Science of the Total Environment 487 (2014) 350360

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Impact evaluation of greengrey infrastructure interaction on built-space

integrity: An emerging perspective to urban ecosystem service
Abhishek Tiwary a,, Prashant Kumar b,c
a
b
c

Faculty of Engineering and the Environment, University of Southampton, Higheld, Southampton SO17 1BJ, United Kingdom
Department of Civil and Environmental Engineering, Faculty of Engineering and Physical Sciences (FEPS), University of Surrey, Guildford GU2 7XH, United Kingdom
Environmental Flow (EnFlo) Research Centre, FEPS, University of Surrey, Guildford GU2 7XH, United Kingdom

H I G H L I G H T S

Greengrey interaction, i.e. impact of urban greening on built-up space is studied.

A lateral ecosystem function of GI in built-space integrity is identied.
Material surface recession for limestone and steel is computed under inuence of GI.
Material loss for steel is estimated to be over 5 times higher than for limestone.
GI species selection and seasonal variation inuence integrated ecosystem service.

a r t i c l e

i n f o

Article history:
Received 28 November 2013
Received in revised form 8 March 2014
Accepted 9 March 2014
Available online 3 May 2014
Editor: Simon Pollard
Keywords:
Air pollution
Building integrity
CFD
Doseresponse function
Ecosystem service
Green infrastructure

a b s t r a c t
This paper evaluates the role of urban green infrastructure (GI) in maintaining integrity of built-space. The latter
is considered as a lateral ecosystem function, worth including in future assessments of integrated ecosystem
services. The basic tenet is that integrated greengrey infrastructures (GGIs) would have three inuences on
built-spaces: (i) reduced wind withering from ow deviation; (ii) reduced material corrosion/degeneration
from pollution removal; and (iii) act as a biophysical buffer in altering the micro-climate. A case study is presented,
combining the features of computational uid dynamics (CFD) in micro-environmental modelling with the emerging science on interactions of GGIs. The coupled seasonal dynamics of the above three effects are assessed for
two building materials (limestone and steel) using the following three scenarios: (i) business as usual (BAU),
(ii) summer (REGEN-S), and (iii) winter (REGEN-W).
Apparently, integrated ecosystem service from greengrey interaction, as scoped in this paper, has strong
seasonal dependence. Compared to BAU our results suggest that REGEN-S leads to slight increment in limestone
recession (b10%), mainly from exacerbation in ozone damage, while large reduction in steel recession (up to
37%) is observed. The selection of vegetation species, especially their bVOC emission potential and seasonal
foliage prole, appears to play a vital role in determining the impact GI has on the integrity of the neighbouring
built-up environment.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Incorporating green infrastructure (GI) into the urban built-space
is gaining popularity as a cost-effective and long term measure for
mitigating climate change impacts associated with proliferating grey infrastructure globally (CABE, 2010; Hamdouch and Depret, 2010; Llauss
and Roe, 2012; MEA, 2005; Schfer and Swilling, 2013; Thaiutsa et al.,
2008). In essence, this is being achieved by utilising their ecosystem
functions i.e. facilitating interactions between ecosystem structure and

Corresponding author. Tel.: +44 7866187059.

E-mail addresses: a.tiwary@soton.ac.uk (A. Tiwary), p.kumar@surrey.ac.uk (P. Kumar).

http://dx.doi.org/10.1016/j.scitotenv.2014.03.032
0048-9697/ 2014 Elsevier B.V. All rights reserved.

processes that underpin the capacity of an ecosystem to provide goods

and services (Defra, 2011; TEEB, 2012). The UK National Ecosystem Assessment (NEA, 2011) has identied the following four broad categories
of ecosystem services i.e. benet people obtain directly or indirectly
from ecosystems: (i) supporting (i.e. facilitating habitats for species);
(ii) provisioning (i.e. generating resources); (iii) regulating (i.e. moderating climatic and biological effects), and (iv) cultural (i.e. recreational
and aesthetic). Exploring the potentials of quantitative and qualitative
approaches for assessing ecosystem services is a relatively new science,
developing rapidly through a combination of numerical modelling and
spatial analysis tools (Busch et al., 2012; Scholz and Uzomah, 2013).
Among the regulating services of GI, the majority of efforts till date
have been concentrated on assessing the direct benets, for example,

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

ecological and human health implications. The application of ecosystem

service values to a new area such as built-space integrity is a novel
contribution to knowledge and understanding. Such knowledge
development is vital for fostering an inclusive greengrey urban (and
landscape) planning, with the consideration for the extended ecosystem service to facilitate sustainable urban futures.
Ample efforts have gone in determining the role of vegetation on
urban microclimates, with numerous studies applying detailed
physical as well as CFD simulations to assess the modications
to pollution concentrations through coupled effects of building
morphology and vegetation on pollutant dispersion. These studies
fall under two schools of thinking, depending on the buildingvegetation biophysical interactions. One, projecting their positive
inuence by considering them as pollutant sinks (e.g., ltration and
absorption of particulates and NOx; Buccolieri et al., 2011; Tiwary
et al., 2006, 2009, 2013a,b). Two, elucidating their negative inuence
as obstacles to airow i.e. hampering the mixing of pollutants in
poorly ventilated areas close to streets and reduced air exchange
with the above-roof ambient environment (Gromke, 2011; Vos
et al., 2012; Wania et al., 2012).
The majority of vegetation studies on buildings have focussed
mainly on the assessment of thermal comfort (Ali-Toudert and Mayer,
2007; Berkovic et al., 2012; Berry et al., 2013; Santamouris, 2012; Yu
and Hien, 2006) and reduced building energy demands (Akbari et al.,
2001; Bouyer et al., 2011; Yang et al., 2012). A more recent study
evaluated the role of urban green commons comprising mainly of
collectively managed parks, community gardens and allotment areas
in developing resilience and environmental stewardship in cities
(Colding and Barthel, 2013). However, to our knowledge, no dedicated
assessment of the impact of GI on the integrity of the surrounding
grey infrastructure, including bridges, car parks and historical buildings, through their coupled aerodynamic and biophysical interactions
have been conducted so far. Developing such understanding is pertinent
to the on-going emphasis on enhancing GI investments as a tool in
large scale climate change adaptation strategies. Moreover, this would
aid holistic assessment of GIs by integrating all relevant sciences to sustain ecosystem services (Lundy and Wade, 2011; McMinn et al., 2010).
The relevance of such study is greater now in the face of recent
projections suggesting accentuations in the theoretical building dose
response functions (DRFs; the metrics commonly used to assess
integrated exposure of building materials due to air pollutants and
meteorological parameters.) under air pollution and changing environment, mainly owing to the altered micro-meteorological prole and
chemical withering of building materials (including concrete, steel,
stone, wood) under changing weather patterns (Brimblecombe and
Grossi, 2008; Kumar and Imam, 2013). Such impacts need to be
understood fairly swiftly, for both inner city and free-eld environments, in the context of the modications brought by the upcoming
GI interventions.
The aim of this study is to enhance the understanding of the role of
urban GI in ameliorating the micro-meteorological parameters and
pollutant concentrations in an urban space, and the impact of these
alterations on the material recession of surrounding built structures,
such as building walls and bridges. Essentially, the modelling approach
applied here is somewhat a hybrid assessment of what people have
seen until now in individual pockets. The case study demonstrates the
ecosystem services (or disservices) from GI in terms of their impact of
built-space integrity, which has not been adequately accounted for in
the conventional evaluation of their ecosystem functions so far. In
particular, the following three inuences of GI on the existing builtspace are assessed: (i) as quasi bluff bodies in modifying the wind
elds and withering; (ii) in reducing ambient pollution, and (iii) in
altering the micro-climate. All these collectively inuence the integrity
of neighbouring built-spaces. The study envisages promoting designing
of cohesive greengrey infrastructures (GGIs) as future of sustainable
city planning.

351

2. Methodology
2.1. Environmental modelling case study
The case study is designed to assess the role of GI for two contrasting
seasonal conditions (summer and winter), typically representative of
temperate climes. These were developed to understand the role of
varying microclimatic effects from GI intervention on the integrity of
inner-city built infrastructure both historical and new constructions.
Keeping this in mind, the scenarios covered solid limestone wall
structures (traditional buildings in European cities) and carbon steel
structures (modern buildings). The domain comprised of a busy street
canyon environment, exposed to trafc emissions, to ascertain the
level of intervention offered by GIs in modifying the following two key
factors inuencing building integrity: (i) microclimate (wind, temperature, humidity), and (ii) pollutant prole (source/sink).
2.1.1. Base case
As a rst step, a base case model was developed for business-asusual (BAU) scenario. A fast response building-resolved Lagrangian
dispersion modelling platform, QUIC Quick Urban and Industrial
Complex v5.81, with computational speeds and model complexities in
between a Gaussian and a CFD model, was applied (Nelson and
Brown, 2010). Its appropriateness for this task was ascertained based
on its recent applications in urban ow simulations around built-up
area (Hanna et al., 2011; Zwack et al., 2011). The modelling platform
comprises of three sequential components a city builder, a ow
simulator (QUIC-URB or QUIC-CFD), and a dispersion calculator
(QUIC-PLUME).
The QUIC model domain used a nested gridding with inner
domain of 300 m 300 m 20 m (length breadth height), mainly
covering the grey infrastructure (buildings, bridges and car parks)
(shown in Fig. 1). This was centred in an outer domain spanning
1000 m 1000 m 20 m, allowing for evolution of the ow in the
urban boundary layer to satisfy the guidelines for applications of CFD
to simulate urban ows (Franke et al., 2007; Tominaga et al., 2008).
The wind elds and pollutant dispersion for BAU were computed for
a typical inner-city street environment, comprising of cross-streets
lined with buildings, car parks (CP1, CP2) and over-bridges (B1B4)
(Fig. 1a). The foot bridges (B1, B2) are located close to the cross-street
intersection and the two cantilever car bridges (B3, B4) are located on
approach to the two car parks, adhering to the design specication for
over-bridges (DMRB, 2004). The meteorological inputs were acquired
from a local weather station, including wind speed, ambient temperature, relative humidity, and ambient pressure.
As explained in Section 2.1.2, the wind direction was intentionally
kept static at 210. The road emissions were modelled as line sources
for a typical European street environment (Table 1).
The simulation time period was set to allow the model to converge
on a steady state solution. Pollutant concentrations for BAU were
determined by quantifying the number of particles passing through a
constant grid volume (5 m 5 m 2 m) during the time period of
interest. Concentrations were calculated on 1-min average basis in
each grid volume. Pollutant concentrations were not calculated until
the rst released particles had passed completely over the domain and
exited the downwind side (starting at 300 s). This step ensured the
model computations to surpass evolutionary phase of the plume in
order to output steady state concentration (Nelson and Brown, 2010).
Overall, 766,500 QUIC particles were released over the entire 2000 s
simulation.
2.1.2. Inclusion of green infrastructure
Two important considerations were made while introducing the GI
for inuencing both the microclimate and the resulting pollutant
concentrations: (i) selection of vegetation species, and (ii) location of
the plantations. Use of large urban trees has been recommended in

352

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

Fig. 1. Planar view of the model domain (a) status quo (BAU, grey infrastructure only)
with cross-streets showing location of foot bridges (B1, B2) and cantilever car bridges
(B3, B4) across multi-storey car parks (CP1, CP2) respectively [note z = 10 m];
(b) Modied model domain for regenerated (REGEN, grey + green infrastructure) showing the location of the proposed vegetation patches (V1, V2) [shaded green, includes a
combination of shrubs and trees; buildings are colour-segregated on the basis of height].

the urban landscaping literature of the UK Construction Industry community to obtain higher benets (CIRIA, 2012). An earlier investigation
reported net annual benet of planting large tree species as 44% greater
than for medium tree species and 92% greater than for a small tree species (McPherson et al., 1999). However, large trees in close vicinity of
built structures tend to pose damage to the built environment due to
vigorous root growth. In this study we applied the following combination of three vegetation species with distinct seasonal characteristics
and vertical foliage proles to test the dynamic role of vegetation buffers
(their approximate area percentages provided alongside) deciduous
trees: Sycamore maple (Acer pseudoplatanus) (40%); deciduous hedgerow: Hawthorn hedge (Crataegus monogyna) (20%); coniferous tree:
Douglas r (Pseudotsuga menziesii) (40%). The hedgerows and trees
were allocated uniform heights of 2 m and 15 m respectively, which is
typical for inner city plantations in Europe. The idea was mainly to assess the microclimatic and pollution source/sink effects of deciduous
species (Sycamore and Hawthorn) with negligible foliage in winter
month to ascertain the holistic evaluation of GI effects. Our species

selection corroborates with a recent tree survey, reporting Sycamore

maple as the most abundant tree species in temperate and oceanic
climate (typically over 35% of the mix) (Scholz and Uzomah, 2013).
Further, the opted combination has been applied to assess the role
of new planting in PM10 capture and its human health benets for
London (Tiwary et al., 2009).
To simulate the regeneration (REGEN) scenarios, the BAU model
domain was modied to include two vegetation patches (V1 and V2),
away from streets and in the available open spaces upwind of the two
car parks CP1 and CP2 respectively (assuming the prevailing wind
enters the model domain in the lower left corner) (Fig. 1b). V1 and V2
were modelled respectively as high and low density vegetation canopy
buffer spaces, close to existing grey-infrastructure, using two different
arrangements of hedges and trees, typical of urban GI and commonly
found in temperate climes. While the area percentage of the three
selected species for both V1 and V2 were kept similar, the species
were grouped to test different congurations V1 was composed of
two rectangular blocks, comprising of conifers in the central part and
surrounded by deciduous trees and hedgerows in the outer ring. On
the other hand, V2 was composed of a line of conifer trees giving a
wind break effect, with a row of deciduous species (hedges and trees)
located immediately upwind of CP2 (Fig. 2).
Altogether two regeneration scenarios (summer, REGEN-S and winter, REGEN-W) were simulated using representative, and somewhat
contrasting, meteorological and vegetation parameters. The seasonal
variations to input congurations were adequately parameterised
summer was characterised by denser foliage and mild meteorological
conditions (low wind speed, high temperature, and low humidity);
winter was characterised by lower foliage in the crown and the ground
layers (Fig. 3) and aggressive meteorological conditions (high wind
speed, low temperature, high humidity) (Table 2). For the sake of generalisation, summer was considered as between April and September
and winter as between October and March; the micrometeorological
parameters for 2012 was applied to model the surface recession. This
year was chosen since 2012 has been recorded as a wet year in the UK
mainland, with nearly 800 mm rainfall in the midlands, the highest for
the last 6 years (about 30% more than the average year) with high
number of rain days.
The vegetation patches representing the GI in this study were
introduced upwind of the buildings and bridges assessed and away
from the street geometry (i.e. the pollutant source). Such practice follows recommendations from recent literature suggesting roadside
urban vegetation to be accentuating the pollutant concentrations
(mainly from localised sources, including trafc), owing to reduced
ventilation and poor mixing of the pollutants (Buccolieri et al., 2011;
Gromke and Ruck, 2009; Vos et al., 2012). This was ensured by keeping
the wind direction static at 210, which enabled the receptor locations
(CP1, CP2, B3, B4) used to assess the vegetation effects to remain downwind of the vegetation patches (V1 and V2) over the entire model run.
Owing to the lack of an all-inclusive vegetation modelling tool,
which can allow estimation of the required parameters for DRF calculations, inclusion of GI was evaluated in two steps. In the rst step, the
BAU set up in QUIC was modied using its vegetation modelling
features (Pardyjak et al., 2009) to simulate the two REGEN scenarios,
albeit with limited success since it does not allow for explicit resolution
of individual vegetation components (e.g., leaves, stems) of canopies.
These are parameterised in terms of their bulk attenuation coefcient
(Nelson and Brown, 2010), which can be either chosen from a library
of attenuation coefcients for a list of species ranging from orchards to
single/mixed species forests (Cionco, 1978), or can be customised. Essentially, this is an extension of the windbreak model (Raupach et al.,
2001), capable of simulating one-way interactions in terms of the bulk
drag effects of vegetation as bluff bodies on the mean air ow and
pollutant deposition. Due to underperformance of almost 4050% of
the deciduous species included in the vegetation buffers in winter, the
corresponding attenuation coefcients for winter period were kept

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

effectively 40% lower than that of the summer months (QUIC library
value of 4.03 for maple-r stand in REGEN-S and 2.42 for only r
stand in REGEN-W were used).
The QUIC model allowed evaluation of only two of the three vegetation effects on buildings scoped within this study one, bluff-body
effect, and the other, pollution reduction potential. It does not have
any mechanism to simulate the dynamic biophysical interactions between the vegetation components and the built-structure. Therefore,
in the next step, a 3D prognostic microclimate model, coupling the
principles of computational uid dynamics and thermodynamics
(ENVI-met; Bruse, 2013), was applied to evaluate the alteration in
the local microenvironment from inclusion of GI. Its capabilities of
modelling plantatmosphere interactions in city environments, simulating aerodynamics, thermodynamics and the radiation balance in
complex urban structures have been established through several
studies (Bruse and Fleer, 1998; Peng and Elwan, 2012; Rosheidat et al.,
2008; Spangenberg et al., 2008; Vos et al., 2012; Wania et al., 2012).
The model implements computational schemes of a conventional CFD
model into a detailed vegetation canopy module to capture the
two-way interactions of local vegetation on the wind eld and microclimate both the forward effect on the wind-eld and the thermodynamic feedbacks of the vegetation on the ambient air according to
position of the sun, urban geometry, vegetation, soils and various construction materials by solving thermodynamic and plant physiological
equations. This enabled more realistic description of the exchange
processes between the built- and green-infrastructure. Appropriate to
the need of our application, the numerical schemes further incorporate
these feedbacks while simulating the diffusion and deposition of pollutants (Steyn and Rao, 2010).
One limitation faced was that ENVI-met is designed for microscale
modelling so only a sub-set of the QUIC model domain, covering
110 m 70 m 20 m, with a grid resolution of 5 m 5 m 2 m
was selected for the simulation of two-way exchanges. The latter grid
cell size was chosen to make the computational steps consistent with
the QUIC model simulation (see Section 2.1.1); typical resolutions available in ENVI-met range between 0.5 m and 10.0 m (Bruse, 2013). Further,
to minimise the boundary effects, which may distort the output data, the
model uses an area of nesting grids around the core of the model to move
the model boundary away from the area of interest (Bruse, 2013). For this
purpose, the central portion of the QUIC domain covering the main features of analysis, including vegetation patches (V1, V2) and the studied
receptor location (CP1, CP2, B3, B4), were selected (Fig. 2).
Following the QUIC approach, the trafc emissions in ENVI-met
were modelled as cumulative line source emissions per lane. Representative foliage proles for V1 and V2 were provided using the generic
parameters in the plant database for hedge (2 m) and trees (15 m).
The corresponding leaf area density (LAD) (m2/m3) proles for the
three vegetation species for the two seasons, applied to the 10 layers
in the ENVI-met plan model (layer depth, z = 2 m), are shown in
Fig. 3. The maximum LAD for REGEN-S and REGEN-W is approximately

353

0.8 and 0.2 respectively (Table 2); the upright line in the lower most
layer of REGEN-S at LAD = 0.8 represents the uniform deciduous hedgerow. It is noteworthy that the species composition (i.e. area percentages
of the three species) of V1 and V2 is kept uniform over the two seasons
and the only difference is in the spatial distribution of the LADs due
to foliage loss in winter (as shown later in Fig. 4). This is meant for
evaluating the altered effects (if any) of the reduced GI intervention
in REGEN-W. The local meteorological variables applied to the base
case model were obtained from a weather station in Leicester and the
upper air radiosonde data, accessed from the homepage of University
of Wyoming (UWYO, 2013), for the closest sounding station at Watnall
near Nottingham (station reference number: 03354; Latitude: 53;
Longitude: 1.25; Altitude above mean sea level: 117 m; ~ 30 miles
from the study site). Representative summer and winter scenarios
were run as simulations for an entire day starting from daylight hours
(24 h from 0600 h GMT).
2.2. Building doseresponse evaluation
Two different building materials limestone and carbon steel
widely used in Europe, have been considered for the evaluation of
material recession in unsheltered environmental conditions for the
three scenarios (BAU, REGEN-S and REGEN-W). Evaluation of the impact of multipollutants and meteorological conditions on built-space
was assessed using the DRF approach. The DRFs serve as a tool for
assessing the material recession rate as a consequence integrated exposure of building materials to air pollutants (mainly NO2, SO2, PM10, O3
and CO2) and meteorological parameters (primarily ambient temperature, relative humidity, wind eld, pH). A number of DRF models are
available in the published literature and the ones selected for our estimates, as summarised in a recent review article by Kumar and Imam
(2013), are listed in appendix Table A.1 (Appendix A). The choice of
using more than one model is for comprehensiveness, essentially to
capture the range of variation in the estimates. For instance, four types
of DRF models are used for estimating the recession rate of limestone,
developed by Lipfert (1989), Tidblad et al. (2001), Kucera et al. (2007)
and Screpanti and De Marco (2009). Likewise, carbon steel DRFs are
used, which were developed by Kucera et al. (2007) and Noah's Ark
(2006).
In our study the driving parameters for DRF evaluation affected by
GI interventions at the four earmarked receptor locations include pollutant concentrations (NO2, SO2, PM10, O3) and prevalent meteorology
(ambient temperature, humidity, wind eld). These were obtained for
each of the three scenarios from the micro-environmental modelling
steps; air temperatures were rounded off to the nearest whole number
(Table 3). Dry deposition velocities for HNO3 are based on the values
reported in the literature and assumed to be 0.38 and 0.32 cm s1 respectively (Kumar and Imam, 2013; Sabboni et al., 2006). A uniform
CO2 concentration was applied to the DRF assessments (383 ppm) for
all four receptor locations. This can be argued to be acceptable since

Table 1
Descriptors for road properties used in the BAU model set-up as shown in planar view of Fig. 1a.
Road link ID

Width (m)

Link length (m)

Link 1 (L1)
Link 2 (L2)
Link 3 (L3)
Link 4 (L4)
Link 5 (L5)
Link 6 (L6)
Link 7 (L7)
Link 8 (L8)

20
19
21
20
18
15
27
33

87.5
82.6
143.5
98
118
109
102.2
92.7

UTMC Geo-referencing coordinate system.

Start coordinatesa

End coordinatesa

North

East

North

East

303563
303620
303470
303509
303492
303445
303352
303283

457591
457611
457642
457422
457441
457862
457472
457440

303465
303672
303465
303541
303465
303455
303465
303352

457543
457641
457543
457308
457543
457711
457543
457472

Building height (m)

Direction (N)

7
6
5
6
7
6
10
6

25
25
273
287
287
273
30
30

354

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

Fig. 2. ENVI-met simulation domain. The left panel shows portion of the QUIC model domain used for extended modelling in ENVI-met in the inset. (Gridded red mesh, separated by
white space, represents two-lane trafc on each road; different shades of green mesh represent vegetation patches: Light green deciduous species; dark greenconifer species.)

our aim was to analyse the relative effect of pollutant concentrations on

structural material in the presence of vegetation. CO2, largely being inert
and abundantly available, is expected to remain spatially uniform for
the four receptor sites. Likewise, the pH was also assumed to remain
uniform as 5.2; representative Lipfert value of 18.8 was applied to the
estimation following Brimblecombe and Grossi (2008). The maritime
inuence on the Karst effect was ignored, given the study site was
located away from sea in the midlands. Likewise, the estimates were
made for clean precipitation, given that deposition of sea salt aerosol
has maximum effect within the rst 100 m (Bonazza et al., 2009),
which was considered negligible for the case study site in the UK
midlands.
It is worth noting that extensive model validation (i.e. crosscomparison) exercise was not scoped within this study, mainly owing
to the complexities in setting up a dedicated eld measurement
campaign (or a wind tunnel experiment) for validating the modelled
parameters alongside. The model scenarios were developed using a
set of static vegetation and meteorological parameters, without

inclusion of all possible uncertainties therein. This begs a level of prudence while interpreting the results in the following sections as absolute values, accommodating for the uncertainties likely to propagate
from individual modelling stages into the nal outcome. While we
have incorporated the level of variations in the predictions of vegetation
effects on building integrity from different DRF models for the two
seasons (as error bars in Fig. 5), our results should be considered only
as overall estimates of the impacts such interactions may have to highlight the need for their inclusion in future integrated ecosystem
assessments.
3. Results and discussions
3.1. Evaluation of environmental parameters
A comparison table has been generated (Table 3) for the modelled
micro-environmental parameters (air temperature, relative humidity,
wind speed) and pollutant concentrations (NO2, SO2, PM10, O3) output

Fig. 3. Leaf area density proles (LAD; m2/m3) for the three vegetation species (maple, r and hawthorn) used in 10 layers of ENVI-met model for the two scenarios: (a) REGEN-S,
(b) REGEN-W.

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

355

Table 2
Initial conguration data applied in the ENVI-met model scenarios.
BAU (summer '12)

REGEN-S (summer '12)

REGEN-W (winter '12)

Atmosphere
Simulation date/time range
(GMT)
Wind speed at reference height (10 m above ground) [m s1]
Wind direction (degrees)
Initial air temperature [K]
Relative humidity at 2 m [%]
Specic humidity at 2500 m [g water/kg air]a
Perceptible water [mm]

09 Aug 2012
(06002400 h)
5.2
210
283
78
5.5
5.2

09 Aug 2012
(06002400 h)
5.2
210
283
78
5.5
5.2

01 Jan 2012
(06002400 h)
3.7
210
270
94
7.9
20.6

Buildings
Albedo of walls
Albedo of roofs

0.4
0.3

0.4
0.3

0.4
0.3

Vegetation (see Fig. 3)

Maximum leaf area density (LAD) [m2 m3]
Tree crown [m]
Shrubs/hedge [m]

N/A
N/A
N/A

0.8
15
2

0.2
15
2

Source: University of Wyoming (UWYO, 2013).

at four strategically selected receptor locations two car parks

(CP1, CP2) and two bridges (B1, B3) (see Fig. 1 for spatial references
of these receptors). CP1 and CP2 were considered suitable as the
two built structures immediately downwind of the high density and
the low density vegetation patches (V1 and V2, respectively); B1
represented a cross-street location downwind and away from trees
(i.e. unperturbed site); B3 represented a deeper canyon location in
side street L5, downwind of car park CP1. For all these receptor locations
the simulation outputs were obtained for the three modelled scenarios
(BAU, REGEN-S and REGEN-W). It is noteworthy that BAU can only be
directly compared with REGEN-S owing to similarity in underlying
meteorology, whereas REGEN-W had inherently dominant winter
characteristics in both foliage prole and meteorology. This pattern of
model comparison is adopted hereafter throughout the discussion.
Preliminary results from this assessment indicate inclusion of GI to
be largely affecting humidity and wind elds, with only marginal
inuence on the ambient air temperature. Compared to BAU the relative
humidity downwind of dense vegetation (V1) is found to be about 10%
higher for REGEN-S and the corresponding value is about 20% higher for
REGEN-W. Overall, REGEN-S showed lowering while REGEN-W shows
slight increment of air temperature at CP1 and CP2 compared to B1;
the relative reductions being nearly two-fold higher closer to high density patch (V1) compared to low density patch (V2). This is attributable
to the fact that vegetation can lower the temperature of the air and can
increase the humidity of the air during hot summer. These observations
are consistent with previous studies (Spangenberg et al., 2008;
Yang et al., 2012), suggesting the cooling effects of urban vegetation.
However, the reported trends are based on pure modelling exercise,
which is subjected to numerous uncertainties both during evaluation
of the individual parameters and from their application in the model
formulations. As a consequence, these estimates should only be treated
as a pathway towards developing any strategic implementation plan for
future GIs. Nonetheless, we demonstrated successful implementation of
this tiered modelling approach in assessing the impacts of urban green
on built-up environment, giving some vital insights into the green
grey interactions in the inner city environment.
For REGEN-W, regions with high density patch (V1) and low
density patch (V2) were respectively 1.1 C and 0.8 C (i.e. slightly
higher air temperature than BAU). Such warming, instead of cooling in
sub-zero temperatures with low sunlight (hence reduced or negligible
evapotranspirative cooling), is owing to the fact that shading and evaporative cooling effect of the vegetation are hugely reduced in winter,
which is benecial for buildings. This has also been observed in another

study for winter air temperature simulations (Yang et al., 2012) and
attributable mainly to the inactive evaporation from vegetation in low
sunlight regime, augmented by the discounted contributions of lost
foliage from deciduous trees in winter.
The second half of Table 3 lists the concentration distribution for
a number of regulated pollutants at the selected receptor locations
which are considered crucial for estimation of surface recession of
limestone and carbon steel (Section 3.2). A general spatial and seasonal
pattern for pollutant distribution was noted for the chosen receptor
locations. This essentially reected the compounding effects of the
underlying model mechanism, with strong association with proximity
to the street geometry, meteorology and vegetation source/sink effects.

Fig. 4. Spatial plot of regeneration scenarios for wind speed change (%) output showing
the seasonal dependence on meteorological and vegetation effect. Upper panel summer
(REGEN-S); lower panel winter (REGEN-W). (Darker green vegetation (both V1 and
V2) in REGEN-S represents additional deciduous foliage, leading to higher effective LAD).

356

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

For example, NO2 concentrations at bridge locations, being closer to the

road sources, were higher than off-road car park locations; B1 showed
higher values than B3 because of being located downwind of the intersection. Although the SO2 concentrations remained slightly higher close
to road sources (B1, B3) compared to off-road sites (CP1, CP2); the SO2
loadings were found to be fairly uniform, mainly owing to the fact that
modern vehicles have marginal sulphur emissions. It is worth noting
that the winter concentrations bear resembling distribution prole, except showing higher values across the whole model domain. This is possibly due to the lowering of the boundary layer during colder months,
leading to localised enhancement of pollution at these sites. The concentrations for REGEN-S were generally lower than for BAU for most of the
pollutants accounting for the sink terms, except for ozone. Slight increments were observed for the latter, especially at off-street locations
(CP1, CP2), possibly from enhanced ozone photochemistry in the presence of bVOC active broad-leaved maple during the summer. However,
for REGEN-W the corresponding concentrations were much lower compared to BAU, which could be due to lack of precursor bVOCs and low
solar radiation.

Table 3
Microclimate and pollutant concentrations (hourly average) at different receptor locations
for the three scenarios modelled [Note: Reference height = 10 m; Geo-reference origin
(UTMX, UTMY): 457,350 m, 3,033,000 m; Relative coordinates of receptors (+x, +y):
CP1 (252, 148), CP2 (123, 160), B1 (198, 188), B3 (253, 168)].

Annual rainfall [mm]

Air temperaturea [K]
CP1
CP2
B1
B3
Relative humiditya [%]
CP1
CP2
B1
B3
Wind speeda [m s1]
CP1
CP2
B1
B3
NO2 [gm3]ac
CP1
CP2
B1
B3
SO2 [gm3]d,e
CP1
CP2
B1
B3
PM10 [gm3]bd
CP1
CP2
B1
B3
O3 [gm3]a,c,e
CP1
CP2
B1
B3

BAU

REGEN-S

REGEN-W

602

602

962

285
285
285
285

279
281
285
282

273
274
272
272

72
72
72
72

80
78
72
75

89
95
83
87

4.6
4.6
4.8
4.7

1.0
2.4
4.3
2.4

1.3
1.9
2.6
1.3

6.98E05
3.09E05
1.64E04
1.19E04

6.77E05
3.00E05
1.55E04
1.12E04

8.17E05
3.37E05
2.21E04
1.97E04

3.87E06
3.20E06
4.78E06
4.23E06

3.79E06
3.14E06
4.73E06
4.21E06

5.26E06
4.15E06
7.33E06
5.47E06

1.83E05
1.67E05
1.90E05
1.63E05

1.65E05
1.46E05
1.73E05
1.47E05

1.91E05
1.85E05
2.02E05
1.76E05

8.24E05
6.87E05
7.02E05
6.44E05

9.06E05
7.26E05
7.72E05
7.08E05

3.37E05
2.42E05
1.64E05
1.17E05

CP = car park; B = bridge.

a
From ENVI-met.
b
Only trafc source.
c
Accounts for additional sources/sinks under vegetation effects during summer for
O3, NO2 (Tiwary et al., 2013a) and PM10 (McDonald et al., 2007).
d
From QUIC.
e
Trafc + urban background.

3.2. Evaluation of building integrity

Utilising the micro-environmental parameters obtained from previous steps, the building integrity was evaluated in terms of surface
material recession based on DRF. As described in Section 2.2 this exercise was limited to limestone and carbon steel in the study, restricted
by the availability of doseresponse formulations for these two materials extensively in the literature. The resulting surface recession
stimates at the four receptor locations CP1, CP2, B1 and B3 for these
two materials are compared from the available models in Figs. A.1
and A.2, respectively. The surface recession is estimated in terms of
depth of material loss (m) in a year. These can be converted to
annual mass of material loss per unit area (g m2) by multiplying the
material surface recession (m yr1) with the density of carbon steel
(~ 7850 kg m 3) or lime stone (~ 2160 kg m 3; for type II medium
density). For example, this approach gives ~ 106.23 and 18.95 g of
material loss per m2 area per year for 13.52 and 8.77 m yr1 of surface
recession in BAU (CP1) for carbon steel and lime stone, respectively.
Surface recession (m yr1) in this particular case is ~1.54 times higher
for carbon steel compared with lime stone, but mass of material loss
comes out ~ 5.60 times higher for carbon steel than those for lime
stone, because of much higher density of the former material. The
approximation of the material loss can be made accordingly for values
presented for other scenarios in Fig. 5.
It is obvious from these gures that different DRF models provide
variable results, which can be explained by the sensitivity of these
models towards the various pollutants. In these models SO2, NO2, and
O3 are considered as important corrosive gases; SO2 maintains a nonlinear relationship with corrosion and its corrosive effect is maximum
at a temperature of about 911 C (Kucera et al., 2007). However,
given the emission source was restricted to urban trafc SO2 is not
found to be a dominant pollutant in our case which leaves NO2 and O3
as major contributors to the recession rates. Based on the model
parameterisation (Table A.1) the DRF estimates for surface recession
are found to be inuenced in the following order by the underlying factors considered in this study (see Table 3): Limestone Rain NO2 N
O3 N SO2 N PM10. Carbon steel air temperature N SO2 N PM10. Relative
humidity has similar implications for all the materials and scenarios
included in the study. Broadly, the models for limestone are based
on the Lipfert function approach which has greater sensitivity to
precipitation/rain (typical Lipfert function value used is 18.8 rain).
This dominates the whole surface recession estimates for limestone as
clearly noted in Fig. A.1. On the other hand, carbon steel has not got
NO2 and O3 effect, not because these do not affect it, but because these
are not part of available DRFs. The surface recession estimates for steel
are more sensitive to corrosive effects of pollutants (peaking at air
temperature of about 282284 K) and therefore show much wider
spatial variation for all the DRF models included in this assessment
(Fig. A.2).
To show the relative changes from GI interventions, the surface
recession estimates for limestone and carbon steel at the four receptor
locations have been obtained as average from all available models
(Fig. 5). Shown alongside in the same plots are the standard deviations,
demonstrating the level of variations in the predictions from different
models. For consistency, the observed values have to be compared
separately in two sets, as follows BAU is compared with REGEN-S as
they are both using summer conditions (except the latter scenario
incorporates additional biophysical effects of the introduced vegetation
on the microenvironment and pollutant source/sink); REGEN-S is
compared with REGEN-W to evaluate the seasonal dynamics in surface
recession estimates, including the relative changes arising from the
coupled vegetation-microenvironment effects (specically the inuence of reduced foliage from leaf shedding by deciduous species
maple and hawthorn) as well as the inuence of the altered boundary
layer on pollutant concentrations in sub-zero temperatures. The two
sites immediately downwind of the vegetation patches (CP1 and CP2)

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

Fig. 5. Average recession of (a) lime stone and (b) carbon steel for all the three scenarios
[note: only positive values of standard deviation are added for the clarity of gures].

show more prominent inuences (limestone up to 10% increase; steel

2837% decrease), whereas the site away from the vegetation patch
(B1) has only marginal inuence (limestone b2% increase; steel
up to 7% decrease). It is noteworthy that the effects observed at B1 for
steel in REGEN-W is arising from ambient wind conditions, away from
vegetation effects (i.e. unperturbed site), which is much higher in winter (see Table 3). The observed increase for limestone surface recession
in the presence of vegetation is attributed to slight enhancement in
ground level ozone from introduction of bVOC active vegetation in
REGEN-S. This demonstrates the importance of species selection in
optimising the ecosystem functions of GI on limestone buildings by
avoiding exacerbation of ground level ozone during summer. On the
other hand, steel is found to have greater reduction in surface recession
estimates through GI intervention in summer, primarily owing to its independence from ozone damage.
However, some caution is required while interpreting the trends
observed in Fig. 5 as apart from vegetation effects there are additional
inuences incorporated in the model arising purely as artefacts of
contrasting meteorological parameters between REGEN-S (which is
same as BAU) and REGEN-W, which cannot be associated with vegetation as such. For example, evaluation of the seasonal effects alongside
GI intervention shows reverse effects on surface recession values for

357

limestone and carbon steel at B3 over the two contrasting seasons,

which can be explained using the model parameters presented in
Table 3 (much lower air temperature in winter months compared to
BAU). Further, for limestone the average values at all the four receptors
are slightly higher during winter (i.e. REGEN-W N REGEN-S) (Fig. 5a)
whereas for carbon steel the corresponding values are much lower
during winter (Fig. 5b). This is an interesting observation, useful to
both research and planning communities, to take into account the
varying seasonal inuence of GI on different building materials. The observed winter enhancement of limestone material recession is primarily
owing to heavier rain over the winter months compared to the summer
months during the simulation period (see Table 2) combined with twofold effects on exacerbation of ground level pollutant concentrations
one, due to loss of the vegetation sink from foliage loss by deciduous
trees and hedges; two, from lowering of the atmospheric boundary
layer in sub-zero temperatures (see Table 3). On the other hand, the observed summer enhancement of steel recession is primarily attributed
to favourable temperature range of 282284 K, maximising the corrosive effects of acidic pollutants.
Apparently, due to the high density of built-up areas in the core
model domain (about 82% including roads, bridges and buildings;
Fig. 1), the overall surface recession reduction from the two vegetation
patches (V1 and V2) is not substantial, albeit indicative of the potential
for additional inuence such intervention would hold for integrated
greengrey infrastructure planning at the city-region levels. While our
study mainly focussed on evaluating the role of different species on
the basis of seasonal parameterisation of LAD, it revealed some inherent
characteristics of GI which are strongly dependent on their species
composition, including inhibition of particulate sink over winter,
enhancement of ozone formation potential and wind speed reduction
over summer. This is going to be enhanced further on the basis of
appropriate vegetation selection, mainly the mix of evergreens with deciduous species to compensate for the seasonal effects in the face of climate change catering to both warmer summers and harsher winters.
Weighing all the negative and positive inuences of GI (both existing
and planned) in the urban ecosystem against each other is nearimpossible and, as we showed through this example of estimating
building integrity, is heavily marred by non-availability of all-inclusive
model formulations. The thrust of the majority of such evaluations is
currently on improving air quality and/or thermal comfort, and conservation of building surfaces, as highlighted through this study, is very
much an emerging perspective of greengrey interactions (which is
envisaged to get more intense with further increase in GGIs). However,
it is recommended that these ndings be used only to get broader
insight into this integrated urban ecosystem service; further scrutiny
of detailed evaluation should take into account the uncertainty aspect
of these interactions.
4. Conclusions
This study evaluated the role of modied urban microenvironment
through inclusion of GI on building integrity; the metrics adopted is
material surface recession of limestone and carbon steel. The dynamic
seasonal characteristics in meteorology and foliage prole (for 60%
deciduous component in the simulated vegetation) have been incorporated through two case study scenarios (for summer and winter) to portray the varying degrees of impacts over a year. The assessment has
been conducted in two stages rst, utilising CFD modelling capabilities to quantify the aerodynamic features and bio-physical interactions
between the grey and the green components of the city. In the
next step, adequate model parameterisation from available literature
was applied to estimate the coupled effects of pollutants and microenvironmental variables on building material recession.
Our ndings show the inuence of GI on built-space integrity in
terms of differentiating the four receptor locations used in the assessment. For example, the two sites immediately downwind of the

358

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

vegetation patches (CP1 and CP2) show prominent changes in the summer scenario with GI intervention, REGEN-S compared to the BAU scenario. However, contrasting seasonal inuences of GI on the surface
recession rates of the two building materials have been noted. Slight increment in the surface recession is observed for limestone during winter whereas large reductions are found in recession of carbon steel
during summer. This is quite revealing, as most GI assessments till
date would assume only the positive inuences of vegetation as windbreak and pollution sink, overlooking their pollution source contributions in affecting neighbouring built-space. The latter gains relevance
in our study since both Sycamore maple and Douglas r (making up
to 80% of the vegetation buffer in our case study) are active sources of
bVOC emissions over summer months. Given limestone recession is
strongly inuenced by ground-level ozone, availability of ozone precursors (through photochemical interactions of bVOC emissions with NO2
from trafc) explains the observed increment. Hence, our integrated assessment of GI intervention on built-space integrity (Fig. 5) has shed
light on their varying, and apparently reciprocal effects on the two
building materials, primarily inuenced by the bio-physical characteristics of the constituent vegetation species and meteorological factors.
The former gains relevance in summer months in terms of the enhanced
bVOC emissions, serving as ozone precursor (a major contributor to surface recession for limestone). The latter gains relevance in winter
months in exacerbating pollutant concentrations under harsher meteorology in temperate climes (primarily owing to lowering of the atmospheric boundary layer during the colder months). While our
modelling exercise provides broader insight and overall estimates of
the interactions between the greengrey infrastructure and integrity
of built-up space, studies focusing on detailed model validation exercises are needed for accurate estimations and for reducing the levels
of uncertainty in the results.

Our study has shown the relevance of GI for future sustainability of

greengrey infrastructure. We encapsulated the plausibility of a lateral
ecosystem function of GI in built-space integrity, beyond the direct
human benets identied under the regulating services of GI under
the generic ecosystem service variable local climate and air quality
regulation (LCAR; LCAR accounts for the effects of trees and other plants
in lowering the temperature by providing shade and inuence water
availability (e.g., evapotranspiration); regulating air quality by removing pollutants from the atmosphere (e.g., ltration and absorption of
particulates and NOx)). As a natural next step, this would warrant
quantication of the lateral ecosystem functions offered by these
initiatives in future urban environments, which are currently not
taken into account as part of ecosystem service (NEA, 2011). Our results
also highlight some of the challenges faced in spatial modelling of ecosystem services. More research is therefore recommended to develop
the ecosystem service assessment approach further into a numerical
model.

Acknowledgements
We acknowledge the technical assistance from Matt Nelson and
Michael Brown at the Los Alamos National Laboratory, USA in setting
up the QUIC model and from Prof. Michael Bruse, University of Mainz,
Germany in setting up the ENVI-met model. Acknowledgments
are also due to Peter Vos at Flemish Institute for Technological
Research (VITO), Belgium in clarifying few issues in modelling the
micro-environmental feedbacks from vegetation intervention in built
space using ENVI-met. We would also like to acknowledge the invaluable contributions from the anonymous reviewers towards further
enhancement of this paper.

Appendix A

Table A.1
Summary of DRFs used for our estimates; table adopted from Kumar and Imam (2013). Please note that ML and R stand for mass loss by corrosion attack in g m2 and surface recession or
thickness loss in m (N1-year exposure) or m yr1 (1-year exposure), respectively. Gaseous and ion concentrations are annual mean in g m3 and mg lit1. Dcl is chloride deposition
(mg m2 day1) and Rh60 = (Rh60) when Rh N 60; otherwise 0. Rn is precipitation in m yr1; VdS and VdN are deposition velocities (cm s1) for SO2 and HNO3, respectively.
Material

Doseresponse function

Source

Carbon steel

[0.033Rh + 0.040T]
R = 1.58[SO2]0.52 e[0.02Rh + fCs(T)] + 0.166Rn[H+] + 0.0761 PM10 + 0.102D0.33
Cl e
fCs(T) = 0.150(T 10) when T 10C
fCs(T) = 0.054(T 10) when T N 10C
R = 1.77[SO2]0.52 e[0.20Rh + fws(T)] + g(Cl, Rh, T)
ML = 29.1 + t0.6 (21.7 + 1.39[SO2]0.6 Rh60 efWs(T) + 1.29 Rn[H+] + 0.593 PM10)
fws(T) = 0.150(T 10) when T 10C
fws(T) = 0.054(T 10) when T N 10C
R = 2.7 [SO2]0.48 e0.018T t0.96 + 0.019 Rn[H+] t0.96
R = 3.1 + t(0.85 + 0.0059 Rh60 [SO2] + 0.078 Rh60 [HNO3] + 0.054Rn[H+] + 0.0258 PM10)
[HNO3] = 516 e3400/(T + 273) ([NO2][O3] Rh)0.5
R = 18.8 Rn + 0.016 [H+] Rn + 0.18 (VdS [SO2] + VdN [HNO3])
a
R = 3.1 + t (0.85 + 0.0059[SO2] Rh60 + 0.054 Rn[H+] + 0.078 (516 e3400/(T + 273) ([NO2] [O3] Rh)0.5 Rh60) + 0.0258 PM10)

Noah's Ark (2006)

Portland
limestone

Nomenclature:
ML = mass loss (g m2).
R = surface recession (m yr1).
SO2 = sulphur dioxide (g m3).
NO2 = nitrogen dioxide (g m3).
O3 = ozone (g m3).
PM10 = particulate matter 10 m in diameter (g m3).
T = ambient temperature (C).
t = time (years).
fCs(T) = correction factor depending on temperature ().
fws(T) = correction factor depending on temperature ().
Rh = relative humidity (%).
VdS = deposition velocity of sulphur dioxide, SO2 (cm s1).
VdN = deposition velocity of nitric acid, HNO3 (cm s1).
Dcl = chloride deposition (mg m2 day1).
Rn = precipitation (m yr1).

Kucera et al. (2007)

Tidblad et al. (2001)

Kucera et al. (2007)
Lipfert (1989)
Screpanti and De Marco (2009)

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

Fig. A.1. Recession of lime stone for all the three scenarios, using four different models.

359

Fig. A.2. Recession of carbon steel for all the three scenarios, using two different models.

360

A. Tiwary, P. Kumar / Science of the Total Environment 487 (2014) 350360

References
Akbari H, Pomerantz M, Taha H. Cool surfaces and shade trees to reduce energy use and
improve air quality in urban areas. Sol Energy 2001;70:295310.
Ali-Toudert F, Mayer H. Effects of asymmetry, galleries, overhanging faades and
vegetation on thermal comfort in urban street canyons. Sol Energy 2007;81:74254.
Berkovic S, Yezioro A, Bitan A. Study of thermal comfort in courtyards in a hot arid
climate. Sol Energy 2012;86:117386.
Berry R, Livesley SJ, Aye L. Tree canopy shade impacts on solar irradiance received by
building walls and their surface temperature. Build Environ 2013;69:91100.
Bonazza A, Messinaa P, Sabbionia C, Grossi CM, Brimblecombe P. Mapping the impact
of climate change on surface recession of carbonate buildings in Europe. Sci Total
Environ 2009;407:203950.
Bouyer J, Inard C, Musy M. Microclimatic coupling as a solution to improve building
energy simulation in an urban context. Energy Build 2011;43:154959.
Brimblecombe P, Grossi C. Millennium-long recession of limestone facades in London.
Environ Geol 2008;56:46371.
Bruse Mhttp://www.envi-met.com, 2013. [accessed in Feb 2013].
Bruse M, Fleer H. Simulating surfaceplantair interactions inside urban environments
with a three dimensional numerical model. Environ Model Software 1998;13:
37384.
Buccolieri R, Salim SM, Leo LS, Di Sabatino S, Chan A, Ielpo P, et al. Analysis of local scale
treeatmosphere interaction on pollutant concentration in idealized street canyons
and application to a real urban junction. Atmos Environ 2011;45:170213.
Busch M, Notte AL, Laporte V, Erhard M. Potentials of quantitative and qualitative
approaches to assessing ecosystem services. Ecol Indic 2012;21:89103.
CABE. Urban green nation: building the evidence base. London: UK Commission for
Architecture and the Built Environment; 2010.
Cionco RM. Analysis of canopy index values for various canopy densities. Bound-Layer
Meteorol 1978;15:8193.
CIRIA. The benets of large species trees in urban landscapes: a costing, design and
management guide. London: Construction Industry Research and Information
Association; 2012.
Colding J, Barthel S. The potential of Urban Green Commons in the resilience building of
cities. Ecol Econ 2013;86:15666.
Defra. Overarching impact assessment for the Natural Environment White Paper. London:
UK Department of the Environment, Food and Rural Affairs; 2011 [url-http://www.
archive.defra.gov.uk/environment/natural/documents/newp-ia-110607 [accessed in
Feb 2012]].
DMRB. Design manual of roads and bridges design criteria for footbridges; London.
url- http://www.dft.gov.uk/ha/standards/dmrb/vol2/section2/bd2904.pdf, 2004.
[accessed in Jul 2012].
Franke J, Hellsten A, Schlnzen H, Carissimo B. Best practice guideline for the CFD simulation of ows in the urban environment, COST Guidelines. Hamburg: European
Science Foundation; 2007.
Gromke C. A vegetation modeling concept for building and environmental aerodynamics
wind tunnel tests and its application in pollutant dispersion studies. Environ Pollut
2011;159:20949.
Gromke C, Ruck B. Effects of trees on the dilution of vehicle exhaust emissions in urban
street canyons. Int J Environ Waste Manage 2009;4:22542.
Hamdouch A, Depret M-H. Policy integration strategy and the development of the green
economy: foundations and implementation patterns. J Environ Plann Manage 2010;
53(4):47390.
Hanna S, White J, Trolier J, Vernot R, Brown M, Gowardhan A, et al. Comparisons of
JU2003 observations with four diagnostic urban wind ow and Lagrangian particle
dispersion models. Atmos Environ 2011;45:407381.
Kucera V, Tidblad J, Kreislova K, Knotkova D, Faller M, Reiss D, et al. UN/ECE ICP materials
doseresponse functions for the multi-pollutant situation. Water Air Soil Pollut Focus
2007;7:24958.
Kumar P, Imam B. Footprints of air pollution and changing environment on the sustainability of built infrastructure. Sci Total Environ 2013;444:85101.
Lipfert FW. Atmospheric damage to calcareous stones: comparison and reconciliation of
recent experimental ndings. Atmos Environ (1967) 1989;23:41529.
Llauss A, Roe M. Green infrastructure planning: cross-national analysis between the
North East of England (UK) and Catalonia (Spain). Eur Plan Stud 2012;20(4):
64163.
Lundy L, Wade R. Integrating sciences to sustain urban ecosystem services. Prog Phys
Geogr 2011;35:65369.
McDonald AG, Bealey WJ, Fowler D, Dragosits U, Skiba U, Smith RI, et al. Quantifying the
effect of urban tree planting on concentrations and depositions of PM10 in two urban
conurbations. Atmos Environ 2007;41:845567.
McMinn WR, Yang Q, Scholz M. Classication and assessment of water bodies as adaptive
structural measures for ood risk management planning. J Environ Manage 2010;
91(9):185563.
McPherson EG, Simpson JR, Peper PJ, Qingfu X. Benet-cost analysis of Modesto's
municipal urban forests. J Arboric 1999;25(5):23548.

MEA. Millennium ecosystem assessment: ecosystems and human well-being. Washington:

Island Press; 2005.
NEA. The UK national ecosystem assessment: synthesis of the key ndings. United
Nations Environment Programme World Conservation Monitoring Centre. Cambridge
Press; 2011.
Nelson M, Brown M. The QUIC start guide (V5.6), LA-UR-10-01062. Los Alamos National
Laboratory; 2010.
Noah's Ark. Global climate change impact on built heritage and cultural landscapes. Overview of the expected negative and positive consequences of global environmental
changes on deterioration of materialsProject Report No. SPPI-CT-2003-501837NOAH'S ARK; 2006.
Pardyjak E, Amatul UN, Nelson MA, Brown MJ. Development of a vegetation deposition
model for a fast response urban Lagrangian dispersion model. Eighth symposium
on the urban environment. Phoenix Arizona: American Meteorological Society; 2009.
Peng C, Elwan FA. Bridging outdoor and indoor environmental simulation for assessing
and aiding sustainable urban neighbourhood design. Int J Archit Res 2012;6:7290.
Raupach MR, Woods N, Dorr G, Leys JF, Cleugh HA. The entrapment of particles by windbreaks. Atmos Environ 2001;35:337383.
Rosheidat A, Hoffman D, Bryan H. Visualising pedestrian comfort using ENVI-met. Third
National Conference of IBPSA-USA, Berkeley, California, July 30August 1; 2008.
Sabboni C, Cassar M, Brimblecombe P, Tidblad J, Kozlowski R, Drdacky M, et al. Global climate change impact on building heritage and cultural landscapes. In: Gomez-Heras,
Vazquez-Calvo, editors. Heritage, weathering and conservation Fort, Alvarez de
Buergo 0-415-41272-2; 2006. p. 395401.
Santamouris M. Cooling the cities a review of reective and green roof mitigation
technologies to ght heat island and improve comfort in urban environments.
Sol Energy 2012. http://dx.doi.org/10.1016/j.solener.2012.07.003.
Schfer A, Swilling M. Valuing green infrastructure in an urban environment under
pressure the Johannesburg case. Ecol Econ 2013;86:24657.
Scholz M, Uzomah VC. Rapid decision support tool based on novel ecosystem service
variables for retrotting of permeable pavement systems in the presence of trees.
Sci Total Environ 2013;458460:48698.
Screpanti A, De Marco A. Corrosion on cultural heritage buildings in Italy: a role for
ozone? Environ Pollut 2009;157:151320.
Spangenberg J, Shinzato P, Johansson E, Duarte D. Simulation of the inuence of
vegetation on microclimate and thermal comfort in the city of Sao Paulo. Rev SBAU
2008;3:119.
Steyn DG, Rao ST, editors. Air pollution modelling and its application. Springer Science, B.
V; 2010. http://dx.doi.org/10.1007/978-90-481-3812-8.
TEEB. The Economics of Ecosystems and Biodiversity (TEEB) manual for cities: ecosystem
services in urban management. The Ecological and Economic Foundation, chapter 1;
2012. p. 19 [url-http://www.teebweb.org [accessed in Oct 2012]].
Thaiutsa B, Puangchit L, Kjelgren R, Arunpraparut W. Urban green space, street tree and
heritage large tree assessment in Bangkok, Thailand. Urban For Urban Green 2008;
7:21929.
Tidblad J, Kucera V, Mikhailov A, Henriksen J, Kreislova K, Yates T, et al. UN ECE ICP
materials: doseresponse functions on dry and wet acid deposition effects after
8 years of exposure. Water Air Soil Pollut 2001;130:145762.
Tiwary A, Morvan HP, Colls JJ. Modelling the size-dependent collection efciency of
hedgerows for ambient aerosols. J Aerosol Sci 2006;37:9901015.
Tiwary A, Sinnett D, Peachey C, Chalabi Z, Vardoulakis S, Fletcher T, et al. An integrated
tool to assess the role of new planting in PM10 capture and the human health
benets: a case study in London. Environ Pollut 2009;157:264553.
Tiwary A, Namdeo A, Fuentes J, Dore A, Hu X-M, Bell M. Systems scale assessment of the
sustainability implications of emerging green initiatives. Environ Pollut 2013a;183:
21323.
Tiwary A, Chatterton T, Namdeo A. Co-managing carbon and air quality: pros and cons of
local sustainability initiatives. J Environ Plann Manage 2013b. http://dx.doi.org/10.
1080/09640568.2013.802677.
Tominaga Y, Mochida A, Yoshie R, Kataoka H, Nozu T, Yoshikawa M, et al. AIJ guidelines
for practical applications of CFD to pedestrian wind environment around buildings.
J Wind Eng Ind Aerodyn 2008;96:174961.
UWYO. Atmospheric soundings. USA: Department of Atmospheric Sciences, University of
Wyoming; 2013 [url: http://weather.uwyo.edu/upperair/sounding.html. [accessed in
March 2013]].
Vos PEJ, Maiheu B, Vankerkom J, Janssen S. Improving local air quality in cities: to tree or
not to tree? Environ Pollut 2012;183:11322.
Wania A, Bruse M, Blond N, Weber C. Analysing the inuence of different street vegetation on trafc-induced particle dispersion using microscale simulations. J Environ
Manage 2012;94:91101.
Yang X, Zhao L, Bruse M, Meng Q. An integrated simulation method for building energy
performance assessment in urban environments. Energy Build 2012;54:24351.
Yu C, Hien WN. Thermal benets of city parks. Energy Build 2006;38:10520.
Zwack LM, Hanna SR, Spengler JD, Levy JI. Using advanced dispersion models and mobile
monitoring to characterise spatial patterns of ultrane particles in an urban area.
Atmos Environ 2011;45:48229.