flow among different parts of theree, transpiration was calculated from the average of three pairs ofrobes located in each tree at the same height (about 0.8 m), atqual intervals around the trunk.

. Empirical findings

The courtyards were monitored over a 45-day period duringuly–August 2007, with comprehensive measurements taken of

icro-meteorological variables and of water consumption. Thendings presented in this section focus on individual microclimaticarameters measured within the courtyards.

Typical temperatures during this period ranged from 20 to 33 ◦C,ith an average relative humidity of 35% at 14:00 and up to 90% atight. Wind speed ranged from a maximum of 6 m s−1 in late after-oon to a minimum of 1 m s−1 during the night. Prevailing windirections ranged from north to northwest, with a light southwestreeze during early morning hours.

Each landscape configuration was monitored for a period of ateast 3–4 successive days. To account for minor differences in ambi-nt conditions between periods, measured data were normalizedelative to a common reference dataset taken from the adjacenteteorological station. This procedure is considered appropriate

or the case at hand because the summer period at Sde-Boqer is cli-atically stable, with clear sunny conditions prevailing throughout

he season. Courtyard air temperature at a given hour was adjustedroportionally based on the ratio between the simultaneously mea-ured ambient temperature at the adjacent meteorological station,nd the average ambient temperature for that hour over the entiretudy period:

normal,i(t) = Ti(t)

(Tmet,av(t)Tmet,i(t)

)(1)

here Ti(t) is the site-measured air temperature at time t on mea-urement day i, Tmet,i(t) is the meteorological air temperature atime t on measurement day i, and Tmet,av(t) is the meteorological airemperature at time t averaged over all the days of the experiment.

Water consumption was normalized according to the samerocedure, based on daily pan evaporation at the meteorologicaltation. This procedure is adequate because the rate of evapo-ranspiration under well-irrigated and calm conditions is mainlyffected by solar radiation and ambient air temperature (Jensent al., 1963), and thus is proportional to their daily variations at aepresentative meteorological station.

.1. Air temperature pattern

Table 2 summarizes the air temperature data for the variousandscape strategies at a height of 1.5 m, in the early morning, at

oon, in mid-afternoon and at night. The data are normalized rela-ive to the average of the measurement period (July 1–August 15).

The strategy providing the maximum cooling effect (measuredt 14:00) was the combination of trees over grass, which registeredtemperature of 32.2 ◦C—compared with 34.3 ◦C in the exposed

able 2ormalized air temperature [◦C] for the six study strategies (“central” observationoint, 1.5 m height), Sde-Boqer data, July–August 2007.

ase study 600 1200 1400 2400

xposed bare 21.5 32.2 34.3 21.7xposed grass 20.4 31.8 33.8 21.6esh bare 21.6 33.7 35.1 22.0esh grass 20.7 33.2 34.5 21.6

rees bare 21.3 31.3 32.5 22.1rees grass 20.6 30.7 32.2 21.7

Fig. 3. Normalized diurnal air temperature series for the six landscape strategies,“inner” observation point, 1.5 m height.

bare court. The highest maximum temperature was recorded inthe mesh-covered courtyard (“mesh bare”), which at 35.1 ◦C wasnearly a full degree higher than the daily maximum in the samecourtyard when exposed to the sky (“exposed bare”). The diurnalair temperature patterns of the six study strategies are illustratedin Fig. 3.

Differences between the different locations and heights in thecourtyards, as indicated by the horizontal and vertical temperatureprofiles, were minor (up to 0.5 K at noontime), indicating a high rateof mixing of air within the courtyards.

3.2. Humidity pattern

The ambient relative humidity was typical for the region—about35% at noon, rising up to 80–90% at night, corresponding to an abso-lute humidity of 13–15.5 g m−3. The different landscaping strategiesresulted in only minor differences in this humidity level beingobserved between the courtyards.

Table 3 shows the humidity pattern in g m−3 at three heightsof 0.5, 1.5 and 2.5 m in the six studied cases. The differencesrange between 12.0 and 14.0 g m−3 among the different heights.Humidity tends to increase slightly when the readings are madeclose to the vegetation, either at 0.5 m height above the grass

or at 2.5 m height, near the trees’ canopy. The small magnitudeof the differences in humidity reinforces previous observations(Oke, 1989) that the effects of transpiration from trees are feltmostly immediately above the canopy rather than in the volumebelow.

L. Shashua-Bar et al. / Landscape and Ur

Table 3Normalizeda absolute humidity [g m−3] for the six study strategies (“central” obser-vation point, three heights, hour 14:00), Sde-Boqer data, July–August 2007.

Case Height 0.5 m Height 1.5 m Height 2.5 m

Bare ground casesExposed bare 11.7 13.2 13.4Shade bare 12.1 13.8 13.9Trees bare 13.7 13.1 14.5

Grass casesExposed grass 13.4 13.2 13.2Shade grass 13.8 13.3 13.2Trees grass 12.6 12.1 14.1

a Based on normalized wet bulb temperature.

Table 4Averaged wind speed [m s−1] at three wind situations (central observation point,1.5 m height), Sde-Boqer data, July–August 2007.

Case study 600 1200 1400 2400

Exposed case 0.4 2.5 3.1 0.9MT

3

dohr1tass5t

3

hfttmcmstsc

TAl

C

EEMMTT

represents the daily estimate of the ET. The graph shows that thecombined total ET of the trees and the grass in their shade (“Grassunder trees”) is still much less than the daily ET of the exposedgrass.

esh coverage 0.3 1.1 1.3 0.4rees coverage 0.2 0.3 0.4 0.2

.3. Wind speed

The two courtyards were protected from the dominant windirections during the day, such that in the exposed court (with-ut tree or mesh cover), wind speeds were reduced to aboutalf of those in the surrounding open space. Table 4 summa-izes the average wind speed in the courtyards at a height of.5 m, for three configurations exhibiting substantial differences:he bare exposed court, the court covered with a shade meshnd the court with trees. The data in Table 3 indicate that bothhading strategies tend to reduce the wind speed inside thepace substantially relative to that of the exposed court, by about0% in the case of mesh coverage and by 80% in the case ofrees.

.4. Ground temperature

An important element affecting both air temperature anduman thermal comfort is the radiant temperature of solid sur-

aces within the courtyard. Table 5 summarizes the ground surfaceemperatures for the different landscape strategies. The main pointo notice in this table is the high mid-day temperature of the pave-

ent in the exposed bare court (reaching a maximum of 55 ◦C),ompared to the maximum air temperature of 33.5 ◦C. The pave-ent temperatures under the mesh and under the trees are reduced

ubstantially, reaching maxima of 39 and 37 ◦C respectively, while

hose for grass exposed and shaded under the mesh are about theame as the air temperature. The grass under the trees was muchooler—only 27 ◦C.

able 5veraged ground surface temperatures [◦C] for the six study strategies (“central”

ocation).

ase 600 1200 1400 2400

xposed bare 21.6 53.4 54.9 23.7xposed grass 18.3 36.7 34.9 18.0esh bare 23.0 35.7 38.9 25.1esh grass 21.6 30.9 32.2 20.5

rees bare 22.8 32.1 36.4 24.5rees grass 21.0 26.4 27.0 20.7

ban Planning 92 (2009) 179–186 183

3.5. Water consumption

Compared to standard pan evaporation recorded at the nearbymeteorological station for the period of July and August, the waterlost by evapotranspiration (ET) from the trees was relatively low.The ratio was 0.14 for the Prospis juliflora, and 0.05 for Tipuanatypu, both of which are low compared to the pan coefficients esti-mated for mature plants by Kremmer and Galon (1996) of 0.2 and0.3, respectively. The pan evaporation coefficient of grass in thefully exposed courtyard was found to be 0.72, while in a prelim-inary test (in late May), grass of the same species situated in anexposed open area yielded a coefficient of approximately 0.8. Differ-ences between the two coefficients reflect the courtyard’s effect onground-level evapotranspiration, presumably due to the restrictedsky view factor and reduced exposure to solar radiation and airflow.

Fig. 4 shows results of the estimated ET as measured for thegrass in three configurations: exposed, under shading mesh andunder trees (using lysimeters) and for the central Prospis julifloratree (using the sap flow method). It can be seen from the hourlyET graph that the plants’ transpiration follows the daily pattern ofthe solar radiation and of air temperature; accordingly, the maxi-mum difference between the configurations occurs during the noonhours. In the cumulative ET graph, the value at the end of the day

Fig. 4. Hourly and cumulative ET for the exposed grass and for the central tree atthe studied sites, averaged data, Sde-Boqer data, July–August 2007.

184 L. Shashua-Bar et al. / Landscape and Ur

Fm

d

4

pmbiv

TC(

C

(((((((

TTe

ig. 5. Hourly differences in air temperature (K) between courtyards treated withesh and grass (a) or trees and grass (b), relative to the base case (“exposed bare”).

An analysis of the daily ET in relation to the cooling effect for theifferent study cases is presented in Section 4.

. Cooling efficiency

The cooling effect is expressed here by the reduction in air tem-

erature yielded by a given landscape strategy, relative to thateasured at the same time in the base case configuration (“exposed

are”). The normalized hourly temperature differences are shownn Fig. 5, which illustrates the diurnal pattern of the effect (negativealues indicate cooling and positive values indicate relative heat-

able 6ooling efficiency calculated for the six landscape strategies, based on the ratio between t6:00–18:00), Sde-Boqer data, July–August 2007.

ase Cooling effect

�T [K] �QH [kJ m−2]

1) Mesh bare +0.9 +7562) Mesh grass +0.2 +1863) Trees bare −1.1 −9814) Trees grass −1.6 −13555) Exposed grass −0.3 −2926) Grass (under mesh)* −0.7 −5707) Grass (under trees)** −0.5 −374

he cooling efficiency ratio is multiplied by −100 to yield a positive percentage for net cohe values for cases (6) and (7) above were not measured directly, but were obtained inffect from (4) to (3).

ban Planning 92 (2009) 179–186

ing). The maximum cooling effect is obtained in the daytime hoursby the strategy with trees over grass.

Compared to the base case, shade trees created an averagereduction of 1.7 K in normalized Tmax (the maximum daily value,occurring at 14:00), whereas the shade mesh, counter-intuitively,caused an increase in Tmax of 0.8 K. The irrigated lawn reduced Tmax

by only 0.5 K when added to the bare courtyard, and contributedto a total reduction of 2.2 K at this hour when combined with thetrees. The grass also reduced the relative heating effect of the shademesh from +0.8 to +0.2 K. It may also be seen that the largest rel-ative temperature reductions in fact occurred during the morninghours, reaching as much as 2.5 K in the courtyard with trees andgrass.

Table 6 summarizes the cooling effect and total daily water usein the courtyards according to the different landscaping strategies.The daily water requirement of exposed grass was relatively large(6.1 l m−2), while its cooling effect was negligible (maximum ofonly −0.5 K). Grass under trees required only 3.0 l m−2 of water perday, and when combined with the trees’ transpiration (1.0 l m−2),yielded a total of 4.0 l m−2 per day. Thus, adding trees over grassreduces the total water consumption of the vegetation when com-pared to exposed grass alone. A similar reduction appears in thecase of the shade mesh over grass (4.4 l m−2 vs. 6.1 l m−2). Shadingthe grass thus resulted in a saving of 28% in the case of grass undera shade mesh, and of 51% for the grass under trees.

Adding shade not only reduced the water consumption of grass,but also increased its daily average cooling effect, from 0.3 to 0.7 K(under the mesh) and to 0.5 K (under the trees). Thus in both casesthe shading treatments created a synergetic effect, increasing the“efficiency” of the evaporative cooling with respect to water use.

Quantitatively, the cooling efficiency of the landscaping strate-gies was calculated as the ratio between the sensible heat removedfrom the space (�QH) and the amount of water supplied to it, withthe latter expressed as the equivalent latent heat of evaporation(�QE)—both given in kJ m−2 (Table 6):

�QH

�QE= �T�CpHcAch

LV ET(2)

The sensible heat �QH was calculated from the average differ-ence in daytime air temperature �T with respect to the base case(K), the volumetric heat capacity of the air based on its density �(kg m−3) and specific heat Cp (kJ kg−1 K−1), the height Hc of the aircolumn enclosed by the courtyard walls (m) and the exchange rateof the courtyard air volume as a product of the hourly exchange

rate Ac (estimated as 20 air changes per hour) and the number ofdaytime hours h per day (estimated as 12), under the given court-yard conditions. The latent heat was calculated from the latentheat of vaporization LV (2422 kJ kg−1 at 35 ◦C) and the total dailyamount of water evaporated ET (kg m−2), which was estimated per

he daily average cooling effect and water consumption for the total daytime period

Water use Cooling efficiency

ET [kg m−2] �QE [kJ m−2] �QH/�QE [%]

– – –4.4 7,531 −2.51.0 1,720 57.04.0 6,784 20.06.1 10,339 2.84.4 7,531 7.63.0 5,064 7.4

oling.directly as follows: *calculated cooling effect from (2) to (1); **calculated cooling

nd Ur

um

cpcaAl

esegwcbltc

awicdtaetccameem“itvtri

Fra

L. Shashua-Bar et al. / Landscape a

nit of landscaped area from the lysimeter and sap flow measure-ents.The last column in Table 6 shows the estimated cooling effi-

iency, with the ratio in Eq. (2) multiplied by −100 to yield a positiveercentage for net cooling. This indicator is used as a means ofomparing the relative cooling obtained from evaporating a givenmount of water in each of the different landscaping strategies.s such it can be used to judge the relative merits of the various

andscaping strategies, in addition to their effect on air temperature.It is clear from Table 6 that shade trees provide by far the high-

st cooling efficiency (57%), while the combination of grass andhade trees yields an efficiency of 20%. These results also show thextent to which shading can improve the net cooling efficiency ofrass—which on its own has a cooling efficiency of only 2.8%, buthen added to a courtyard that is shaded (by either trees or mesh)

ontributes to cooling with a net efficiency of about 7.5%. It shoulde emphasized that the cooling efficiency values given for the six

andscape strategy are indicative of their relative magnitude, andhat the precise values would obviously vary under different siteonditions.

Calculation of the cooling efficiency as defined above requiresn estimate of the air change rate in the courtyard. This datumas not measured directly, and is in fact quite difficult to obtain

n an open space. However, even while the absolute value of theooling efficiency is not known, the relative performance of theifferent configurations does not depend on this value. Modifyinghe (assumed) air change rate merely changes the magnitude offixed factor in the equation, resulting, for each landscape strat-

gy, in a unique linear relationship between the air change rate andhe cooling efficiency. Fig. 6 shows the range of cooling efficien-ies obtained for different hourly rates of air exchange Ac (in airhanges per hour—ACH), assuming the same (measured) �T. Their change rates represented in the graph range from 1 ACH to aaximum of 35 ACH. It is seen in Fig. 6 that by increasing the air

xchange rate the cooling efficiency increases proportionally forach strategy under the same conditions. The cooling efficiencyust be limited, by definition, to positive numbers: a negative

cooling efficiency” is obtained if the landscaping strategy resultsn a net heating effect, in spite of the evaporation of water, as inhe case of grass below a shading mesh. It is also possible to obtain

ery high values of the cooling efficiency, simply by assigning (arbi-rarily) high air exchange rates. Such values, though, are clearly notealistic, since such high rates would tend to erase any differencesn measured air temperature.

ig. 6. The effect on calculated cooling efficiency of different assumed air changeates in the courtyards, leaving all other measured inputs unchanged. (A value of 20ir changes per hour was used to calculate the values in Table 6.)

ban Planning 92 (2009) 179–186 185

5. Conclusions

This study presents empirical findings regarding the coolingeffect and water use of six landscaping strategies in a hot-aridregion. The focus is on vegetation in the urban environment, ata physical scale of small semi-enclosed spaces such as courtyardsor patios. The study introduces a criterion for judging the meritsof a landscape strategy in an arid region by computing its coolingefficiency with respect to water consumption per unit of groundarea.

Keeping in mind these limitations, the following conclusions areproposed:

1. The combination of shade trees over grass was predictably foundto be the most effective landscape strategy in terms of the coolingprovided, with the maximum air temperature reduced by up to2 K.

2. Somewhat unexpectedly, a shade mesh providing the sameamount of shade as the trees did not cool the air at the stud-ied sites, but rather caused a small relative heating effect (up to0.9 K).

3. Planting grass resulted in a smaller reduction in air temperature,yet consumed large amounts of water. Shading the grass, eitherby trees (preferably) or by a shade mesh, had the synergetic effectof increasing the cooling effect and reducing water consumption.

4. Both trees and mesh have the potential to improve outdoor ther-mal comfort in areas where they provide shade, as indicated bythe fact that in both cases the surface temperature of the shadedground was reduced substantially. Introducing grass under treesor under the mesh further reduces the ground temperature, thuscontributing even more to human comfort.

5. Trees provide by far the most efficient means of reducing outdoorair temperature, as measured by water consumption.

Acknowledgements

Dr. Shashua-Bar’s work was supported by fellowships from theBlaustein Center for Scientific Cooperation at Ben Gurion Universityof the Negev and from the Planning and Grants Committee of theIsrael Council for Higher Education.

Mr. Wolfgang Motzafi-Haller was responsible for the prepara-tion, installation and ongoing maintenance of the experimentalinstruments, and meteorological data from the adjacent weatherstation were provided by Mr. David Klepatch of the Energy and Envi-ronmental Physics Dept. of the Blaustein Inst. for Desert Research.

The authors are grateful to Prof. Pedro Berliner of the WylerDepartment of Dryland Agriculture at the Jacob Blaustein Institutesfor Desert Research for his assistance and insight over the courseof the study, and to Dr. Shabtai Cohen from the Department ofSoil, Water and Environmental Sciences at the Agricultural ResearchOrganization—Volcani Center for his help in applying the sap flowmethod of measuring evapotranspiration in trees and for his adviceon the experimental setup.

References

Alexandri, E., Jones, P., 2008. Temperature decreases in an urban canyon due to greenwalls and green roofs in diverse climates. Build. Environ. 43 (4), 480–493.

Bernatzky, A., 1982. The contribution of trees and green spaces to a town climate.Energ. Buildings 5, 1–10.

Bitan, A., Rubin, S., 1994. Climatic Atlas of Israel for Physical and EnvironmentalPlanning and Design. Ramot Publishing Company, Tel Aviv.

Bonan, G.B., 2000. The microclimates of a suburban Colorado (USA) landscape and

implications for planning and design. Landscape Urban Plan. 49, 97–114.

Brutsaert, W., 1982. Evaporation Into the Atmosphere. Kluwer Academic Publishers,Dordrecht/Boston/London.

Bruins, H.J., Berliner, P.R., 1998. Aridity, climatic variability and desertification. In:Bruins, H.J., Lithwick, H. (Eds.), The Arid Frontier: Interactive Management ofEnvironment and Development. Kluwer, Dordrecht.

1 nd Ur

C

CD

E

E

F

G

G

G

G

H

J

K

LM

M

M

O

O

P

P

P

Evyatar Erell is Associate Professor at the Jacob Blaustein Institutes for Desert

86 L. Shashua-Bar et al. / Landscape a

hang, C., Li, M., Chang, S., 2007. A preliminary study on the local cool-island intensityof Taipei city parks. Landscape Urban Plan. 80, 386–395.

hen, Y., Wong, N.H., 2006. Thermal benefits of city parks. Energy Build. 38, 105–120.imoudi, A., Nikolopoulou, M., 2003. Vegetation in the urban environment: micro-

climate analysis and benefits. Energy Build. 35, 69–76.rell, E., Williamson, T., 2006. Simulating air temperature in an urban street canyon

in all weather conditions using measured data from a reference meteorologicalstation. Int. J. Climatol. 26, 1671–1694.

liasson, I., 2000. The use of climatic knowledge in urban planning. Landscape UrbanPlan. 48, 31–44.

erguson, B.K., 2007. Water conservation methods in urban landscape irrigation: anexploratory overview 1. J. Am. Water Resour. Assoc. 23 (1), 147–152.

ash, J.H.C., Granier, A., 2007. A new method of sap flow measurement in tree trucks.In: Gash, J.H., Shuttleworth, W.J. (Eds.), Evaporation. IAHS Press, Wallingford.

rimmond, C.S.B., Isard, S.A., Belding, M.J., 1992. Development and evalua-tion of continuously weighting mini-lysimeters. Agric. Forest Meteorol. 62,205–218.

rimmond, C.S.B., 2007. Urbanization and global environmental change: local effectsof urban warming. Geogr. J. 173, 83–88.

uhathakurta, S., Gober, P., 2007. The impact of the phoenix urban heat island onresidential water use. J. Am. Plann. Assoc. 73 (3), 317–329.

emsely, A.R., Poole, I., 2004. The Evolution of Plant Physiology: From Whole Plantsto Ecosystems. Elsevier Academic Press, 492 pp.

ensen, M.E., Asce, M., Haise, H.R., 1963. Estimating evapotranspiration from solarradiation. J. Irr. Drain. Div.-ASCE 89 (RI4), 15–41.

remmer, O., Galon, I., 1996. Water Consumption Coefficients for Gardening andLandscape Irrigation. SHAHAM, Ministry of Agriculture, Israel (in Hebrew).

ankford, B., 2006. Localizing irrigation efficiency. Irrig. Drain. 55, 345–362.cPherson, E.G., Rowntree, A.R., Wagar, J.A., 1994. Energy-efficient landscapes. In:

Bradley, G. (Ed.), Urban Forest Landscapes—Integrating Multidisciplinary Per-spectives. University of Washington Press, Seattle/London.

eir, I., Pearlmutter, D., Etzion, Y., 1995. On the microclimatic behavior of two semi-attached courtyards in hot dry region. Build. Environ. 30, 563–572.

oller, M., Tanny, J., Li, Y., Cohen, S., 2004. Measuring and predicting evap-otranspiration in an insect-proof screenhouse. Agric. Forest Meteorol. 127,35–51.

ke, T.R., 1989. The micrometeorology of the urban forest. Phil. Trans. Royal Soc. B324, 335–349.

nmura, S., Matsumoto, M., Hokoi, S., 2001. Study on the evaporative cooling effectof roof lawn gardens. Energy Build. 33, 653–666.

aparelli, A., Kurban, A., Cunsulo, M., 1996. Strategies for bioclimatic design in an

urban area of an arid zone: San Juan (Argentina). Landscape Urban Plan. 34,19–25.

earlmutter, D., Berliner, P., Shaviv, E., 2006. Physical modeling of the pedestrianenergy exchange within the urban canopy. Build. Environ. 41, 783–795.

earlmutter, D., Kruger, E.L., Berliner, P., 2009. The role of evaporation in the energybalance of an open-air scaled urban surface. Int. J. Climatol. 29, 911–920.

ban Planning 92 (2009) 179–186

Perry, C., 2007. Efficient irrigation; inefficient communication; flowed recommen-dations. Irrig. Drain. 56, 367–378.

Potchter, O., Cohen, P., Bitan, A., 2006. Climatic behavior of various urban parks dur-ing hot and humid summer Mediterranean city of Tel Aviv. Israel. Int. J. Climatol.26, 1695–1711.

Potchter, O., Goldman, D., Kadish, D., Iluz, D., 2008. The oasis effect in an extremelyhot and arid climate: the case of southern Israel. J. Arid Environ. 72 (9),1721–1733.

Saaroni, H., Bitan, A., Ben Dor, E., Feller, N., 2004. The mixed results concerning the‘oasis effect’ in a rural settlement in the Negev desert, Israel. J. Arid Environ. 58(2), 235–248.

Shashua-Bar, L., Hoffman, M.E., 2000. Vegetation as a climatic component in thedesign of an urban street: an empirical model for predicting the cooling effectof urban green areas with trees. Energy Build. 31, 221–235.

Shashua-Bar, L., Hoffman, M.E., Tzamir, Y., 2006. Integrated thermal effects of genericbuilt forms and vegetation on the UCL microclimate. Build. Environ. 41, 343–354.

Spronken-Smith, R.A., Oke, T.R., 1998. The thermal regime of urban parks in twocities with different summer climates. Int. J. Remote Sens. 19, 2085–2104.

Spronken-Smith, R., Oke, T.R., Lowry, W.P., 2000. Advection and the surface energybalance across an irrigated urban park. Int. J. Climatol. 20, 1033–1047.

Stabler, L.B., Martin, C.A., Brazel, A.J., 2005. Microclimates in a desert city wererelated to land use and vegetation index. Urban Forestry Urban Greening 3 (3–4),137–214.

Steemers, K., 2003. Energy and the city: density, buildings and transport. EnergyBuild. 35, 3–14.

Svensson, M.K., Eliasson, I., 2002. Diurnal air temperatures in built-up areas in rela-tion to urban planning. Landscape Urban Plan. 61, 37–54.

Takebayashi, H., Moriyama, M., 2007. Surface heat budget on green roof and highreflection roof for mitigation of urban heat island. Build. Environ. 42, 2971–2979.

Limor Shashua-Bar is a post-graduate fellow at the Jacob Blaustein Institutes forDesert Research, Ben-Gurion University of the Negev. Her research has focused onthe microclimatic effects of vegetation in the urban environment.

David Pearlmutter is an architect and senior researcher at the Jacob Blaustein Insti-tutes for Desert Research of Ben-Gurion University in Israel. His research on urbanmicroclimate and other energy-related issues in the built environment has been pub-lished widely, and he is currently the editor of Urban Climate News—the quarterlynewsletter of the International Association for Urban Climate.

Research, Ben-Gurion University of the Negev. His work has focused on urban micro-climate and on passive cooling of buildings. He is the author of several books onplanning and environmental issues, including, Urban Clustering: The Benefits andDrawbacks of Location (Ashgate, 2001) with B. Portnov, and Roof Cooling Techniques—ADesign Handbook (Earthscan, 2006) with S. Yannas and J.L. Molina.