Content uploaded by Serdar Selim
Author content
All content in this area was uploaded by Serdar Selim on Nov 18, 2022
Content may be subject to copyright.
Integrated Environmental Assessment and Management —Volume 00, Number 00—pp. 1–9
Received: 31 May 2022
|
Revised: 6 September 2022
|
Accepted: 19 October 2022 1
Ecosystem Services
Determination of the optimum number of sample points
to classify land cover types and estimate the contribution
of trees on ecosystem services using the I‐Tree Canopy tool
Serdar Selim,*Burçin Dönmez, and Ali Kilçik
Department of Space Sciences and Technologies, Faculty of Science, Akdeniz University, Antalya, Turkey
Abstract
The process of producing information about dynamic land use and land cover and ecosystem health quickly with high
accuracy and low cost is important. This information is one of the basic data used for sustainable land management. For this
purpose, remote sensing technologies are generally used, and sampling points are mostly assigned. Determination of the
optimum number of sampling points using the I‐Tree Canopy tool was the main focus of this study. The I‐Tree Canopy tool
classifies land cover, revealing the effects of tree cover on ecosystem services, such as carbon (C) sequestration and storage,
temperature regulation, air pollutant filtering, and air quality improvement, with numerical data. It is used because it is
practical, open source, and user‐friendly. This software works based on sampling point assignment, but it is unclear how
many sampling points should be assigned. Therefore, determining the optimum number of sample points by statistical
methods will increase the effectiveness of this tool and guide users. For this purpose, reference data were created for
comparison. Then, 31 I‐Tree Canopy reports were created with 100‐point increments up to 3100. The data obtained from the
reports were compared with the reference data, and statistical analysis based on Gaussian and a second‐order polynomial fit
was performed. At the end of the analysis, the following results were obtained; the results of this study demonstrated that the
optimum number of sample points for a 1‐ha area is 760 ± 32 from the comparison of the real area and I‐Tree Canopy results.
Similar results from the Gaussian fit of annually sequestered and stored C and carbon dioxide (CO
2
) amounts in trees and the
reduction in air pollution in grams were obtained as 714 ±16. Therefore, we may conclude that taking more than 800 sample
points will not be statistically significant. Integr Environ Assess Manag 2022;00:1–9. © 2022 SETAC
KEYWORDS: Gaussian fit; land cover classification; random sample point; remote sensing; tree canopy
INTRODUCTION
Land use and land cover information, such as forest, set-
tlement, agriculture, water surfaces, impervious surfaces,
open‐green areas, is important and must be obtained with
high accuracy for environmental management applications at
various scales (Hundera et al., 2020; Zhang et al., 2019). Using
this information, researchers determine the causes of land use
dynamics, predict their results, and reveal their effects on
ecosystem services (Veldkamp & Verburg, 2004). They are
also important sources of information used to monitor con-
tinuous changes to the earth's surface and to understand
ecological interactions (Patino & Duque, 2013). In parallel with
the rapid developments in sensor technology, high spatial
resolution images using remote sensing offer important op-
portunities for extracting information from land cover in detail
(Zhao et al., 2016). In addition, classifying land cover from
satellite images is still a problem due to the spectral and
spatial complexity of the images as well as the difficulty and
variety of the methods used (Lu & Weng, 2006). Various
methods have been developed to solve these problems using
geographical information systems. Techniques such as pixel‐
and object‐based simple classification (Deng et al., 2019;
Toure et al., 2018), fuzzy classification (Barakat et al., 2019),
artificial neural network and/or artificial immune systems
(Zhang et al., 2019; Zhong et al., 2006), decision trees (Chen
et al., 2018; Talukdar et al., 2020), support vector machines
(Mustafa et al., 2018; Wu et al., 2018), machine learning al-
gorithms (Abdi, 2020; Camargo et al., 2019), deep learning
(Wu et al., 2018; Zhang et al., 2018), and histogram estimation
classifiers (Liu et al., 2019) are widely used. These techniques
provide information about land cover classes. This information
is used socioeconomically (Cheng et al., 2018; Gounaridis
et al., 2019), socioculturally (Rimal et al., 2018; Tu et al.,2018),
and ecologically (Clerici et al., 2019; Zou et al., 2021) in the
evaluation and management of lands (Arowolo et al., 2018;
Wellmann et al., 2018). From an ecological perspective, de-
termining and estimating vegetation in land cover classes
(Vargas et al., 2019) and classifying and calculating its effects
on ecosystem services (Sun et al., 2019) are important issues
on the world agenda within the scope of global climate
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.DOI: 10.1002/ieam.4704
*Address correspondence to serdarselim@akdeniz.edu.tr
Published 25 October 2022 on wileyonlinelibrary.com/journal/ieam.
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
change, rapid urbanization, and population growth (Bhan
et al., 2021; Wei et al., 2021; Yee et al., 2021). Therefore,
determining the vegetation in land cover is the first step in
ecological studies, environmental planning, and ecological
sustainability (Honeck et al., 2018; Svoray et al., 2013).
Vegetation cover plays a critical role in stabilizing the cli-
mate and carbon (C) storage and regulating the hydrological
cycle (Köhl et al., 2015; Sankey et al., 2021). In particular,
forest cover provides a habitat for various plant and animal
species. It also provides many ecosystem services for all
populations (Miura et al., 2015; Sujetovienė&Daba-
šinskas, 2022). Several important ecosystem services are
provided by forest cover (Van Wensem et al., 2017). These
services are absorbed C and CO
2
within the scope of the
climate, produced oxygen (O
2
), holding nitrogen dioxide
(NO
2
), ozone (O
3
), sulfur dioxide (SO
2
); and in the scope of the
hydrological cycle, avoided runoff, evaporation, interception,
and transpiration (Bösch et al., 2018; Jenkins& Schaap, 2018).
The sustainability and management of these services depend
primarily on the determination of vegetation, its ecological
benefits, and the development of environmental manage-
ment strategies (Leal Filho et al., 2021).
In recent years, the I‐Tree Tools software suite has been
used by many researchers around the world, both in land
cover classification and in calculating the environmental
benefits and ecosystem services of vegetation (Hirabayashi,
2014; Jacobs et al., 2014; Omodior et al., 2021; Parmehr
et al., 2016; and references therein). In addition, this software
produces accurate results based on the operator's ability to
accurately interpret aerial or satellite imagery and detect the
presence or absence of tree cover at each sample point
(Parmehr et al., 2016). This software, developed by the US
Department of Agriculture (USDA) Forest Service in 2006, is
designed to evaluate forest resources in urban and rural areas,
demonstrate the environmental benefits of forests, calculate
the effect of vegetation on ecosystem services, and improve
environmental quality and human health (Nowak, 2021). This
freely available software is used by more than 510000 re-
searchers in 159 countries as of 2020 and is cited in more than
1000 scientific papers (Nowak, 2021). Scientific studies using
this software generally focus on understanding and evaluating
local ecosystem services provided by vegetation and classi-
fying land cover (Omodior et al., 2021; Raum et al., 2019; Ross
et al., 2020; Walters & Sinnett, 2021), determining environ-
mental management strategies for sustainable use of forests
and green lands (Hirabayashi et al., 2011; Hwang &
Wiseman, 2020; Song et al., 2020), calculating air pollution
reduction trends of green areas (Hirabayashi, 2014; Ma et al.,
2020; Pace et al., 2021; Wu et al., 2019), determining land use
change (Ersoy Tonyaloğlu & Atak, 2021), and calculating the
material value of the ecological benefits provided by vege-
tation (Hilde & Paterson, 2014; Qian et al., 2019).
I‐Tree Canopy, an I‐Tree tool, is used frequently in the in-
ternational arena due to its ease of use and data require-
ments. According to Nowak (2021), this tool allows users to
easily photointerpret Google aerial images to produce stat-
istically reliable estimates of tree and other cover types along
with calculations of the uncertainty of their estimates. This tool
appears to provide a quick and inexpensive means to esti-
mate tree cover and other cover types accurately. The I‐Tree
Canopy tool works in three stages: selecting the field, de-
fining the ecological values of the vegetation per unit area,
and determining random sampling points in the relevant re-
gions. Area selection can be made by the user via Google
Earth, which is the base map of the software. Predetermined
values can be used by the USDA Forest Service to calculate
the ecological value of vegetation per unit area, and these
values can also be determined by users for any selected area
using I‐Tree Eco. Randomly sampled points are manually
classified as “with trees”and “without trees”to estimate the
vegetation cover of a predefined area. Online and stock sat-
ellite imagery is used via Google Maps to determine the land
cover of random points. Estimated vegetation cover is given
as a percentage or area (m
2
), and its contribution to eco-
system services is calculated. The accuracy of the estimated
values depends on the operator's ability to correctly interpret
the satellite images and detect the presence or absence of
trees at each sample point (Parmehr et al., 2016). However,
the identification and classification of random sample points
may increase the data collection time and cost. Although the
number of random sampling points suggested by I‐Tree
Canopy software is between 500 and 1000 (https://canopy.
itreetools.org), the high accuracy and minimum error the user
seeks is directly proportional to the increase in the number of
points. To achieve high accuracy, many random sampling
points are needed, and there is no upper limit. This means
that much time and money will be needed for the user to
achieve high accuracy and precision. There must be a balance
between accuracy and time spent for thepractical application.
The lack of comprehensive and statistical research in the lit-
erature on determining the optimum number of sampling
points for land cover classification, including the contribution
of trees to ecosystem health using I‐Tree Canopy, has moti-
vated this study.
This study focused on the I‐Tree Canopy software. I‐Tree
Canopy is a free and practical software that can calculate the
contribution of tree cover to ecosystem services depending
on land cover types while making land use classifications.
Additionally, the European Green Deal has reported that
ecosystem services should be mapped and assessed for
ecosystems and their services (Maes et al., 2013), and I‐Tree
Canopy can provide all this information.
To determine the land cover classification and the con-
tribution of the tree cover in an ecosystem, the optimum
number of sample points is determined. For this purpose, on‐
screen digitation was produced with the help of remote
sensing (RS) and geographical information system (GIS)
technology specific to the study area, and these data were
used as a starting point. Although this software recommends
taking 500–1000 sampling points for a reliable analysis, in the
literature, different sampling numbers are used by different
researchers. In this study, each report produced from 100
points gave data, including its statistical error. After 31 re-
ports, the level of statistical error difference between each
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.wileyonlinelibrary.com/journal/ieam
2Integr Environ Assess Manag 00, 2022—SELIM ET AL.
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
report remained nonsignificant. Then, 31 I‐Tree Canopy re-
ports were used with 100‐point increments up to 3100 in the
study. Each report was compared with the reference data. As
a result, the optimum number of sample points was de-
termined with high accuracy. It is expected that this study will
be a guide for the data collection stage for the scientific
research community using this software.
MATERIALS AND METHODS
Study area
The study area is located in the Döşemealtıdistrict in
Antalya, which is regulated for urban development, where
the land use cover is analyzed. The geographical coor-
dinates of the study area are 36°58′43.31″N and 30°35′
49.43″E, and it covers an area of 1 ha (Figure 1). It is a
moderately urbanized area located in southern Turkey. The
presence of urban forests in the region is also noteworthy.
The study area was divided into six classes, considering the
existing land cover classes and the I‐Tree Canopy cover
classes. These classes are listed as impervious buildings, grass
and/or herbaceous, impervious surface, soil and/or bare
ground, tree and/or shrub, and water. The presence of land
cover belonging to each class was effective in the selection of
the study area (Table 1).
Data preparation
Data generation was done in two stages as reference and
random point sampling datasets (Figure 2). The reference
data were produced by processing the Google Earth satellite
image of September 2022 using ArcGIS 10.4.1 software
(https://www.esri.com/en-us/home). Georeference processes
were done by transferring the satellite image whose corner
coordinates were defined to the software. The image is div-
ided into land use classes with on‐screen digitation. High‐
precision manual classification is verified by a land survey. As
a result of the classification, the percentage and area of each
class were obtained. Random point sampling data were ob-
tained using the I‐Tree Canopy tool of the same image. Then,
explanations for each land use class and the presence or
absence of trees are defined after determining the study area
boundaries. Therefore, appropriate tree utility values were
assigned to each land cover class. For this purpose, I‐Tree Eco
software was used, and analyses conducted in the region
were referenced in the scientific literature (Ersoy Tonyaloğlu &
Atak, 2021; Selim & Atabey, 2020). Altogether, 3100 points
were assigned to the study area in the I‐Tree Canopy tool,
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.DOI: 10.1002/ieam.4704
FIGURE 1 Location of the study area
TABLE 1 Land cover classes of the study area
Number Cover class Description
1 Impervious buildings Areas with impervious cover occupied by buildings
2 Grass and/or herbaceous Grass and other herbaceous ground cover
3 Impervious surface Other impervious cover (e.g., pavements, roads, concrete areas)
4Soiland/or bare ground Soil surface and bare areas without vegetation
5 Tree and/or shrub Areas covered with trees and tall bush vegetation
6 Water surface Artificial and natural water surface without vegetation
OPTIMUM NUMBER OF SAMPLE POINTS OF I‐TREE CANOPY—Integr Environ Assess Manag 00, 2022 3
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
provided that a report was generated for every 100 points
and the land cover class was defined for each point.
Methods
The method was based on the classification and com-
parative analysis of the current satellite image of the same
area using RS techniques and random sample points taken
from I‐Tree Canopy.
To obtain the optimum number of sample points for the 1‐
ha (see Figure 1) area, the variation in the investigated pa-
rameters (land cover classes, sequestered and stored
CandCO
2
amount in trees, and the air pollution in grams in
the study area) were analyzed. All the abovementioned pa-
rameters for 100‐point resolution (from 100 to 3100) were
produced. Then, all these variations were plotted with respect
to the sample points with errors that were taken directly from
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.wileyonlinelibrary.com/journal/ieam
FIGURE 2 Reference data (left) and random point sampling data (right) of the study area
FIGURE 3 Total area of each class with error in the study area
4Integr Environ Assess Manag 00, 2022—SELIM ET AL.
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
the I‐Tree Canopy tool reports. To describe the optimum
number of sampling points, Gaussian fit (Equation 1) was
applied to the peak points of sequestered and stored C and
CO
2
amounts in trees and the air pollution in grams in the
study area.
⎡
⎣
⎢⎛
⎝⎞
⎠⎤
⎦
⎥
μσ σπ
μ
σ
()= −−
Gx x
,, 1
2exp 1
2
,
2(1)
where μis the mean value and σis the standard deviation of
the applied fit.
ANALYSIS AND RESULTS
Comparison of I‐Tree Canopy with remote sensing
classifications
In the first step of our analysis, the areas of different classes
with the number of sample points are compared (Figure 3).
The distribution of the total area of each class reveals that
each exhibits rather different behavior from one another. It
was difficult to obtain a certain number of sampling points.
However, the distribution of all plots does not indicate serious
saturation after approximately 700 sample points, as shown in
Figure 3. To obtain a more reliable determination of the
necessary sample point average, the percentage difference of
all land classes for each set is calculated using the real area
obtained from RS classifications (Table 2). Then, the combined
average error of each class is plotted for each set, as shown in
Figure 4. It can be seen from this figure that the error has a
minimum of approximately 800 and 3000 points. Therefore, a
second‐order polynomial fit is applied for points around the
first minimum. The necessary number of points obtained from
the fit for minimum error in the area estimation is 760 ± 32
with 1 standard deviation. Taking more datapoints will not
reduce the error level, and therefore, taking more datapoints
will not be necessary (Table 3).
Determination of optimum number of sampling points
To investigate the variation in annually sequestered and
stored C and CO
2
amounts in trees, each parameter is plotted
against the number of sample points, as shown in Figure 5.
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.DOI: 10.1002/ieam.4704
TABLE 2 Real area of each cover class calculated using on‐screen
digitation method
Cover class Area (m
2
)
Impervious buildings 528.06
Grass and/or herbaceous 1596.03
Impervious surface 2029.84
Soil and/or bare ground 1187.34
Tree and/or shrub 4603.17
Water surface 55.56
FIGURE 4 Average error in all land classes calculated with I‐Tree Canopy and the real area of each class given in Table 2
TABLE 3 Number of necessary sampling points calculated from the
Gaussian fit for tree benefits estimates (carbon and CO
2
equivalent)
Description Number of points
Sequestered annually in trees (carbon) 722 ± 16
Sequestered annually in trees (CO
2
equivalent)
715 ± 16
Stored in tree (carbon) 715 ± 16
Stored in tree (CO
2
equivalent) 714 ± 16
OPTIMUM NUMBER OF SAMPLE POINTS OF I‐TREE CANOPY—Integr Environ Assess Manag 00, 2022 5
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
It can be seen that both annually sequestered and stored
C and CO
2
amounts in trees exhibit similar behavior in
that all datasets have a local maximum for approximately
700 datapoints. Therefore, a Gaussian fit is made around
these points. The obtained fit results are listed in Table 4.
A similar analysis is made for the reduction in air pollution
in grams in the study area. As shown in Figure 6, all selected
parameters (CO, NO
2
,O
3
,SO
2
, PM2.5, and PM10) exhibit
similar behavior without exception. Therefore, again, a
Gaussian fit was done, and the results are also shown in
Table 4.
DISCUSSION AND CONCLUSIONS
It is important to have reliable, accurate, and up‐to‐date
information for the sustainability of the environment (Martins
et al., 2021). This contributes to the implementation of
decision‐making strategies related to a better understanding
of ecological dynamics, management, and sustainability of
the environment (Fassnacht et al., 2016). In general, RS im-
ages and GIS software are used for this purpose, which are
expensive to purchase and require expertise for analysis
(Miraki et al., 2021). In addition, RS data for monitoring and
assessing land use often cannot keep up with the rapid
change in the urban environment (Musakwa & Van Nie-
kerk, 2013). Cities have undergone rapid change due to both
reconstruction and the expansion of concretion over natural
areas as a result of socioeconomic transformation (Mansour
et al., 2020). Therefore, acquiring up‐to‐date, accurate, and
practical ecological information provides important in-
formation to managers in the planning phase. The free I‐Tree
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.wileyonlinelibrary.com/journal/ieam
FIGURE 5 Annually sequestered and stored carbon and carbon dioxide (CO
2
) amount in trees
TABLE 4 Number of necessary sampling points calculated from the Gaussian fit for tree benefits estimates (air pollution)
Abbreviation Description Number of points
CO Carbon monoxide removed annually 714 ± 16
NO
2
Nitrogen dioxide removed annually 714 ± 16
O
3
Ozone removed annually 714 ± 16
SO
2
Sulfur dioxide removed annually 714 ± 16
PM2.5 Particulate matter <2.5 μm removed annually 714 ± 16
PM10 Particulate matter >2.5 μm and <10 μm removed annually 714 ± 16
6Integr Environ Assess Manag 00, 2022—SELIM ET AL.
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Canopy tool provides a low‐cost, fast, and repeatable
method to evaluate tree cover and explain its impact on
ecosystem services with numerical data. It is also stated that
this software is promising in predicting and evaluating the
benefits generated by ecosystem services, providing high
accuracy; data produced largely overlap with results obtained
by image classification techniques (Olivatto, 2019; Raza-
ghirad, 2021; Richardson & Moskal, 2014). However, there
are some restrictions on the use of this method. The most
obvious and time‐consuming limitation is the inability to de-
termine the appropriate number of sampling points. Several
examples can be given as follows: Omodior et al. (2021) used
an average of 280 random points for vegetation cover clas-
sification; Jacobs et al. (2014) used 1000 random points to
classify landscape features across Australia; 500 points were
used to examine the tree cover differences in the study area
between 2008 and 2011 by Hwang and Wiseman (2020);
Richardson and Moskal (2014) used 1000 points to evaluate
the urban forest cover; Ersoy Tonyaloğlu and Atak (2021)
used 10 608 points in a study investigating the effects of land
use and land cover change on potential regulating ecosystem
services. It has been seen that a different number of sample
points are used regardless of the size of the land (Benjamin
et al., 2015; Walters & Sinnett, 2021; Zho, 2017). Even in the
user's guide of the I‐Tree Canopy software, it is emphasized
that the number of random sample points increases the ac-
curacy, but no information for the upper limit is given (https://
canopy.itreetools.org/). The motivation of this study was to
determine the optimum number of random sample points for
this software that appeals to a large number of users and
produces important information about regulatory ecosystem
services. Therefore, it can be used quickly, practically, and
with high accuracy.
The results of this study demonstrated that the optimum
number of sample points for a 1‐ha area is 760 ± 32 from the
comparison of the real area and I‐Tree Canopy results.
Similar results from the Gaussian fit of annually sequestered
and stored C and CO
2
amounts in trees and the reduction in
air pollution in grams were obtained as 714 ± 16. It is seen
that points taking more than 800 samples did not create a
statistically significant difference. Therefore, the time–cost
balance can be achieved with the optimum number of
points determined within the scope of this study.
There is no specific recommendation in the literature re-
garding the number of points that should be assigned in
different sizes and in different land covers. It is envisioned
that the optimum number of points may vary statistically to
reach maximum accuracy with area size. This requires a
study to determine the optimum number of sampling points
in different area sizes and different land textures. The de-
termination of the optimum number of points to be as-
signed in a unit area (e.g., for m
2
or km
2
) by analyzing areas
of different sizes is the subject of future study.
ACKNOWLEDGMENT
There are no funders to report for this submission.
AUTHOR CONTRIBUTIONS
Serdar Selim: Conceptualization; Data curation; Formal
analysis; Investigation; Methodology; Resources; Software;
Supervision; Visualization; Writing—original draft; Writing—
review & editing. Burçin Dönmez: Data curation;
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.DOI: 10.1002/ieam.4704
FIGURE 6 Reduction in air pollution in grams in the study area
OPTIMUM NUMBER OF SAMPLE POINTS OF I‐TREE CANOPY—Integr Environ Assess Manag 00, 2022 7
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Investigation; Methodology; Software; Validation; Visual-
ization; Writing—original draft; Writing—review & editing. Ali
Kilçik: Data curation; Formal analysis; Methodology; Vali-
dation; Writing—original draft; Writing—review & editing.
DATA AVAILABILITY STATEMENT
Data, associated metadata, and calculation tools are
available from corresponding author Serdar Selim
(serdarselim@akdeniz.edu.tr).
ORCID
Serdar Selim http://orcid.org/0000-0002-5631-6253
Burçin Dönmez http://orcid.org/0000-0002-9730-1998
Ali Kilçik http://orcid.org/0000-0002-0094-1762
REFERENCES
Abdi, A. M. (2020). Land cover and land use classification performance of
machine learning algorithms in a boreal landscape using Sentinel‐2 data.
GIScience & Remote Sensing,57(1), 1–20.
Arowolo, A. O., Deng, X., Olatunji, O. A., & Obayelu, A. E. (2018). Assessing
changes in the value of ecosystem services in response to land‐use/land‐
cover dynamics in Nigeria. Science of the Total Environment,636,597–609.
Barakat, A., Ouargaf, Z., Khellouk, R., El Jazouli, A., & Touhami, F. (2019).
Land use/land cover change and environmental impact assessment in
Béni‐Mellal District (Morocco) using remote sensing and gis. Earth Sys-
tems and Environment,3(1), 113–125.
Benjamin, A. R., Hochmair, H. H., Gann, D., & Fu, Z. J. (2015). Miami‐Dade
urban tree canopy analysis (Vol. 45). GIS Center.
Bhan, M., Gingrich, S., Roux, N., Le Noë, J., Kastner, T., Mate, S., & Erb, K. H.
(2021). Quantifying and attributing land use‐induced carbon emissions to
biomass consumption: A critical assessment of existing approaches.
Journal of Environmental Management,286, 112228.
Bösch, M., Elsasser, P., Franz, K., Lorenz, M., Moning, C., Olschewski, R., &
Weller, P. (2018). Forest ecosystem services in rural areas of
Germany: Insights from the national TEEB study. Ecosystem Services,
31,77–83.
Camargo, F. F., Sano, E. E., Almeida, C. M., Mura, J. C., & Almeida, T. (2019).
A comparative assessment of machine‐learning techniques for land use
and land cover classification of the Brazilian tropical savanna using ALOS‐
2/PALSAR‐2 polarimetric images. Remote Sensing,11(13), 1600.
Chen, W., Zhang, S., Li, R., & Shahabi, H. (2018). Performance evaluation of
the GIS‐based data mining techniques of best‐first decision tree, random
forest, and naïve Bayes tree for landslide susceptibility modeling. Science
of the Total Environment,644, 1006–1018.
Cheng, Y., Yu, L., Xu, Y., Liu, X., Lu, H., Cracknell, A. P., Kanniah, K., &
Gong, P. (2018). Towards global oil palm plantation mapping using
remote‐sensing data. International Journal of Remote Sensing,39(18),
5891–5906.
Clerici, N., Cote‐Navarro, F., Escobedo, F. J., Rubiano, K., & Villegas, J. C.
(2019). Spatio‐temporal and cumulative effects of land use‐land cover and
climate change on two ecosystem services in the Colombian Andes.
Science of the Total Environment,685, 1181–1192.
Deng, Z., Zhu, X., He, Q., & Tang, L. (2019). Land use/land cover classification
using time series Landsat 8 images in a heavily urbanized area. Advances
in Space Research,63(7), 2144–2154.
Ersoy Tonyaloğlu, E., & Atak, B. K. (2021). Impact of land cover change on
urban tree cover and potential regulating ecosystem services: The case of
Aydın/Turkey. Environmental Monitoring and Assessment,193(11), 1–9.
Fassnacht, F. E., Latifi, H., Stereńczak, K., Modzelewska, A., Lefsky, M., Waser,
L. T., & Ghosh, A. (2016). Review of studies on tree species classification
from remotely sensed data. Remote Sensing of Environment,186,64–87.
Gounaridis, D., Chorianopoulos, I., Symeonakis, E., & Koukoulas, S. (2019). A
Random Forest‐Cellular Automata modelling approach to explore future
land use/cover change in Attica (Greece), under different socio‐economic
realities and scales. Science of the Total Environment,646, 320–335.
Hilde, T., & Paterson, R. (2014). Integrating ecosystem services analysis into
scenario planning practice: Accounting for street tree benefits with i‐Tree
valuation in Central Texas. Journal of Environmental Management,146,
524–534.
Hirabayashi, S. (2014). i‐Tree Canopy air pollutant removal and monetary value
model descriptions. The Davey Institute. Retrieved April 15, 2022,
from: https://www.chesapeakebay.net/channel_files/43362/i-tree_canopy_
air_pollutant_removal_and_monetary_value_model_descriptions.pdf
Hirabayashi, S., Nowak, D., Endreny, T. A., Kroll, C., & Maco, S. (2011). i‐Tree:
Tools to assess and manage structure, function, and value of community
forests. American Geophysical Union, Fall Meeting Abstracts,2011,
B21B‐0263.
Honeck, E., Castello, R., Chatenoux, B., Richard, J. P., Lehmann, A., &
Giuliani, G. (2018). From a vegetation index to a sustainable development
goal indicator: Forest trend monitoring using three decades of earth
observations across Switzerland. ISPRS International Journal of Geo‐
Information,7(12), 455.
Hundera, H., Mpandeli, S., & Bantider, A. (2020). Spatiotemporal analysis of
land‐use and land‐cover dynamics of Adama District, Ethiopia and its
implication to greenhouse gas emissions. Integrated Environmental As-
sessment and Management,16(1), 90–102.
Hwang, W. H., & Wiseman, P. E. (2020). Geospatial methods for tree canopy
assessment: A case study of an urbanized college campus. Arboriculture
& Urban Forestry,46(1), 51–65.
Jacobs, B., Mikhailovich, N., & Delaney, C. (2014). Benchmarking Australia's
urban tree canopy: An i‐Tree assessment. Institute for Sustainable Futures.
Jenkins, M., & Schaap, B. (2018). Forest ecosystem services. Background
analytical study 1 (41 p.). Background study prepared for the thirteenth
session of United Nations Forum on Forests.
Köhl, M., Lasco, R., Cifuentes, M., Jonsson, Ö., Korhonen, K. T., Mundhenk,
P., & Stinson, G. (2015). Changes in forest production, biomass and
carbon: Results from the 2015 UN FAO Global Forest Resource Assess-
ment. Forest Ecology and Management,352,21–34.
Leal Filho, W., Azeiteiro, U. M., Balogun, A. L., Setti, A. F. F., Mucova, S. A.,
Ayal, D., & Oguge, N. O. (2021). The influence of ecosystems services
depletion to climate change adaptation efforts in Africa. Science of the
Total Environment,779, 146414.
Liu, H., Wang, F., Yang, S., Hou, B., Jiao, L., & Yang, R. (2019). Fast semi-
supervised classification using histogram‐based density estimation for
large‐scale polarimetric SAR data. IEEE Geoscience and Remote Sensing
Letters,16(12), 1844–1848.
Lu, D., & Weng, Q. (2006). Use of impervious surface in urban land‐use
classification. Remote Sensing of Environment,102(1–2), 146–160.
Ma, J., Cheng, J. C., Xu, Z., Chen, K., Lin, C., & Jiang, F. (2020). Identification
of the most influential areas for air pollution control using XGBoost and
Grid Importance Rank. Journal of Cleaner Production,274, 122835.
Maes, J., Teller, A., Erhard, M., Liquete., C., Braat, L., Berry, P., Egoh, B.,
Puydarrieux, P., Fiorina, C., Santos, F., Paracchini, M. L., Keune, H.,
Wittmer, H., Hauck, J., Fiala, I., Verburg, P. H., Condé, S., Schägner, J. P.,
San Miguel, J., …Bidoglio, G. (2013). Mapping and assessment of eco-
systems and their services. An analytical framework for ecosystem as-
sessments under action 5 of the EU biodiversity strategy to 2020.
European Union.
Mansour, S., Al‐Belushi, M., & Al‐Awadhi, T. (2020). Monitoring land use and
land cover changes in the mountainous cities of Oman using GIS and CA‐
Markov modelling techniques. Land Use Policy,91, 104414.
Martins, J. A. C., Menezes, G., Gonçalves, W., Sant'Ana, D. A., Osco, L. P.,
Liesenberg, V., & Junior, J. M. (2021). Machine learning and SLIC for tree
canopies segmentation in urban areas. Ecological Informatics,66, 101465.
Miraki, M., Sohrabi, H., Fatehi, P., & Kneubuehler, M. (2021). Individual tree
crown delineation from high‐resolution UAV images in broadleaf forest.
Ecological Informatics,61, 101207.
Miura, S., Amacher, M., Hofer, T., San‐Miguel‐Ayanz, J., & Thackway, R.
(2015). Protective functions and ecosystem services of global forests in
the past quarter‐century. Forest Ecology and Management,352,35–46.
Musakwa, W., & Van Niekerk, A. (2013). Implications of land use change for
the sustainability of urban areas: A case study of Stellenbosch, South
Africa. Cities,32, 143–156.
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.wileyonlinelibrary.com/journal/ieam
8Integr Environ Assess Manag 00, 2022—SELIM ET AL.
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Mustafa, A., Rienow, A., Saadi, I., Cools, M., & Teller, J. (2018). Comparing
support vector machines with logistic regression for calibrating cellular
automata land use change models. European Journal of Remote Sensing.
51(1), 391–401.
Nowak, D. J. (2021). Understanding i‐Tree: 2021 summary of programs and
methods (General Technical Report NRS‐200‐2021). Northern Research
Station, US Department of Agriculture, Forest Service, Madison, WI.
Olivatto, T. F. (2019). Using i‐Tree Canopy to estimate and value ecosystem
services of air pollutant removal. In Y. Iano, R. Arthur, O. Saotome, V.
Vieira Estrela, & H. Loschi (Eds.), Proceedings of the 3rd Brazilian Tech-
nology Symposium. BTSym 2017. Springer.
Omodior, O., Eze, P., & Anderson, K. R. (2021). Using i‐Tree Canopy vege-
tation cover subtype classification to predict peri‐domestic tick presence.
Ticks and Tick‐borne Diseases,12(3), 101684.
Pace, R., Guidolotti, G., Baldacchini, C., Pallozzi, E., Grote, R., Nowak, D. J., &
Calfapietra, C. (2021). Comparing i‐Tree Eco estimates of particulate
matter deposition with leaf and canopy measurements in an urban
Mediterranean Holm Oak forest. Environmental Science & Technology,
55(10), 6613–6622.
Parmehr, E. G., Amati, M., Taylor, E. J., & Livesley, S. J. (2016). Estimation of
urban tree canopy cover using random point sampling and remote
sensing methods. Urban Forestry & Urban Greening,20, 160–171.
Patino, J. E., & Duque, J. C. (2013). A review of regional science applications
of satellite remote sensing in urban settings. Computers, Environment
and Urban Systems,37,1–17.
Qian, W., Zhang, Z., & Ping, W. (2019). An assessment of ecosystem services
of urban green spaces based on i‐Tree. Journal of Landscape Research,
11(1), 53–56.
Raum, S., Hand, K. L., Hall, C., Edwards, D. M., O'Brien, L., & Doick, K. J.
(2019). Achieving impact from ecosystem assessment and valuation of
urban greenspace: The case of i‐Tree Eco in Great Britain. Landscape and
Urban Planning,190, 103590.
Razaghirad, B. (2021). Urban tree canopy assessment using geospatial tech-
nologies: A case study of the town of Lincoln, Ontario (thesis). Brock
University, St. Catharinas, CA.
Richardson, J. J., & Moskal, L. M. (2014). Uncertainty in urban forest canopy
assessment: Lessons from Seattle, WA, USA. Urban Forestry & Urban
Greening,13(1), 152–157.
Rimal, B., Zhang, L., Keshtkar, H., Haack, B. N., Rijal, S., & Zhang, P. (2018).
Land use/land cover dynamics and modeling of urban land expansion by
the integration of Cellular Automata and Markov chain. ISPRS Interna-
tional Journal of Geo‐Information,7(4), 154.
Ross, S., Jean‐Philippe, S., Clatterbuck, W., Giffen, N., Herold, J., & Zobel, J.
(2020). i‐Tree Eco analysis of landscape vegetation on remediated areas
of Oak Ridge National Laboratory. Open Journal of Forestry,10(04), 412.
Sankey, T., Belmonte, A., Massey, R., & Leonard, J. (2021). Regional‐scale
forest restoration effects on ecosystem resiliency to drought: A synthesis
of vegetation and moisture trends on Google Earth Engine. Remote
Sensing in Ecology and Conservation,7(2), 259–274.
Selim, C., & Atabey, S. (2020). Determination of the benefits of urban road
plantings: A case study of Atatürk Boulevard (Antalya). Journal of Agri-
cultural Faculty of Bursa Uludag University,34, 235–247.
Sujetovienė, G., & Dabašinskas, G. (2022). Interactions between changes in
land cover and potential of ecosystem services in Lithuania at temporal
and spatial scale. Ecological Complexity,49, 100984.
Song, P., Kim, G., Mayer, A., He, R., & Tian, G. (2020). Assessing the eco-
system services of various types of urban green spaces based on i‐Tree
Eco. Sustainability,12(4), 1630.
Sun, Y., Xie, S., & Zhao, S. (2019). Valuing urban green spaces in mitigating
climate change: A city‐wide estimate of aboveground carbon stored in
urban green spaces of China's capital. Global Change Biology,25(5),
1717–1732.
Svoray, T., Perevolotsky, A., & Atkinson, P. M. (2013). Ecological sustainability
in rangelands: The contribution of remote sensing. International Journal
of Remote Sensing,34(17), 6216–6242.
Talukdar, S., Singha, P., Mahato, S., Pal, S., Liou, Y. A., & Rahman, A. (2020).
Land‐use land‐cover classification by machine learning classifiers for sat-
ellite observations—a review. Remote Sensing,12(7), 1135.
Toure, S. I., Stow, D. A., Shih, H. C., Weeks, J., & Lopez‐Carr, D. (2018). Land
cover and land use change analysis using multi‐spatial resolution data and
object‐based image analysis. Remote Sensing of Environment,210,
259–268.
Tu, S., Long, H., Zhang, Y., Ge, D., & Qu, Y. (2018). Rural restructuring at
village level under rapid urbanization in metropolitan suburbs of China
and its implications for innovations in land use policy. Habitat Interna-
tional,77, 143–152.
Van Wensem, J., Calow, P., Dollacker, A., Maltby, L., Olander, L., Tuvendal, M.,
& Van Houtven, G. (2017). Identifying and assessing the application of
ecosystem services approaches in environmental policies and decision
making. Integrated Environmental Assessment and Management,13(1),
41–51.
Vargas, L., Willemen, L., & Hein, L. (2019). Assessing the capacity of ecosystems
to supply ecosystem services using remote sensing and an ecosystem ac-
counting approach. Environmental Management,63(1), 1–15.
Veldkamp, A., & Verburg, P. H. (2004). Modelling land use change and
environmental impact. Journal of EnvironmentalManagement,72(1–2), 1–3.
Walters, M., & Sinnett, D. (2021). Achieving tree canopy cover targets: A case
study of Bristol, UK. Urban Forestry & Urban Greening,65, 127296.
Wei, C., Lin, Q., Yu, L., Zhang, H., Ye, S., & Zhang, D. (2021). Research on
sustainable land use based on production–living–ecological function: A
case study of Hubei Province, China. Sustainability,13(2), 996.
Wellmann, T., Haase, D., Knapp, S., Salbach, C., Selsam, P., & Lausch, A.
(2018). Urban land use intensity assessment: The potential of spatio‐
temporal spectral traits with remote sensing. Ecological Indicators,85,
190–203.
Wu, J., Wang, Y., Qiu, S., & Peng, J. (2019). Using the modified i‐Tree Eco
model to quantify air pollution removal by urban vegetation. Science of
the Total Environment,688, 673–683.
Wu, W., Li, A. D., He, X. H., Ma, R., Liu, H. B., & Lv, J. K. (2018). A comparison
of support vector machines, artificial neural network and classification tree
for identifying soil texture classes in southwest China. Computers and
Electronics in Agriculture,144,86–93.
Yee, S. H., Paulukonis, E., Simmons, C., Russell, M., Fulford, R., Harwell, L., &
Smith, L. M. (2021). Projecting effects of land use change on human well‐
being through changes in ecosystem services. Ecological Modelling,440,
109358.
Zhang, C., Sargent, I., Pan, X., Li, H., Gardiner, A., Hare, J., & Atkinson, P. M.
(2019). Joint deep learning for land cover and land use classification.
Remote Sensing of Environment,221, 173–187.
Zhang, P., Ke, Y., Zhang, Z., Wang, M., Li, P., & Zhang, S. (2018). Urban land
use and land cover classification using novel deep learning models based
on high spatial resolution satellite imagery. Sensors,18(11), 3717.
Zhao, B., Zhong, Y., & Zhang, L. (2016). A spectral–structural bag‐of‐features
scene classifier for very high spatial resolution remote sensing imagery.
ISPRS Journal of Photogrammetry and Remote Sensing,116,73–85.
Zho, R. (2017). Assessing the changes in urban tree canopy cover, ecosystem
services, streamflow and pollutants in Corpus Christi, Texas using I‐Tree
applications (dissertation). Texas A&M University, Kingsville, TX.
Zhong, Y., Zhang, L., Huang, B., & Li, P. (2006). An unsupervised artificial
immune classifier for multi/hyperspectral remote sensing imagery.
IEEE Transactions on Geoscience and Remote Sensing,44(2), 420–431.
Zou, T., Chang, Y., Chen, P., & Liu, J. (2021). Spatial‐temporal variations of
ecological vulnerability in Jilin Province (China), 2000 to 2018. Ecological
Indicators,133, 108429.
Integr Environ Assess Manag 2022:1–9 © 2022 SETAC.DOI: 10.1002/ieam.4704
OPTIMUM NUMBER OF SAMPLE POINTS OF I‐TREE CANOPY—Integr Environ Assess Manag 00, 2022 9
15513793, 0, Downloaded from https://setac.onlinelibrary.wiley.com/doi/10.1002/ieam.4704 by Akdeniz University, Wiley Online Library on [18/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
