The model for root system architecture (RSA) has started in the early 1970s 1. It contains two [612142]

Introducere
The model for root system architecture (RSA) has started in the early 1970s [1]. It contains two
important spatial configurations (i.e., geometry and topology) of the plant illustrated in three
dimension visualization. This is illustrated in the works of Gregory [2] and Danjon and Reubens [3]
who then investigated the architecture including the following determinants of every root system in
the soil, such as shape, size, ori entation, spatial location (geometry) and physical branching patterns
(topology). {(1)Dupuy, L., P.J. Gregory and A.G. Bengough, Root growth models: Towards a new
generation of continuous approaches. J. Exp. Bot., 61(8), 21312143 (2010). (2) Gregory, P.J., Plant
Roots: Growth, Activity and Interaction with Soils. Blackwell, Oxford, UK (2006); (3) Danjon, F. and B.
Reubens, Assessing and analyzing 3D architecture of woody root systems, a review of methods and
applications in tree and soil stability, resource acquisition and allocation. Plant Soil, 303(1 -2), 1 -34
(2008).
Danjon and Reubens [3] identified two common types of structural anchorage namely coarse roots
and fine roots. The first one (larger than 2 mm in diameter, e.g., structural roots or tap roots either at
thinner or deeper layer) is able to anchor in strong strata, and the latter one (smaller than 2 mm in
diameter, e.g., absorbing roots) has a low bending stiffness for the absorption of moisture and
nutrients without any secondary growth [14,15] {(14) Reubens, B., J. Poesen, F. Danjon, G.
Geudens and B. Muys, The role of fine and coarse roots in shallow slope stability and soil erosion
control with a focus on root system architecture: A review. Trees -Struct. Funct., 21(4), 385 -402 (2007).
(15) Day, S.D., P.E. Wiseman, S.B. Dickinson and J.R. Harris, Contemporary concepts of root system
architecture of urban trees. Arbori. Urban For., 36(4), 149 -159 (2010)}
Sometimes, these coarse roots may become thicker buttress roots near the tru nk whereas some may
expand and plunge deeper as sinker roots [16] {Ghani, M.A., A. Stokes and T. Fourcaud, The effect of
root architecture and root loss through trenching on the anchorage of tropical urban trees (Eugenia
grandis Wight). Trees -Struct. Fun ct., 23(2), 197 -209 (2009).} . Jim [17] {Jim, C.Y., Old stone walls as an
ecological habitat for urban trees in Hong Kong. Landsc. Urban Plann., 42(1), 29 -43 (1998). }
described those having both roots can reach the standard dimensions or larger, but only fine roots
may restrict the potential sizes.
Regarding the resolution of individual coarse roots using GPR, it was influenced by root size, depth
and spacing. Other investigators [11,18,20,22] { (11) Hruska, J., J. Cermak and S. Sustek, Mapping tree
root systems with ground penetrating radar. Tree Physiol., 19(2), 125 -130 (1999).}; {(18) Cermak, J., J.
Hruska, M. Martinkova and A. Prax, Urban tree root systems and their survival near houses analyzed
using ground penetratin g radar and sap flow techniques. Plant Soil, 219(1 -2), 103 -116 (2000)}; {(20)
Stokes, A., T. Fourcaud, J. Hruska, J. Cermak, N. Nadyezdhina, V. Nadyezhdina and L. Praus, An
evaluation of different methods to investigate root system architecture of urban tr ees in situ:
Groundpenetrating radar. J. Arbori., 28(1), 2 -10 (2002)}; {(22) Cox, K.D., H. Scherm and N. Serman,
Groundpenetrating radar to detect and quantify residual root fragments following peach orchard
clearing. HortTechnology , 15(3), 600 -607 (2005) } reported that coarse roots around 3 and 5 cm in
diameter could be detected, with an error of 1 to 2 cm between radar reflectors and excavated data.
Those roots were between 11 and 114 cm at burial depths. Most of roots were situated in the
uppermost 15 to 40 cm of soil [23] { Niltawach, N., C.C. Chen, J.T. Johnson and B.A. Baertlein, A
numerical study of buried biomass effects on ground -penetrating radar performance. IEEE T. Geosci.
Remote, 42(6), 1233 -1240 (2004).}. The above results provided a basis fo r individual root or biomass
assessments with acceptable accuracy.

Instrumente
The major applied non -destructive techniques in root system investigations are pulling tests, ground
penetration radar(Guo et al. 2013), and electric resistivity tomography(Am ato et al. 2008; Zenone et
al. 2008). A comparison of ground penetration radar and pulling tests (Danjon –Reubens 2007) has
been reported

Metodologie
1) Pulling -2016
The pulling or winching test is a method for investigating the root -soil system and root a nchorage.
Assessing the safety of trees and evaluating stem conditions are other applications of pulling, as well
as serving as calibration for measurements in real wind situations. Duringapulling test (Figure 1)a
cable is generallyplacedon the treeas clos eas possibleto the crown center or half of the trunk. The
other end of the cable should be anchored to another tree or other object, and the applied force
should be measured. During the test, a pulling force is applied through the cable to bend the tree an d
the inclination, elongation, and dislocation of the ground are monitored and measured. (Elongation is
usually measured along the trunk at higher levels as seen in Figure 1.As the stem bends, it elongates
as well. Conditions of the trunk can be evaluated. Dislocation of the ground is caused by the roots
asthey move and/or break during the pulling. Measurements can be performed on the ground near
the winched tree.)

The pulling test usually continues untilthe inclination of the trunk at ground level reache s 0.020 -0.6
degrees. The collected force -inclination data should be filtered according to the relaxation of the
tree. (Winching stresses the tree, causing inclination; although relaxation of the wood material of the
tree does occur. For the evaluation of s afety, only the inclinations immediately responsive to the
pulling are used. Inclination data that are collected during the relaxation effects should be
neglected.) (Divóset al. 2009; Sani et al. 2012; James et al. 2013).The difference between trees can be

stated clearly, even for the same species at the same field; the measured pulling moment ranges
between 0.07 and 0.6 degrees. The forceusedand the caused moment were higher for trees whose
tilt was 0.07, which refers to more enormous anchorage strength th an treesthattilted 0.6 degrees for
weaker winching. (James et al. 2013) Thepullingof a treeshould be less than 0.25 -0.6degrees to avoid
injuring the roots.(This range is quite great according to the difference of trees.)Coutts (1983)
foundthatthe firstsoun d indicatingroot breaking startedata0.5 tilt. Rahardjo and colleagues
(2014)established a model of tree felling based on their field experiences. They also verified roots
breaking during uprooting and stated root cross -sectional area and root plate radius to be relevant
for root anchorage evaluations. Modeling focusing on root failure (shape and stresses) was
presented byLundstorm and colleagues (2007B).

2) Acoustic detection – 2016
Acoustic detection of roots is a technique inspired by other acoustic techniques (time -of-
flight measurement, tomography; Grabianowski et al. 2006; Wang 2013) used for tree
evaluation. The method has been developed at the University of West Hungary and the new
method has beentested in theuniversity’sbotanical garden (Divós20 08). The physical
background of the measurement is the velocity -difference in wood and soil. The
acousticsignal’s velocity in soil is about 250 -400 m/s depending on soil type and moisture
content,while the velocity in the roots is between 2000 and 4000 m/s (Bucur 1995;Divóset al.
2009). Based on these differences, modelshave beenconstructedto estimate the acoustic
signal’s path in both pure soil and the soil containing roots. After successful experiments, a
set-up for field usage has been designed. The devic e consists of a transmitter, a receiver, and
a timemeasuring component. The transmitter is needle -like and should be placed onto the
trunk at ground level, while the receiver is a long metal spike (30 cm or longer), which has a
good couplingfor the soil(Fi gure 3)(Divós et al. 2009; Buza –Göncz 2015). At realization of
acoustic detection, a very short signal (an ‘acoustic flash’) is sent from the transmitter to the
tree and to the roots. The time measure turns on. When the receiver detects the arrival of
the acoustic signal,thetime measure stops.Whena root is nearby (closer than 10 cm),the
travel time decreases significantly. The depth of the root can be estimated as well. This
technique can find roots from 4 cm diameter and can separate two roots from each ot her if
there is at least 20 cm distance between them (Divóset al. 2009). The maximum depth for
detection is 50 cm.
The main limitation of the pulling test is its static nature of loading. Wind usually comes in
gusts andisdynamicin nature.The pulling test a lso requiresheavyequipment. For an overall
evaluation of a tree, a pulling test should be performed in two perpendicular directions,
which may not be possible due to the surroundings found in some environments (Brudi –
Wassenaer 2002,Saniet al 2012).

The measure is usually performed alonga circle around the trunk, with a step of 10 -15 cm.
The measurement is repeated alongwider and wider circles. The direction of the roots can be
marked by sand or typed to a program. An example of root detection is seen in Figure 4 and
5. Acomparison of predictions of the theoretical model and the measured data were
performed; contrastingofevaluated root system and real root situationwas completed as
well. Theroots were excavated and the 3D model of the evaluated root sys tem fit well to the
roots found by uncovering.

Limitations of the presented technique are clear; roots with diameters smaller than 4 cm are
not seen by acoustic detection, and the maximum depth of the measure is about 0.5 m.
Nevertheless, compared to oth er available techniques like ground penetration radar (GPR) or
electric resistivity tomography (ERT), acoustic root detection has a remarkable advantage;
the rootsofthe measured tree aredetected. Signals of GPR and ERT both change in the
presence of a root . However, GPR and ERT signals also change even if tubes, rocks, or other
materials are buried (Hruska – Cermak 1999; Cermak et al. 2000; Stokes et al. 2002; Hirano et
al. 2

3) Ground penetrating radar (GPR) – 2014 – Georadar
In recent years, GPR gained po pularity in forestry, thanks to its application in water content
estimation, root stress evaluation, root biomass modelling, and roots location.
Since 1999, GPR has provided an important method for non -invasive study of plant root
system, including coarse root mapping [4 –7] {4. Bassuk, N.; Grabosk, J.; Mucciardi, A.; Raffel,
G. Ground penetrating radar accurately locates tree roots in two soil media under pavement.
Arboric. Urban For. 2011, 37, 160 –166. 5. ˇCermák, J.; Ka, J.H.; Martinková, M.; Prax, A.
Urban tree root systems and their survival near houses analyzed using ground penetrating
radar and sap flow techniques. PlantSoil 2000, 219, 103 –116. [CrossRef] 6. Hruska, J.;
ˇCermák, J.; Šustek, S. Mapping tree root systems with ground -penetrating radar. Tre e
Physiol. 1999, 19, 125 –130. [CrossRef] [PubMed] 7. Isaac, M.E.; Anglaaere, L.C.N. An in situ
approach to detect tree root ecology: Linking ground -penetrating radarimagingtoisotope –
derivedwateracquisitionzones.Ecol. Evol.2013,3,1330 –1339. [CrossRef][PubMe d]} , root
system architecture reconstruction [8,9]
{8.Stokes,A.;Fourcaud,T.;Hruska,J.;Cermak,J.;Nadyezdhina,N.;Nadyezhdin,V.;Praus,L.Anevalu
ationof different methods to investigate root system architecture of urban trees in situ. I.
Ground -penetrating radar. J. Arboric. 2002, 28, 2 –10. 9. Wu, Y.; Guo, L.; Cui, X.; Chen, J.; Cao,
X.; Lin, H. Ground -penetrating radar -based automatic reconstruction of three -dimensional
coarse root system architecture. Plant Soil 2014, 383, 155 –172. [CrossRef]} , and root
diameter or biomass estimation [10 –15] {10. Butnor, J.R.; Doolittle, J.A.; Kress, L.; Cohen, S.;
Johnsen, K.H. Use of ground -penetrating radar to study tree roots in the southeastern United States.
Tree Physiol. 2001, 21, 1269 –1278. [CrossRef] [PubM ed] 11. Cui, X.; Guo, L.; Chen, J.; Chen, X.; Zhu, X.
Estimating tree -root biomass in different depths using ground -penetrating radar: Evidence from a
controlled experiment. IEEE Trans. Geosci. Remote Sens. 2013, 51, 3410 –3423. [CrossRef] 12. Cui, X.H.;
Chen, J.; Shen, J.S.; Cao, X. Modeling tree root diameter and biomass by ground -penetrating radar. Sci.
China Earth Sci. 2011, 5, 711 –719. [CrossRef] 13. Guo, L.; Lin, H.; Fan, B.; Cui, X.; Chen, J. Impact of root
water content on root biomass estimation usi ng ground penetrating radar: Evidence from forward
simulations and field controlled experiments. Plant Soil 2013, 371, 503 –520. [CrossRef] 14. Hirano, Y.;
Dannoura, M.; Aono, K.; Igarashi, T.; Ishii, M.; Yamase, K.; Makita, N.; Kanazawa, Y. Limiting factors in
the detection of tree roots using ground -penetrating radar. Plant Soil 2009, 319, 15 –24. [CrossRef] 15.
Stover, D.B.; Day, L.F.P.; Butnor, J.R.; Ke, B.G. Effect of elevated CO2 on coarse -root biomass in Florida
scrub detected by ground -penetrating rada r. Ecology 2007, 88, 1328 –1334. [CrossRef] [PubMed]} .
Unlike optical images, raw GPR radargrams provide insufficient geometrical information of
buried targets and are always disturbed by soil clutter noise. Hence, locating and identifying
root objects in GP R radargrams is a prerequisite step in GPR data processing.

Tree roots generally show hyperbolic patterns on GPR images, which are similar to other
linear objects such as pipes and cables. Methods for automatic recognition of these linear
objects in GPR images can be classified mainly into three types: machine learning based
methods, clustering based methods, and Hough transform (HT) based methods [16] 16.
Chen, H.; Cohn, A. Probabilistic robust hyperbola mixture model for interpreting ground
penetrating r adar data. In Proceedings of the International Joint Conference on Neural
Networks (IJCNN), Shanghai, China, 6 –9 June 2010; pp. 1 –8.
Although a single root and a single pipe share similar hyperbolic pattern on GPR image, root
systems possess more complexch aracteristicsthanundergroundpipesorcables:
(1)thesizeanddepthofrootsareuncertain; (2) the directions and angles of root stretching are
variable; (3) the distribution range of root systems is not certain; and (4) the soil
environment where roots grow is mor e complex. These factors render the automatic
identification of root systems more difficult than pipes in GPR images.
Radar (GPR) fulfils rather well these requirements, being capable to produce a detailed map
of the subsurface using radio waves generated j ust above the ground surface (Jol, 2009;
Daniels, 2004; Annan 2004). The interaction between the electromagnetic waves and the
buried targets, produces an electromagnetic image similar to ultrasound images used in
diagnostic medicine. The main advantage o f GPR is its capability to detect, in a non –
destructive way, the vertical and horizontal dielectric anomalies associated with natural or
manmade subsurface variability like: lithology, water content and bulk density changes,
voids or buried objects, liqui d or solid waste disposal, etc.
GPR is a geophysical method which uses electromagnetic (em) waves, typically in the
frequency range 10 -3000 MHz, to image structures and features buried in the ground. The
physical principle of GPR detection, is based on the dielectric contrast between the buried
target and the background material. Such a contrast can be produced in several ways: by
spatial changes in the physical -chemical property of the sediments or the soil, changes in
water content and bulk density of the material, or even by the presence of different objects
in the subsoil like voids, rocks and boulders, wood, or manmade materials like metal and
plastic targets. In general, if there is a detectable contrast between different subsurface
objects, this can g enerate strong signal reflections which can be clearly identified on a GPR
image (Annan, 2004; and Jol, 2009). The equipment used in all GPR systems consists of four
main elements: a transmitting unit; a receiving unit; a control unit; and a display unit. The
transmitter produces a short duration, high voltage pulse. This pulse is applied to the
transmitting antenna (Tx), which radiates it into the ground. The receiving antenna (Rx)
collects the signals coming from the material under investigation, which a re amplified and
formatted for display, by the control unit, as depicted in Fig. 1. The radar measures the signal
amplitude vs. time (two -way travel time), for each position of the Tx -Rx on the ground. The
data are collected moving the Tx -Rx system along a profile, so that a bi -dimensional radar
cross -section having the two -way travel time on the Y axis and the antennas position on the
X axis is obtained for each collected profile. To estimate the depth of the target (i.e. to
convert time in depth) a simple calibration technique can be used, as illustrated in Annan,
2004, even though in complicated subsurface scenarios more refine algorithms should be
applied, (Annan, 2004; and Jol, 2009). The velocity and the attenuation of the radar signals
depend on the
electromagnetic properties of the soil, which can be frequency -dependent quantities. In
particular, in common geo -materials the maximum investigation depth decreases rapidly
with increasing frequency due to the signal attenuation; this explain why almost a ll

subsurface radar systems operate at frequencies lower than 3 GHz. The antenna frequency
also affects the vertical and the horizontal resolution achievable in a GPR image. The shorter
is the time pulse width (i.e. the higher is the antenna frequency) , t he higher is the
resolution.

Figure 1. A schematic reconstruction of a GPR system for root investigations. The four main
elements are the transmitting unit, the receiving unit, the control unit, and the display unit. A
detectable contrast between diffe rent subsurface objects generates strong signal reflections
which can be clearly identified on a GPR image.

In order to obtain a XY image of the subsurface, the radar data should be collected in multi –
profile mode, where the profiles are acquired para llel to each other, at a fixed distance. This
technique allows to create XY time (or depth) slices, in which the lateral geometry of the
targets can be identified, i.e. Jol 2009. A further processing of the data collected on a XY
grid, allows a pseudo -3D visualization of the subsurface, throughout the countering of the
anomalies generated by the electromagnetic contrast between the target and the
background material. The final results is a three -dimensional representation of the
subsurface using the isosur faces to display a surface of constant data value in three
dimensions (Pettinelli et al., 2011).
Basedonbothcontrolledandinsituexperimentsforrootdetectionusing GPR, this study presents,
to our knowledge, the first attempt on comprehensive analysis of facto rs for root object
automatic recognition based on GPR. A two -level framework of the limiting factors has been
applied, as shown in Figure 8, namely root growth level (root properties, root distribution
and soil) and signal level (signal strength, signal in completeness, signal interference and
clutter noise). The root growth factors influence the GPR signal, which in turn affects the
recognition of root by the RHT algorithm.

Conclusions
Non -destructive tests can provide quantifiable information about the i nner conditions of
trees. As the trees are not injured during the tests, repeated measurements and long term
monitoring are possible as well. Safety is an important topic as tree failure can cause
property damage or harm to humans
One of the advantages of acoustic root detection is its ability to find the root of a selected
tree, but it can also be utilized as a tool for safety evaluations. Predicting the conditions of
root systems is an area where complex knowledge is needed to get useful results.
The non -invasive nature of this geophysical method makes it appealing in all those
applications where the common techniques used require the destruction of the samples.
GPR is capable to create an electromagnetic image of the targets buried in the soil, allowing a
detailed 3D reconstruction of their position and form. This is particularly important in roots
system architecture study, where the alternative is the excavation or coring of the roots.
Moreover, a part from very conductive soils, where the attenuation dr astically reduces the
maximum penetration depth of the radar pulses, this method can be successfully applied to
every type of material from snow and ice to asphalt or dry sand. It is important to notice,
however, that this technique is not able per se to d efine the nature of the object, e.g. metal,
wood or rock, but can still be used as a reconnaissance method before performing any
destructive test. Another important advantage of this type of technique is its fast real -time
acquisition, which allow to cre ate a very large georeferenced (using a D -GPS) data volume.
This aspect is particularly important for long term monitoring, where the evolution and
development of the root system should be followed for months or years.
Itwasfoundthattherootdistributionands oilheterogeneityfactorsintherootgrowthlevelarecruci
al in the recognition of roots by RHT, because these factors lead to incomplete and
interfering reflections. Suitable conditions are also proposed for root recognition by RHT
algorithm. Further improvements are still required to address these specific limiting factors
for root recognition