![Maps of estimated moisture content derived from EMI mapping on Field 21 during 2011. Th e ECa values have been temperature corrected and converted into moisture content using to the calibrated Eq. [1], see Fig. 9. Th e arrow in Fig. 12c points at the area covered by dense and tall spontaneous vegetation on 30 Mar. 2011.](https://www.researchgate.net/profile/Antonino-Soddu-Pirellas/publication/232815116/figure/fig3/AS:668983950835744@1536509611999/Maps-of-estimated-moisture-content-derived-from-EMI-mapping-on-Field-21-during-2011-Th-e_Q320.jpg)
Content uploaded by Antonino Soddu Pirellas
Author content
All content in this area was uploaded by Antonino Soddu Pirellas
Content may be subject to copyright.
www.VadoseZoneJ ournal.org
Noninvasive Monitoring of Soil StaƟ c
CharacterisƟ cs and Dynamic States:
A Case Study Highligh ng Vegeta on
Eff ects on Agricultural Land
In this paper we present the results of seasonal monitoring and irriga on tests performed
on an experimental farm in a semiarid region of Southern Sardinia. The goal of the study is
to understand the soil–vegeta on interac ons and how they can aff ect the soil water bal-
ance, par cularly in view of possible clima c changes. We used long-term electromagne c
induc on (EMI) me lapse monitoring and short-term irriga on experiments monitored
using electrical resis vity tomography (ERT) and EMI, supported by me domain refl ec-
tometry (TDR) soil moisture measurements. Mapping of natural γ-ray emission, texture
analysis, and laboratory calibra on of an electrical cons tu ve rela onship on soil sam-
ples complete the dataset. We observe that the growth of vegeta on, with the associated
below-ground alloca on of biomass, has a signifi cant impact on the soil moisture dynamics.
It is well known that vegeta on extracts a large amount of water from the soil par cularly
during summer, but it also reduces evapora on by shadowing the soil surface. Vegeta on
represents a screen for rainfall and prevents light rainfall infi ltra on but enhances the wet-
ng process by facilita ng the infi ltra on and the ground water recharge. In many cases,
the vegeta on creates a posi ve feedback system. In our study, these mechanisms are well
highlighted by the use of noninvasive techniques that provide data at the scale and resolu-
on necessary to understand the hydrological processes of the topsoil, also in their lateral
and depth spa al variability. Unlike remote sensing techniques, noninvasive geophysics
penetrates the soil subsurface and can eff ec vely image moisture content in the root zone.
We also developed a simple conceptual model capable of represen ng the vegeta on–soil
interac on with a simple enough parameteriza
on that can be fulfi lled by measurements
of a noninvasive nature, available at a large scale and evidences possible relevant develop-
ments of our research.
Abbrevia ons: EMI, electromagne c induc on; ERT, electrical resis vity tomography; TDR, me do-
main refl ectometry.
Upscaling knowledge on soil moisture dynamics and vegetation growth into the
soil from the small scale of a single root and soil structure (see e.g., Javaux et al., 2008) to
the larger eld scale is still a partially unexplored and challenging task that has relevant
implication in the interdisciplinary elds of ecohydrology and geoecohydrology. e form
of root growth is an important aspect of the study of vegetation in arid areas, but the plant
root system is not easily accessible and far less studied. e structure and function of roots
are expected to evolve for optimal uptake of water leading eventually to competition among
di erent species (Cody, 1986 and reference therein).
In terms of coupled dynamics, the soil supports the plant growth and, conversely, the
below-ground architecture of plants can a ect the soil structure and thus its physical char-
acteristics having an indirect impact on the subsurface water uxes. Soil moisture balance
and biomass balance are strongly interconnected due to their two-way interaction and the
positive and negative feedbacks that take place between the dynamics of water and the
vegetation growth.
Preferential infiltration and soil moisture redistribution have been indicated as the
two major processes influencing the establishment and persistence of spontaneous
vegetation cover and vegetation patterns in arid and semiarid land (HilleRisLambers
et al., 2001; Rietkerk et al., 2002). Furthermore, Ursino (2007) demonstrated that
the plant survival strategy determines which of the above mentioned hydrological
processes is more important. Vegetation that consumes less water relies more on pref-
erential infiltration for surviving under scarce mean annual rainfall and leads to a
This paper presents the results of sea-
sonal monitoring and irriga on tests in a
semiarid region of Sardinia, using mainly
electromagne c induc on and electrical
resis vity tomography me lapse moni-
toring. The vegeta on has a signifi cant
impact on the soil moisture dynamics,
changing infi ltra on and evapotranspira-
on pa erns.
G. Cassiani, J. Boaga, M. Rossi, and M. T. Perri,
Dipar mento di Geoscienze, Università degli Studi
di Padova, Via Gradenigo 6, 35131 Padova, Italy; R.
Deiana, Dipar mento di Beni Culturali, Università
degli Studi di Padova, Piazza Capitaniato 7, 35139,
Padova, Italy; N. Ursino, Dipar mento ICEA, Uni-
versità degli Studi di Padova, Via Loredan 20, 35131
Padova, Italy; G. Vignoli, King Fahd Univ. of Petro-
leum and Minerals, Earth Sciences Dep., 31261
Dhahran, Saudi Arabia; currently Department
of Geoscience, HydroGeophysics Group, Aarhus
University, Aarhus, Denmark; M. Blaschek and R.
Du mann, Ins tute for Landscape Ecology and
Geoinforma on, Dep. of Geography, Univ. of Kiel,
Ludewig-Meyn-Strasse 14, 24098 Kiel, Germany;
S. Meyer and R. Ludwig, Dep. of Geography, Univ.
of Munich, Luisenstr. 37, 80333 Munich, Germany;
A. Soddu, AGRIS Sardegna, Viale Trieste 111, 09100
Cagliari, Italy; P. Dietrich and U. Werban, UFZ - Hel-
mholtzCentre for Environmental Research, Dept.
Monitoring and Explora on Technologies, Per-
moserstr. 15, Leipzig, Germany. *Corresponding
author (giorgio.cassiani@unipd.it).
Vadose Zone J.
doi:10.2136/vzj2011.0195
Received 16 Dec. 2011.
Special Section: Soil–Plant–
Atmosphere Continuum
Giorgio Cassiani*
Nadia Ursino
Rita Deiana
Giulio Vignoli
Jacopo Boaga
MaƩ eo Rossi
Maria Teresa Perri
Michael Blaschek
Rainer DuƩ mann
Swen Meyer
Ralf Ludwig
Antonino Soddu
Peter Dietrich
Ulrike Werban
© Soil Science Society of America
5585 Guilford Rd., Madison, WI 53711 USA.
All rights reserved. No part of this periodical may
be reproduced or transmi ed in any form or by any
means, electronic or mechanical, including pho-
tocopying, recording, or any informa on storage
and retrieval system, without permission in wri ng
from the publisher.
www.VadoseZoneJ ournal.org
scenario where more soil moisture is lost due to runoff and
leakage (Ursino, 2005, 2007, 2009).
Noninvasive techniques can play a key role in the hydrological
investigation of the near surface, as they provide spatially exten-
sive imaging that complements the more traditional hydrological
point measurements (e.g., Vereecken et al., 2006). A number of
studies have appeared in the recent literature, particularly focused
on ground-penetrating radar (GPR) and ERT. e use of these
techniques has been increasingly focused on their ability to mea-
sure, albeit indirectly, changes in moisture content (e.g., Binley et
al., 1996; Michot et al., 2003; Strobbia and Cassiani, 2007; Deiana
et al., 2008) and solute concentration (e.g., Cassiani et al., 2006)
by conducting time-lapse measurements. Recently, ERT has also
been used to image the root zone geometry, with some degree
of success (Werban et al., 2008; al Hagrey and Petersen, 2011).
Frequency-Domain EMI is also widely used in soil mapping, to
determine soil salinity (e.g., Corwin and Lesch, 2005), subsurface
morphology (e.g., Comas et al., 2004), and texture (e.g., Jung et
al., 2005; Trianta lis and Lesch, 2005) thanks to its noncontact
capability of producing georeferenced data quickly and inexpen-
sively. Applications of time-lapse EMI to determine soil moisture
changes and study soil hydrological processes have been limited to
date (Kachanoski et al., 1988; Sheets and Hendrickx, 1995; Abdu
et al., 2008; Robinson et al., 2009) and deserve further attention
(Robinson et al., 2008).
On the other hand, suitable modeling techniques are necessary to
exploit the information content of eld data and answer critical
questions about basic mechanisms. Flow through porous media
have been extensively investigated on the base of the continuum
theory, by formulating the soil moisture balance equations at
the scale of the representative elementary volume (Bear, 1972;
Richards, 1931). e growing interest in interdisciplinary sciences
that consider the spatial and temporal evolution of di erent species
conditioned to the hydrology of the ecosystem where the growth
lead in the last 10 to 15 yr to the frequent adoption of minimal
bucket models formulated at a much larger scale for the soil mois-
ture balance instead of the more rigorous approach based on the
continuum theory. Evaporation evapotranspiration, in ltration,
lateral subsurface ow, and leakage are the building blocks of the
soil water balance at the eld scale, and they are all in uenced by
vegetation roots and soil properties. (Eagleson, 2002; Nuttle, 2002;
Porporato and Rodriguez-Iturbe, 2004).
The Mediterranean ecosystem is characterized by dry summer
and wet winter producing water stress conditions for the
vegetation that grows during summer. Grasses that are char-
acterized by high water demand during winter and spring are
more drought adapted, even though there is anyway a lag time
between soil moisture accumulation during the rainy season
and soil moisture use during the growing season. Soil moisture
storage and thus soil moisture availability during the growing
season depends on the partitioning between infiltration and
runoff, the flow field within the root zone and the depth of
the root zone that represents the volume where water is stored.
Minimal models based on soil moisture and biomass balance
(such as the one that will be introduced here), have a steady
state solution that does not depend on the depth of the root
zone when the hydrologic forcing (rainfall and evapotranspira-
tion) is set at its constant annual average. But when the time
lag between water storage and water release comes into play, the
root zone depth becomes crucial for the soil moisture balance
and thus for the survival of selected vegetation species. At an
intra-seasonal time scale for random rainfall pulses the water
balance and the optimum root depth are governed by the rain-
fall frequency and intensity (Milly, 1993; Porporato et al., 2004;
Guswa, 2008, 2010). At the catchment scale of our experiment,
the focus switches from species survival and adaptation to other
hot topics related to land and water management (Jackson et al.,
2009). On an average annual basis, major issues concerning the
connectivity of the soil moisture paths above ground through
overland flow and below ground through lateral flow and the
connectivity of the soil surface with the water table should be
raised to clarify how annual precipitation, evapotranspiration
and water yield are altered by a land use change. In a specific
ecohydrological context the interrelation between infiltration,
plant growth and water yield depends on biological variables
(leaf area index, rooting depth, and seasonality of plant activity),
climate and soil texture (Huxman et al., 2005; Newman et al.,
2006). The soil structure is often related to vegetation growth
and to the soil heterogeneity that the vegetation growth transfer
on the soil structure (e.g., Flury et al., 1994).
In this paper we address, at least partly, the general question of
what could be the impact of vegetation and particularly of root
architecture on soil properties and indirectly on the water cycle,
according to the experimental evidence collected via noninvasive
techniques. In more in detail, we discuss the interconnection
between soil moisture paths, soil moisture path connectivity and
vegetation cover. We do this by comparing noninvasive eld obser-
vations with the results of a very simple conceptual model. e
experimental study was conducted at an agricultural experimental
farm located in Sardinia, Italy, as part of the part of the EU-FP7
CLIMB project (Ludwig et al., 2010), focused on the analysis
of climate change impacts on the hydrology of Mediterranean
basins. The main goal of the experimental activities within
CLIMB are to collect information about the hydrologic behavior
of Mediterranean catchments, ranging from the small soil pro le
scale to the larger catchment scale.
e interpretation of collected data has been supported by an
ecohydrological minimal model, incorporating knowledge on the
relevant feedbacks between root architecture and dynamic water
balance in cultivated soils.
www.VadoseZoneJ ournal.org
6
Site DescripƟ on
e study site is located at the San Michele farm in the Rio Mannu
Catchment (Southern Sardinia). e basin ranges in elevation
from 62 to 842 m asl with an average of 295.5 m asl. e basin is
mainly covered by agriculture elds and grassland, while only a
small percentage of its area is occupied by forests in the southeast of
the basin. e farm area has a gentle topography and is part of the
Campidano plain. e soils in the area are brown soils, regosols and
vertisoils or marls, with outcrops of sandstones and conglomerates.
e oodplain is characterized by alluvial soils, predominantly
gravelly or sandy gravel. e yearly mean total precipitation in the
farm area is about 500 mm with an estimated mean runo of about
200 mm. e hydrological regime is characterized by wet periods
from October to April, where more than 90% of the rainfall is
accumulated, and very dry summers (May–September). e yearly
average temperature is 16°C. Groundwater is thought to provide a
negligible contribution to the stream ow. e San Michele farm
is an agronomic research eld covering an area of 4.36 km
2
and
managed by AGRIS, a research agency of Regione Autonoma della
Sardegna government. Part of the Azienda (~2 km
2
) is located in a
hilly area with Maquis shrubland veg-
etation that is also a center of wildlife
animal restocking. In the northeastern
part, the Azienda is delimited by the
San Michele hill. At the bottom of
this hill, the Rio Flumineddu joins the
Rio Mannu. e farm has been used
for decades to investigate agricultural
genetics for a more e cient farming of
durum wheat (Triticum turgidum L.)
in climatic conditions with frequent
drought periods, such as the Sardinian
ones. Today, the experimental agricul-
tural elds are destined to open- eld
Mediterranean cultivations, particu-
larly important for the economy of
the island.
e soil in the farm is characterized by
silty-sandy agricultural soil overlying
a formation with abundant calcium
carbonate nodules. A plow layer about
30 cm thick is maintained across the
site by land preparation each year,
irrespective of the soil being used for
cultivation. No soil pan layer is present,
thanks to the careful cultivation prac-
tices, including periodic deep plowing.
Figure 1 shows a satellite view of the
San Michele farm evidencing the areas
where measurements have taken place.
Traditional and noninvasive methods
have been applied to monitor changing moisture content condi-
tions on selected elds since 2009.
In two eld campaigns (October 2010 and March 2011) about
300 soil samples from three depths (0–30, 30–60 and 60–90 cm)
were collected over the Rio Mannu catchment. Forty-three of these
samples from 0–30 cm depth were taken over Field 21. e cor-
responding grain size distribution is summarized in Table 1, while
interpolated maps of clay, silt and sand weight percentages and
CaCO
3
percentages are shown in Fig. 2. e spatial variations of
grain size fractions is consistent with the geomorphologic features
of the area: for example, sand prevails in the southern part of Field
21, in correspondence of the original bed of the creek owing from
east (see Fig. 1). However these variations are subtle, with all frac-
tions having ranges around 10%.
In October 2010 a collaboration between the EU FP7 projects
CLIMB and iSOIL (Werban et al., 2010) brought a UFZ eld crew
to the eld site and allowed, among other things, for a rapid map-
ping of natural γ-ray emission over nearly the entire farm. e most
prominent evidence from this γ-ray survey is a strong correlation
Fig. 1. Satellite view of the San Michele farm in Ussana, near Cagliari, Sardinia. Coordinates are UTM
32. (Image source: Google Earth).
Table 1. Results of grain size analysis on the San Michele farm (Ussana, Cagliari, Sardinia).
Min. Max. Median Mean SD
Clay, % (w/w) 21.43 44.52 34.43 33.50 5.76
Silt, % (w/w) 19.08 34.87 25.55 26.00 3.89
Sand, % (w/w) 31.02 55.31 36.70 40.51 7.26
C
org
, % (w/w) 0.52 1.19 0.79 0.83 0.21
www.VadoseZoneJ ournal.org
between low total dose rate and a whiter soil color, which in turn
is associated to a high calcium carbonate content (Fig. 2), derived
from erosion of calcite nodules (Fig. 3). Note that the di erent soil
color is visible in satellite images (e.g., in Fig. 3) solely because Field
21 was maintained free of vegetation for long periods as part of a
remote sensing ground-truthing experiment.
6
IrrigaƟ on Experiment:
Results and Discussion
In May 2010, a controlled irrigation experiment was undert aken
on Field 23 (see map in Fig. 1) where the eastern plot was cultivated
with alfalfa (Medicago sativa L.) even though several other invasive
species coexisted with it at the time of the experiment. e western
plot was le bare, but few spontaneously growing species survived
there. A satellite view of Field 23 at the time of the experiment is
shown in Fig. 4a, from which it is apparent how the vegetation
cover is substantially di erent in the two plots separated by a dirt
road. e vegetation density can also be appreciated by the photo-
graphs in Fig. 4c and 4d.
Figure 5 shows the typical root architecture of the di erent species
that cover the two plots. e vegetation in the cultivated site is
uniform with high density and preferential allocation of biomass
above ground; the roots are shallow. is fact does not allow indi-
cation, though, that the roots are not able to extract water from
deeper in the soil pro le via suction. On the contrary, the vegeta-
tion cover in the bare soil appears sparse above ground but the
allocation of biomass is, in this case, preferential below ground.
Spontaneous vegetation growing in the bare soil presents deep
roots that spread both vertically and horizontally.
No di erence in soil texture or color is apparent between the
two plots, as evidenced in Fig. 4b, where we show the natural γ
Fig. 2. Interpolated grain size percentages and carbonate content on Field 21.
www.VadoseZoneJ ournal.org
Fig. 3. Comparison between satellite image of Field 21 on 29 Oct. 2002 (bare soil image source: Google Earth) and the total γ-ray emission map
obtained from soil mapping.
Fig. 4. Setup of the area where the irrigation experiment was conducted in May 2010: (a) satellite view of Field 23 (24 May 2010, image source: Google
Earth) showing the vegetated and the bare plots; (b) total γ ray emission map of the same area; (c) photo of the vegetated; and (d) of the bare soil plots.
www.VadoseZoneJ ournal.org
emission map collected in October 2010:
note that the total dose rate variation is
within a very narrow range (compare Fig.
4b with the variations observed in Field
21 (Fig. 3b).
Before the irrigation experiment, we
conducted an EMI survey in the area,
covering both the vegetated and the bare
plots, using a GF Instruments CMD1
sonde in low-penetration con guration,
corresponding to an estimated total
depth of investigation of 75 cm. The
results (Fig. 6) show a strong di erence
in average electrical conductivity of these
top 75 cm in the two plots. e vegetated
plot was considerably more resistive (on
average a factor of 2) than the bare plot
just a few meters away. This result is
somewhat surprising as the bare soil has
a crusty appearance, evidently as a result
of evaporation from the top layer, while
the soil in the vegetated plot was much
wetter at the surface probably thanks
to the shading provided by the veg-
etation against direct sunlight. On the
other hand, given that the soil texture
is the same in both plots, we could
only attribute the di erence in appar-
ent electrical conductivity to the e ect
of vegetation, and its interaction with
the local hydrology. During the period
October 2009 to September 2010 the
mean annual rainfall at the site was
500 mm, while the temperature ranged
between 10 and 30°C. e estimated
mean annual evapotranspiration was
between 100 and 200 mm (Agenzia
Regionale per la Protezione dellAm-
biente Sardegna, 2011).
To obtain detailed information about
the system’s changes as a result of a
controlled irrigation, we deployed two
electrical resistivity tomography (ERT)
lines, one in each plot. Each line was
composed of 24 electrodes spaced 20
cm, for a total length of 4.6 m each, and
an expected depth of investigation not
exceeding 1 m. ese lines were le in
place throughout the experiment until
4 d a er irrigation. Time-lapse measure-
ments were taken periodically, using a
Fig. 5. Field 23: Typical root length (le ) in the cultivated plot and (right) in the bare soil where
spontaneous and sparse vegetation grows.
Fig. 6. e results of the small-scale electromagnetic induction survey conducted on 18 May 2010 on
the part of Field 23 used for the irrigation experiment. e survey was conducted with a GF Instruments
CMD1 sonde in horizontal loop con guration, with a nominal penetration depth of 0.75 m. Both the
vegetated and the bare soil plots were surveyed. e gure also shows the location of the two electrical
resistivity tomography lines used for irrigation monitoring and of the xed time domain re ectometry
probes, in both plots. At the same time TRASE measurements were run at random georeferenced loca-
tions on both elds to measure the average moisture content in the top 0.10 m.
www.VadoseZoneJ ournal.org
dipole-dipole skip 0 scheme (i.e., with
dipole length equal to one electrode
spacing) and full acquisition of recipro-
cals to estimate the data error level (see
e.g., Monego et al., 2010). Consistently,
the data inversion used an Occam
inversion approach as implemented in
the Pro leR/R2/R3 so ware package
(Binley, 2011) accounting for the error
level estimated from the data them-
selves. The skip 0 scheme allows for
the highest achievable resolution and
still produces signi cant signal/noise
ratios given the short dipole distances
used in these acquisitions.
e noninvasive monitoring has been
complemented by (i) fixed TDR
probes (32 cm and 50 cm) located
next to the ERT lines (Fig. 6), peri-
odically monitored with a Tektronix
1502 instrument; and (ii) a roaming
10-cm TRASE probe (Soilmoisture
Equipment Corporation), that unfor-
tunately failed a er the background
measurements on 19 May 2010. Figure
7 shows the results of the ERT moni-
toring (lines 1 and 2) of the irrigation experiment. In particular,
the background images (19 May 2010) show that indeed the ERT
pro le in the vegetated plot (line 1) is substantially more resistive
than the equivalent pro le (line 2) in the bare soil. In fact, the two
pro les are nearly mirrored.
1.
In the vegetated plot, a (relatively) resistive subsoil underlies a
thin, more conductive soil layer at the top. Our tentative expla-
nation is that (i) the top layer is moist, as it is shaded from direct
sunlight by the canopy, while (ii) the deeper soil is relatively dry,
as roots manage to extract soil moisture content from a depth in
the 1-m range, mostly through suction, as the roots themselves
are relatively shallow (Fig. 5).
2.
In the bare plot, a more conductive soil layer underlies a thin,
more resistive top layer. Here our tentative hypothesis is that
the topsoil is strongly a ected by evaporation due to direct sun-
light and exposure to the warm air (no continuous canopy is
present), thus protecting the underlying soil from evaporation
(the top crust has reduced hydraulic conductivity).
e ERT background results are totally consistent, also quanti-
tatively, with the EMI maps in Fig. 6 particularly if we consider
that the EMI data refer to an average over the soil down to 75 cm.
e night before the experiment (between 19 May and 20 May)
a natural 13-mm rainfall event occurred over the farm. e irri-
gation experiment took place overnight between 20 May and 21
May 2010: about 42 mm of irrigated water was applied to both
vegetated and bare soils covering the entire area surrounding the
two ERT lines.
e e ects of both natural rainfall and irrigated water is shown,
for both ERT lines, in Fig. 7. ree aspects are clearly noticeable.
1.
e 13-mm natural rainfall seemed to cause little to no e ect
on the ERT images. is may be slightly surprising, but it can
be explained by (i) the rainfall interception on the canopy for
line 1, and (ii) the large runo fraction caused by the soil crust
on line 2.
2.
e 42-mm controlled irrigation causes a dramatic change in
the images of ERT line 1: suddenly they become very similar
to the background image along line 2. A possible explanation
is that the large rate of in ltrating water replenished the water
de cit caused by evapotranspiration in the vegetated plot.
3.
e 42-mm controlled irrigation has no noticeable e ect on
ERT line 2: is can be again explained by the low hydraulic
conductivity of the soil crust covering the bare plot, drawing
most irrigated water into runo onto the road and/or locally
ponded water that is rapidly evaporated in absence of a protect-
ing canopy.
ese phenomena seem to point toward the existence of well
de ned mechanisms associated to the presence of vegetation. e
presence of roots, albeit short and thin, is clearly capable of enhanc-
ing the in ltration capacity of the vegetated plot: this is a positive
feed-back mechanism that enhances the replenishment of the
water reservoir available to vegetation. Indeed, the background
Fig. 7. Sequence of electrical resistivity tomography lines collected over lines 1 and 2 (see Fig. 6 for loca-
tion) before and a er the irrigation was applied to both vegetated and bare soil plots in Field 23.
www.VadoseZoneJ ournal.org
ERT and EMI images demonstrate that vegetation can e ectively
draw water from a depth exceeding half a meter; that is the depth
replenished as a consequence of the intense arti cial irrigation
input. On the other hand, the absence of an extensive vegetal cover
on the western plot causes the formation of a low permeability dry
topsoil layer that protects the underlying soil from evaporation and
maintains a high moisture content in the underlying soil layer. On
the other hand, the impermeable dry topsoil also limits in ltration
into the subsoil.
e above discussion is necessarily limited to qualitative terms. We
made an attempt to quantify the meaning of the ERT monitoring
by calibrating a suitable constitutive relationship linking electri-
cal conductivity and moisture content. Given the relatively large
ne grain fraction (see Table 1) it is necessary to account for grain
surface conductivity in the constitutive relationship (consider, e.g.,
Brovelli and Cassiani, 2011). A classical approach, albeit somewhat
limited, is to use the Waxman and Smits (1968) model in its unsat-
urated porous media form:
n
S
W
S
FS
⎛⎞
σ
⎟
⎜
σ= σ +
⎟
⎜
⎟
⎜
⎝⎠
[1]
where σ is the bulk electrical conductivity (σ = 1/ρ, where ρ is the
electrical resistivity derived e.g., from the ERT inversion); σ
W
is the
pore water electrical conductivity; σ
S
is the equivalent grain surface
conductivity; S is water saturation (0 ≤ S ≤ 1); n is referred to as
the saturation exponent; F is Archie’s formation factor, that can be
expressed as F = 1/φ
m
, where φ is porosity, and m is the so-called
Archie’s cementation exponent. To reduce the number of indepen-
dent parameters we assumed m = n. is assumption is justi ed as
we did not seek to characterize the spatial variability of porosity,
and in absence of these data an independent calibration of m is
impossible. In addition, as we are interested only in the variation of
electrical conductivity with saturation, any other parameterization
that describes the formation factor would be e ectively equivalent.
We used Eq. [1] to model the transformation from electrical con-
ductivity to water saturation. To x the model parameters we rst
decided to try and honor the in situ moisture content measure-
ments provided by TDR and TRASE probes in the vegetated plot,
where moisture content changes are extreme. is was done by (i)
taking the horizontal averages of the line 1 ERT resistivity images
in Fig. 7, thus constructing one-dimensional resistivity pro les as
a function of depth only; (ii) Monte Carlo simulations exploring
the parameter space and identifying the optimal parameter set
that transforming the one-dimensional resistivity pro les above,
satis es, in a least squared sense, the TDR data on 19 May and on
24 May 2010, that is, at the beginning and end of the irrigation
experiment, and the TRASE data on 19 May.
e result of this tting procedure is shown in Fig. 8. Note
that it has not been able to reproduce the TRASE average value on
19 May. is is hardly surprising, as the TRASE data are limited
to the 10 cm depth of the TRASE probe, while the resolution of
the ERT-derived moisture content estimations cannot be any ner
than the 20 cm electrode spacing (actually, lower than that). e
TDR data are reasonably well tted by Eq. [1] with the optimal
parameters: φ = 0.391; n = m = 2.42; σ
W
= 6.48 × 10
−2
S/m; σ
S
= 0.982 S/m.
Several soil samples from the top 30 cm were used for laboratory
measurements of electrical resistivity as a function of water satu-
ration. Tap water was used to saturate the samples during these
experiments, with an average water conductivity equal to 4.9 ×
10
−2
S/m. e water extracted during desaturation, a er con-
tact with the sample soil, had an electrical conductivity equal to
6.6 × 10
−2
S/m, i.e., very close to the σ
W
value obtained by t-
ting the ERT data to the eld TDR measurements using Eq. [1].
Desaturation and resistivity measurement were performed similar
to the procedure described in Cassiani et al. (2009). e labora-
tory measurements were conducted at 20°C, i.e., very similar to
the in situ temperature in late May 2010 at the San Michele site
(18°C). Figure 9 shows the comparison between the Waxman and
Smits (1968) relationship (Eq. [1]), tted to the eld data, and the
laboratory data. While a nonnegligible scatter is clearly present in
the laboratory results, the calibrated Eq. [1] is largely compatible
with the laboratory results.
Fig. 8. Calibration of electrical resistivity tomography inversion results
against in situ time domain re ectometry and TRASE measurements
of moisture content over the vegetated plot. e curves of moisture
content as a function of depth are obtained taking the horizontal aver-
ages of the line 1 electrical resistivity tomography resistivity images
of 19 May and 24 May 2010 in
Fig. 7, that is, at the beginning and
end of the irrigation experiment, and transforming resistivity into
moisture content values with Eq. [1] calibrated on time domain
re ectometry and TRASE data. e horizontal dotted line marks
the maximum depth considered reliable for the electrical resistivity
tomography–derived pro les, that is, the one that correspond to a cor-
rect mass balance of in ltrated water.
www.VadoseZoneJ ournal.org
Figure 8, however, demonstrates the existence of one remaining
pending issue: the integral of moisture content change between
19 May and 24 May 2010, taken from the ground surface to the
maximum depth of about 92 cm (about 100 mm) exceeds the total
known in ltrated water (55 mm). In fact a large percentage (nearly
50%) of the moisture content change integral lies in the bottom
30 cm of the pro le, i.e., the region where surface ERT resolu-
tion is poorest. Mass balance issues from ERT inversion of tracer
test monitoring data is known to exist and can lead to severe bal-
ance errors (e.g., Singha and Gorelick, 2005): is phenomenon
is well established to be a consequence of resolution limitations
of geophysical data (Cassiani et al., 1998; Day-Lewis et al., 2005).
Since the poorest resolution in ERT images is necessarily present
at depth, it is reasonable to trust the ERT results down to a depth
compatible to the need to honor mass balance, i.e., to roughly 60
cm depth (see Fig. 8). All water mass that appears to be deeper
than this depth is probably only caused by an extrapolation of
the electrically more conductive layer put in place by the irriga-
tion between 30 and 60 cm depth. To give some con rmation of
this hypothesis, we ran a synthetic exercise relevant to the ERT
acquisition on the vegetated eld on 24 May. First we considered
a one-dimensional resistivity versus depth pro le obtained from
horizontally averaging the two-dimensional ERT inverted section
(the one at the bottom of Fig. 7). is pro le is the one used to
derive the moisture content pro le in Fig. 8. We extended this
pro le to larger depth, as needed for synthetic modeling and we
generated a forward dataset with the same con guration as the
eld acquisition. is dataset was then inverted with the same data
error (5%): the result is shown in Fig. 10 as synthetic (a). We then
modi ed the pro le at depths larger than 62 cm adopting a larger
resistivity value, roughly equal to the maximum observed close to
the surface, and the corresponding inverted image is also shown in
Fig. 10 as synthetic (b). Both synthetic results are compared with
the real inverted data: even though the test is not conclusive, the
images in Fig. 10 seem to indicate that the true data are likely to
correspond to a situation intermediate between the two synthetic
(one-dimensional) cases, con rming the low resolution character-
istics of the inversion at depths larger than about 60 cm.
Given the above, we accepted the Waxman and Smits relationship
with the parameters tted to the eld TDR data, and corroborated
Fig. 10. Sensitivity analysis with respect to the actual resistivity pro le below 0.63 m, that is, the depth down to which the electrical resistivity tomog-
raphy inversion is considered reliable. On the le the two pro les used in the synthetic forward/inverse modeling: (a) is the inverted pro le on 24
May (see
Fig. 8) extrapolated to larger depth; (b) is the same pro le down to 0.63 m, but with higher resistivity deeper than that. On the right the
corresponding synthetic electrical resistivity tomography inversion results compared against the actual inverted electrical resistivity tomography image.
Fig. 9. Laboratory data on soil samples from the San Michele farm
(diamonds) compared against the eld-calibrated Waxman and Smits
relationship (Eq. [1]).
www.VadoseZoneJ ournal.org
by the laboratory measurements, as our best estimate of the rela-
tionship between electrical resistivity and moisture content, and
thus converted all ERT sections collected during the irrigation
experiment (Fig. 7) into moisture content sections (Fig. 11). e
most prominent features of this result are (i) the large di erence
in moisture content observed between vegetated and bare elds,
and (ii) the dramatic e ect that irrigation has on the vegetated
plot, apparently restoring the subsoil moisture content condi-
tions to a situation similar to the one observed in the bare plot.
We argue that the pre-irrigation di erence is mainly caused by the
evapotranspirative e ect of the vegetation, and that irrigation only
replenishes the depleted soil reservoir in the soil top 60 cm.
6
Seasonal Monitoring: Results
and Discussion
We also performed the time-lapse hydrogeophysical monitoring
of the soil hydrological dynamics under natural conditions. e
technique we utilized (EMI) is most sensitive to changes in the
hydrological state of the system, particularly to its moisture con-
tent. However temperature changes cannot be neglected. e
changes considered are naturally induced by the di erent forcing
conditions (e.g., precipitation, evapotranspiration) that strongly
depend on the season. e ultimate aim is to translate the time and
space dependent geophysical data into quantitative estimates of the
hydrological state variables, distributed in space and time, that in
turn shall be part of the dataset for hydrological model calibration.
Monitoring started in May 2009 and is currently ongoing
(December 2011). e dataset is composed of a good number of
repeated frequency-domain electromagnetic measurements par-
ticularly focused on Field 21. e acquisition has been performed
using a GF Instruments CMD1 electro-magnetometer in low
penetration con guration (horizontal loops) with a nominal pen-
etration of about 0.75 m. Most of the measurements have been
performed manually and are automatically georeferenced using a
Trimble GPS with decimetric precision in horizontal positioning.
Note that, given the high soil electrical conductivity, GPR does
not produce usable results, as the signal is quickly attenuated in
the subsurface. erefore neither structural information nor time-
lapse measurements of soil moisture content are possible using this
technique.
Temperature e ects have been accounted for by correcting the EMI
electrical conductivity readings (EC
a
) according to the relation-
ship proposed by Sheets and Hendrickx’s (1995) and discussed in
detail by Ma et al. (2011):
Fig. 11. Sequence of electrical resistivity tomography lines collected over lines 1 and 2 before and a er the irrigation, converted into estimates of mois-
ture content according to the calibrated Eq. [1] (see
Fig. 9).
www.VadoseZoneJ ournal.org
()
/26.815
25
0.4470 1.4034
T
a
EC EC e
−
⎡⎤
=+
⎢⎥
⎣⎦
[2]
where EC
25
is the EC
a
standardized at 25°C and T is the soil tem-
perature (°C). Estimates of soil temperature have been derived
from average monthly air temperatures recorded on site. Similarly
the bulk conductivity in the calibrated Eq. [1] can be normalized
to 25°C to transform EC
a
readings into estimated average moisture
content in the top 0.75 m of soil. Figure 12 shows a sequence of
such estimated moisture content maps on Field 21 from September
2010 to October 2011. During this period the eld was maintained
free of vegetation by herbicide application. is practice was inter-
rupted between the beginning of February and the end of March
2011, and consequently a mixture of spontaneous species grew with
a prevalence of oilseed rape.
e moisture content maps in Fig. 12 show three main features.
1. In all seasons the spatial patterns of moisture content are well
de ned and repeatable, showing an area of higher moisture con-
tent to the western side of the site. e higher moisture content
is e ectively correlated with the ne soil fraction (clay+silt: see
Fig. 2) while it is not so obviously correlated with γ-ray emission
that is strongly controlled by the abundance of CaCO3 (Fig.
3) thus probably masking the more subtle ne fraction e ect.
2.
Moisture content changes substantially over time, and as
expected is much lower over summer (below 8%) than in
winter–spring (as high as 22%).
3.
In March 2011 the most dense spontaneous vegetation is con-
centrated in the area where higher moisture content is always
observed, i.e., in the region of finer soil texture. Here the
canopy reached more than 2 m in height, making it di cult
even to walk through this area. Elsewhere in Field 21 the veg-
etation appeared to be sparse and large patches of bare soil still
remained, similar to the situation of the bare soil plot in Field
23. Note that the dense vegetation is here correlated with higher
rather than lower moisture content in the soil as observed in
Field 23 in May 2010. However this is not surprising, as (i) we
are here considering data related to an earlier period (March
versus May) with a di erent evapotranspiration demand, and
(ii) the ner soil texture in the western area of Field 21 is a factor
controlling the vegetation growth.
Figure 13 shows another interesting phenomenon linked to
the interaction between vegetation and soil state. Here the esti-
mated moisture content maps in May 2009, May 2010, and May
2011 are compared. e EMI surveys were conducted roughly at
the same date. For May 2010 this corresponds to the time of the
irrigation experiment in Field 23.
In January–February 2010 the south-eastern portion of the eld
was seeded with wheat, and in May 2010 this was fully grown. On
the contrary, in both May 2009 and May 2011, the whole eld was
bare. e wheat area is marked by the thick line in Fig. 13. e
presence of wheat has a distinct e ect on moisture content patterns,
reducing moisture content basically to the level observed in the veg-
etated plot of Field 23 at the same time (10–12%) and erasing any
pattern observable in 2009 and 2011 over the same area, and likely
linked to soil texture subtle patterns. e remaining part of Field
21, always le bare, shows the same features in May 2009, 2010,
and 2011. Note that Field 21 was, up to May 2011, not irrigated.
6
Model
e phenomena observed in the eld irrigation experiment and long-
term monitoring, particularly the interactions between soil moisture
content and vegetation, call for the development of a mechanistic
model capable at least of capturing the main controlling features.
In this section, we would like to provide a tentative theoretical
framework that could support our hypothesis that the signi cant
di erences that characterize the biomass and water dynamics in
bare and vegetated plots of Field 23 could be ascribed to the water
stress that originates from the scarce wettability of the bare soil, that
is just partially contrasted by the vegetation growth. Furthermore,
we would discuss what could cause the very di erent soil moisture
patterns observed in the two noncultivated plots of Field 21 and 23
where the vegetation density and the soil saturation are respectively
very high and quite low, nally we would discuss how the soil- vege-
tation feedbacks could impact the water budget and the connectivity
of the soil surface with the soil layers below the root zone.
We do not try to invert measurements but rather we try to translate
soil moisture patterns into evidence of relevant processes in act
and justify the link and relevance of these processes to the growth
of di erent species, on the base of mass balance consideration.
e following experimental evidences may be easily incorporated
within simple biomass and water balance evaluation.
1.
High biomass density corresponds to high soil moisture avail-
ability (at Field 21) and conversely a patchy distribution of
vegetation (corresponding to low average biomass density) is
associated to the persistent soil moisture de cit (at Field 23).
2.
A feedback seems to exist between in ltration and vegetation
growth according to the soil moisture patterns observed at Field
23, if the scarce vegetation cover of the bare plot is associated
with increased runo and reduced in ltration.
3.
A linkage seems to exist between the dynamics of soil moisture
and the depth of the soil layer that participates to the dynamics
of soil moisture (it could be expected to be related with the root
depth). e linkage is evidenced by the comparison between the
soil moisture patterns observed at the cultivated site (no water
stress) and those observed at the bare plot of Field 23.
e conceptual model proposed in the following consists of
two mass balance equations: one for the soil moisture and one for
biomass (see Fig. 14 for a conceptual scheme of the model uxes).
e rate of variation of the biomass density B is the di erence
between the growth rate of the biomass G and the mortality rate
M due to water scarcity.
d
d
B
GM
t
= − [3]
www.VadoseZoneJ ournal.org
Fig. 12. Maps of estimated moisture content derived from EMI mapping on Field 21 during 2011. e ECa values have been temperature corrected and
converted into moisture content using to the calibrated Eq. [1], see
Fig. 9. e arrow in Fig. 12c points at the area covered by dense and tall spontane-
ous vegetation on 30 Mar. 2011.
www.VadoseZoneJ ournal.org
e time derivative of the soil moisture volume nHS , where S is
the soil saturation, n the soil porosity and H the root depth, equals
the in ltration rate I minus evapotranspiration ET and leakage L
.
d
ET
d
S
nH I L
t
= −− [4]
According with our intuitive interpretation of the experimental
data, we assume that in ltration is impeded in the absence of
vegetation and furthermore, that in ltration and evapotranspi-
ration both increase with increasing biomass density. e soil
vegetation interaction a ects leakage as well. We simulate the
possible occurrence of fast ow from the soil surface to the water
table by increasing the soil conductivity and thus L, under the
hypothesis that roots create fast ow path to the detriment of
soil moisture storage.
Parameters and functions are taken from literature and our specu-
lation on the relevance of the soil vegetation feedbacks will serve
to outline speci cally dedicated further experiments. Functions
(G, M, I, ET, L), function parameters and related references are
reported in Table 2. Equation [3] and [4] have been integrated on
daily time scale for several years to lose memory of an arbitrary
initial condition and average estimate of the dimensionless biomass
density b = B/K
b
and of soil saturation S are presented in Fig. 15 as
a function of H, for di erent hypothetical environmental scenarios
(see Table 2 for parameter de nition).
When the partitioning of rainfall into in ltration and runo is
linked to the vegetation growth by a positive feedback between
in ltration and biomass growth (as conjectured for the Field 21)
the gain in storage capacity achieved by the plant that develops
Fig. 14. Conceptual scheme of the model used in this paper.
Fig. 13. Maps of estimated moisture content
derived from EMI mapping on Field 21 in May
2009, 2010, and 2011. e ECa values have
been temperature corrected and converted into
moisture content using to the calibrated Eq. [1].
Note the clear e ect of wheat grown in 2010 in
the southern part of the eld. Satellite image
source: Google Earth.
www.VadoseZoneJ ournal.org
deep roots, does not represent an advantage for the plant itself as
roots grow deeper. Above a certain threshold H, b, and S become
less sensitive to further increase in root depth.
e “critical” root depth at which plant cover no longer a ects the
soil moisture balance and thus the water yield depends on plant
physiology and phenology (through ET) and on the soil vegeta-
tion interaction and feedback that alter the soil wettability and
conductivity (and thus I, and L). Modeled and observed (at the
bare plot of Field 23) “critical” root depth are in the same order of
magnitude of about 500 mm.
e occurrence of a water stress situation is associated to the occurrence
of a patchy vegetation cover (low S corresponds to low b). According to
a simple mass balance approach, water stress may be attributed to the
physiological high water demand (when ET
0
= 500 mm yr
−1
cases: c
and d (from Fig. 15 and Table 2), the e ective evapotranspiration cor-
responds to the literature data taken here as reference values due to the
low biomass density, but it is locally very high) or to a disadvantageous
partitioning of rainfall into runo due to the lack of soil wettability
where the vegetation does not grow (
0
0/1
() 0/0.8
b
Ib I
=
=
) or to
enhanced leakage when the root growth coincide to the establishment
of a fast connection of the soil surface with the water table (simulated
by setting n = 0.5 in cases a and c). e model outcome demonstrates
that (i) vegetation may a ect the soil structure and enhance leakage
and/or in ltration producing a negative or positive feedbacks, (ii) pref-
erential allocation of biomass below ground determines and increment
of soil storage capacity but does not always turn into an advantage for
biomass growth.
Table 2. Model function functionality, parameter de nition, parameter values and related references.
Model functions Parameters Parameter values References
b
() 1
B
GB rB
K
⎛⎞
⎟
⎜
⎟
= −
⎜
⎟
⎜
⎟
⎜
⎝⎠
r = speci c growth rate and
K
b
= carrying capacity
r = 1 yr
−1
K
b
= 5 kg m
−2
Verhulst (1838)
Lieth and Whittaker (1975)
1
1
−
=
+
S
MmB
S
m = mortality rate m = 1 yr
−1
b
B
IPg
K
⎛⎞
⎟
⎜
⎟
=
⎜
⎟
⎜
⎟
⎜
⎝⎠
()
0
1sin2=+πPP t
P
0
= average annual precipitation P
0
= I = 500 mm yr
−1
Agenzia Regionale per la Protezione
dell’Ambiente Sardegna (2011)
b
1
b
=
+
B
K
g
B
k
K
k
1
= feedback parameter k
1
= 0.25 HilleRisLambers et al. (2001)
Ursino (2007)
()
0b
ET ET 1 cos2 tBK=+π
ET
0
= average annual
evapotranspiration
ET
0
= 500 mm yr
−1
(c; d)
ET
0
= 200 mm yr
−1
(a; b)
Agenzia Regionale per la Protezione
dell’Ambiente Sardegna (2011)
Baudena et al. (2012)
L = K
s
S
n
K
s
= soil saturated conductivity
n = soil empirical parameter
K
s
= 300 mm yr
−1
n = 0.5 (a; c)
n = 5 (b; d)
Brooks and Corey (1964)
Kim et al. (1996)
Ursino (2009)
Fig. 15. Model predictions. Average biomass density and soil satu-
ration for di erent control volume depth H. Dashed line: soil
saturation; continuous line: dimensionless biomass density. Di erent
scenarios (concerning the soil-water-vegetation interaction) have been
simulated. Black lines: positive feedback between vegetation growth
and in ltration. Gray lines: wettable soil, and no feedback g = g(1).
Curves a and c: low storage capacity due to fast leakage through the
root zone toward the water table n = 0.5. Curves a and b: low evapo-
transpiration ET
0
= 200 mm yr
−1
, with high and low leakage n = 0.5
(a) and n = 5 (b), respectively. Curves c; d: high evapotranspiration
(leading to possible water stress) ET
0
= 500 mm yr
−1
with high and
low leakage n = 0.5 (c) and n = 5 (d), respectively.
www.VadoseZoneJ ournal.org
us, the “critical” root depth of about 500 mm (observed at the
bare plot of Field 23) may be the signature of vegetation adaptation
to climate and/or soil water scarcity.
e vegetation growth indirectly a ects locally the average soil
saturation and globally the connectivity between the soil moisture
paths. Increasing
H
due to adaptation or physiological charac-
terization turns into an advantage for the vegetation (and to an
increase in soil saturation) when no water stress occurs (case b)
or when the water stress may be contrasted by a positive feedback
between in ltration and biomass growth (cases a and d) despite
the formation of a soil surface sealing crust in the absence of veg-
etation and the either high ET (case d) or high L (case a). When
ET and L are both high (case c) deep rooted species are penalized
(case c). If the soil surface remains wettable and the conjectured
crust does not develop over the bare soil patches (gray lines with
no feedback
() (1) 0.8gb g==
), moderate and severe water scar-
city conditions (cases a and c) turns into more favorable scenarios,
leading at least to the conversion of a patchy vegetation cover into
a uniform vegetation cover (such as in case a) if the roots can grow
deep enough. e fact that the vegetation cover at Field 21 and 23
is di erent despite the soil texture is similar, may be attributed to
species physiology (including the di erent root depth that a ects
the annual water balance) and phenology and/or to the vegetation-
soil interaction that a ects the soil structure, the soil wettability
and the vertical connectivity between the soil surface and the deep
soil layers. is means that our tentative mechanistic explanation
for such di erent water dynamics at the observed sites may not
be unique. Corresponding to a di erent possible explanation, the
expected impact of the local soil vegetation interaction on global
soil moisture balance changes.
6
Conclusions
is paper presents the results of long-term monitoring and irri-
gation tests performed on the San Michele experimental farm in
Southern Sardinia, an area of semiarid climate where concerns
exist about future impacts of possible climatic changes on the
already stressed water balance. e collected data allowed us to
draw some interesting conclusions.
1.
Noninvasive techniques, and particularly soil mapping with
electromagnetics and γ-ray emission, provide data at the scale
and resolution necessary to understand the hydrological pro-
cesses of the topsoil, in their spatial variability. Unlike remote
sensing techniques, noninvasive geophysics penetrates the soil
subsurface and can e ectively image moisture content in the
rst meter or so, that is, in the active layer where moisture con-
tent and vegetation strongly interact.
2.
Careful calibration of these noninvasive techniques, in our
case on the basis of a controlled irrigation experiment, leads
to producing quantitative estimates of moisture content at the
scale and resolution needed by large-scale hydrological models.
3.
e evidence collected at the San Michele farm using nonin-
vasive techniques point toward a strong interaction between
vegetation and soil water dynamics, with important feed
backs between biota and water in ltration and ex ltration
from the soil.
4.
e collected evidence calls for a conceptual model capable
of representing the vegetation–soil interaction, and that has
simple enough parameterization that can be ful lled by mea-
surements of a noninvasive nature, available at a large scale. One
such key parameter is the thickness of the active layer.
In our experimental site we observed that the growth of veg-
etation, the associated below ground allocation of biomass and the
architecture of root have a signi cant impact on the soil moisture
dynamics. e presence of the roots in the vadose zone has been
observed to be crucial in the water balance due to the fact that the
below ground biomass a ects the soil structure and its response to
the hydrologic forcing. We reasonably conjectured that the vegeta-
tion that spontaneously grows in the eld that was le bare alters the
structure of the compact layer that seals the soil surface otherwise.
Indeed, the bare soil reacts slowly to rainfall and irrigation and its
reaction is con ned within the upper soil layers, despite the fact that
the vegetation grows deeper roots as compared to the neighbor crop.
By enhancing in ltration, the vegetation grows locally and pro ts
of the increased soil moisture availability and at the same time
exploits the scarce water resources creating the conditions that
limit its own growth. In many arid sites of the word the mosaic
vegetation cover (dense and patchy vegetated patterns within a bare
background) represents the reaction of the ecosystem to this com-
plex interplay between soil, biomass and hydrological constrain.
We observed a quite uniform, although scattered, vegetation cover
that we would attempt to explain on the base of the data collected
and the outcome of a minimal model.
Although encouraging, the numerical results presented here shall
be considered as preliminary investigations only. Indeed our
modeling approach is very simple, as it roughly describes the time
variability of the hydrological forcing, the soil characterization
and the plant physiology. e model ignores daily time scale pro-
cesses, and the di erent dynamics that characterizes the vegetation
growth above and below ground. Nevertheless, despite the many
simplifying hypothesis, it suggest that accounting for ecohydro-
logical feedback may provide an explanation for some physiological
aspect of vegetation growth and soil moisture dynamics.
Geophysics evidenced relevant di erences in soil moisture patterns
that may be linked to the vegetation growth although our tentative
mechanistic explanation for such di erences may not be unique.
What reason determines the strong di erences in observed soil
moisture patterns at di erent elds certainly deserved deeper
investigation to address the main water and land use management
issues concerning the water availability and the water quality.
Acknowledgments
www.VadoseZoneJ ournal.org
We acknowledge funding from the EU FP7 Collaborative Projects CLIMB (“Cli-
mate Induced Changes on the Hydrology of Mediterranean Basins– Reducing Un-
certainty and Quantifying Risk”) and iSOIL (“Linking geophysics, soil science and
digital soil mapping”).
References
Abdu, H., D.A. Robinson, M. Seyfried, and S.B. Jones. 2008. Geophysical imaging
of watershed subsurface pa erns and predic on of soil texture and water
holding capacity. Water Resour. Res. 44:W00D18 10.1029/2008WR007043.
doi:10.1029/2008WR007043
al Hagrey, S.A., and T. Petersen. 2011. Numerical and experimental mapping of
small root zones using op mized surface and borehole resis vity tomogra-
phy. Geophysics 76(2):G25–G35 10.1190/1.3545067.
Agenzia Regionale per la Protezione dell’Ambiente Sardegna. 2011. Analisi
Agro meteorologica e climatologica (O obre 2009–Se embre 2010). Ca-
gliari, Italy.
Baudena, M., I. Bevilacqua, D. Canone, S. Ferraris, M. Previa , and A. Provenza-
le. 2012. Soil water dynamics at a midla tude test site: Field measurements
and box modeling approaches. J. Hydrol. 414–415:329–340.
Bear, J. 1972. Dynamics of fl uids in porous media. American Elsevier, New York.
Binley, A. 2011. Free resis vity so ware. h p://www.es.lancs.ac.uk/people/
amb/Freeware/freeware.htm (accessed 4 Dec. 2011).
Binley, A., S. Henry-Poulter and B. Shaw. 1996. Examina on of solute transport
in an undisturbed soil column using electrical resistance tomography. Wa-
ter Resour. Res. 32(4): 763–769.</jrn>
Brooks, R.H., and A.T. Corey. 1964. Hydraulic proper es of porous media. Hydro.
Paper No. 3 Colorado State Univ., Fort Collins.
Brovelli, A., and G. Cassiani. 2011. Combined es ma on of eff ec ve electri-
cal conduc vity and permi vity for soil monitoring. Water Resour. Res.
47:W08510. doi:10.1029/2011WR010487
Cassiani, G., G. Boehm, A. Vesnaver, and R. Nicolich. 1998. A geosta s cal
framework for incorpora ng seismic tomography auxiliary data into hy-
draulic conduc vity es ma on. J. Hydrol. 206:58–74. doi:10.1016/S0022-
1694(98)00084-5.
Cassiani, G., V. Bruno, A. Villa, N. Fusi, A.M. Binley. 2006. A saline tracer test
monitored via me-lapse surface electrical resis vity tomography. J. Appl.
Geophys. 59:244–259.
Cassiani, G., A. Kemna, A. Villa, and E. Zimmermann. 2009. Spectral induced
polariza on for the characteriza on of free-phase hydrocarbon contamina-
on in sediments with low clay content. Near Surf. Geophys. 7:547–562.
doi:10.3997/1873-0604.2009028.
Cody, M.L. 1986. Roots in plant ecology. Trends Ecol. Evol. 1:76–78.
doi:10.1016/0169-5347(86)90022-4.
Comas, X., L. Slater, and A. Reeve. 2004. Geophysical evidence for peat basin
morphology and stra
graphic controls on vegeta on observed in a north-
ern peatland. J. Hydrol. 295:173–184. doi:10.1016/j.jhydrol.2004.03.008.
Corwin, D.L., and S.M. Lesch. 2005. Characterizing soil spa al variability with
apparent soil electrical conduc vity: I. Survey protocols. Comput. Electron.
Agric. 46:103–133. doi:10.1016/j.compag.2004.11.002.
Day-Lewis, F.D., K. Singha, and A.M. Binley. 2005. Applying petrophysical models
to radar travel me and electrical resis vity tomograms: Resolu on-depen-
dent limita ons. J. Geophys. Res. 110:B08206. doi:10.1029/2004JB003569.
Deiana R., G. Cassiani, A. Villa, A. Bagliani and V. Bruno. 2008. Model calibra-
on of a water injec on test in the vadose zone of the Po River plain using
GPR cross-hole data. Vadose Zone J. 7:215–226. doi:10.2136/vzj2006.0137.
Eagleson, P.S. 2002. Ecohydrology: Darwinian expression of vegeta on form
and func on. Cambridge Univ. Press, New York.
Flury, M., H. Flühler, W.A. Jury, and J. Leuenberger. 1994. Suscep bility of soils
to preferen al fl ow of water: A fi eld study. Water Resour. Res. 30(7):1945–
1954 doi:10.1029/94WR00871.
Guswa, A.J. 2008. The infl uence of climate on root depth: A carbon cost benefi t
analysis. Water Resour. Res. 44:W02427. doi:10.1029/2007WR006384.
Guswa, A.J. 2010. Eff ect of plant uptake strategy on the water−op mal root
depth. Water Resour. Res. 46:W09601. doi:10.1029/2010WR009122.
HilleRisLambers, R., M. Rietkerk, F. van den Bosch, H.H.T. Prins, and H. de Kroon.
2001. Vegeta on pa ern forma on in semi-arid grazing systems. Ecology
82:50–61. doi:10.1890/0012-9658(2001)082[0050:VPFISA]2.0.CO;2
Huxman, T.E., B.P. Wilcox, D.D. Breshears, R.L. Sco , K.A. Snyder, E.E. Small, K.
Hul ne, W.T. Pockman, and R.B. Jackson. 2005. Ecohydrological implica on
of woody plant encroachment. Ecology 86:308–319. doi:10.1890/03-0583.
Jackson, R.B., E.G. Jobbagy, and M.D. Nose o. 2009. Ecohydrology bearings—
Invited commentary. Ecohydrology in a human-dominated landscape. Eco-
hydrology 2:383–389. doi:10.1002/eco.81.
Javaux, M., T. Schroder, J. Vanderborght, and H. Vereecken. 2008. Use of a
three-dimensional detailed modeling approach for predic ng root water
uptake. Vadose Zone J. 7:1079–1088. doi:10.2136/vzj2007.0115.
Jung, W.K., N.R. Kitchen, K.A. Sudduth, R.J. Kremer, and P.P. Motavalli. 2005. Re-
la onship of apparent soil electrical conduc vity to claypan soil proper es.
Soil Sci. Soc. Am. J. 69:883–892. doi:10.2136/sssaj2004.0202.
Kachanoski, R.G., E.G. Gregorich, and I.J. Van Wesenbeeck. 1988. Es ma ng spa-
al varia ons of soil water content using non-contac ng electromagne c in-
duc ve methods. Can. J. Soil Sci. 68:715–722. doi:10.4141/cjss88-069.
Kim, C.P., J.N.M. Stricker, and P.J.J.F. Torfs. 1996. An analy cal framework for
the water budget of the unsaturated zone. Water Resour. Res. 32(12):3475–
3484. doi:10.1029/95WR02667.
Lieth, H., and R.H. Whi aker. 1975. Primary produc vity of the biosphere. Ecol.
Stud., Vol. 14. Springer-Verlag, New York.
Ludwig, R., A. Soddu, R. Du mann, N. Baghdadi, S. Benabdallah, R. Deidda, et
al. 2010. Climate-induced changes on the hydrology of Mediterranean ba-
sins—A research concept to reduce uncertainty and quan fy risk. Fresenius
Environ. Bull. 19(10A):2379–2384.
Ma, R., A. McBratney, B. Whelan, B. Minasny, and M. Short. 2011. Comparing
temperature correc on models for soil electrical conduc vity measure-
ment. Precis. Agric. 12:55–66. doi:10.1007/s11119-009-9156-7.
Michot, D., Y. Benderi er, A. Dorigny, B. Nicoullaud, D. King, and A. Tabbagh.
2003. Spa al and temporal monitoring of soil water content with an irri-
gated corn crop cover using surface electrical resis vity tomography. Water
Resour. Res. 39(5):1138. doi:10.1029/2002WR001581.
Milly, P.C.D. 1993. An analy c solu on of the stochas c storage prob-
lem applicable to soil water. Water Resour. Res. 29(11):3755–3758
doi:10.1029/93WR01934.
Monego, M., G. Cassiani, R. Deiana, M. Pu , G. Passadore, and L. Al ssimo.
2010. Tracer test in a shallow heterogeneous aquifer monitored via me-
lapse surface ERT. Geophysics 75(4):WA61–WA73. doi:10.1190/1.3474601.
Newman, B.D., B.P. Wilcox, S.R. Archer, D.D. Breshears, C.N. Dahm, C.J. Duff y,
et al. 2006. Ecohydrology of water-limited environments: A scien fi c vision.
Water Resour. Res. 42:W06302. doi:10.1029/2005WR004141.
Nu le, W.K. 2002. Ecohydrology’s past and future in focus. Eos Trans. AGU
83:205. doi:10.1029/2002EO000138.
Porporato, A., E. Daly, and I. Rodriguez-Iturbe. 2004. Soil water balance
and ecosystem response to climate change. Am. Nat. 164:625–632.
doi:10.1086/424970.
Richards, L.A. 1931. Capillary conduc on of liquids through porous mediums.
Physics 1(5):318–333. doi:10.1063/1.1745010.
Rietkerk, M., M.C. Boerlijst, F. van Langevelde, R. HilleRisLambers, J. van de
Koppel, and L. Kumar. 2002. Self-organiza on of vegeta on in arid ecosys-
tem. Am. Nat. 160:524–530. doi:10.1086/342078.
Robinson, D.A., A. Binley, N. Crook, F. Day-Lewis, P.T. Ferré, V.J.S. Grauch, et
al. 2008. Advancing process-based watershed hydrological research using
near-surface geophysics: A vision for, and review of, electrical and mag-
ne c geophysical methods. Hydrol. Processes 22:3604–3635. doi:10.1002/
hyp.6963.
Robinson, D.A., I. Lebron, B. Kocar, K. Phan, M. Sampson, N. Crook, et al. 2009.
Time-lapse geophysical imaging of soil moisture dynamics in tropical del-
taic soils: An aid to interpre ng hydrological and geochemical processes.
Water Resour. Res. 45:W00D32. doi:10.1029/2008WR006984
Sheets, K.R., and J.M.H. Hendrickx. 1995. Noninvasive soil water content mea-
surement using electromagne c induc on. Water Resour. Res. 31:2401–
2409. doi:10.1029/95WR01949.
Singha, K., and S.M. Gorelick. 2005. Saline tracer visualized with three-dimen-
sional electrical resis vity tomography: Field-scale spa al moment analysis.
Water Resour. Res. 41:W05023. doi:10.1029/2004WR003460.
Strobbia, C., and G. Cassiani. 2007. Mul -layer GPR guided waves in shallow
soil layers for the es ma on of soil water content. Geophysics 72(4):J17–
J29. doi:10.1190/1.2716374.
Triantafi lis, J., and S.M. Lesch. 2005. Mapping clay content varia on using elec-
tromagne c induc on techniques. Comput. Electron. Agric. 46:203–237.
doi:10.1016/j.compag.2004.11.006.
Ursino, N. 2005. The infl uence of soil proper es on the forma on of unstable
vegeta on pa erns on hillsides of semiarid catchments. Adv. Water Resour.
28:956–963. doi:10.1016/j.advwatres.2005.02.009
Ursino, N. 2007. Modeling banded vegeta on pa erns in semiarid regions:
Interdependence between biomass growth rate and relevant hydrological
processes. Water Resour. Res. 43:W04412. doi:10.1029/2006WR005292.
Ursino, N. 2009. Above and below ground biomass pa erns in arid lands. Ecol.
Modell. 220:1411–1418. doi:10.1016/j.ecolmodel.2009.02.023.
Vereecken, H., A. Binley, G. Cassiani, I. Kharkhordin, A. Revil, and K. Titov (eds).
2006. Applied hydrogeophysics. Springer-Verlag, Berlin.
Verhulst, P.F. 1838. No ce sur la loi que la popula onsuit dans son accroisse-
ment. Corr. Math. Phys. 10:113–117.
Waxman, M.H., and L.J.M. Smits. 1968. Electrical conduc vi es in oil-bearing
shaly sands. Soc. Pet. Eng. J. 8:107–122.
Werban, U., S.A. al Hagrey, and W. Rabbel. 2008. Monitoring of root-zone water
content in the laboratory by 2D geoelectrical tomography. J. Plant Nutr. Soil
Sci. 171(6):927–935. doi:10.1002/jpln.200700145.
Werban, U., T. Behrens, G. Cassiani, and P. Dietrich. 2010. iSOIL: An EU
project to integrate geophysics, digital soil mapping, and soil science.
In: R.A. Viscarra Rossel, A. McBratney, and B. Minasny, editors, Proximal
soil sensing. Progress in Soil Sci. 1. Springer, Berlin. p. 103–110.
doi:10.1007/978-90-481-8859-8_8.
