Tree Physiology 19, 125--130
© 1999 Heron Publishing----Victoria, Canada
Mapping tree root systems with ground-penetrating radar
JIRI HRUSKA,1 JAN CERMÁK2,3 and SVATOPLUK SUSTEK2
1
Geofyzka a.s., Division of General Geophysics, Jecna 29a, P.O. Box 62, 61246 Brno, Czech Republic
2
Institute of Forest Ecology, Mendel University of Agriculture and Forestry, Zemedelska 3, 61300 Brno, Czech Republic
3
Author to whom correspndence should be addressed
Received June 6, 1997
Summary A ground-penetrating radar (GPR) technique was
used to study the three-dimensional distribution of root systems
of large (DBH = 14 to 35 cm) oak trees (Quercus petraea
(Mattusch.) Liebl.) in relatively dry, luvisoil on loamy deluvium and weathered granodiorite. We used a pulse EKKO 1000
GPR system, a profile grid of 0.25 × 0.25 meters, at 0.05 m
intervals, and a signal frequency of 450 MHz, to assure resolution of about 3 cm in both directions (further increases in
resolution up to 1 cm are possible with the system). Coarse root
density was 6.5 m m −2 of stand area and 3.3 m m −3 of soil
volume. Maximum rooting depth of the experimental oaks was
2 m, and the root ground plan was significantly larger (about
1.5 times) than the crown ground plan. Based on earlier studies
of Quercus robur L. from floodplain forests, where the extent
of the root systems was much smaller (root ground plan:crown
ground plan ratio of 0.6), we conclude that the high root ground
plan:crown ground plan ratio indicates less favorable conditions of water supply at the experimental site than in the
floodplain forest. The ground-penetrating radar system is noninvasive and allows relatively rapid and repeated measurements
of the distribution of coarse root systems of trees.
Keywords: forest stands, ground-penetrating radar, large
trees, Quercus, root skeleton, three-dimensional distribution.
Introduction
Various methods have been used to study root systems of
plants (Zygurovskaya 1958, Reijmerink 1964, Schuurman and
Goedwaagen 1965, Melhuish 1968, Bohm 1979, Cappy and
Brown 1980, Vogt and Persson 1991). Studies of large trees are
more complicated because their root systems are large and
complex and include thin absorbing roots as well as coarse
roots (Kolesnikov 1962, Jenik and Sen 1964, Sutton and Tinus
1983). Coarse tree roots are usually analyzed by archeological
methods, i.e., by excavating root systems from soils (Jenik
1957, 1967, 1978, Kutschera 1960, Kolesnikov 1972, Vyskot
1976, Santantonio et al. 1977, Kasyanenko 1980, Carlson et al.
1988). Fine roots have been studied by sequential soil core
sampling techniques, and by using root-free ingrowth cores
(McQueen 1968, Persson 1983, Vogt et al. 1987, Persson et al.
1990). The disadvantage of all excavation methods is that they
are destructive, and core sampling precludes study of entire
tree root systems. Several kinds of nondestructive methods
(radioactive tracers, combining soil water content and sap
flow) have been used to estimate the extent and depth of root
systems of forest trees (Woods 1969, Cermák et al. 1980,
Cermák and Kucera 1990). The limitation of these methods is
that they provide little or no resolution of root structure.
Ground-penetrating radar has proved useful in several archeological studies (Cammarano and Piro 1997) including some
studies that focused on tree roots (Papamarinopoulos et al.
1997). We undertook a study to determine whether the technique is sensitive enough to be used for mapping tree root
systems in forest stands.
Materials and methods
Experimental site
The experimental stand is located on a small plain near the
middle of a wide ridge dividing two small watersheds near
Sobesice, north of the city of Brno, at an altitude of 360 m. The
luvisoil is on loamy deluvium and weathered granodiorite. The
shallow surface horizon (5 to 7 cm), which is rich in humus,
overlies yellow-brown loamy soil (down to 30 cm) and loamy
to clayish red-brown soil (down to 75 cm). Weathered granodiorite occurs below 115 cm. The soil, which was well supplied with nutrients, had a pH (H2O) at the surface of 7.1, and
a pH of 5.0 in the deeper layers. The clay fraction in the surface
layers decreased with depth from 19 to 9%. There is no underground water available. The forest site is classified as Carpineto--Quercetum with typical herbaceous species Poa
nemoralis L., Poa pratensis L. ssp. angustifolia, Brachypodium sylvaticum (Huds.) p. Beauv. and Festuca heterophylla L.
with admixture of thermophillous and oligotrophic species.
Long-term mean annual temperature is 7.8 °C and annual
precipitation is 583 mm (360 mm falling during the growing
season) (Vasicek 1984).
Forest stand and sample trees
The stand was composed mostly of 50-year-old oak (Quercus
petraea (Mattusch.) Liebl.) trees with a few trees about 100
years old. Maximum stand height was 18 m with a basal area
of 37.9 m ha −1. An experimental square plot (6 × 6 m) was
chosen comprising two sample trees, approximately repre-
126
HRUSKA, CERMÁK AND SUSTEK
senting the range of tree sizes in the stand. The main biometric
parameters of the two sample trees are given in Table 1. In
addition to the sample trees, there was one small fresh stump
and two old mostly decomposed stumps within the experimental square.
Crown ground plan (crown area projected on the ground
surface) was measured geodetically. Crown volume, which
was taken to be ellipsoid, was calculated from crown length
and radius.
Ground-penetrating radar measurement
Ground-penetrating radar (georadar, GPR) measurements are
made on a profile line or system of lines. A transmitter and a
receiver of the signal are situated on the line of the profile to
be investigated (Figure 1). Their distance, configuration and
step of measurement depend on the nature, size and depth of
the objects to be investigated. Electromagnetic waves are used
as a signal. The choice of frequency is influenced by the nature
of material being investigated. The transmitted signal, received
back after being reflected from the underground bodies on a
profile (the cross section of underground space) is digitized
and displayed on a computer screen. All of the georadar cross
lines are gradually displayed. The file of acquired data is
processed with standard geophysical signal processing software. The processing system makes it possible to highlight
certain parts of the cross section under study, while suppressing other parts. The resulting profiles give a picture characterizing the exact position of underground bodies with respect
to the depth in the cross section and their relation relative to
other objects lying above and below. The profiles also trace the
position of strata and their boundaries. Examples of boundaries include contacting surfaces of different materials, strata
surfaces, and various local bodies, such as cables, pipes, stones
and roots.
Depth of penetration by GPR is influenced by the choice of
the frequency, because waves with lower frequency penetrate
deeper. In principle, depth of penetration depends on the geological composition of the section, namely on the attenuation
coefficient of electromagnetic waves in the rock strata. Penetration depth can reach tens of meters under suitable conditions
but it may be less than several meters under unfavorable
conditions. Resolution increases with the increase in signal
frequency. For low frequencies (wavelength in meters), resolution is in the order of meters, whereas for the highest
frequencies (wavelength in millimeters) resolution is in millimeters. The accuracy of detection of body position depends on
the measurement step, and on the depth and density of signal
sampling.
In the present study, ground-penetrating radar measurements were made in two directions within a 6 by 6 m square,
with a 0.25 × 0.25 m profile grid, at 0.05 m intervals, and a
signal frequency of 450 MHz. A pulse EKKO 1000 GPR
system was used (Sensors and Software Inc., Mississauga,
Ontario, Canada). Altogether 300 m of lines were scanned on
50 profiles. Software packages Ekko Tools 42 (Sensors and
Software Inc., Mississauga, Ontario, Canada) and Reflex 3
(K.J. Sandmeier, Karlsruhe, Germany) were used for signal
processing and evaluation. Both packages are standard geophysical (i.e., georadar and seismic) signal processing packages. A procedure called migration was performed on the
signal files after signal amplification and noise reduction, by
which scattered signal energy was traced back to its sources
(i.e., the roots). By this means, anomalous signal sources were
removed, making individual roots visible. This procedure
served as a basis for further processing, which consisted in
making correlations between lines. In the next step, the root
system of the large oak tree was analyzed in detail by applying
depth correlations of georadar indications from single profiles
to develop a three-dimensional picture of the root system. Root
length was estimated with the DIPS image analysis system
(Brno, Czech Republic). A complete field investigation of the
Figure 1. Schematic illustration of ground-penetrating radar detection
of local underground anomalies as specific objects or interfaces. Recorded position of the antenna and travel time of pulses shown in lower
part of figure are main input data for the GPR system.
Table 1. Main biometric parameters of oak trees at the experimental plot. (Abbreviation: DBH = diameter at breast height.)
Sample tree
DBH
(cm)
Basal area
(cm2)
Tree height
(m)
Crown radius
(m)
Crown length
(m)
Crown projected
area (m2)
Crown volume
(m3)
Timber volume
(m3)
Large
Small
35.2
13.9
984
152
17.6
15.2
2.7
1.2
7.8
6.2
22.5
4.4
117.3
18.4
0.939
0.109
TREE PHYSIOLOGY VOLUME 19, 1999
MAPPING ROOT SYSTEMS WITH PENETRATING RADAR
area took about 6 h, and data evaluation took about 30 h (that
is, complete mapping of the experimental plot was done within
a week). Additionally, the root system was excavated and
photographed and root lengths and diameters were measured
to verify the radar data.
Results and discussion
Measurement with GPR
Samples of georadar profiles after initial signal processing are
shown in Figure 2. Numerous manifestations of roots of several individual trees were observed in the primary records. The
presence of several different root systems complicated the
evaluation of the data; however, this difficulty allowed us to
assess the powers of detection and discrimination of the GPR
system and its associated software. Other belowground objects
(e.g., stones) did not seriously hamper evaluation of the roots.
Under the conditions used, the resolution of the GPR system
was sufficient to separate roots that were 3 to 4 cm in diameter.
Diameters of roots detected by the GPR system corresponded
to measured diameters of excavated roots with an error of
between 1 and 2 cm. The GPR system determined the length
of individual roots, from stem to the smallest detectable width,
with an error of about 0.2 to 0.3 dm. Applying higher frequencies together with smaller measurement intervals improved the
resolution and accuracy to less than 1 cm (data not shown).
127
Root system of oak trees
The ground plan picture of oak root systems showed a relatively sparse network of coarse roots (Figure 3). Root planar
density and root spatial density expressed as root length per
unit of area and per unit of rooted volume of the experimental
plot, respectively, are presented in Table 2.
In addition to the root system of the main sample tree, roots
of the second minor tree and the remains of two root systems
belonging to the two rotten stumps were distinguished in the
sample plot. In addition, root systems of some nearby trees
growing outside the experimental plot penetrated the area. The
main biometric data for root system size compared to crown
size for the two experimental trees (%cr ) are given in the
Table 3.
In both trees, the root system ground plan radius (calculated
as the mean of the radii of the longest coarse roots in eight
directions including length outside the ground plan shown in
Figure 3. Ground plan view of all tree root systems at the experimental
plot. Length of roots per stand area is visible, where numbers characterize measured profiles (PF). Old decomposed stumps are marked
white.
Table 2. Planar density and spatial density of oak tree root systems at
the experimental plot.
Figure 2. Ground penetrating radar profiles (1 to 3) at the experimental
plot with numerous tree root indications. Recorded time is converted
to depth.
Root planar density (m m −2)
Root spatial density (m m −3)
Living
Dead
Total
Living
Dead
Total
5.4
1.1
6.5
2.7
0.6
3.3
TREE PHYSIOLOGY ON-LINE at http://www.heronpublishing.com
128
HRUSKA, CERMÁK AND SUSTEK
Table 3. Main biometric data of oak tree root systems at the experimental plot.
Sample tree
Large
Small
Root length
Mean root density
−2
Mean root radius
−3
Root ground plan area
Rooted volume
(m)
Planar (m m )
Spatial (m m )
(m)
(%cr)
(m )
(%cr)
(m3)
(%cr)
82
21
2.32
2.80
1.16
(1.75)
3.35
1.55
125
130
35.3
7.5
157
170
70.6
6.6
60
36
Figure 3) was about 10 times the stem diameter (± 10%).
Relative to stem diameter, oak roots can be longer than this
under certain soil conditions (Jenik 1957); for example, Kutschera (1960) found more extensive root systems in oak trees
growing on fluvial gravel. Enlargement of a large oak root
system was observed in sandy loam soil in response to drought
(Cermák et al. 1981, 1986).
Radar scanning provided a three-dimensional picture of the
root systems. Figure 4 shows a ground plan, as well as a front
and side view of the root system of the large sample tree. The
main horizonal, heart and tap roots are also visible. The data
indicate a maximum rooting depth of 2 m at a mean coarse root
spatial density (expressed as root length per unit volume) of
about 1.2 m m −3. There were no large roots between the soil
surface and a depth of 0.2 m belowground or in the subsoil
below 2.2 m. The vertical distribution of root spatial density
sharply increased with soil depth below 0.2 m and reached a
maximum at about 1.6 m. Below this depth, root density
gradually decreased to zero (Figure 5). The radial distribution
of root spatial density of the oak trees (i.e., root distances from
the stem) is represented by a curve in Figure 6. The maximum
number of large roots occurred at a distance from the stem
equal to about 56% of crown radius (Figure 6). This pattern
2
differed from that in shallow-rooting species such as spruce
(Henderson et al. 1983), although variation in coarse root
system distribution in trees of different species is large (Strong
and La Roi 1983). Based on the mean diameter of large roots
visible on the records, which was about 5 cm, coarse root
surface area was about 13 m2, coarse root area index was about
0.37, and root volume was about 0.16 m3. Given the resolution
of the GPR, these values are only useful for comparative
Figure 5. Vertical distribution of roots of the large experimental oak
tree (total length of coarse roots is about 82 m). Coarse root distribution is approximated by the equation: y = a(b − x)exp((−(b − x)c)/d );
r 2 = 0.96; a = 14.2; b = 2.25; c = 6.22; and d = 25.4.
Figure 4. Ground plan, front, and side views of the root system of the
large experimental oak tree.
Figure 6. Radial distribution of roots of experimental oak trees, east-west and south--north cross sections were evaluated (total length of
coarse roots is about 82 m). Radial distribution is approximated by the
equation: ; y = a(b − x)cexp(−d(b − x)2); r 2 = 0.96; a = 6.53; b = 3.25;
c = 0.51; and d = 0.0735.
TREE PHYSIOLOGY VOLUME 19, 1999
MAPPING ROOT SYSTEMS WITH PENETRATING RADAR
purposes, because total absorbing root area index was substantially higher than the measured coarse root area index.
Maximum root density occurs in the middle layer of fertile
soils of floodplain forests and in deep loamy soils, whereas oak
roots usually develop a deep bell-like root system with maximum root density close to the surface in nutrient-poor soils
(Jenik 1957). Compared with first-order coarse roots, lateral
roots of the experimental oaks reached similar depths at different distances from the stem indicating that the layer of weathered granodiorite below 2 m is well drained and does not hold
water or nutrients penetrating from the upper horizons.
Not all of the soil below the tree crown contained roots, even
though root systems of neighboring trees overlapped by about
70%; however, there appeared to be no large pockets of soil
free of coarse tree roots when the entire stand was considered
(i.e., including roots of neighbor trees), indicating that roots
were distributed rather uniformly in the soil (the distribution is
even more uniform when both coarse roots and fine absorption
roots are considered; Persson et al. 1990). In contrast, a large
proportion of soil between trees is void of tree roots at sites
with supra-optimal plant water supply and a high water table
(e.g., in floodplain forests, where root systems are small relative to crowns, and reach only about 60% of the crown projected area; Jenik 1957, Vyskot 1976). At such sites, root-free
soil in direct contact with underground water represents significant storage for stand water balance such that the stand is
only partially dependent on water from precipitation (Cermák
and Prax 1999). In the experimental stand with no belowground water, where available soil water and root density have
significant effects on actual and potential transpiration relationships (Stewart et al. 1985), the root systems were extensive, allowing the trees to survive on water derived from
precipitation only.
Conclusions
Ground-penetrating radar measurement as a method of mapping tree roots has several advantages over other methods.
(1) It is capable of scanning root systems of large trees under
field conditions in a short time.
(2) It does not disturb soils or damage the trees examined
and causes no harm to the environment.
(3) Being noninvasive, it allows repeated measurements that
reveal long-term root system development.
(4) It allows observation of root distribution beneath
streams, lakes, rocks, roads and buildings.
(5) Its accuracy is sufficient to resolve coarse roots with
diameters from less than 1 cm to 3 cm or more.
(6) It can characterize roots at both the individual tree and
stand levels, facilitating correlations with tree- and stand-level
measurements of physiological processes (e.g., sap flow and
eddy correlation) in complex ecological studies.
Acknowledgment
The study was supported by the Projects VS 96077 and CGA
501/94/0954. The authors thank Dr. Tomas Krejzar from Mendel
University, Brno, for excellent technical assistance with the DIPS
image analysis system.
129
References
Bohm, W. 1979. Methods of studying root systems. Springer-Verlag,
Berlin, 188 p.
Cammarano, F. and S. Piro. 1997. Application of GPR method to
locate and reconstruct archeological structures in the S. Cecilia
Basilica (Roma, Italy). In 1st Intl. Workshop: Electric, Magnetic
and Electromagnetic Methods Applied to Cultural Heritage. Ostuni,
Italy, p 24.
Cappy, J.J. and D.A. Brown. 1980. A method for obtaining soil-free,
soil solution grown plant root systems. Soil. Sci. Soc. Am. J.
44:1321--1323.
Carlson, W.C., C.A. Harrington, P. Farnum and S.W. Hallgren. 1988.
Effects of root severing treatments on loblolly pine. Can. J. For. Res.
18:1376--1385.
Cermák, J. and J. Kucera. 1990. Water uptake in healthy and ill trees,
under drought and hypoxia and non-invasive assessment of the
effective size of root systems. In Proc. COST 612 Workshop: Above
and Below-Ground Interactions in Forest Trees in Acidified Soils.
Ed. H. Persson. Simlangsdalen, Sweden, pp 185--195.
Cermák, J. and A. Prax. 1999. Water balance of the floodplain forests
in southern Moravia considering rooted and root-free compartments under contrasting water supply and its ecological consequences. Tree Physiol. Volume 19. In Press.
Cermák, J., J. Huzulak and M. Penka. 1980. Water potential and sap
flow rate in adult trees with moist and dry soil as used for the
assessment of the root system depth. Biol. Plant. (Praha) 22:34--41.
Cermák, J., J. Kucera, M. Stepankova and A. Prax. 1981. Measurement and evaluation of transpiration flow rate in fully grown trees
growing near buildings and evaluation of soil hydrolimits in soils of
the housing Development of Hodin-Bazantnice during the growing
season 1981. Res. Rep. Inst. of Forest Ecology, Agr. Univ., Brno,
59 p. In Czech.
Cermák, J., A. Prax and J. Kucera. 1986. Ecological conditions of
permanent coexistence of fully grown trees and constructions in
housing developements. In Proc. Zakladani (staveb) na objemove
nestalych zeminach se zohlednenim vlivu vegetace. Dum techniky
CSVTS Brno, pp 31--78. In Czech.
Henderson, R., E.D. Ford, E. Renshaw and J.D. Deans. 1983. Morphology of the structural root system of Sitka spruce. 1. Analysis
and quantitative description. Forestry 56:121--135.
Jenik, J. 1957. Root systems of oak trees, Quercus robur L. and
Q. petraea Liebl. Rozpr. Cesk. Akad. 67:1--88. In Czech.
Jenik, J. 1967. Root adaptations in West African trees. J. Linn. Soc.
Lond. Bot. 60:25--29.
Jenik, J. 1978. Roots and root systems in tropical trees: morphogenic
and ecologic aspects. In Tropical Trees as Living Systems. Eds.
P.B. Tomlinson and M.H. Zimmermann. Cambridge Univ. Press,
Cambridge, pp 323--349.
Jenik, J. and D.N. Sen. 1964. Morphology of root systems in trees: a
proposal for terminology. Proc. 10th Intl. Bot. Congr. Edinburgh, pp
393--394.
Kasyanenko, A.I. 1980. Kornevaya sistema podvoyev plodovych
derevyev. (Root systems of fruit trees.) Naukova Dumka, Kiev,
220 p. In Russian.
Kolesnikov, V.A. 1962. The root system of fruit plants. Publ. Mir,
Moscow, USSR, 270 p. In Russian.
Kolesnikov, V.A. 1972. Metody izucheniya kornevoi sistemy drevesnykh rasteniy. (Methods of studying the root systems of woody
plants.) Monograph, Lesnaya Promyshlennost, Moscow, USSR,
152 p.
Kutschera, L. 1960. Wurzelatlas mitteleeuropäischer Ackerunkrauter
und Kulturpflanzen. DLG-Verlag, Frankfurt am Main, 574 p.
TREE PHYSIOLOGY ON-LINE at http://www.heronpublishing.com
130
HRUSKA, CERMÁK AND SUSTEK
McQueen, D.R. 1968. The quantitative distribution of absorbing roots
of Pinus sylvestris and Fagus sylvatica in a forest succession.
Oecol. Plant. 3:83--99.
Melhuish, F.M. 1968. A precise technique for measuring of roots and
root distribution in soils. Ann. Bot. 32:15--22.
Papamarinopoulos, St.P., M. Papaioannou and P. Stefanopoulos. 1997.
Explanation of a religious miracle at a Byzantine Church with
geophysical methods at South Greece. In 1st Intl. Workshop: Electric, Magnetic and Electromagnetic Methods Applied to Cultural
Heritage. Ostuni, Italy, 25 p.
Persson, H. 1983. The distribution and productivity of fine roots in
boreal forests. Plant Soil 71:87--101.
Persson, H., K. Ahlstrom, A. Clemensson-Lindell and H. Majdi. 1990.
Experimental approaches to the study of air pollution on trees---root studies. In Proc. CEC and SUAS Above and Below-Ground
Interactions in Forest Trees in Acidified Soils. Ed H. Persson.
Simlangsdalen, Sweden, pp 8--16.
Reijmerink, A. 1964. A new method for recording root distribution.
Meded. Dir. Tuinbouw 27:42--49. In Dutch with English summary.
Santantonio, D., R.K. Herrmann and W.S. Overton. 1977. Root
biomass studies in forest ecosystems. Pedobiologia 17:1--31.
Schuurman, J.J. and M.A.J. Goedwaagen. 1965. Methods for the
examination of root systems and roots. 2nd Edn. Center for Agric.
Publ. and Doc., Wageningen, The Netherlands, 86 p.
Stewart, D.W., L.M. Dwyer and R.L. Desjardins. 1985. The effect of
available soil water and root density on actual and potential transpiration relationships. Can. Agric. Eng. 27:7--11.
Strong, W.L. and G.H. La Roi. 1983. Root system morphology of
common boreal forest trees in Alberta, Canada. Can. J. For. Res.
13:1164--1173.
Sutton, R.F. and R.W. Tinus. 1983. Root and root system terminology.
For. Sci. Suppl., Vol. 24, 138 p.
Vasicek, F. 1984. The characteristics of biogenocenoses. In Ecophysiological and Ecomorphological Studies of Individual Trees in the
Spruce Ecosystem of the Drahanska Vrchovina Uplands (Czechoslovakia). Ed. F. Vasicek. Folia Universitatis Agriculturae, Faculty
Silviculturae, Agric. Univ., Brno, Czech Republic, pp 4--7.
Vogt, K.A. and H. Persson. 1991. Measuring growth and development
of roots. In Techniques and Approaches in Forest Tree Ecophysiology, Eds. H. Lassoie and T.M. Hinckley. CRC Press, Baton Roca,
FL, pp 477--502.
Vogt, K.A., D.J. Vogt, E.E. Moore, B.A. Fatuga, M.R. Redlin and R.L.
Edmonds. 1987. Conifer and angiosperm fine-root biomass in relation to stand age and site productivity in Douglas-fir forests. J. Ecol.
75:857--870.
Vyskot, M. 1976. Tree story biomass in lowland forests in South
Moravia. Rozpravy CSAV 86 (10), Academia Praha, 186 p.
Woods, F.W. 1969. Root extension of forest trees: a method of investigation using radioactive tracers. In Root Growth. Ed. W.J.Whittington. Butterworths, London, pp 413--417.
Zygurovskaya, L.N. 1958. Anatomical and physiological research of
absorbing, growing and conducting roots of woody species.Trudy
Inst. Lesa 41:5--31. In Russian.
TREE PHYSIOLOGY VOLUME 19, 1999