Ann. For. Res. 61(2): 189-202, 2018 ANNALS OF FOREST RESEARCH [629559]

189
Ann. For. Res. 61(2): 189-202, 2018 ANNALS OF FOREST RESEARCH
DOI: 10.15287/afr.2018.1189 www.afrjournal.org
Data collection methods for forest inventory: a
comparison between an integrated conventional
equipment and terrestrial laser scanning
Bogdan Apostol1§, Serban Chivulescu1, Albert Ciceu1, Marius Petrila1, Ionut-Silviu
Pascu1,2, Ecaterina Nicoleta Apostol1, Stefan Leca1, Adrian Lorent1,2, Mihai Tanase1,3,4,
Ovidiu Badea1,2
Apostol B., Chivulescu S., Ciceu A., Petrila M., Pascu I.-S., Apostol E.N.,
Leca S., Lorent A., Tanase M., Badea O., 2018. Data collection methods for
forest inventory: a comparison between an integrated conventional equipment and
terrestrial laser scanning. Ann. For. Res. 61(2): 189-202.
Abstract. This study aims to present a comparison analysis of two data col –
lection methods that can be used in order to obtain reference ground truth
data for forestry – a conventional method that uses specific equipment such
as Field Map system, caliper and vertex inclinometer and a modern meth –
od based on terrestrial laser scanning (TLS) technology. The research was
conducted in six circular Permanent Plots (PPs) with an area of 500 square
meters each, within thinning and selected cuttings stands of sessile oak
(Quercus petraea (Matt.) Liebl.), common beech ( Fagus sylvatica L.) and
Norway spruce ( Picea abies L. Karst.), all situated in the Southern Car –
pathians (Mihăești, Mușeteși and Vidraru Forest Districts). Using the con –
ventional method, the dendrometric tree characteristics such as height, di –
ameter at breast height (dbh) and tree position were directly recorded in the
field. As a modern method for data collection, a Faro Focus3D X 130 HDR
terrestrial laser scanning device was used to scan each plot and to extract
the dbh and height of the trees. In this regard, two scanning approaches were
used – single scan (SS) and multiple scan (MS). In order to compare the two
data acquisitions methods, we applied a Strengths, Weaknesses, Opportuni –
ties, Threats (SWOT) analysis on the basis of which we could establish the
pros and cons of using the two methods. Therefore, one can choose the most
advantageous method for obtaining the reference data for forestry, in terms
of equipment acquisition cost, personnel skills and qualifications, data col –
lection working time, accuracy of the data recorded, post processing time,
labor costs. Although the use of TLS in forest inventory is a technology with
high potential, further investigations need to be done, especially in the case
of automatic extraction of the tree height. For accurate reference ground
data for forest inventory purposes, we still recommend using the conven –
tional methods although they are time consuming.
Keywords: Field Map system, terrestrial laser scanning, forest inventory
Authors. 1“Marin Drăcea” National Research and Development Institute in
Forestry, 128 B-dul. Eroilor, V oluntari, Ilfov, 077190 Romania | 2Faculty of , ,
,
^
,,^,
,

190
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85Introduction
The essential role of the forest ecosystems is
to provide resources and ecological services,
making them important for biodiversity con –
servation, soil protection and mitigation of cli –
mate change (Trumbore et al. 2015, Tubiello et
al. 2015, Cabo et al. 2018).
The assessment of stand characteristics with
high accuracy is a key aspect that has impli –
cations for the forest management activities,
forest fire modeling and carbon stock estima –
tion (Keenan et al. 2015, MacDicken 2015,
Cabo et al. 2018). Methods for measuring the
structural characteristics of the forest stands
have rapidly evolved from the conventional
to the modern ones. The conventional meth –
ods of measuring the structural characteristics
of the forest have the ability to provide direct
measurement, but the cost of producing them
is rather high. In this regard, getting fast data
at minimal cost became a necessity. Thereby
one of the highest precision data acquisition
devices that could be used in forestry is the ter –
restrial laser scanner (TLS) (Wang et al. 2017).
Lately, the laser scanning technology was used
for different activities such as mapping (Blair
et al. 1999, Asner et al. 2014), photography
(Niska et al. 2010, Wang et al. 2017) and sur –
veying (Boehm et al. 2013).
Nowadays, TLS represents the most ad –
vanced method of determining field informa –
tion (Zemánek et al. 2017). Compared with
airborne and spaceborne laser scanning sys –
tems, the terrestrial laser scanning devices
are used more locally in order to capture the
details of the objects (Wang et al. 2017) and
to collect the spatial information of the forest
(Watt & Donoghue 2005) as well as its struc -tural characteristics (Maas et al. 2008, Lovell
et al. 2011, Pueschel et al. 2013).
The aim of the study is to present a com –
parative analysis of two data collection meth –
ods that can be used in order to obtain refer –
ence ground data in forestry: a conventional
method that use specific equipment – caliper
and vertex inclinometer and a GIS recording
with a field computer (Field Map system) and
a modern method based on terrestrial laser
scanning technology. The main objective was
the analysis of the pros and cons of using the
two methods, as a support to choose the most
advantageous method to obtain reference data
in forestry.
Materials and methods
The research was conducted in six circular
Permanent Plots (PP) with an area of 500 m2
each, within thinning and selected cuttings
stands of sessile Sessile oak ( Quercus pet –
raea (Matt.) Liebl.), common beech ( Fagus
sylvatica L.) and Norway spruce ( Picea abies
L. Karst. ), all located in the Southern Carpath –
ians (Mihăești, Mușătești and Vidraru Forest
Districts) (figure 1). Each circular plot is part
of a Permanent Sample Area (PSA) with a size
of 1 ha that were installed in the framework of
EO-ROFORMON project (http: //www.eo-ro –
formon.ro ).
As conventional method in our study we
used the Field Map system (www.fieldmap.
cz), a combination of electronical caliper and
vertex inclinometer to measure the trees char –
acteristics (dbh, height and spatial position of
trees) within each plot. Field Map system is
a specialized equipment and software used Silviculture and Forest Engineering, “Transilvania” University of Brașov, 1
Șirul Beethoven, 500123, Romania | 3Department of Geology, Geography, and
Environment, University of Alcalá de Henares, Spain | 4School of Ecosystem
and Forest Sciences, University of Melbourne, Australia.
§ Corresponding author: Bogdan Apostol (bogdanap_ro@yahoo.com)
Manuscript received November 4, 2018; revised December 26, 2018; accepted
December 29, 2018; online first December 31, 2018.

191
Apostol et al. Data collection methods for forest inventory …in forest inventories, which use a laser range
finder, an inclinometer, an electronic compass
mounted on a monopod and connected to a
software (Field Map Data Collector) installed
on a rugged tablet computer (V opěnka &
Černý 2006). The software is used to structure
the database characteristics, to record the field
measurements and for the geospatial process –
ing. In order to get reference ground data this
technology was used in remote sensing stud –
ies based both on passive (Bernasconi et al.
2017, Brovkina et al. 2018) and active sensors
(Brovkina et al. 2017, Tockner et al. 2017).
In each plot and by the use of the Field Map
system, we directly assessed the tree position
and the species. For a better productivity, den –
drometric tree characteristics (dbh, tree height)
were measured using a caliper and a Vertex IV
inclinometer and the data were imported in
Field Map Data Collector software. For further
geospatial analysis, the central coordinates of
each plot were recorded with a Trimble Ge –
oXH 6000 GNSS receiver equipped with a
Zephyr external antenna. The GNSS data were
differential post processed using EUREF (Eu –
ropean reference frame) stations and used to
convert the Field Map measured data from lo –
cal to global coordinates.
The use of the terrestrial laser scanning tech –
nology to extract the dendrometric character –
istics of the trees is considered to be a modern
method, which could have favorable implica -tions in forestry practice. We used a Faro Fo –
cus3D X 130 HDR Terrestrial Laser Scanning
device to scan the trees within each plot (fig –
ure 2). This version of the TLS is a phase-shift
based scanner, the distance between the scan –
ner and the target being calculated accordingly
with the shift in phase of the returned modula –
tions (Newnham et al. 2011).
In our study, two scanning approaches were
used – a single scan (SS) and multiple scan
(MS) (figure 3). In multiple scan approach,
the TLS stations were placed in the plot center
and according the cardinal directions (north,
east, south, and west) at a distance of 15 m
from the center of the plot. The most import –
ant parameters that can be set up before the
scanning operation are the resolution and the
quality. Based on literature (Stanley 2013),
the chosen values of theese, considered for the
forested areas were: ¼ for resolution and 3x
for the quality. Thereby, all the trees near the
border of the circular plot were recorded. For
co-registration purpose, in each plot there were
placed 7 white spheres with a 14 cm diameter,
so that from each TLS station should be seen
at least four of them. The co-registration of
the TLS point clouds was made with the Faro
Scene software (http://www.faro.com). Fur –
ther, the point clouds were post processed with
Computree software (http://www.computree.
onf.fr) , to extract the ground level (the digital
terrain model – DTM) of each plot (Othmani et Study area Figure 1

192
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85al. 2011), and the dbh and height of its trees. A
randomized Hough transformation ( Simonse
et al. 2003, Aschoff & Spiecker 2004 ) was ap –
plied as an automatic method that allows de –
tection of tree trunks as clusters of points. The
tree dbh was estimated by fitting a geometric
circle into a trunk slice located at 1.25-1.35
above the ground level. The tree height was
calculated as the difference between the low –
est point (ground level) and the highest point
inside of one cluster.
The TLS estimated dbh, height and positions
of the trees within each plot were clipped ac –
cording to the 500 m2 plot border. The result -ed data were then compared
with the Field Map inventory
data at the plot level, without
making the individual tree
identification between the
datasets. To eliminate the ob –
servations which may induce
errors to the final results – e.g.
outliers or measurement er –
rors – we used different meth –
ods to clean the data. On the
Field Map inventory data ob –
tained with the caliper (dbh)
and Vertex IV (heights), a
statistic approach where all
values below the first quan –
tile and above third quantile
would be considered outliers
could not be applied. This
is was choosed because the
analyzed stands have high –
er coefficients of variation
of dbh and an associated in –
verse-J shaped distribution,
characteristic to uneven-aged
stands or where the selected
cuttings are applied. In order
to eliminate the outliers in
an objective way, we used
the ratio between height (m)
and dbh (cm). This index is
a good indicator of trees me –
chanic stability; we accepted
an interval of 0.6 and 1.7 for
eliminating trees with unreal height or dbh
as such values of the index were studied be –
fore (Grudnicki 2004) for different production
classes and species. We also eliminated all the
trees having a top break, being dead or having
merged stems. For the trees obtained by the
TLS method, we eliminated all values above
the maximum and below the minimum height
and the dbh values which had been assessed in
the Field Map inventory.
To compare the two acquisitions methods
and its output data, we also applied a Strengths,
Weaknesses, Opportunities, Threats (SWOT)
Faro Focus3D X 130 HDR Terrestrial Laser Scanning
(TLS) – in scan position (plot SM15TM)Figure 2
TLS scanning approaches: a) single scan (SS) b) mul –
tiple scan (MS)Figure 3

193
Apostol et al. Data collection methods for forest inventory …Equipment
usedPersonnel
skills and
qualificationsActivities
Field Map
system,
caliper
and vertex
inclinometerForestry
engineer
& 2-4
forestry
techniciansField works
– Select the plot location and create the Field Map project
– Measure the location and label each tree within the plot
– Measure the dbh of each tree within the plot
– Measure the height of each tree within the plot
– Write the measured tree dbh’s and heights
– Use the GNSS receiver to collect the coordinates of the plot center
– Import the measured tree dbh’s and heights into the Field Map
database
– Check for the errors and correct it if any
Office works
– Perform the differential correction of the coordinates recorded by the
GNSS receiver
– Georeference the plot measurement
Faro Focus3D
X 130 HDR
TLSForestry
engineer
& 1-2
forestry
techniciansSingle scan
Field works
– Start/Stop the scanning process
Office works
– Post process the TLS raw data
(single scan)Multiple scans
Field works
– Select the locations of the TLS
stations
– Place the spheres needed for co-
registration process
– Start/Stop the scanning process
Office work:
– Post process the TLS raw data
(multiple scans)analysis. Several aspects where considered in
this analysis: the equipment acquisition cost,
the personnel skills and qualifications, the data
collection working time, the type of the data
recorded, the post processing time, the accura –
cy of the data recorded and the labor costs.
Results
Equipment acquisition costs and personnel
skills
Regarding the cost, we can say that the con –
ventional method uses much cheaper equip –
ment than the modern method. Lately, the
price of a Faro Focus3D X130HDR that in –
cludes a dedicated post processing software
(Faro Scene software) is about 48,000 Euros,
which is 2.8 times more than the total price of the Field Map system, with caliper and vertex
inclinometer. With the technological advance,
we are convinced that terrestrial lasers will
become more affordable in terms of price and
thus would be used as much as possible in for –
ested areas.
In terms of personnel skills and qualifica –
tions, the method based on TLS technology
has the advantage of a less numerous field
team, but the processing of TLS raw data re –
quires a highly skilled staff (Table 1).
Data collection, working time
The results on the necessary working time for
recording the field data indicated that the use
of TLS was faster than the classic method (Ta –
ble 2, Table 3). The effective scanning time for
a single scan was 7 minutes and 47 seconds.
Positioning the spheres in the plot for further
Comparison of measurement and processing activities Table 1

194
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85co-registration of the scans during the multiple
scan approach could be a time-consuming ac –
tivity. By contrast, using the traditional meth –
od in the young stands, where the number of
trees is quite high, the working time can reach
11 hours (Sessile oak stand with forest thin –
ning).
Types of data recorded
The Field Map Data Collector directly meas –
ured in the field the tree positions and their
crown projections and further recorded in a lo –
cal coordinate system (figure 4a). The caliper was used to measure the tree dbh and the Ver –
tex inclinometer for tree heights; afterwards
all the measurements were transferred to the
Field Map Data Collector. One important issue
concerning the Field Map use: it provides di –
rectly in the field the correlation between dbh
and trees heights (figure 4b) and thus offer the
possibility to remeasure the trees for which
errors seemed likely. In the case of terrestrial
laser scanning, are recorded point clouds and
images in natural colors (RGB) (figure 5), and
those point clouds need to be post-processed
at office.Working time for Field Map system, caliper and vertex inclinometer Table 2
Stand
compo-
sition,
forestry
works
(plot code)Number
of trees
within
the plotEstimated time for the activity (ho urs) Total
time
(hours)Create the Field
Map project
and measure the
location of each
tree (dbh > 6 cm)
within the plotMeasure
the crown
projection of
each tree (dbh
> 6 cm) within
the plotMeasure the
dbh (>6cm)
of each tree
within the
plotMeasure
the height
of each tree
within the
plotCompleting
and importing
dbh-height
data
Sessile oak,
thinning
(SG1 RM)135 4 – 2 4 1 11
Sessile oak,
selected
cuttings
(SG11TM)35 1 1 0.5 1 0.5 4
Common
beech,
thinning
(SF4RM)64 2 – 1 3 1 7
Common
beech,
selected
cuttings
(SF10TM)31 1 1 0.5 1 0.5 4
Norway
spruce,
thinning
(SM7RM)92 3 – 2 3 1 9
Spruce,
selected
cuttings
(SM15TM)32 1 1 0.5 1 0.5 4

195
Apostol et al. Data collection methods for forest inventory …Working time for Faro Focus3D X 130 HDR TLS Table 3
Stand composition, forestry work (plot code)Single scan
(minutes)Multiple scans
(minutes)
Sessile oak, thinning (SG1RM) 15 60-90
Sessile oak, selected cuttings (SG11TM) 15 60
Common beech, thinning (SF4RM) 15 60-90
Common beech, selected cuttings (SF10TM) 15 60
Norway spruce, thinning (SM7RM) 15 60-90
Norway spruce, selected cuttings (SM15TM) 15 60
Field Map system a) tree crown projections b) correlation between trees dbh – height Figure4
TLS recorded data: a) Single scan point cloud (Faro Scene software) (SS) b) Multiple
scans point cloud (Faro Scene software)Figure 5

196
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85Post-processing time
The conventional method presents the advan –
tage that the resulted data are in GIS already,
measured in local coordinates – simple and
easy to use immediately, with minimum or no
further office processing method. By contrast,
TLS data need to be post-processed in order
to get values for the tree dbh and height. The
post processing time could be up to 10 hours in
the case of young stands, with a higher density
of trees (Sessile oak forest stand – thinning)
(Table 4).
Accuracy of the data recorded
By comparing averages of the dbh for each
plot measured both by conventional and TLS
methods, in the case of single scan approach
we obtained deviations in absolute values,
ranging from 0.4 cm (common beech thinning
stand – SF4RM) to 6.4 cm (common beech se –
lected cuttings stand – SF10TM). When using
a multiple scan approach, the deviations were
higher, ranging between 0 cm (spruce select –
ed cuttings stand – SM15TM) and 13.7 cm
(SF10TM) (Table 5, figure 6).
Within the stands with high dbh coefficients
of variation, the differences between the aver –
ages, both dbh and height, of the two popula -tions are high, suggesting a lower accuracy for
the TLS trees detection, and for their dbh and
height, than in stands with a low variability of
dbh. In the case of a coefficient of variation of
dbh less than 35-40%, the difference between
the average dbh obtained by the two methods
(conventional and modern) is relatively low
(less than 2 cm in both approaches, single and
multiple scans).
The tree height was underestimated when
used both the TLS methods (single and multi –
ple approach), with a lower value in the case of
the multiple approach. The deviation between
the average height of each plot, measured both
by conventional and TLS methods, were high –
er than in the case of dbh, ranging between 4.8
m (sessile oak thinning stand – SG1RM) and
18.5 m (SF4RM), for single scan approach,
respectively between 3.2 m (SG1RM) and
11.4 m (sessile oak selected cuttings stand
– SG11TM) in the case of the multiple scan
approach (Table 5, Figure 7). The underes –
timations of the height by the TLS measure –
ment could be related to the treetops, which
are sometimes not visible due to occlusions of
the other tree crown or because their position,
nearby the scanner station.
Code of the
permanent
plotEstimated time for the activity (hours)
Total
(hours)TLS scans
co-registration
(Faro Scene
software)Extracting the
Digital Terrain
Model (DTM)
(Computree
software)Point cloud
segmentation
(Computree
software)Extract trees charac-
teristics (dbh, height,
positions) (Computree
software)
SG1RM 2 1 4-5 2 10
SG11TM 1.5-2 1 3 2 8
SF4RM 2.5 1 4-5 2 10
SF10TM 1.5-2 1-1.5 3 2 8
SM7RM 2.5-3 1 4 1.5-2 10
SM15TM 2 1 3-3.5 1.5 8 Post processing time for Faro Focus3D X 130 HDR TLS system Table 4
Note. Abbreviations: * The code of permanent plot is according to Table 2.

197
Apostol et al. Data collection methods for forest inventory …Labor costs
Labor cost for one of the measured plot was
slightly higher when we used the convention –
al method (approximatively 1.7 times high –
er than TLS method). Considering all of the
above-mentioned aspects, the SWOT analysis
is presented in Table 6.Discussion
The study intended to compare two acquisi –
tions methods, to obtain reference ground truth
data for forestry and to analyze its suitability
by applying a SWOT analysis.
Modern technology, such as TLS, is reduc –
ing the fieldwork time, but the data provided Accuracy of the recorded data Table 5
Note. Abbreviations: * The code of permanent plot is according to Table 2.Code of
permanent
plot*Field Map system, caliper
and vertex inclinometerFaro Focus3D X 130 HDR
TLS – SSFaro Focus3D X 130 HDR
TLS – MS
(cm)Coeficient
of variation
of the dbh
measured in
the field
(%)
(m)
(cm)
(m)∆dbh
(cm)∆h
(m)
(cm)
(m)∆dbh
(cm)∆h
(m)
SG1RM 12.5 28% 15.7 11.0 11.0 1.5 4.8 11.2 12.5 1.3 3.2
SG11TM 30.0 78% 25.5 26.3 9.5 3.7 16.1 23.4 14.1 6.6 11.4
SF4RM 19.8 37% 26.5 20.2 8.0 -0.4 18.5 18.0 15.5 1.8 10.9
SF10TM 21.0 96% 19.0 27.4 11.4 -6.4 7.6 34.7 13.1 -13.7 5.9
SM7RM 16.1 47% 15.0 18.6 8.6 -2.5 6.4 18.3 9.7 -2.2 5.3
SM15TM 30.6 31% 22.8 29.2 10.6 1.4 12.2 30.6 14.8 0.0 8.0
Boxplot of dbh measured in the field (FM) and the TLS estimated – TLSS (single scan),
TLSM (multiple scan) a) SG1RM, b)SG11TM, c) SF4RM, d) SF10TM, e) SM15TM
f) SM7RMFigure 6

198
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85should be processed in order be usable and
useful. Referring to the importance of using
TLS technology in forestry, Wang et al. 2017
considered the new technology very useful in
forest inventory, but the high price of the de –
vices is still a drawback.
In this study, the working time required in a
multiple scan approach, considering both ac –
tivities – placing the spheres in the field and
scanning – was approximately 60-90 minutes
per plot. Bauwens et al. (2016) indicated, for
a multiple scan approach and only for positing
the spheres within the plot, a required time of
40 minutes. They used in their study the Faro
Focus3D 120 device, which is similar to the
TLS used by us. In the present study, the es –
timated working time necessary in the use of
conventional method ranged between 4 hours
(sessile oak, common beech and Norway
spruce selected cuttings stands) and 11 hours
(sessile oak, common beech and Norway
spruce thinning stands). On the working time
and the TLS practical application, Wezyk et al. (2007) pointed that the TLS methods are suit –
able for small areas, in monitoring and model –
ling the growth of the trees and stands, while
for large sampling areas are not very success –
ful, due to the time consuming for automatic
data processing.
A study of Weiß (2009) indicates that the
multiple scan approach is superior to single
scan one, revealing better results in terms of
dbh estimation and trees recognition; still, ful-
ly automatic tools are further needed for these
approaches to be used for practical forest in –
ventory purposes .
Using the single scan (SS) approach, we ob –
tained, at plot level, a deviation between the
mean dbh of the Field Map measurements and
the mean dbh estimated by automatic TLS
method ranging from -6.4 cm (SG10TM) to
3.7 cm (SG11TM), with an average of -0.5 for
all the six measured plots. The results for dbh
using the multiple scan (MS) approach were
in the interval -13.7 cm (SF10TM) and 6.6 cm
(SG11TM), with an average of -1.0 cm. Within
Boxplot of height measured in the field (FM) and the TLS estimated – TLSS (single
scan), TLSM (multiple scan) a) SG1RM, b) SG11TM, c) SF4RM, d) SF10TM, e) SM –
7RM, f) SM15TMFigure 7

199
Apostol et al. Data collection methods for forest inventory …the stands with a low coefficient of variation of
dbh (less than 35-40%), the difference between
the average dbh obtained with the two methods
(conventional and modern) is relatively low,
being less than 2 cm. Using a phase-shift TLS
device (Faro LS 800 HE80) and SS in 3 plots (2 plots with mixed forest tree species and 1
pure beech forest plot) with a radius of 15 m,
Maas et al. (2008) obtained a bias between -0.7
and 1.6 cm of dbh. In the beech forest plot,
the dbh bias was underestimated with 1.6 cm,
while in our study we obtained only an overes – HELPFUL HARMFUL
Field Map TLS Field Map TLSInternal origin Strengths Weaknesses
Experienced teams
and well-established
procedures
Proven high
performance, water proof
and durable equipment
Relative reduced cost of
equipment
Reduced processing
office work (30 min)
Direct measuring,
suitable for various
usages as reference data
Works well in conditions
of thick shrubs/
understory layer
GIS ready, simple and
small size output data,
ready for use after
measurements
Versatility in multiple
forestry related fields
dealing with geospatial
technologies: forest
inventory, forest
design, forest genetics,
forest ecology, forest
protection, etc.Unrivaled detailed
3D panoramic
reconstruction of the
environment at a certain
moment
Simple to use in the
field with a small team
(1 or 2 men)
Reduced time for field
work (between 10-20
min/scan)
Suitable for monitoring
purposes, change
detectionMuch field workforce
needed, especially for
large areas
Requires high skills for
the field staff
Relatively long
working time on the
field (4-11 hours/plot),
more physical effort
High staff costs
Very high size of the
output data – request
large storage capacities
Demands very high
performance computers
for data processing
Long time processing
office work (8-10 h/
plot)
Requires high
skilled staff for post-
processing
Very expensive
equipment
Sensitive to
environmental
conditions (wind, rain,
cold, dust, shocks etc.)
The precision of TLS
derived estimated data
is lower than the direct
measurements
Difficult to use in
conditions of thick
lower layer, and hard
to record the tree
tops in dense canopy
conditions
The software and
processing algorithms
are not mature yetSWOT analysis Table 6

200
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85
External originOpportunities Threats
Increasing demand for
use in various projects
Further diversification
of expertise as a result
of technological
developments and
growing stakeholders
demands
Collaborations and
exchange of experience
with international
specialists/ institutes
Low exposure for work
automationPossibilities to use
in complex projects,
including virtual reality
Improved software and
processing algorithms
could bring very fast
work automation for
tree detection/extraction
and simplicity in data
usage
The equipment offers
are increasing and costs
are decreasingLack of qualified
personnel for field
work
The software/
hardware components
and alternative
technologies are
evolving/changing
rapidlyLack of qualified
personnel for post-
processing
High resistance
to change and
modernization of the
personnel
Insufficient funds
allocated to
endowments and
training
timation of 0.4 cm. Using a Riegl LMS-Z420i
TLS device and MS method they obtained for
another beech forest plot a dbh bias of 0.9 cm,
better with 0.9 cm than the one obtained in our
study.
Brolly & Kiraly (2009) applied SS in a
mixed forest (sessile oak, hornbeam, beech,
larch and spruce species), in one plot with a ra –
dius of 30 m. The estimation bias for dbh was
from -1.6 to 0.5 cm (i.e. the dbh was measured
by 3 methods). In other study, Liang & Hyyp –
pä (2013) scanned 5 plots of 10 m radius in a
boreal forest with Scots pine, Norway spruce
and birch with densities between 605 and
1,210 stems ha-1; the reference measurements
included trees with at least 5 cm dbh and they
reported biases from -0.2 to 0.8 cm for the SS
approach, while using a specific MS method
the dbh bias was ranging from 0.1 to 0.7 cm.
In their work, for SS approach applied in the
dominant Norway spruce stand (plot 5) the dbh
bias obtained was better with 0.8 cm than the
one obtained in our study (SM15TM plot). In
the case of MS method, while the dbh bias ob –
tained in their study was 0.6 cm, we obtained
no difference between the mean values of the
dbh.
In a previous estimations of the tree height based on the TLS, Liang et al. (2016) indicated
the difficulty of identifying treetops in dense for –
ested plots. With the SS approach, we obtained,
at plot level, a high deviation between the mean
height based on Field Map measurements and
the mean height estimated by automatic TLS
method ranging from 4.8 m (SG1RM) to 18.5m
(SF4RM) with an average of 10.9 m for all the
six measured plots. Likewise, following the
MS approach, the biases were between 3.2 m
(SG1RM) and 11.4 m (SG11TM), with an av –
erage of -1.0 m. The very high deviations ob –
tained with the both approaches indicate that
TLS underestimate the tree height. Still, there
are necessary more investigations to get accu –
rate tree height data using this technology. Reli –
able height measurements based on TLS seems
to be possible in sparse forested plots (Fleck et
al. 2011, Huang et al. 2011). Liang & Hyyppä
(2013) obtained a height bias between -1.3 and
2.15 m with the SS approach, and between -0.34
and 2.11 m using the MS method. Olofsson et
al. (2014) reported the use of the SS method on
16 plots, with a resulted bias of -0.1 m, while in
another study (Huang et al. 2011) the bias value
was of -0.3 m for heights estimates (a single
plot with a density of 212 stems/ha).(continuation) Table 6

201
Apostol et al. Data collection methods for forest inventory …Conclusions
The difference between the average dbh ob –
tained by the two methods (conventional and
modern) is relatively low, being less than 2 cm
in stands with a low coefficient of variation of
dbh.
In the case of tree heights, the comparison
between the two methods revealed a consider –
able underestimation in both the TLS methods
(the single and the multiple approach), with a
less value for the later.
Although it is a technology with a higher
potential, for a large use of TLS in forest in –
ventory further investigations are required, es –
pecially in the case of automatically extraction
of the tree height. To obtain accurate reference
ground data for forest inventory purposes, we
still recommend the use of the conventional
methods, although they are time consuming.
Acknowledgements
This research was made possible through the
funding provided by the Romanian Ministry
of Research and Innovation (MCI) – EO-RO –
FORMON Project (http://www.eo-roformon.
ro).
References
Aschoff T., Spiecker H., 2004. Algorithms for the auto –
matic detection of trees in laser scanner data. The Inter –
national Archives of Photogrammetry, Remote Sensing
and Spatial Information Sciences, Freiburg, Deutsch –
land, V ol. XXXVI, Part 8/ W2, pp. 66- 70.
Asner G.P, Mascaro J., 2014. Mapping tropical forest
carbon: Calibrating plot estimates to a simple LiDAR
metric. Remote Sensing of Environment 140: 614–624.
DOI: 10.1016/j.rse.2013.09.023
Bauwens S., Bartholomeus H., Calders K., Lejeune P.,
2016. Forest Inventory with Terrestrial LiDAR: A
Comparison of Static and Hand-Held Mobile Laser
Scanning. Forests 7(6): 1-17. DOI: 10.3390/f7060127
Bernasconi L., Chirici G., Marchetti, M., 2017. Biomass
estimation of xerophytic forests using visible aerial
imagery: Contrasting single-tree and area-based ap –
proaches. Remote Sensing 9(4): 1-12. DOI: 10.3390/rs9040334
Blair J.B., Rabine D.L., Hofton M.A., 1999. The Laser
Vegetation Imaging Sensor: a medium-altitude, digiti –
sation-only, airborne laser altimeter for mapping veg –
etation and topography. ISPRS Journal of Photogram –
metry and Remote Sensing 54: 115–122. DOI: 10.1016/
S0924-2716(99)00002-7
Boehm H.D., Liesenberg V ., Limin S.H., 2013. Multi-tem –
poral airborne LIDAR-survey and field measurements
of tropical peat swamp forest to monitor changes. IEEE
Journal of Selected Topics in Applied Earth Observa –
tions and Remote Sensing 6: 1524–1530. DOI: 10.1109/
JSTARS.2013.2258895
Brolly G., Kiraly G., 2009. Algorithms for stem mapping
by means of terrestrial laser scanning. Acta Silvatica et
Lignaria Hungarica 5: 119–130.
Brovkina O., Cienciala E., Surový P., Janata P., 2018. Un –
manned Aerial Vehicles (UA V) for assessment of qual –
itative classification of Norway Spruce in temperate
forest stands. Geo-Spatial Information Science 21(1):
12–20. DOI: 10.1080/10095020.2017.1416994
Brovkina O., Novotny J., Cienciala E., Zemek F., Russ R.,
2017. Mapping forest aboveground biomass using air –
borne hyperspectral and LiDAR data in the mountainous
conditions of Central Europe. Ecological Engineering,
100: 219–230. DOI: 10.1016/j.ecoleng.2016.12.004
Cabo C., Ordó-ez C., López-Sánchez C.A., Armesto J.,
2018. Automatic dendrometry: Tree detection, tree
height and diameter estimation using terrestrial la –
ser scanning. International Journal of Applied Earth
Observation and Geoinformation 69: 164-174. DOI:
10.1016/j.jag.2018.01.011
Fleck S., Mölder I., Jacob M., Gebauer T., Jungkunst H.F.,
Leuschner C., 2011. Comparison of conventional eight-
point crown projections with LIDAR-based virtual
crown projections in a temperate old-growth forest. An –
nals of Forest Science 68: 1173–1185. DOI: 10.1007/
s13595-011-0067-1
Grudnicki F., 2004. Coeficientul de zveltețe și stabilitatea
individuală a arborilor de molid [Standerness index and
the individual stability of Norway spruce trees]. Buco –
vina Forestieră 12(1-2): 75-87.
Huang H., Li Z., Gong P., Cheng X., Clinton N., Cao C.,
Ni W., Wang L., 2011. Automated methods for measur –
ing DBH and tree heights with a commercial scanning
lidar. Photogrammetric engineering and remote sensing
77: 219–227. DOI: 10.14358/PERS.77.3.219
Keenan R.J., Reams G.A., Achard F., de Freitas J.V .,
Grainger A., Lindquist E., 2015. Dynamics of global
forest area: results from the FAO global forest resourc –
es assessment 2015. Forest Ecology and Management
352: 9–20. DOI: 10.1016/j.foreco.2015.06.014
Liang X., Hyyppä J., 2013. Automatic stem mapping by
merging several terrestrial laser scans at the feature and
decision levels. Sensors 13: 1614–1634. DOI: 10.3390/
s130201614
Liang X., Kankare V ., Hyyppä J., Wang Y ., Kukko A.,
Haggrén H., Yu X., Kaartinen H., Jaakkola A., Guan F.,

202
Ann. For. Res. 61(2): 189-202, 2018 Research article / INCDS85Holopainen M., Vastaranta, M., 2016. Terrestrial laser
scanning in forest inventories. ISPRS Journal of Pho –
togrammetry and Remote Sensing 115: 63–77. DOI:
10.1016/j.isprsjprs.2016.01.006
Lovell J.L., Jupp D.L.B., Newnham G.J., Culvenor D.S.,
2011. Measuring tree stem diameters using intensi –
ty profiles from ground-based scanning lidar from a
fixed viewpoint. ISPRS Journal of Photogrammetry
and Remote Sensing 66: 46–55. DOI: 10.1016/j.isprs –
jprs.2010.08.006
MacDicken K.G., 2015. Global forest resources as –
sessment 2015: What, why and how? Forest Ecology
and Management 352: 3–8. DOI: 10.1016/j.fore –
co.2015.02.006
Maas H.G., Bienert A., Scheller S., Keane E., 2008.
Automatic forest inventory parameter determina –
tion from terrestrial laser scanner data. International
Journal of Remote Sensing 29:5: 1579-1593. DOI:
10.1080/01431160701736406
Newnham G., Armstrong J., Muir J., Goodwin N., Culve –
nor D., Pushel P., Nystrom M., Johansen K., Tindall D.
2011. Evaluation of terrestrial laser scanners for mea –
suring vegetation structure. CSIRO Australia. Sustain –
able Agriculture Flagship, Manuscript ID: EP 124571.
Niska H., Skon J., Packalen P., Tokola T., Maltamo M.,
Kolehmainen M., 2010. Neural networks for the predic –
tion of species-specific plot volumes using airborne la –
ser scanning and aerial photographs. IEEE Transactions
on Geoscience and Remote Sensing 48: 1076–1085.
DOI: 10.1109/TGRS.2009.2029864
Olofsson K., Holmgren J., Olsson H., 2014. Tree stem
and height measurements using terrestrial laser scan –
ning and the RANSAC algorithm. Remote Sensing 6:
4323–4344. DOI: 10.3390/rs6054323
Othmani A., Piboule A., Krebs M., Stolz C., Lew Yan V oon
L. 2011. Towards automated and operational forest in –
ventories with t-lidar. 11th International Conference on
LiDAR Applications for Assessing Forest Ecosystems
(SilviLaser 2011), October 2011, Hobart, Australia.
Pueschel P., Newnham G., Rock G., Udelhoven T., Werner
W., Hill J., 2013. The influence of scan mode and circle
fitting on tree stem detection, stem diameter and volume
extraction from terrestrial laser scans. ISPRS Journal of
Photogrammetry and Remote Sensing 77: 44–56. DOI:
10.1016/j.isprsjprs.2012.12.001
Simonse M., Aschoff T., Spiecker H., Thies M., 2003. au -tomatic determination of forest inventory parameters
using terrestrial laser scanning. In: Proceedings of the
ScandLaser Scientific Workshop on Airborne Laser
Scanning of Forests, pp. 251–257.
Stanley T., 2013. Assessment of the FARO 3D focus la –
ser scanner for forest inventory. BSc thesis, Faculty of
Health, Engineering and Sciences, University of South –
ern Queensland, Toowooba, QLD, Australia, 86 p.
Tockner A., Leisch F., Nothdurft A., 2017. Automatic
mapping of forest stands based on three-dimensional
point clouds derived from terrestrial laser-scanning.
Forests 8(8): 1-19.
Trumbore S., Brando P., Hartmann H., 2015. Forest health
and global change. Science 349 (6250): 814–818. DOI:
10.1126/science.aac6759
Tubiello F.N., Salvatore M., Ferrara A.F., House J., Fed –
erici S., Rossi S., Prosperi P., 2015. The contribution
of agriculture, forestry and other land use activities to
global warming, 1990–2012. Global Change Biology
21(7): 2655–2660. DOI: 10.1111/gcb.12865
V opěnka P, Černý M (2006). GIS aided Statistical Forest
Inventory in Transcarpathia, Ukraine. ArcNews 27(4).
Wang P., Bu G., Li R., Zhao R., 2017. Automated low-
cost terrestrial laser scanner for measuring diame –
ters at breast height and heights of forest trees. arX –
iv:1702.02235.
Watt P.J., Donoghue D.N.M., 2005. Measuring forest
structure with terrestrial laser scanning. International
Journal of Remote Sensing, 26(7): 1437-1446. DOI:
10.1080/01431160512331337961
Weiß J., 2009. Application and statistical analysis of ter –
restrial laser scanning and forest growth simulations to
determine selected characteristics of douglas-fir stands.
Folia Forestalia Polonica 51(2): 123–137.
Wezyk P., Koziol K., Glista M., Pierzchalski M., 2007.
Terrestrial laser scanning versus traditional forest in –
ventory: First results from the Polish forests. In: Proc.
Laser scanning 2007 and Silvilaser 2007. ISPRS Com –
mission III Workshop, V ol. 36, Part 3, Espoo, Finland,
Sept. 12–14, 2007, pp. 424-429. Web: http://www.
isprs.org/proceedings/XXXVI/3-W52/final_papers/
Wezyk_2007.pdf. Accessed: 10.2018.
Zemánek T., Cibulka M., Pelikán P., Skoupil J., 2017.
The use of terrestrial laser scanning for determining
the driver’s field of vision. Sensors 17(9): 2098. DOI:
10.3390/s17092098