Urban Forestry & Urban Greening 43 (2019) 126366

Contents lists available at ScienceDirect

Urban Forestry & Urban Greening

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

Review

A review of urban forest modeling: Implications for management and future T

research
⁎
Jian Lina, , Charles N. Krollb, David J. Nowakc, Eric J. Greenfieldc
a
Graduate Program in Environmental Science, State University of New York College of Environmental Science and Forestry (SUNY-ESF), 321 Baker, 1 Forestry Dr.,
Syracuse, NY, 13210, USA
b
Department of Environmental Resources Engineering, SUNY-ESF, Syracuse, NY, 13210, USA
c
USDA Forest Service, 5 Moon Library, SUNY-ESF, Syracuse, NY, 13210, USA

A R T I C LE I N FO A B S T R A C T

Handling Editor: N. Nilesh Timilsina Urban forest modeling is becoming increasingly complex, global, and transdisciplinary. Increased modeling of
Keywords: urban forest structure and function presents an urgent need for comparative studies to assess the similarities and
Urban forestry differences between modeling techniques and applications. This paper provides a systematic review of 242
Comparative studies journal papers over the past two-decades, and identifies 476 case studies. We assess model case studies among
Multi-scale different locations, units and scales, compare the ability and functional capacity of the models and different
Ecosystem services tools, compare papers published in different disciplines, and identify new emerging topics in the field of urban
Social-ecological system forest modeling. Conclusions from this analysis include: (1) the spatial distribution of case studies is primarily
clustered around the US, Europe, and China, with the most popular units to model being streets and parks; (2)
the most commonly used model types are the i-Tree toolset, ENVI-met, computational fluid dynamic models, and
the Hedonic price model; (3) uncertainty assessment of urban forest models is limited; (4) spatially explicit
models are critically important for estimating of ecosystem services as well as for environment management; (5)
most case studies focus on biophysical benefits with few studies estimating economic and social benefits; and (6)
linkages between urban forests and their social-psychological and health effects are less common due to sub-
jectivity and uncertainty in expressing and quantifying human cultures, attitudes and behaviors. Based on a
comparison of different models and a syntheses of case studies, we make suggestions for future research con-
necting urban forestry and urban ecosystems, model development, and ecosystem services. Such knowledge is
critical for policy- and decision-makers, and can help improve urban forest planning, design and management.

1. Introduction heterogeneous, complex adaptive social-ecological system (Wu, 2014),

which aims for not only environmental functionality, but also social
A term first used in 1965 (Gerhold, 2007), “urban forestry” has equity and economic viability (BES LTER, 2018). Compared to tradi-
become increasingly transdisciplinary in terms of theories (from both tional forestry, “forestry” in the urban context focuses on additional
physical and social sciences), methods (e.g., Geographic Information services to advance urban sustainability. As a demographic trend and
Systems, remote sensing, monitoring, and modeling), and participants land transformation process (Pickett et al., 2001), urbanization creates
(e.g., researchers, government officials, citizens, and volunteers). Many many environmental issues (e.g., Duh et al., 2008; Grimmond, 2007;
definitions of urban forestry have been given, and the definition and Poumanyvong and Kaneko, 2010); these issues make the design of
terminology harmonization is challenging (Konijnendijk et al., 2006). sustainable urban forestry (Fazio, 2003) particularly challenging.
However, several widely-used definitions, such as those provided by The morphological characteristics (e.g., leaf area, stem diameter),
Jorgensen (1986), Society of American Foresters (Helms, 1998), functions (e.g., photosynthesis, evapotranspiration), and structure (e.g.,
Konijnendijk et al. (2006), and Nowak et al. (2010), all emphasize species composition, spatial pattern) of trees provide a wide range of
urban forestry’s comprehensive nature, which involves scientific, ecosystem services (ES) and benefits that can alleviate the adverse ef-
management, and planning elements. In this article, we look at urban fects of urbanization (Nowak and Dwyer, 2007). Many cities have es-
forestry in a general way. Literally, “urban forestry” consists of two tablished substantial programs to increase their tree canopy coverage
parts “urban” and “forestry”. An “urban” system is a spatially (Morani et al., 2011; McPherson et al., 2011). However, simply

⁎
Corresponding author.
E-mail addresses: [email protected] (J. Lin), [email protected] (C.N. Kroll), [email protected] (D.J. Nowak), ejgreenfi[email protected] (E.J. Greenfield).

https://doi.org/10.1016/j.ufug.2019.126366
Received 8 August 2018; Received in revised form 3 June 2019; Accepted 7 June 2019
Available online 10 June 2019
1618-8667/ © 2019 Elsevier GmbH. All rights reserved.
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

increasing tree canopy itself does not guarantee the provision of ex- attention in the literature (Escobedo et al., 2011; Gómez-Baggethun and
pected ES. For example, Vos et al. (2013) have shown that it may not be Barton, 2013). The differentiation between ecosystem function and
a viable solution to alleviate a local air pollution hotspot by using urban service has been well-established, with the former emphasizing eco-
vegetation, and Wu (2014) indicated that urban greening may lead to system processes (means) while the latter focusing on specific outputs
unintended environmental injustice issues such as ‘ecological gentrifi- or products (ends) (Escobedo et al., 2011; Roy et al., 2012). In this
cation’. study, we focus on ES that can be derived from forest structure and
To better manage urban forests and maximize tree benefits, several function. Following the classification scheme of urban forest ES pro-
models have been developed and implemented. These models have vided by Nowak and Dwyer (2007), we expressed them in three value-
been applied in case studies on individual locations and provide us with domains: biophysical, social and economic.
knowledge about urban tree services and benefits. Although there is
evidence of a global trend of increased urban landscapes and ecological 3. Study methods
structural homogenization (Wu, 2014; Turner and Gardner, 2015), each
city is still unique, and the ES provided by urban forests change with Model practices and case studies of urban forests in academic
forest characteristics and environmental conditions. Findings for one English-language journals were reviewed during the past two-decades
city can be quite different compared to those of another city, and the (1996–2017). Here we use the term “case study” to refer to one simu-
current global distribution of urban forest case studies tends to cluster lation at one location employing either numerical or statistical models.
within specific regions. To be comprehensive, objective and accurate, a systematic quantitative
There are limited comparative studies of urban forest ecosystem literature review was first performed (Petticrew, 2001). Two worldwide
models. Of interest here is summarizing and generalizing findings scholarly electronic databases, Google Scholar and Scopus, were em-
across a wide range of case studies to identify trends and gaps in urban ployed in this study. Keywords or combination of keywords used for the
forest modeling. Such knowledge is critical for urban forests research search included: ‘urban tree/forest/vegetation/green roof’, ‘ecosystem
and management. By reviewing urban forest modeling over the past services/benefits’, and ‘model/tool’. For each identified paper, articles
two-decades, the goal of this paper is to facilitate a better under- of related or similar topics were identified via: (1) references within the
standing of model characteristics and uses, and integrate different paper, (2) ‘related articles/documents’ function in Google Scholar and
model practices and case studies to advance our knowledge of urban Scopus, and (3) articles that cited the paper. Although this step was
forestry and inform future research and management. mainly implemented based on Google Scholar and Scopus, other scho-
larly electronic databases were involved because search results often
2. Key terms and concepts led to different links (e.g., Science Direct, Research Gate, Springer Link,
and individual journal websites). While our literature search was not
The urban forest contains all trees, shrubs, lawns, and pervious soils exhaustive, we believe we’ve captured a majority of journal articles on
in urban areas (Escobedo et al., 2011; Roy et al., 2012). Our review here this topic.
focuses on trees and shrubs in different urban areas (e.g., street, park, After identifying journal articles, the following items were extracted
and residential area), as well as their local site and environmental from each paper: (i) year of publication, (ii) case study location, (iii)
conditions. Green roofs, green infrastructure, and green space (Rowe, model(s), (iv) input data, (v) title, (vi) author(s), (vii) journal, (viii)
2011) are all different, but related concepts, and they include various discipline, and (ix) topics and ES. A spatio-temporal analysis was then
vegetative components. They are also included in this review if their performed using (i) year of publication and (ii) case study location. For
study focuses on the structure and benefits of urban trees and shrubs. this analysis, each paper was grouped by continent and major climatic
There are many definitions of interdisciplinarity and transdisci- zone to determine the distribution and pattern of urban forest studies.
plinary. We differentiate them based on participants and final goals. Following the work of Roy et al. (2012), the continents included were
Here interdisciplinary studies refer to the involvement of several aca- North America, South America, Europe, Asia, Australia, and Africa; and
demic disciplines under a common research goal to create new the climatic zones were tropical, dry, subtropical, temperate, and con-
knowledge. Alternatively, transdisciplinary studies involve not only tinental. Other space-based analyses included identifying the scale of
academic researchers but also non-academic participants (e.g., the each study performed (e.g., city, region, nation), and the unit for each
public and policy-makers) for the purpose of solving real-world pro- case study (e.g., park, street, neighborhood, community, district, wa-
blems (Tress et al., 2005). tershed). Next, comparisons among models and among disciplines were
A model is a simplified description of a real system with inputs, key conducted using (iii) model(s), (iv) input data, and (viii) discipline. For
components of the system and their relationships, and outputs con- each model, the total numbers of papers and citations (how many times
strained within specific spatial boundary (Jones, 2013). A model can be that particular paper has been cited) were calculated. In addition, as
developed based on either mechanistic approaches or empirical re- input datasets are part of any model, each paper was also characterized
lationships, or a hybrid of both. The models considered in this study based on the acquisition sources of the input datasets. Each journal was
must be able to describe urban forest structure (e.g., size, species grouped into a specific field, and a comparison among fields was con-
composition, spatial configuration) (Nowak et al., 2008), and function ducted. We grouped journals into fields based on journal description
(e.g., various ES) in highly complex systems. They use forest structure, and the topics of the identified papers from journals. Finally, compar-
as well as other site and environmental parameters, as input variables isons between ES were investigated using (ix) ES topics.
to estimate ES as model outputs. We focus on numerical and statistical
models since they are used extensively to quantify forest derived ES. To 4. Results
link more directly to management implications and limit the scope of
the analyses reviewed, models focusing entirely on forest structure and We identified 242 relevant papers and 476 case studies over the
dynamics (e.g., growth, mortality) are excluded. As input datasets are a time period 1996–2017 (see Supplementary Material for a list of pa-
necessary part of any model, characteristics of input datasets are also pers), with more than half of the papers published during the past 6
explored from the perspective of data acquisition approaches: bottom- years (2012–2017). There are more case studies than publications be-
up approaches mainly consist of field surveys and sampling while top- cause some papers include several case studies. Citation numbers, pri-
down approaches rely mainly on remotely sensed data. marily conducted between the period of November 2017 to January
Since the release of the UN’s Millennium Ecosystem Assessment 2018 based on Google scholar, show a relatively exponential-type
(MEA) (MEA, 2005) and The Economics of Ecosystems and Biodiversity growth pattern over time (Fig. 1), reflecting the increasing number of
(TEEB) report (TEEB Foundations, 2010), ES have gained broader publications, activities and influences of this field.

2
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

Fig. 1. The number of publications and citations yearly from 1996 to 2017
Fig. 3. Summary statistics (percentage) of 242 original papers in different sub-
(citation counting was conducted between the period of November 2017 to
categories: continents, climatic zones, scale, and input data sources. The per-
January 2018 based on Google scholar).
centages for each continent and climatic zone are calculated based on number
of case studies, while percentages for each scale and input data source are based
4.1. Place-based, comparative studies on number of papers.

Among the papers examined, a total of 476 model practices and case socioeconomic sub-regions, and other similar units, while microscale
studies were identified globally (Fig. 2): North America (66.6%), includes green roofs, buildings, parks, streets and other similar settings.
Europe (14.5%), Asia (11.1%), Australia (3.6%), South America (2.7%), Most of the studies were conducted at city, local and microscale levels,
and Africa (1.5%). Another way to express the global distribution of while some studies have been made at watershed, regional and national
case studies is to classify case studies by climatic zones: tropical (2.8%), levels.
dry (7.4%), subtropical (4.9%), temperate (44.9%), and continental Inside the city, a variety of geographies have been employed in case
(40.0%) (Fig. 3). The global distribution of case studies was uneven, studies, depending on the study purpose and discipline. Each discipline
with a majority of studies focused on urbanizing regions of temperate may identify a geographical unit or the most salient features associated
and continental climatic zones in the US, Europe and China; there were with the unit differently (Grimm et al., 2000), such as a watershed
comparatively few studies of urban forest modeling in South America, (hydrology), land use or land cover types (geography), neighborhood or
Australia, and Africa. community (social science), and street canyon or building block (en-
With regards to scale, there were 8 papers conducted at a national ergy science). For the local scale, the most studied units were districts/
level, 9 at a regional level, 61 at a city level, 8 at a watershed level, 49 communities with a total of 28 case studies; within the microscale,
at a local scale level, and 107 at a microscale level (Fig. 3). Both local streets, parks, and green roofs received the most attention, with the
and microscale levels are scales smaller than a city level. Local scale numbers of case studies being 58, 22 and 25, respectively.
includes neighborhoods, communities, districts, planning zones,

Fig. 2. Global distribution of urban forest case studies.

3
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

Table 1
Distribution of urban forest modeling papers among different fields.
Fields Journal Title No. of Papers Field Total

Environment Environmental Pollution 16 53

Atmospheric Environment 12
Journal of Environmental Management 7
Science of the Total Environment 5
Environmental Modelling & Software 2
Environmental Science & Technology 1
International Journal of Environment and Pollution 1
Environmental Management 1
Environmental Science and Pollution Research 1
Environment and Behavior 1
Journal of Environmental Planning and Management 1
International Journal of Environmental Science and Development 1
Atmospheric Pollution Research 1
Procedia Environmental Sciences 1
Ambio 1
AIMS Environmental Science 1
Forestry and Arboriculture Urban Forestry & Urban Greening 34 48
Journal of Arboriculture 5
Arboriculture and Urban Forestry 3
Journal of Forestry 2
iForest-Biogeosciences and Forestry 1
Frontiers of Forestry in China 1
Journal of Sustainable Forestry 1
Forests 1
Energy Building and Environment 16 37
Energy and Buildings 15
Solar Energy 3
Energy Procedia 1
Applied Energy 1
Building Simulation 1
Landscape Landscape and Urban Planning 28 28
Ecology Urban Ecosystems 7 14
Ecological Modelling 2
International Journal of Biodiversity Science, Ecosystem Services & Management 2
Ecological Applications 1
Ecosystem Services 1
Ecosystems 1
Meteorology and Climatology Theoretical and Applied Climatology 4 12
Meteorologische Zeitschrift 2
Atmosphere 2
International journal of climatology 1
Boundary-layer meteorology 1
Advances in Meteorology 1
Journal of Applied Meteorology and Climatology 1
Economics Ecological Economics 4 11
Journal of Forest Economics 1
Journal of Environmental Economics and Management 1
Land Economics 1
The Appraisal Journal 1
Australian Journal of Agricultural and Resource Economics 1
The Journal of Real Estate Finance and Economics 1
Forest Policy and Economics 1
Geography Urban Geography 1 3
Moravian Geographical Reports 1
Chinese Geographical Science 1
Other Sustainable Cities and Society 6 36
Sustainability 5
JAWRA Journal of the American Water Resources Association 3
Cities 2
Journal of Wind Engineering and Industrial Aerodynamics 2
Land Use Policy 2
Advances in Urban Rehabilitation and Sustainability 1
Journal of Sound and Vibration 1
Journal of Contemporary Water Research & Education 1
Agriculture and Agricultural Science Procedia 1
Remote Sensing of Environment 1
Spatial Demography 1
International Journal of Sustainable Development and Planning 1
Transportation Research Part D: Transport and Environment 1
Book Chapter 4
Official publication from USDA, National Recreation and Park Association and ENVI-met 4

4
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

4.2. Field-based analyses

Spatially explicit or
Sixty-nine journals were identified over a wide range of fields

Yes for specific

(Table 1), revealing the transdisciplinary nature of this topic. Three

modules
fields interact closely and contribute the largest number of papers on
this topic (in parenthesis are the number of papers and percentages,

Yes

Yes

Yes

Yes
Yes

Yes
not

No

No
No
respectively): environment (53, 21.9%), forestry (48, 19.8%), and en-
ergy (37, 15.3%). The reason that the environmental field occupied the
largest number of papers is due to the contribution from two journals:

Uncertainty assessments (No,

Environmental Pollution (16, 6.6%) and Atmospheric Environment (12,
5.0%). Thirty-four papers were published in Urban Forestry & Urban
Greening, which makes forestry the next most common field. This field

limited, developed)
was followed by energy, with the largest contributions from Building
and Environment (16, 6.6%) and Energy and Buildings (15, 6.2%).
Other fields that also contribute to this topic were landscape (28,

Limited

Limited

Limited

Limited

Limited
Limited

Limited
Limited

Limited
11.6%), ecology (14, 5.8%), economics (11, 4.5%), climatology (12,

No
5.0%), and geography (3, 1.2%) (Table 1). This topic attracts attention
from not only scientists, but also urban planners and policy makers,
leading to papers in urban planning and management journals (e.g.,

User programming knowledge required

Journal of Environmental Management, Environmental Management).

4.3. Urban forest models

(low, medium, high)

Urban forest case studies have been analyzed and simulated using a
wide range of models (Table 2). In terms of numerical models, they can
be roughly divided into two categories: general-purpose models (ENVI-
met, computational fluid dynamics (CFD), Green Cluster Thermal Time

Medium

Medium
Medium

Medium

Medium
None
High
Constant (Green CTTC), DOE-2 building-energy simulation program
Low

Low

Low
(DOE-2), and Solar and Longwave Environmental Irradiance Geometry
Free & open source

(SOLWEIG)), and urban forest-specific models (i-Tree, CITYgreen). The

detailed description of these models can be found in the Supplementary
Material to this paper. Yes / Yes

Yes / Yes
Yes / No

Yes / No

Yes / No
No / No

No / No
No / No
i-Tree is the most dominant model used in urban forest modeling

Yes /No
or not

None
(Table 2). i-Tree and ENVI-met are toolsets, including various sub-tools
or modules (Table 3). Of the various i-Tree toolsets, Eco (formerly
Eco, Hydro, Streets, Vue, Species, Canopy, Design, & Landscape:

UFORE) was implemented most frequently, although case studies can Lohmeyer’s Microscale Flow and Dispersion Model (MISKAM):
ANSYS’ Fluent https://www.ansys.com/products/fluids/ansys-
Open Field Operation and Manipulation (OpenFOAM) http://

also be found using Streets (formerly STRATUM), Hydro, Canopy, and

Atmospheric, Vegetation, Soil, and Built environment &

Species. The next widely used models are ENVI-met and CFDs. For
ENVI-met application, the typical approach is based on a scenario
comparison of designed or real landscapes (e.g., with/without trees,
CHAM’s PHOENICS http://www.cham.co.uk/
Building system: http://www.envi-met.com/

tree configuration, tree-building spatial layouts) (e.g., Skelhorn et al.,

2014; Salata et al., 2015; Morakinyo and Lam, 2016). CFD is a collec- http://www.urban-climate.net/content/
tion of models that are based on the fundamental laws of fluid me-
chanics and thermodynamics. Typical applications of CFD include the
Sub-modules & web references

thermal effects of trees on surrounding buildings and pedestrian en-

https://www.itreetools.org/a

http://www.lohmeyer.de/en

vironments (e.g., Dimoudi and Nikolopoulou, 2003), and removal and

trapping of air pollutants from road traffic due to trees’ deposition ef-
www.openfoam.com
Characteristics of the main numerical urban forest models.

fects, filtering capacity, and aerodynamic effects (e.g., barrier, venti-

http://doe2.com/

lation performance) (e.g., Jeanjean et al., 2015; Amorim et al., 2013).

Detailed principles, processes and parameterizations of CFDs can be
found in Buccolieri et al.’s (2018) review of urban tree CFD modeling.
fluent

None

None

Unlike i-Tree, which emphasizes the impact of different tree aspects,

The last access of weblink is July 2018.
2008; Version 2016a (as of
2004; Version 1712 (As of

Table 2
1996; Version 19.1 (As of
1994; Version 4.3 (As of

1989; Version 6.3 (As of

1978; Version 2.3 (As of

Initial release & current

1996; Version 5 (As of

Summary statistics of urban forest models.

1981; Version 2018

Citations Country Case studies Publications

1996; Version 6

March 2004)
2002; None

i-Tree 8461 21 264 76

May 2018)

Sept 2016)
July 2014)

July 2017)
Nov 2017)

Dec 2017)

ENVI-met 2614 18 50 43
version

CFD 2206 8 35 35
CITYgreen 305 2 8 6
Green CTTC 881 3 7 7
Green CTTC

DOE-2 1658 2 24 5
CITY-green

SOLWEIG
ENVI-met

SOLWEIG 222 3 4 4
Models
Table 3

DOE-2
i-Tree

Hedonic price model 2996 10 40 32

CFD

Others 2710 10 44 34
a

5
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

Table 4 case study was found utilizing the vegetated urban canopy model (Lee,
Number of case studies assessing urban forest ES. 2011), and the CHIMERE air quality model (Alonso et al., 2011).
1996-2010 2011-2017 1996- Regarding statistical models, 45 papers and 60 case studies were
2017 identified over the study period. Three characteristics can be sum-
Ecosystem services (#/year) (#/year) Total # marized. First, statistical models often have a strong economic focus,
and consider issues such as an urban forest’s impact on property values
Physical/Biological Benefits
Removal of Air Pollutants
(Donovan and Butry, 2010), rental rates (Laverne and Winson-
Remove course particulate matter (PM10) 1.2 5.1 54 Geideman, 2003), and energy savings (Pandit and Laband, 2010).
Remove ozone (O3) 1.2 4.8 52 Second, 18 out of 45 papers adopted a spatially explicit approach. Even
Remove nitrogen dioxide (NO2) 0.9 4.8 47 for some models adopting a non-spatial approach, they considered
Remove carbon monoxide (CO) 0.9 3.7 40
spatial effects indirectly by employing location or distance factors as
Remove sulfur dioxide (SO2) 0.7 3.6 36
Remove fine particulate matter (PM2.5) 0.0 2.4 17 predictor variables (Tyrväinen, 1997; Laverne and Winson-Geideman,
Remove volatile organic compounds 0.5 1.0 15 2003; Morancho, 2003). Finally, among the 45 papers focusing on
(VOCs) statistical models of urban forests, 32 papers used Hedonic price
Remove elemental carbon (EC) 0.0 0.3 2
modeling, a method to estimate the contribution of ecosystem or en-
Remove nitrogen monoxide (NO) 0.0 0.2 1
Remove ultraviolet (UV) radiation 0.0 0.2 1
vironmental services to the value of a property (Sander et al., 2010)
Temperature and Microclimatic (Table 2).
Modifications Two characteristics of models are also investigated: spatial ex-
Lower air temperature 1.4 3.0 42 plicitness and uncertainty. A model is spatially explicit when the inputs,
Provide tree shade 1.0 0.7 20
outputs or processes vary spatially (Turner and Gardner, 2015). ENVI-
Reduce urban heat island (UHI) 0.1 1.4 12
Provide evaporative and transpiration 0.3 0.4 8 met, CFD, SOLWEIG, i-Tree Design and i-Tree Landscape are spatially
cooling explicit models, while other models investigated are generally not
Provide park cool effect 0.1 0.9 8 spatially explicit (Table 3). Uncertainty, due to incomplete information
Regulate wind 0.1 0.9 8 or the lack of knowledge of underlying processes, is a fundamental
Reduce incoming solar radiation 0.0 0.3 2
Carbon storage and sequestration
characteristic of any model (Wu et al., 2006). Uncertainty is generally
Carbon storage and sequestration 0.9 3.5 39 insufficiently evaluated or overlooked in current urban forest models
Storm water regulation (Table 3). Uncertainty assessments are usually something added after
Reduce runoff 0.9 1.9 26 the model has already been developed. For example, in models such as
Improve water quality 0.0 0.3 2
ENVI-met, Green CTTC, and SOLWEIG, only model output uncertainty
Other
Wildlife and biodiversity 0.1 0.2 3 (or prediction error) is assessed and expressed as the discrepancy be-
Noise effect 0.0 0.2 1 tween the model predictions and observations (e.g., Wu and Chen,
Economic Benefits 2017; Shashua-Bar and Hoffman, 2002; Lindberg and Grimmond,
Reduce building energy use (e.g., heating, 1.5 2.3 39 2011). In addition, only specific kinds of uncertainty are typically as-
and cooling)
Increasing property value or rent price 1.3 2.3 36
sessed. For example, in i-Tree, only sampling error of field plot data is
Aesthetic quality 0.1 0.4 5 evaluated while other kinds of uncertainties (e.g., model structure and
Social Benefits parameter uncertainty) are ignored, resulting in the underestimation of
Thermal comfort/heat stress 0.1 2.0 15 the overall uncertainty (Nowak et al., 2013). None of the papers address
Crime rate 0.1 0.6 5
uncertainty in communication of model output to the public and de-
Human health and disease 0.1 0.3 3
Environmental inequality 0.0 0.3 2 cision-makers.
In terms of acquiring input datasets, 164 papers employed only
bottom-up approaches, while 78 papers used the top-down approaches
ENVI-met and CFDs also simulate the impacts of street and building relying on remotely sensed imagery (Fig. 3), including aerial photo-
characteristics (e.g., sky view factor, road traffic volume, canyon geo- graphs, AVHRR, Landsat, MODIS, LiDAR, NLCD, TRMM, IKONOS, and
metry, and ground and building materials) (e.g., Wania et al., 2012; Tan QuickBird imagery. Fifty-four of the 78 papers were published after
et al., 2016; Salata et al., 2015; Shahidan et al., 2012). As such, ENVI- year 2011, indicating the increasing utilization of remotely sensed
met and CFDs are also employed in the areas of landscape architecture, imagery. A wide range of top-down approaches were employed to de-
building design, and energy and environmental planning (Ambrosini rive different model inputs. For example, MODIS has been used to es-
et al., 2014). timate leaf area index (e.g., Nowak et al., 2014), and high resolution
Although not as widely used as the above-mentioned models, digital imagery and Landsat data have been employed to estimate tree
CITYgreen, Green CTTC, DOE-2, and SOLWEIG are also frequently canopy and land cover types (e.g., Morani et al., 2011; Yang et al.,
employed (Table 2). CITYgreen had many applications from 1996 to 2005).
2006, but became less used afterwards due to model limitations
(Longcore et al., 2004) and probably the increased use of i-Tree tools. 4.4. Ecosystem services estimated with urban forest models
Both DOE-2 and SOLWEIG also have applications in building energy
analysis, emphasizing the impacts of building characteristics (e.g., ES found in the papers examined were classified into three cate-
building layouts, constructions, conditioning systems, and shade pat- gories: biophysical, social and economic (Table 4). Biophysical benefits
terns of walls) on energy usage (Akbari et al., 2001; Lindberg and had 432 case studies, which was much higher than economic benefits
Grimmond, 2011). While other models were also represented, their (80) and social benefits (25), indicating an uneven distribution of case
contributions were minimal. For example, Shadow Pattern Simulator is studies. Of the 432 case studies examining biophysical benefits, air
found in three case studies examining tree’s effect on residential energy pollutant removal was ranked highest with 264 case studies, followed
use and indirectly carbon reduction (e.g., Simpson and McPherson, by temperature and microclimatic modifications (98), carbon storage
1998; Jo and McPherson, 2001). Only two case studies use the fine and sequestration (39), and water regulation (28). There were also
resolution atmospheric multi-pollutant exchange atmospheric transport three case studies analyzing wildlife and biodiversity, and one case
model (e.g., McDonald et al., 2007) and the coupled weather research study focused on noise effects. Regarding economic benefits, the most
and forecasting and urban canopy model (Loughner et al., 2012). One dominant topics were building energy cost reduction (e.g., cooling ef-
fects, heating effects) (39) and increased property values (36), followed

6
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

by aesthetic quality (5). Among social benefits, thermal comfort re- (Boukili et al., 2017). Apart from the magnitude, the direction (e.g.,
ceived the most attention with 15 case studies, followed by reduced from source to sink) can also vary. Based on two studies in Singapore
crime rate (5) and human health and disease (3). and Mexico City, Velasco et al. (2016) concluded that carbon seques-
In terms of new emerging topics, there appears to be an evolution in tration by urban trees are both positive, but when including soil re-
urban forest modeling. While studies of biophysical benefits continue to spiration effects, overall carbon sequestration is negative, i.e. the trees
be most common, studies of fine particulate matter (PM2.5), ultraviolet and soil in Singapore act as a carbon source and not a sink. Soil re-
light, elemental carbon, and water quality appeared only after 2011. spiration is typically ignored due to large areas of impervious surfaces
Assessing the impacts of urban forests on these issues increases the in cities. Even within one city, the impact of location cannot be ne-
diversity of urban forest ES and presents new challenges and opportu- glected. For example, in terms of cooling effects and human thermal
nities in urban forest modeling. Some topics (e.g., urban heat island, comfort, avenue-trees often have the strongest impact, façade greening
park cool effect, thermal comfort, human health and disease) show an has some noticeable effect, and roof greening is mostly ineffective (Ng
increasing rate of study after year 2011, indicating a potential in- et al., 2012; Gromke et al., 2015). Trees also appear to perform dif-
creasing trend in the future. ferently depending on their placement within a unit (such as the lee-
ward, windward, central, and end parts of street canyons) (Moonen
5. Discussion et al., 2013). The compilation of numerous case studies, while uneven,
can give indications of commonalities and ranges of urban forest effects
5.1. Place-based, comparative studies in different cities.

5.1.1. Distribution of case studies 5.1.2. Scale and study unit

The systematic review presented here assesses and compares urban Apart from the uneven spatial distribution of case studies, there is
forest modeling practices among places and across scales. We identified also a gap between local research and global generalizations. Local
that: (1) the spatial distribution of case studies is clustered around scale research is important and the existing literature illustrates and
certain locations (e.g., US, Europe and China and mostly in temperate discusses the need of local forest structure (Escobedo & Nowak, 2009;
and continental climatic zones); (2) most of the studies were conducted Nowak et al., 2013a) and local scale tree design (Nowak et al., 2013b).
at and below city scales, and only a few studies were made at regional However, due to spatial heterogeneity (Escobedo & Nowak, 2009),
or national scales; and (3) within cities, the most popular units were urban trees may have opposing effects at different scales (Vos et al.,
parks and individual streets. The popularity of specific locations, cities 2013), and there is the need for multi-scale approaches (Jeanjean et al.,
and units could be attributed to several factors. The US and Europe are 2015). Caution is needed to generalize findings among different places
highly developed areas while China is one of the most rapidly devel- and scales, but by understanding the physics, chemistry, biology and
oping countries; all have a large number of cities and associated various social structure of urban forests, generalized principles can be devel-
kinds of urban environmental issues. As such, cities in those areas oped to guide urban foresters in designing forest structure to optimize
provide ideal natural laboratories for urban forests studies. In addition, ES.
some models (e.g., ENVI-met, CFD) are designed for microscale simu- Another concept that is related to scale is the study unit. Different
lations, and thus favor units like parks and street canyons. Urban forest units provide different perspectives, and only through integration of a
studies in these areas are generally more comprehensive, and these variety of units can a comprehensive view of urban forestry be
studies have the potential to provide information to support future achieved. For example, focusing on street canyons, the conclusion that
urban forest studies in less-studied regions. roadside trees negatively affect the local air quality may be obtained
These analyses contribute to our understanding of the structure, under certain conditions (Ries and Eichhorn, 2001; Wania et al., 2012).
function, and benefit of urban forests, and the interactions between However, this does not indicate that trees in urban backyards and parks
social and natural systems. Unfortunately, the uneven and fragmented have a similar effect (Vos et al., 2013). More studies are needed to
distribution of case studies may bias our knowledge and understanding integrate different units and scales. Two challenges exist when con-
of urban forestry. Each place is unique in its own way and findings for sidering different study units. First, the increased focus on ecological
one city can be quite different than for another city. For example, units and integration of ecological and political units should be pursued
Nowak et al. (2004) performed computer modeling of air pollution in the future. Existing studies focus mostly on political units (e.g.,
removal by trees in 55 US cities and their results showed that pollution census block groups), while ignoring ecological units such as patches,
removal per unit canopy cover varied significantly from place to place, habitats and ecoregions. Units important to humans are not necessarily
depending on pollution concentrations, length of in-leaf season, amount relevant for tree species or ecological processes, but help convey in-
of precipitation, and other meteorological variables. Overall, manage- formation in units important to managers, planners and politicians. The
ment of urban tree canopy cover could be a viable strategy to improve boundaries of different units, such as watersheds and administrative
air quality. However, Setälä et al. (2013) studied two Finnish cities and districts, may not coincide. In addition, mismatch between units or
concluded that the ability of urban vegetation to remove air pollutants scales of ecological processes and the institutions that are responsible
is minor in northern climates considering the short growing season. Vos for managing them can contribute to decision failures (Cumming et al.,
et al. (2013) conducted a computer simulation and reached the con- 2006). Second, spatially heterogeneous representation of landscapes
clusion that trees can deteriorate air quality at least locally at roadside can be classified as a mosaic, which include patches and corridors with
locations based on summary of 17 scenario simulations of various ve- abrupt discontinuities or boundaries, and gradients with gradual dif-
getation settings. Conclusions about air pollution removal effects are ferences in concentrations (Forman, 1995). Most studies reviewed in
clearly location- and scale-specific, and caution is needed when gen- this paper focus on urban mosaics and ignore gradient approaches. This
eralizing results. Regarding carbon storage, based on the studies of 28 is mainly because the boundaries must be explicitly defined under most
cities and 6 states in the US, carbon density per unit of tree cover varied modeling frameworks. Due to practical need, boundaries are usually
among cities based on tree density, tree size distributions, and species defined where several discontinuities coincide (MEA, 2005). Although
composition, with the general pattern of forested regions having greater gradient areas (e.g., urban-periurban-rural, wildland-urban interface)
carbon densities than grassland or desert regions (Nowak et al., 2013a). have been intensively studied in ecology (Openshaw, 1984), geography
In terms of carbon sequestration, depending on which models you (Kwan, 2012), and even urban forestry (Zipperer et al., 1997) using
employ (e.g., i-Tree Streets, allometric equations from Urban Tree Da- approaches such as landscape metrics, spatial statistics, and transect
tabase, or other empirical equations), the differences among the mag- analyses (Luck and Wu, 2002; Kong and Nakagoshi, 2006), few studies
nitudes of carbon sequestration estimates can be up to a factor of 2 incorporate these ideas or principles in urban forest modeling.

7
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

5.2. Field-based analyses where it is developed and calibrated; caution is needed when general-
izing the model to other locations, or when model-based inferences are
Urban forestry has developed rapidly (Fig. 1) due to contributions performed. Future applications of statistical models in urban forestry
from many fields (Table 1). For example, the concept of sustainable should emphasize the use of theoretical guidance towards the selection
urban forestry is largely based on sustainability concepts from the of appropriate model structure and predictor variables.
ecology field (Fazio, 2003); the theories about scale and spatial het-
erogeneity from geography contribute greatly to spatially explicit re- 5.3.3. Spatially explicit modeling
search of urban forests (Escobedo & Nowak, 2009); the laws of fluid Although spatially explicit modeling can increase model complexity
dynamics and thermodynamics from energy science improve our un- and data burden (Turner and Gardner, 2015), the spatial distribution of
derstanding of interactions between surface, vegetation and the atmo- trees and their associated ES is essential for designing effective and
sphere (Bruse & Fleer, 1998); and landscape ecology principles are used equitable policy interventions (TEEB Foundations, 2010). The produc-
in the design and planning of urban green spaces (Zhou et al., 2011). tion, flow and use of ES varies spatially, as do the spatial patterns of
Urban forestry is interdisciplinary by fusing knowledge from several beneficiaries and policy interventions. In addition, apart from the
fields, and transdisciplinary by applying scientific knowledge in policy- number of trees, the spatial composition and configuration of trees can
relevant ways. Transitioning more urban forestry initiatives and studies also affect the ES they provide (e.g., Li et al., 2012; Zhou et al., 2017).
from interdisciplinary to transdisciplinary could be of great benefit. For ENVI-met, CFD and SOLWEIG are designed to be spatially explicit; the
example, with volunteer public participation, the MillionTrees program i-Tree tool suite is also transforming from lumped to spatially explicit
and 10-year cycle street tree census (2015–2016) in New York City have modeling, with two new modules, i-Tree Design and Landscape, that
been implemented more efficiently (NYC Parks, 2018). Discipline- can provide location information at local and landscape scales, re-
bound approaches conflict with the nature of urban forestry because by spectively. Spatially explicit approaches are also often adopted by sta-
definition urban systems are social-ecological, and urban forests pro- tistical models directly by using spatial regression or indirectly by
vide a wide range of ES which are of common interest to multiple employing location or distance factors as predictor variables. Providing
disciplines. Urban forestry not only concerns itself with scientific re- equivalent tree cover per capita (or per land area) and accessibility to
search, but also involves in management, planning, education and green space, especially for underrepresented or disadvantaged groups,
outreach (Moskell et al., 2010; Rae et al., 2010). could be a top priority for future urban forest management programs.
Better quantifying the composition and configuration of trees and its
5.3. Urban forest models influences on ES will also benefit forest management.

5.3.1. Numerical models 5.3.4. Model uncertainties

A wide range of urban forest models exist, each suitable for specific Although the importance of uncertainty in modeling is well re-
applications. i-Tree and ENVI-met are two of the most widely used cognized (Walker et al., 2003), few studies of urban forest modeling
models (Table 2), most likely because they are freely available, do not provide critical information about model uncertainties. For those
require user programming experience, and contain various modules for models that do provide uncertainty information, only specific kinds of
different applications (Table 3). One additional reason that i-Tree is the uncertainties (e.g., sampling error, prediction error) are typically con-
most widely used is that it can be used at new locations or conditions sidered. This may be due to two reasons. First, for existing models that
without the re-calibration of model parameters. This is different from describe complex ecosystem interactions (e.g., i-Tree, ENVI-met), a full
approaches adopted by other models (e.g., ENVI-met, CFD); when ap- and thorough uncertainty assessment (especially quantification and
plying models outside their original modeling domains, new site-spe- reduction) usually involves significant changes to model architecture
cific parameter values must be obtained from measured data. i-Tree (e.g., model assumptions, simplifications, formulations, and para-
eliminates the need of parameter calibration by developing i-Tree da- meterizations). The lack of time and funding given other competing
tabases, that contain tree species and location information for many priorities of model developments limits current uncertainty assess-
countries to support modeling at new locations (see Supplementary ments. Second, although uncertainty assessment methods are well-de-
Material). When site-specific parameters are insufficiently calibrated or veloped (Refsgaard et al., 2007), no method is universally applicable
unavailable, model outputs tend to contain large uncertainties (Walker and effective for all models. Guidance to select appropriate methods for
et al., 2003). CFD models also have many applications for tree tem- specific model types and applications is lacking, plus each method has a
perature effects (e.g., interaction with buildings characteristics), and air learning curve (Pappenberger and Beven, 2006), which further limits
pollution removal effects (e.g., interaction with street characteristics uncertainty assessment. Given the importance of uncertainty analyses,
and road traffic volume). One limitation of CFDs is that they usually especially for those models focused on policy- or decision-making, fu-
require medium to high user programming experience (Table 3). When ture modeling exercises could focus on improving the assessment and
quantifying trees’ thermal and building energy effects is a focus, Green communication of uncertainty. Incorporating uncertainty assessment at
CTTC, SOLWEIG, and DOE-2 are also potential choices. the beginning of problem framing and model framework design, and
tracking and documenting uncertainty throughout model development
5.3.2. Statistical models could significantly reduce overall efforts to incorporate uncertainty
Statistical models tend to be empirical and subjective due to the analyses in urban forest models.
selection of predictor variables and functional forms. It is often the case
that in one paper, several functional forms are developed, the structures 5.3.5. Model comparisons
and forms of statistical models often are identified based on the em- The comparison and integration of numerical models is rare, with
pirical fitting to observational datasets, and comparisons of different only a few studies on model integration (e.g., Tiwary et al., 2009;
fittings are conducted using statistical measurements (e.g., the good- McPherson & Kotow, 2013; Morakinyo and Lam, 2016) and model
ness-of-fit test) and information criteria (e.g., Akaike Information comparison (e.g., Russo et al., 2014; Guidolotti et al., 2016). General
Criterion) (Conway et al., 2010; Sander et al., 2010; Pandit et al., comparisons of models and model outputs may not be useful, and
2013). The resultant best selected model can provide a useful descrip- sometimes can even be misleading. Different models can estimate si-
tion of the system even without physiological or mechanical knowledge milar ES based on different input variables, model assumptions and
(Jones, 2013). The problems of this approach are that (1) across papers, formulations. For example, when estimating trees’ temperature effects,
model forms and explanatory variables can vary widely which makes a CFD model is based on the fundamental laws of fluid mechanics and
comparative studies challenging; and (2) the model may only be valid thermodynamics to simulate the effects of vegetation on transpirational

8
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

cooling and mean air flow and turbulence (e.g., Gromke et al., 2015). which adds additional complexity to model the effects of air pollutant
Contrary to this, Green CTTC employs an energy balance approach exposure. Although challenging, human behaviors and values are well
which quantifies anthropogenic heat-release, reduction of the solar gain modeled in other fields (e.g., economic, political ecology) (Anderies,
due to tree canopy, energy consumption for evapotranspiration, and the 2000; Peterson, 2000); those advanced experiences should benefit fu-
change in the heat stored based on leaf surface temperature (Shashua- ture urban forest modeling. The linkages between forest structure and
Bar & Hoffman, 2004). In this case, the differences in a tree’s tem- biophysical benefits is well understood and modeled, and ES delivery in
perature effects could be due to different modeling approaches rather biophysical terms also provides solid ecological underpinnings to eco-
than a tree’s structure and function. However, this does not mean that nomic and social metrics (TEEB Foundations, 2010). Expanding the
model comparisons should be avoided. Modeling experiences from links to incorporate trees’ social-psychological and health effects, as
other fields (e.g., public health, agriculture crop yield) have shown that well as quantifying and valuing those effects, is a priority area where
the combined information of several models is superior to that of a additional work is needed.
single model (Thomson et al., 2006; Cantelaube and Terres, 2005). The
way models are compared and integrated is important. Model com- 5.4.2. Health-related ES
parisons and integration can be conducted at the decision-making level; Studies regarding the health-related ES of trees increased after
if different models, with dissimilar theoretical foundations, reach si- 2011, compared to the period 1996-2010. These include ‘thermal
milar conclusions about the effects of urban forests, it will increase the comfort/heat stress’ and ‘human health and disease’ from social bene-
confidence of urban forests management decisions based on such si- fits, as well as various kinds of air pollutant removals (e.g., particulate
milar conclusions, especially when uncertainty analyses are lacking. matter, ozone), which have important health implications (Kinney,
2008). Many ES are public goods, and people usually lack direct in-
5.3.6. Input datasets centives to protect and maintain them (TEEB Foundations, 2010). One
Remotely sensed images play an important role in urban forest of the key challenges facing urban forest campaigns is to get the at-
modeling. This is mainly due to increased availability of free remotely tention and involvement of different stakeholders (Zhang et al., 2007),
sensed imagery (Patino and Duque, 2013), and many ready-to-use and human health is one of the few ES that is relevant to almost ev-
image derived products (e.g., vegetation index, leaf area index, tree eryone. Emphasizing the health effects of trees is an effective strategy to
canopy cover) (O’Neil-Dunne et al., 2014; Morani et al., 2011; Yang convey the importance of urban forests and gain support from stake-
et al., 2005). Although there is a trend of increasing utilization of re- holders.
motely sensed imagery, this information mainly serves as input vari-
ables for urban forest models. A closer connection between remote 5.4.3. Ecosystem disservices
sensing and urban forest modeling is needed, which will open up ad- Another aspect that is less studied is ecosystem disservices (EDS) of
ditional possibilities for future research and innovation. Two-dimen- urban forests. Common examples of EDS found in the literature include
sional images may greatly improve our ability to perform spatially biogenic volatile organic compound emissions (Calfapietra et al., 2013),
explicit modeling, and long-term archives of time-series images present increases in potential energy use (Nowak et al., 2017), allergenic effects
an opportunity to improve our understanding of the dynamics of urban (Dobbs et al., 2014), air pollution trapping at road sites (Vos et al.,
forests and the impacts of these changes. In addition, remote sensing 2013), and gentrification (Wolch et al., 2014). Although the adverse
can sometimes aid in validating models (e.g., biomass, LAI) (Lu et al., effects of urban forests have been mentioned and discussed in several
2016; Alonzo et al., 2016), which reduces model uncertainties and in- papers (e.g., McDonald et al., 2007; Buccolieri et al., 2009; Morani
creases the credibility of a model and its outputs. et al., 2011), there are few studies to simulate and quantify EDS (Vos
et al., 2013; Nowak et al., 2013b), let alone integrate EDS in decision-
5.4. Ecosystem services estimated with urban forest models making. Since EDS are often ignored, the overall net benefits of urban
forests may be less than initially estimated. The combined effects of ES
5.4.1. Biophysical, economic and social benefits and disservices and their influence on urban forest management and
Most case studies focus on biophysical benefits while only a few decision making are rarely investigated.
estimate economic and social benefits. This disparity may be due to
significant advances we have achieved in linking forest structure to 5.4.4. ES interactions
function. For instance, we have a good understanding of how a tree’s Regarding individual ES, the majority of papers explore specific
characteristics (e.g., albedo, surface roughness) and biophysiological types of ES while ignoring the interaction among different ES. These
processes (e.g., evapotranspiration, storing carbon) affect temperature interactions can happen at different levels. For instance, for air pollu-
(Bonan, 2008), and how trees uptake and remove air pollution by dry tant removal, most studies estimate the removal of PM10, O3, CO, NO2,
deposition processes (Hirabayashi et al., 2011). We are able to para- and SO2 in parallel. This makes sense for primary gases (e.g., NO2, SO2),
meterize these attributes and formulate these processes explicitly in but not for secondary gases (e.g., ozone), which can be created through
models. However, we have limited capability to simulate economic and complex chemical reactions and interactions (Pickett et al., 2011;
social benefits due to a lack of theory and large subjectivity and un- Morani et al., 2011). Ignoring these interactions will lead to inaccurate
certainty in expressing and quantifying human cultures, values, atti- estimation of net ozone effects (Cabaraban et al., 2013). At a higher
tudes and behaviors in models. For example, trees can provide amenity level, the change of one ES could also affect other kinds of ES. For
services to increase property values (e.g., Payton et al., 2008; Sander example, temperature reductions have implications for energy use, air
et al., 2010). However, amenity services (e.g., aesthetic enjoyment, quality, and human health (Nowak et al., 2014); energy savings can
recreation, intellectual development, and spiritual fulfillment) are in- result in reduced emissions of CO2 and air pollutants (McPherson et al.,
fluenced and shaped by human cultures, knowledge systems, religions, 2017; Nowak et al., 2017). Those interactions, either positive synergies
and social interactions (MEA, 2005). As such, quantifying those benefits (multiple services are enhanced simultaneously) (Bennett et al., 2009)
suffers from large uncertainties and biases. In terms of social benefits, or negative tradeoffs (the provision of one service is reduced as a
for instance, trees can affect human health by reducing air pollutant consequence of the increased use of another) (Turner and Gardner,
concentrations, but valuing the effects suffers from subjectivity due to 2015), are commonplace in ecosystems (MEA, 2005). In contrast to the
the cost of illness, willingness to pay to avoid illness, and productivity above common definition of tradeoffs, Mouchet et al. (2014) also refers
losses associated with health events (Nowak et al., 2014). Different to tradeoffs as various types of compromises, such as management
individuals have different ‘behavioral patterns, dietary patterns and compromises between ES. For instance, are tree species and locations
physiological characteristics (e.g., breathing rates)’ (WHO, 2008), being chosen to maximum or prioritize air pollution removal benefits or

9
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

energy conservation benefits? Urban forest management decisions Alonzo, M., McFadden, J.P., Nowak, D.J., Roberts, D.A., 2016. Mapping urban forest
should consider these interactions and compromises to better avoid structure and function using hyperspectral imagery and lidar data. Urban For. Urban
Green. 17, 135–147.
tradeoffs and enhance synergies. Alonso, R., Vivanco, M.G., González-Fernández, I., Bermejo, V., Palomino, I., Garrido,
J.L., Elvira, S., Salvador, P., Artíñano, B., 2011. Modelling the influence of peri-urban
trees in the air quality of Madrid region (Spain). Environ. Pollut. 159 (8), 2138–2147.
6. Future directions and conclusions Ambrosini, D., Galli, G., Mancini, B., Nardi, I., Sfarra, S., 2014. Evaluating mitigation
effects of urban heat islands in a historical small center with the ENVI-Met® climate
Future directions in urban forest modeling can be organized around model. Sustainability 6 (10), 7013–7029.
Amorim, J., Rodrigues, V., Tavares, R., Valente, J., Borrego, C., 2013. CFD modelling of
three key themes: urban systems, model development, and ES. An urban the aerodynamic effect of trees on urban air pollution dispersion. Sci. Total Environ.
system is a complex and adaptive socio-ecological system which is 461, 541–551.
Anderies, J.M., 2000. On modeling human behavior and institutions in simple ecological
characterized by spatial dependence and heterogeneity. However,
economic systems. Ecol. Econ. 35 (3), 393–412.
urban ecosystems are often modeled non-spatially, which can ignore Bennett, E.M., Peterson, G.D., Gordon, L.J., 2009. Understanding relationships among
some local or microscale effects and interactions due to spatial ar- multiple ecosystem services. Ecol. Lett. 12 (12), 1394–1404.
BES LTER, 2018. The Baltimore Ecosystem Study, Long-Term Ecological Research.
rangements. Future studies of urban forestry could focus on: (1) sum- Retrieved June, 2018 from. http://www.beslter.org/.
marizing and generalizing experiences from well-studied regions (e.g., Bonan, G.B., 2008. Forests and climate change: forcings, feedbacks, and the climate
US, Europe, China) to support urban forest studies in less-studied re- benefits of forests. Science 320 (5882), 1444–1449.
Boukili, V.K., Bebber, D.P., Mortimer, T., Venicx, G., Lefcourt, D., Chandler, M.,
gions; (2) multi-scale approaches to capture interactions among spatial Eisenberg, C., 2017. Assessing the performance of urban forest carbon sequestration
heterogeneity at the local scale, and interaction among cities, suburbs, models using direct measurements of tree growth. Urban For. Urban Green. 24,
212–221.
and their regional background environments at the regional scale; (3) Buccolieri, R., Gromke, C., Di Sabatino, S., Ruck, B., 2009. Aerodynamic effects of trees on
improving linkages between ecological processes and social organiza- pollutant concentration in street canyons. Sci. Total Environ. 407 (19), 5247–5256.
tion scales to improve urban forest modeling, planning, management Buccolieri, R., Santiago, J.-L., Rivas, E., Sánchez, B., 2018. Review on urban tree mod-
elling in CFD simulations: aerodynamic, deposition and thermal effects. Urban For.
and stewardship. With regards to model development, future research Urban Green.
could focus on (1) improving uncertainty analyses, (2) employing Cabaraban, M.T.I., Kroll, C.N., Hirabayashi, S., Nowak, D.J., 2013. Modeling of air pol-
spatially explicit expressions of location information, including issues lutant removal by dry deposition to urban trees using a WRF/CMAQ/i-Tree Eco
coupled system. Environ. Pollut. 176, 123–133.
related to environmental justice, (3) comparing and integrating models Calfapietra, C., Fares, S., Manes, F., Morani, A., Sgrigna, G., Loreto, F., 2013. Role of
at a policy-making level, (4) increasing utilization of remote sensing, Biogenic Volatile Organic Compounds (BVOC) emitted by urban trees on ozone
concentration in cities: a review. Environ. Pollut. 183, 71–80.
(5) increasing model capability to incorporate the effects of human Cantelaube, P., Terres, J.M., 2005. Seasonal weather forecasts for crop yield modelling in
cultures, values, attitudes and behaviors, and (6) increasing mechan- Europe. Tellus A 57 (3), 476–487.
istic understanding and its integration into statistical models. In addi- Cumming, G.S., Cumming, D.H., Redman, C.L., 2006. Scale mismatches in social-ecolo-
gical systems: causes, consequences, and solutions. Ecol. Soc. 11 (1).
tion, more effort could be devoted to model training to engage broader Dimoudi, A., Nikolopoulou, M., 2003. Vegetation in the urban environment: microcli-
audiences and better utilize existing software, and to better commu- matic analysis and benefits. Energy Build. 35 (1), 69–76.
Dobbs, C., Kendal, D., Nitschke, C.R., 2014. Multiple ecosystem services and disservices of
nicate model outputs to stakeholders and decision-makers. In terms of
the urban forest establishing their connections with landscape structure and socio-
ES, emphasis could be put on: (1) expanding linkages between forest demographics. Ecol. Indic. 43, 44–55.
structure and function to incorporate trees’ social-psychological and Donovan, G.H., Butry, D.T., 2010. Trees in the city: valuing street trees in Portland,
Oregon. Landsc. Urban Plan. 94 (2), 77–83.
health effects, (2) quantifying and valuing EDS, and (3) investigating Duh, J.-D., Shandas, V., Chang, H., George, L.A., 2008. Rates of urbanisation and the
ecosystem service synergies and tradeoffs. resiliency of air and water quality. Sci. Total Environ. 400 (1), 238–256.
With well-developed scientific rationales, models serve as a basis for Escobedo, F.J., Kroeger, T., Wagner, J.E., 2011. Urban forests and pollution mitigation:
analyzing ecosystem services and disservices. Environ. Pollut. 159 (8), 2078–2087.
collaboration and knowledge exchange between academic researchers Fazio, J.R., 2003. Urban and Community Forestry: a Practical Guide to Sustainability.
and non-academic participants, and provide a scientific means to Retrieve June 6, 2017 from Arborday website. The National Arbor Day Foundation.
https://www.arborday.org/programs/ucf/english.cfm.
achieve sustainable urban forestry. Urban forest modeling is becoming Forman, R.T., 1995. Land Mosaics: the Ecology of Landscapes and Regions. Cambridge
increasing complex, global, and transdisciplinary, and this trend is University Press, Cambridge, United Kingdom, pp. 81–85.
likely to continue. As such, there is an urgent need for comparative Gerhold, H.D., 2007. Origins of urban forestry. In: Kuser, J.E. (Ed.), Urban and
Community Forestry in the Northeast. Springer, Netherlands.
studies and studies across a wide range of geographic settings. By Gómez-Baggethun, E., Barton, D.N., 2013. Classifying and valuing ecosystem services for
synthesizing case studies from the perspectives of places, units, scales, urban planning. Ecol. Econ. 86, 235–245.
disciplines, tools, and topics, this review provides insights and sug- Grimm, N.B., Grove, J.G., Pickett, S.T., Redman, C.L., 2000. Integrated approaches to
long-term studies of urban ecological systems urban ecological systems present
gestions for future urban forest modeling research. Such knowledge can multiple challenges to ecologists—pervasive human impact and extreme hetero-
improve urban forest planning, design and management, and better geneity of cities, and the need to integrate social and ecological approaches, concepts,
and theory. BioScience 50 (7), 571–584.
support critical policy- and decision-making processes. Grimmond, S., 2007. Urbanization and global environmental change: local effects of
urban warming. Geogr. J. 173 (1), 83–88.
Acknowledgements Helms, J., 1998. The Dictionary of Forestry. Society of American Foresters, Bethesda.
Hirabayashi, S., Kroll, C.N., Nowak, D.J., 2011. Component-based development and
sensitivity analyses of an air pollutant dry deposition model. Environ. Model. Softw.
This research was supported by the United States Department of 26 (6), 804–816.
Jeanjean, A.P., Hinchliffe, G., McMullan, W., Monks, P.S., Leigh, R.J., 2015. A CFD study
Agriculture (USDA) Forest Service with grant 16-DG-11132544-034
on the effectiveness of trees to disperse road traffic emissions at a city scale. Atmos.
recommended by the National Urban and Community Forestry Environ. 120, 1–14.
Advisory Council, grant 11-JV-11242308-112 from the USDA Forest Jo, H.-K., McPherson, E.G., 2001. Indirect carbon reduction by residential vegetation and
planting strategies in Chicago, USA. J. Environ. Manage. 61 (2), 165–177.
Service Northern Research Station, and additional support from the Jones, H.G., 2013. Plants and Microclimate: a Quantitative Approach to Environmental
SUNY ESF Department of Environmental Resources Engineering. Plant Physiology. Cambridge University Press, Cambridge, United Kingdom 1-5 pp.
Jorgensen, E., 1986. Urban forestry in the rearview mirror. Arboric. J. 10 (3), 177–190.
Kinney, P.L., 2008. Climate change, air quality, and human health. Am. J. Prev. Med. 35
Appendix A. Supplementary data (5), 459–467.
Kong, F., Nakagoshi, N., 2006. Spatial-temporal gradient analysis of urban green spaces in
Jinan, China. Landsc. Urban Plan. 78 (3), 147–164.
Supplementary material related to this article can be found, in the Konijnendijk, C.C., Ricard, R.M., Kenney, A., Randrup, T.B., 2006. Defining urban
online version, at doi:https://doi.org/10.1016/j.ufug.2019.126366. forestry–A comparative perspective of North America and Europe. Urban For. Urban
Green. 4 (3-4), 93–103.
Kwan, M.-P., 2012. The uncertain geographic context problem. Ann. Assoc. Am. Geogr.
References 102 (5), 958–968.
Laverne, R.J., Winson-Geideman, K., 2003. The influence of trees and landscaping on
Akbari, H., Pomerantz, M., Taha, H., 2001. Cool surfaces and shade trees to reduce energy rental rates at office buildings. J. Arboricult. 29 (5), 281–290.
use and improve air quality in urban areas. Sol. Energy 70 (3), 295–310. Li, X., Zhou, W., Ouyang, Z., Xu, W., Zheng, H., 2012. Spatial pattern of greenspace affects

10
J. Lin, et al. Urban Forestry & Urban Greening 43 (2019) 126366

land surface temperature: evidence from the heavily urbanized Beijing metropolitan Kaushal, S.S., Marshall, V., McGrath, B.P., Nilon, C.H., 2011. Urban ecological sys-
area, China. Landsc. Ecol. 27 (6), 887–898. tems: scientific foundations and a decade of progress. J. Environ. Manage. 92 (3),
Lee, S.-H., 2011. Further development of the vegetated urban canopy model including a 331–362.
grass-covered surface parametrization and photosynthesis effects. Boundary Layer Pickett, S.T., Cadenasso, M.L., Grove, J.M., Nilon, C.H., Pouyat, R.V., Zipperer, W.C.,
Meteorol. 140 (2), 315–342. Costanza, R., 2001. Urban ecological systems: linking terrestrial ecological, physical,
Lindberg, F., Grimmond, C., 2011. The influence of vegetation and building morphology and socioeconomic components of metropolitan areas. Annu. Rev. Ecol. Evol. Syst.
on shadow patterns and mean radiant temperatures in urban areas: model develop- 32, 127–157.
ment and evaluation. Theor. Appl. Climatol. 105 (3-4), 311–323. Poumanyvong, P., Kaneko, S., 2010. Does urbanization lead to less energy use and lower
Longcore, T., Li, C., Wilson, J.P., 2004. Applicability of CITYgreen urban ecosystem CO2 emissions? A cross-country analysis. Ecol. Econ. 70 (2), 434–444.
analysis software to a densely built urban neighborhood. Urban Geogr. 25 (2), Rae, R.A., Simon, G., Braden, J., 2010. Public reactions to new street tree planting. Cities
173–186. Environ. (CATE) 3 (2011).
Loughner, C.P., Allen, D.J., Zhang, D.-L., Pickering, K.E., Dickerson, R.R., Landry, L., Refsgaard, J.C., van der Sluijs, J.P., Højberg, A.L., Vanrolleghem, P.A., 2007. Uncertainty
2012. Roles of urban tree canopy and buildings in urban heat island effects: para- in the environmental modelling process–a framework and guidance. Environ. Model.
meterization and preliminary results. J. Appl. Meteorol. Climatol. 51 (10), Softw. 22 (11), 1543–1556.
1775–1793. Ries, K., Eichhorn, J., 2001. Simulation of effects of vegetation on the dispersion of
Lu, D., Chen, Q., Wang, G., Liu, L., Li, G., Moran, E., 2016. A survey of remote sensing- pollutants in street canyons. Meteorologische Zeitschrift 10 (4), 229–233.
based aboveground biomass estimation methods in forest ecosystems. Int. J. Digit. Rowe, D.B., 2011. Green roofs as a means of pollution abatement. Environ. Pollut. 159
Earth 9 (1), 63–105. (8), 2100–2110.
Luck, M., Wu, J., 2002. A gradient analysis of urban landscape pattern: a case study from Roy, S., Byrne, J., Pickering, C., 2012. A systematic quantitative review of urban tree
the Phoenix metropolitan region, Arizona, USA. Landsc. Ecol. 17 (4), 327–339. benefits, costs, and assessment methods across cities in different climatic zones.
McDonald, A., Bealey, W., Fowler, D., Dragosits, U., Skiba, U., Smith, R., Donovan, R., Urban For. Urban Green. 11 (4), 351–363.
Brett, H., Hewitt, C., Nemitz, E., 2007. Quantifying the effect of urban tree planting Salata, F., Golasi, I., de Lieto Vollaro, A., de Lieto Vollaro, R., 2015. How high albedo and
on concentrations and depositions of PM 10 in two UK conurbations. Atmos. Environ. traditional buildings’ materials and vegetation affect the quality of urban micro-
41 (38), 8455–8467. climate: a case study. Energy Build. 99, 32–49.
McPherson, E.G., Simpson, J.R., Xiao, Q., Wu, C., 2011. Million trees Los Angeles canopy Sander, H., Polasky, S., Haight, R.G., 2010. The value of urban tree cover: A hedonic
cover and benefit assessment. Landsc. Urban Plan. 99 (1), 40–50. property price model in Ramsey and Dakota Counties, Minnesota, USA. Ecol. Econ.
McPherson, E.G., Xiao, Q., van Doorn, N.S., de Goede, J., Bjorkman, J., Hollander, A., 69 (8), 1646–1656.
Boynton, R.M., Quinn, J.F., Thorne, J.H., 2017. The structure, function and value of Setälä, H., Viippola, V., Rantalainen, A.-L., Pennanen, A., Yli-Pelkonen, V., 2013. Does
urban forests in California communities. Urban For. Urban Green. 28, 43–53. urban vegetation mitigate air pollution in northern conditions? Environ. Pollut. 183,
MEA, 2005. Millennium Ecosystem Assessment. Ecosystems and Human Well-being: 104–112.
Synthesis. Island Press, Washington, DC. Shahidan, M.F., Jones, P.J., Gwilliam, J., Salleh, E., 2012. An evaluation of outdoor and
Morakinyo, T.E., Lam, Y.F., 2016. Simulation study on the impact of tree-configuration, building environment cooling achieved through combination modification of trees
planting pattern and wind condition on street-canyon’s micro-climate and thermal with ground materials. Build. Environ. 58, 245–257.
comfort. Build. Environ. 103, 262–275. Shashua-Bar, L., Hoffman, M.E., 2002. The Green CTTC model for predicting the air
Morancho, A.B., 2003. A hedonic valuation of urban green areas. Landsc. Urban Plan. 66 temperature in small urban wooded sites. Build. Environ. 37 (12), 1279–1288.
(1), 35–41. Simpson, J.R., McPherson, E.G., 1998. Simulation of tree shade impacts on residential
Morani, A., Nowak, D.J., Hirabayashi, S., Calfapietra, C., 2011. How to select the best tree energy use for space conditioning in Sacramento. Atmos. Environ. 32 (1), 69–74.
planting locations to enhance air pollution removal in the MillionTreesNYC initiative. Skelhorn, C., Lindley, S., Levermore, G., 2014. The impact of vegetation types on air and
Environ. Pollut. 159 (5), 1040–1047. surface temperatures in a temperate city: A fine scale assessment in Manchester, UK.
Moskell, C., Broussard Allred, S., Ferenz, G., 2010. Examining volunteer motivations and Landsc. Urban Plan. 121, 129–140.
recruitment strategies for engagement in urban forestry. Cities Environ. (CATE) 3 Tan, Z., Lau, K.K.-L., Ng, E., 2016. Urban tree design approaches for mitigating daytime
(1), 9. urban heat island effects in a high-density urban environment. Energy Build. 114,
Mouchet, M.A., Lamarque, P., Martín-López, B., Crouzat, E., Gos, P., Byczek, C., Lavorel, 265–274.
S., 2014. An interdisciplinary methodological guide for quantifying associations be- TEEB Foundations, 2010. The Economics of Ecosystems and Biodiversity: Ecological and
tween ecosystem services. Glob. Environ. Chang. Part A 28, 298–308. Economic Foundations. Earthscan, London and Washington.
Nowak, D.J., Appleton, N., Ellis, A., Greenfield, E., 2017. Residential building energy Thomson, M., Doblas-Reyes, F., Mason, S., Hagedorn, R., Connor, S., Phindela, T., Morse,
conservation and avoided power plant emissions by urban and community trees in A., Palmer, T., 2006. Malaria early warnings based on seasonal climate forecasts from
the United States. Urban For. Urban Green. 21, 158–165. multi-model ensembles. Nature 439 (7076), 576.
Nowak, D.J., Crane, D.E., Stevens, J.C., Hoehn, R.E., Walton, J.T., Bond, J., 2008. A Tress, B., Tress, G., Fry, G., 2005. Defining Concepts and the Process of Knowledge
ground-based method of assessing urban forest structure and ecosystem services. Production in Integrative Research. Springer, Heidelberg, Germany.
Arboric. Urban For. 34 (6), 347–358. Turner, M.G., Gardner, R.H., 2015. Landscape Ecology in Theory and Practice: Pattern
Nowak, D.J., Dwyer, J.F., 2007. Understanding the benefits and costs of urban forest and Process. Springer-Verlag, New York, New York, USA.
ecosystems. In: Kuser, J.E. (Ed.), Handbook of Urban and Community Forestry in the Tyrväinen, L., 1997. The amenity value of the urban forest: an application of the hedonic
Northeast. Kluwer Academic/Plenum Publishers, New York, pp. 11–22. pricing method. Landsc. Urban Plan. 37 (3-4), 211–222.
Nowak, D.J., Greenfield, E.J., Hoehn, R.E., Lapoint, E., 2013a. Carbon storage and se- Velasco, E., Roth, M., Norford, L., Molina, L.T., 2016. Does urban vegetation enhance
questration by trees in urban and community areas of the United States. Environ. carbon sequestration? Landsc. Urban Plan. 148, 99–107.
Pollut. 178, 229–236. Vos, P.E., Maiheu, B., Vankerkom, J., Janssen, S., 2013. Improving local air quality in
Nowak, D.J., Hirabayashi, S., Bodine, A., Greenfield, E., 2014. Tree and forest effects on cities: to tree or not to tree? Environ. Pollut. 183, 113–122.
air quality and human health in the United States. Environ. Pollut. 193, 119–129. Walker, W.E., Harremoës, P., Rotmans, J., van der Sluijs, J.P., van Asselt, M.B., Janssen,
Nowak, D.J., Hirabayashi, S., Bodine, A., Hoehn, R., 2013b. Modeled PM 2.5 removal by P., Krayer von Krauss, M.P., 2003. Defining uncertainty: a conceptual basis for un-
trees in ten US cities and associated health effects. Environ. Pollut. 178, 395–402. certainty management in model-based decision support. Integr. Assess. 4 (1), 5–17.
Nowak, D.J., Kuroda, M., Crane, D.E., 2004. Tree mortality rates and tree population Wania, A., Bruse, M., Blond, N., Weber, C., 2012. Analysing the influence of different
projections in Baltimore, Maryland, USA. Urban For. Urban Green. 2 (3), 139–147. street vegetation on traffic-induced particle dispersion using microscale simulations.
Nowak, D.J., Stein, S.M., Randler, P.B., Greenfield, E.J., Comas, S.J., Carr, M.A., Alig, J. Environ. Manage. 94 (1), 91–101.
R.J., 2010. Sustaining America’s Urban Trees and Forests: a Forests on the Edge WHO (World Health Organization), 2008. Uncertainty and Data Quality in Exposure
Report. Gen. Tech. Rep. NRS-62. U.S. Department of Agriculture, Forest Service, Assessment Part 1: Guidance Document on Characterizing and Communicating
Northern Research Station, Newtown Square, PA. Uncertainty in Exposure Assessment Part 2: Hallmarks of Data Quality in Chemical
NYC Parks, 2018. The New York City Department of Parks & Recreation. Retrieved June, Exposure Assessment. International Programme on Chemical Safety, Geneva,
2018 from. https://www.nycgovparks.org/. Switzerland.
Openshaw, S., 1984. Ecological fallacies and the analysis of areal census data. Environ. Wolch, J.R., Byrne, J., Newell, J.P., 2014. Urban green space, public health, and en-
Plan. A 16 (1), 17–31. vironmental justice: the challenge of making cities ‘just green enough’. Landsc. Urban
O’Neil-Dunne, J., MacFaden, S., Royar, A., 2014. A versatile, production-oriented ap- Plan. 125, 234–244.
proach to high-resolution tree-canopy mapping in urban and suburban landscapes Wu, J., 2014. Urban ecology and sustainability: the state-of-the-science and future di-
using GEOBIA and data fusion. Remote Sens. 6 (12), 12837–12865. rections. Landsc. Urban Plan. 125, 209–221.
Pandit, R., Laband, D.N., 2010. Energy savings from tree shade. Ecol. Econ. 69 (6), Wu, J., Jones, B., Li, H., Loucks, O.L., 2006. Scaling and Uncertainty Analysis in Ecology:
1324–1329. Methods and Applications. Springer, Dordrecht, The Netherlands 45 pp.
Pappenberger, F., Beven, K.J., 2006. Ignorance is bliss: or seven reasons not to use un- Wu, Z., Chen, L., 2017. Optimizing the spatial arrangement of trees in residential
certainty analysis. Water Resour. Res. 42 (5). neighborhoods for better cooling effects: integrating modeling with in-situ mea-
Patino, J.E., Duque, J.C., 2013. A review of regional science applications of satellite re- surements. Landsc. Urban Plan. 167, 463–472.
mote sensing in urban settings. Comput. Environ. Urban Syst. 37, 1–17. Yang, J., McBride, J., Zhou, J., Sun, Z., 2005. The urban forest in Beijing and its role in air
Payton, S., Lindsey, G., Wilson, J., Ottensmann, J.R., Man, J., 2008. Valuing the benefits pollution reduction. Urban For. Urban Green. 3 (2), 65–78.
of the urban forest: a spatial hedonic approach. J. Environ. Plan. Manag. 51 (6), Zhang, Y., Hussain, A., Deng, J., Letson, N., 2007. Public attitudes toward urban trees and
717–736. supporting urban tree programs. Environ. Behav. 39 (6), 797–814.
Peterson, G., 2000. Political ecology and ecological resilience: an integration of human Zhou, W., Wang, J., Cadenasso, M.L., 2017. Effects of the spatial configuration of trees on
and ecological dynamics. Ecol. Econ. 35 (3), 323–336. urban heat mitigation: a comparative study. Remote Sens. Environ. 195, 1–12.
Petticrew, M., 2001. Systematic reviews from astronomy to zoology: myths and mis- Zipperer, W.C., Sisinni, S.M., Pouyat, R.V., Foresman, T.W., 1997. Urban tree cover: an
conceptions. Br. Med. J. 322 (7278), 98. ecological perspective. Urban Ecosyst. 1 (4), 229–246.
Pickett, S.T., Cadenasso, M.L., Grove, J.M., Boone, C.G., Groffman, P.M., Irwin, E.,

11