International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
Contents lists available at ScienceDirect
International Journal of Applied Earth Observation and
Geoinformation
journal homepage: www.elsevier.com/locate/jag
Investigating the backscatter contrast anomaly in synthetic aperture
radar (SAR) imagery of the dunes along the Israel–Egypt border
Offer Rozenstein a,∗ , Zehava Siegal b , Dan G. Blumberg c , Jan Adamowski a
a
b
c
Department of Bioresource Engineering, McGill University, Macdonald Campus 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Quebec H9X 3V9, Canada
Department of Life Sciences, Ben-Gurion University of the Negev, Negev, 84105, Israel
Department of Geography and Environmental Development, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel
a r t i c l e
i n f o
Article history:
Received 1 November 2015
Received in revised form
12 November 2015
Accepted 15 November 2015
Available online 1 December 2015
Keywords:
Remote sensing
Synthetic aperture radar
Aerial imagery
Surface roughness
Spatial pattern analysis
Soil dielectric permittivity
Land use
a b s t r a c t
The dune field intersected by the Israel–Egypt borderline has attracted many remote sensing studies over
the years because it exhibits unique optical phenomena in several domains, from the visual to the thermal
infrared. These phenomena are the result of land-use policies implemented by the two countries, which
have differing effects on the two ecosystems. This study explores the surface properties that affect radar
backscatter, namely the surface roughness and dielectric properties, in order to determine the cause for
the variation across the border. The backscatter contrast was demonstrated for SIR-C, the first synthetic
aperture radar (SAR) sensor to capture this phenomenon, as well as ASAR imagery that coincides with
complementary ground observations. These field observations along the border, together with an aerial
image from the same year as the SIR-C acquisition were used to analyze differences in vegetation patterns
that can affect the surface roughness. The dielectric permittivity of two kinds of topsoil (sand, biocrust)
was measured in the field and in the laboratory. The results suggest that the vegetation structure and
spatial distribution differ between the two sides of the border in a manner that is consistent with the
radar observations. The dielectric permittivity of sand and biocrust was found to be similar, although they
are not constant across the radar spectral region (50 MHz–20 GHz). These findings support the hypothesis
that changes to the vegetation, as a consequence of the different land-use practices in Israel and Egypt,
are the cause for the radar backscatter contrast across the border.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
Remote sensing imagery is one of the most powerful tools for
studying the surface of the Earth because it is an efficient way to
study vast areas, and areas to which access is restricted or difficult.
Thus, the mosaic of land use and land cover that make up the different ecosystems on the Earth’s surface can be mapped using remote
sensing imagery (Rozenstein and Karnieli, 2011; Levin, 2015).
An interesting example for such use of remote sensing can be
found in the continuous study of the dune field intersected by the
Israel–Egypt border (Fig. 1A) (e.g., Otterman, 1974; Schmidt and
Karnieli, 2000; Roskin et al., 2012). While ground access to this
study site has been restricted at times due to geopolitical circumstances, this area attracted many remote sensing scientists due
to a contrast along the two sides of the political border. The first
remote sensing study of the area showed a difference in spec-
∗ Corresponding author.
E-mail address: [email protected] (O. Rozenstein).
http://dx.doi.org/10.1016/j.jag.2015.11.008
0303-2434/© 2015 Elsevier B.V. All rights reserved.
tral reflectance between the relatively dark Negev dunes in Israel,
and the very bright Sinai dunes in Egypt, based on a Landsat-MSS
image from 1972 (Otterman, 1974). This phenomena was considered to be the result of overgrazing of the vegetation in Sinai, and
simultaneous recovery of the vegetation in the Negev (Otterman,
1981). However, two decades after the origin of this interpretation,
it was rejected, and an alternative explanation was postulated; the
lower reflectance in the Negev was determined to be associated
with the wide-spread cover of the surface by cyanobacteria dominated biocrusts rather than with higher vegetation, which was
considered too sparse to cause this effect on brightness (Karnieli
and Tsoar, 1995; Tsoar and Karnieli, 1996). The high reflectance of
the Sinai dunes was attributed to the absence of biocrusts due to
continuous trampling of the surface by pastoralist activities, preventing biocrust establishment (Meir and Tsoar, 1996). Hence, the
high reflectance in Sinai was associated with the bright sands of the
relatively bare dunes (Karnieli and Tsoar, 1995; Tsoar and Karnieli,
1996). Moreover, a thermal variation across the border occurs and
is also attributed to the same land-use and land-cover differences
(Qin et al., 2001; Qin et al., 2002). The relatively dark biocrusts
14
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
absorb more solar irradiance than the bright sands, which results
in higher land surface temperatures in the Negev than in Sinai during daytime. In addition to the land surface temperature contrast
between exposed and biocrusted dunes, it was shown that because
biocrusts are characterized by higher thermal emissivity (Qin et al.,
2005), the land surface emissivity of the Negev dunes is higher than
the Sinai dunes (Rozenstein and Karnieli, 2015). Recently, spaceborne observations of the fluctuations in the surface emissivity as
a result of water vapor adsorption by the topsoil at night, and their
subsequent evaporation during the day were also shown to differ
between the two environments (Rozenstein et al., 2015).
So far, the exploration of the anthropogenic anomaly created by
the contrasting land use practices in Sinai and the Negev utilized
different spectral regions along the optical spectrum. A backscatter
contrast in synthetic aperture radar (SAR) imagery of the border
was previously suspected to be caused by differences in vegetation
across the border, but this was not established (Blumberg, 1998).
Considering that radar backscatter is affected by very different surface properties than those affecting optical remote sensing images,
the contrast appearing in SAR images points to additional differences between the processes occurring in Sinai and the Negev.
Radar backscatter is mainly influenced by the surface topography, roughness, and dielectric properties, in combination with the
beam incidence angle, wavelength, and polarization (Blumberg and
Greeley, 1993). Yet, the topography of the linear dunes does not
change significantly across the border, and therefore it is not the
cause for the differences in backscatter between Sinai and the
Negev. In light of this, the aim of this paper is to explore the factors
affecting surface roughness (e.g., the vegetation spatial patterns,
and composition) and dielectric properties of the dune surface to
determine the cause for variation in radar backscatter across the
border.
2. Methodology
2.1. Study area
The study area in the Negev–Sinai erg (sand sea) is characterized
by linear dunes arbitrarily intersected by the Israel–Egypt political
border. This borderline that runs in an almost straight line from
the Mediterranean to the Red Sea is based on the 1906 agreement between the British and Ottoman empires, which later on
laid the foundation for a number of agreements between Israel
and Egypt. In spite of their appearance from space, the dunes on
both sides of the border are of the same pedological unit (Roskin
et al., 2011). Since this border was last demarked in 1982 following
the peace agreements between Israel and Egypt, this has restricted
pastoralist activities to Sinai, as conservation policies were
Fig. 1. (A) Location of the arid dunes study site on the Israel–Egypt border; (B) a subset of the SIR-C synthetic aperture radar C-HH band ( = 5.7 cm) over the southern end of
northwestern Negev dunes. The radar backscatter from the Israeli side of the border is higher than from the Egyptian side; (C) a panchromatic aerial image over the southern
end of the dunefield. The vegetation patterns in the areas marked by rectangles on both sides of the border were analyzed and compared.
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
15
Fig. 2. (A) Backscatter in the Negev and in Sinai as measured by SIR-C on 3 October 1994; (B) daily precipitation distribution throughout the winter season of 1993–1994
in Nizzana, Israel, at the southern end of the dune field (Kidron and Yair, 1997). Precipitation measurements show that the next rainy season did not start until October 8,
1994, a few days after the SAR image acquisition, and therefore the soil water content is assumed to be minimal during the space shuttle overpass; (C) backscatter in the
Negev and in Sinai in 15 Envisat-ASAR co-polarized images (top), and daily precipitation for January 2011 to April 2012 in Kadesh-Barnea, Israel (bottom); the backscatter
coefficients in A and C are calculated for equal area polygons on each side of the border.
enforced by Israel in the Negev dunes (Meir and Tsoar, 1996; Tsoar
et al., 2008). The minimal disturbance of the Negev surface during these decades resulted in a stabilization process of the dunes
(Karnieli and Tsoar, 1995) due to the development of biocrusts composed of cyanobacteria, lichens, mosses, green algae, microfungi,
and bacteria (Belnap and Lange, 2001; Pointing and Belnap, 2012).
In parallel to the development of biocrusts, the perennial vegetation cover in the Negev has increased compared to Sinai (Qin et al.,
2006; Seifan, 2009). The rainfall in the area is sporadic and occurs
mainly during the winter season. Since the mid 1990’s until 2009,
mortality of perennials was recorded in the Negev due to a prolonged drought (Siegal et al., 2013); the corresponding changes in
Sinai were not studied. The biocrusted dune surface contain significantly larger proportions of fine clay and silt particles compared
to the underlying sand (Danin and Ganor, 1991; Rozenstein et al.,
2014), and as a result they differ in mineralogical composition. The
sand moisture in the area is very low, between 0 and 2% (Roskin
et al., 2012).
2.2. SAR imagery
Synthetic aperture radar (SAR) imagery from the third shuttle imaging radar (SIR-C) deployed from the space shuttle on 3
October 1994 was the first radar images to clearly show a contrast
in backscatter across the border (Fig. 1B). The SIR-C instrument
was unique since it acquired data for both C-band (5.7 cm) and
L-band (24 cm) in quad-polarization configuration (Evans et al.,
1997). As each frequency-polarization combination is sensitive to
different landscape features, the availability of a multi-band, multipolarization SAR data is advantageous for interpretation of different
ground features (Hess et al., 1995). In addition, a series of 15
co-polarized C-band (5.6 cm) images acquired by the advanced synthetic aperture radar (ASAR) on-board Envisat during 2011–2012
were used to demonstrate the persistence of the radar backscatter
contrast across the border. The images were calibrated, terrain corrected and co-registered. The backscatter from equal area polygons
of 5.5 km on each side of the border was compared.
2.3. Dielectric permittivity measurements
2.3.1. Dielectric permittivity measurements in the field
The temperature corrected real dielectric permittivity of sand
and crust was measured in the field on the afternoon of 24 August
2013 using a POGO Portable Soil Sensor (Stevens Water Monitoring Systems Inc., Portland, Oregon, USA) with an absolute accuracy
of 0.2. This sensor operates at a single frequency of 50 MHz.
16
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
Two modes of measurement were performed: (1) the sensor was
inserted directly into the dune crest, where sand is exposed at the
surface, to the encrusted surface of the dune slopes, and to the same
spots on the slope following the removal of the biocrusts covering
the top few mm of the dune surface; (2) sand from the dune crest
and crumbled biocrusts from the dune slopes were measured in a
1 L glass jar.
2.3.2. Spectral dielectric permittivity measurements
The same sediments that were measured in the field were oven
dried at 105 ◦ C for two days. One kilogram of each sediment was
measured in a glass jar using a commercial open end coaxial probe
kit (Agilent, Model 85070E, USA) connected to a network analyzer
(Hewlett Packard, Model 8720C, USA). The dielectric permittivity
was measured at 1600 points along the entire frequency range of
the Network Analyzer (50 MHz–20 GHz). Following that, 10 mg of
distilled water were added to the sediments to constitute an addition of 1% of the weight, and the measurement was repeated.
2.4. Aerial image
2.4.1. Pre-processing
To evaluate the state of the vegetation across the Israel–Egypt
border at the time of the SIR-C SAR mission, we used an aerial image
from the same year (1994). The negative of the panchromatic aerial
image area taken over the research site on 30 April 1994 at 3pm
was scanned at the highest resolution available (10 ␮m). The image
was geometrically rectified, which resulted in a pixel size of about
29 cm. A subset of the image, encompassing sections of four linear
dunes on both sides of the border was chosen for further analysis of
vegetation patterns (Fig. 1C). In order to remove non-uniform illumination effects, a cross-track illumination correction was applied
using ENVI software (ITT Visual Information Solutions, Boulder, Colorado, USA), but no topographic correction was applied because it
was not shown to improve the classification accuracy (as in Carmel
and Kadmon, 1998).
2.4.2. Binary vegetation classification
Vegetation patches appear as dark spots in the image. Therefore, an iterative-threshold segmentation approach was applied to
classify the vegetation, whereby the image is segmented into areas
of connected pixels by iterating the maximal DN (digital number)
value considered as vegetation. The minimum size for each segment was defined as two pixels, and connectivity was considered
for 8 neighboring pixels. Each of the segmentation products was
examined by two independent expert image interpreters who are
intimately familiar with the research area. The consensus between
the interpreters was that a maximum DN threshold of 130 best
separates vegetation patches from the surrounding background.
Two subsets of equal areas of 744,356 m2 were defined on each
side of the border (Fig. 1C). In order to evaluate the vegetation
classification, an accuracy assessment was performed for a set of
1000 randomly sampled points, 500 on each side of the border.
The land cover at these validation points was visually interpreted
and each point was assigned as “vegetation” or “non-vegetation”.
A confusion matrix was created, and overall accuracy, Kappa
statistic, the user’s accuracy, and producer’s accuracy were calculated (Congalton and Green, 2008).
2.4.3. Spatial statistical analysis
The FRAGSTATS spatial pattern analysis program (McGarigal
et al., 2012) was used to analyze differences in vegetation patches
between the Negev and Sinai subsets. First, the vegetation cover
fraction was calculated as the ratio between the total vegetation
cover to the total landscape area. Next, the patch density was determined as the ratio between the number of patches and the total
landscape area; it is a basic metric for landscape pattern and subdivision. However, it has been criticized for its insensitivity and
inconsistent behavior across a wide range of subdivision patterns
(Jaeger, 2000). Therefore, several other measures were used to
complement it and provide a comprehensive description of differences in spatial patterns of vegetation across the border. One such
indicator is the patch cohesion index (Schumaker, 1996), which is
proportional to the standardized patch perimeter-area ratio and
ranges from 0 to 1. Two additional indices that quantify the degree
of fragmentation were computed from a patch adjacency matrix,
which shows the frequency with which different pairs of patch
types appear side-by-side on the map: the aggregation index (He
et al., 2000) and the clumpiness index (McGarigal et al., 2012).
Additionally, vegetation pixels were converted into polygons
and an average nearest neighbor analysis was performed for each
side of the border (Clark and Evans, 1954; Scott and Janikas, 2010)
using ArcGIS 10.2.2 (ESRI, Redlands, California, USA). This method is
based on the distance between each polygon feature and its closest
neighbor. To overcome a possible bias of the average statistic, the
entire distribution of nearest neighbor distances was examined in
comparison to a hypothetical expected distances distribution of an
equal number of randomly distributed patches over an equal area.
A Z-test was conducted to determine whether the observed nearest neighbor distance distribution is different from the expected
distribution. A nearest neighbor ratio was calculated between the
observed average distance and the expected average distance.
When this ratio is less than 1, the pattern exhibits clustering, and
when it is greater than 1, the distribution leans toward a dispersed
pattern (Mitchell, 2005). All of the metrics described thus far use
only one value to represent each area of interest (i.e., a subset of
Sinai or the Negev). In order to plot the differences between the
two areas of interest across the border in the form of a spatial map,
the vegetation polygons were converted into points representing
their centroids and the density of points was calculated for a circle
of the default size suggested by the program (25.27 m in radius)
around every point.
2.5. Ground photography
In the absence of ground observations concurrent with the space
and aerial coverages of 1994, more recent ground reconnaissance
from the same year as the ASAR imagery were used to provide
insight into differences in the vegetation communities between the
two ecosystems. In addition to repeated visits to the field, digital
still images from high observation points along the Israel–Egypt
border were photographed around noon on 3 February 2011 using
a 200 mm Nikon lens. On that day, visibility was excellent, biocrusts
were green and annuals had started to emerge. These conditions
made it easier to interpret which types of vegetation appear in the
photos. At each location, the camera was pointed toward both Sinai
and the Negev. The interpretation of these images emphasized the
differences in vegetation composition between the two sides of
the border. Since specific species were identified with high certainty mostly for larger plants, the analysis focused on describing
the prevalence of perennial shrubs, annual herbaceous vegetation
and biocrusts in different habitats of the dune (i.e., crest, slopes,
and interdune areas).
3. Results and discussion
3.1. SAR backscatter
The Negev only appears brighter than Sinai in the SIR-C copolarized C-band image (C-HH), while the cross-polarized C-band
image (C-HV) and the L-band images show a similar backscatter
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
across the border (Fig. 2A). The VV polarization was not available
for this image. No significant rain events occurred prior to the SIR-C
image acquisition, and therefore the soil water content is assumed
to be minimal during the space shuttle overpass. The same contrast as the SIR-C co-polarized image is consistently demonstrated
by the ASAR time-series, in a manner independent of the time of
year and proximity to rain events (Fig. 2C).
17
Table 1
Confusion matrix and derived accuracy measures for the vegetation classification in
Sinai and the Negev.
Vegetation
Non-vegetation
Sum
Producer’s accuracy
Vegetation
Non-vegetation
Sum
User’s accuracy
116
7
123
94.31%
4
873
877
99.54%
120
880
1000
96.67%
99.20%
Overall classification accuracy = 98.9%; overall kappa statistic = 0.95.
3.2. Dielectric permittivity of sand and biocrust
The results of in-situ dielectric permittivity on a typical summer
day are presented in Fig. 3. No differences were observed between
the dielectric permittivity of exposed sand and biocrusted sands
(Fig. 3A). This may occur since the biocrusts are only a few mm
thick. However, even when crumbled biocrusts were amassed and
compared to sand, only a minute difference in their dielectric permittivity was found (3.2 for sand and 3.31 for biocrusts as shown
in Fig. 3B). The typical dielectric permittivity of water is 80 and it
ranges from 3 to 8 for non-vegetated dry soil or rock (Blumberg
and Greeley, 1993). While higher surface dielectric permittivity is
usually associated with higher backscatter, it appears that dielectric permittivity differences between the Sinai and Negev dunes
are too small to likely be the cause for the contrast that appears
in SIR-C imagery. This conclusion is in agreement with that of a
previous study (Blumberg, 1998), which used a laboratory time
domain reflectometer to measure sand and biocrust. While the TDR
frequency used in that study is not specified, our in-situ measurements are performed in 50 MHz, and C-band radar frequency is at
about 5.3 GHz. In order to determine the potential for backscatter
differences in different bands, spectral measurements of dielectric
properties for sand and biocrusts was performed (Fig. 4). The results
of this measurement showed that the dielectric permittivity is not
constant throughout this spectral range. The dielectric permittivity
of both sand and biocrust is higher in L-band compared to C-band.
However, the differences between sand and biocrust are very small.
The addition of 1% gravimetric moisture to the samples resulted in
a similar increase in the dielectric permittivity of both sand and
biocrust. Therefore, the spectral measurement of dielectric permittivity reaffirms the conclusion that this property is probably not
the cause for the spectral backscatter differences across the border.
This type of spectral measurement, however, can be useful to other
cases, where the dielectric properties of the surface in different
bands are of interest.
3.3. Vegetation patterns across the border
Our 2011 ground observations along the border showed
markedly different vegetation patterns in Sinai and the Negev
(Fig. 5). According to these observations, in Sinai, the dune crests
were characterized by active sands, and much of it was bare,
unvegetated sands. The Negev dune crests were partially active;
however, they presented more vegetated patches than in Sinai.
Biocrusts appeared in Sinai in small, non-continuous patches, on
the lower dune slopes and interdune areas, while in the Negev
they covered vast, continuous surfaces, even on some dune crests
(in the northern parts of the dune field). Annual herbaceous vegetation and plant litter appeared mostly on crusted surfaces, and
was therefore more common in the Negev. These findings are
in agreement with observations made in previous studies: Dune
stability in the Negev positively influences annual plant germination and growth on stabilized sands by preventing root exposure
and seedling burial by sand movements, as well as increasing soil
nutrients and fertility through carbon and nitrogen fixation by the
biocrusts (Kidron, 2014). While annual herbaceous vegetation generally tends to occupy encrusted surfaces, perennial shrubs tend
to prefer the non-crusted semi-stable dune crest (Li et al., 2002;
Kidron, 2015). It appears that grazing in Sinai reduces the cover of
annuals not only by predation, but also by trampling the biocrusts,
and thus decreasing the area of the habitat where annuals are typically found. The distribution of perennial shrubs in the Negev was
relatively homogenous, while in Sinai it had a patchier appearance.
Furthermore, large Calligonum comosum L’Her shrubs and Tamarix
aphylla (L.) Karsten trees were concentrated in patches on the dune
crests only in Sinai. Therefore, naked-eye and telescopic observations on the ground give the impression that the spatial vegetation
patterns are notably different across the border.
These ground observations are supported by quantitative analysis of the 1994 aerial photograph. The classification of vegetation
patches was found to be 98.9% accurate. The confusion matrix and
additional accuracy measures are presented in Table 1. Vegetation
patch metrics are presented in Table 2. There were more vegetation patches in the Negev than in Sinai. Additionally, vegetation in
the Negev covered more area, and the patch density was greater
than in Sinai. The vegetation cover was estimated at 16% for the
Negev, which is higher than the 7% cover previously found for the
same area in 1994 (Siegal et al., 2013). However, since that study
only included perennial shrubs in their analysis, and our analysis included patches that contain both annuals and perennials, and
may contain herbaceous vegetation and dry vegetative material, it
makes sense for our estimation of vegetation cover to be higher.
Additionally, clumpiness index, cohesion index, and aggregation
index were all higher in the Negev, supporting the field observations of greater subdivision and less physical connection in Sinai.
The nearest neighbor analysis (Table 3) revealed similar results, as
Table 2
Vegetation patch metrics.
Number of patches
Negev
Sinai
39,712
27,906
Vegetation cover (%)
16
7
Patch density [patches/hectare]
Clumpiness index
Cohesion index
Aggregation index
532.95
374.51
0.76
0.73
93.1
88.79
79.82
74.6
Table 3
Results of the nearest neighbor analysis.
Negev
Sinai
Observed mean distance
Expected mean distance
Nearest neighbor ratio
z Score
2.154
2.063
1.963
2.302
1.098 (disperesed)
0.896 (clustered)
41.042
−37.237
18
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
Fig. 3. Real dielectric permittivity measurements: (A) dune crest, where sand is exposed at the surface (sand), encrusted surface of the dune slopes (crust), and the dune
slopes following the removal of the biocrusts covering the top few mm of the dune surface (sand under crust); (B) sediment in a 1 L glass jar filled with either sand from the
dune crest (sand), or crumbled biocrusts from the dune slopes (crust).
the nearest neighbor ratio was found to be 1.098 for the Negev
and 0.896 for Sinai. The deviation of these values from 1 (random
distribution) suggests that the vegetation pattern exhibits dispersion in the Negev, and clustering in Sinai (Mitchell, 2005). These
differences in spatial patterns are presented in Fig. 6, which demonstrates the greater spatial variability of vegetation density in Sinai
compared to the Negev. Therefore, it appears that the vegetation
structure and pattern differ significantly between Sinai and the
Negev and that this variation changes the surface roughness. These
differences in surface roughness are imaged as a radar backscatter
contrast between the two ecosystems.
The intensity of the returned signal is sensitive primarily to
the roughness on a scale comparable to the radar wavelength
(Blumberg and Greeley, 1993). When vegetation is present at the
surface, its effect on backscatter varies as a function of wavelength
(Das and Paul, 2015). Senescent vegetation and dry plant litter can
also influence the backscatter (Li and Guo, 2015). The vegetation in
the research area is relatively low lying, and therefore, it is reasonable for the contrast in backscatter to appear in the C-band
( = 5.7 cm) that interacts with small branches, as opposed to the
L-band ( = 24 cm), which is known to interact with larger vegetation components, such as tree trunks (Mougin et al., 1999). Hence,
if the surface roughness estimation from radar data is desired (e.g.,
Genis et al., 2013), the presence of vegetation at the surface must
also be accounted for. Contrary to previous studies that found that
cross-polarized C-band data best discriminates between active and
fixed dunes as a result of multiple scattering from vegetation (Blom
and Elachi, 1981; Lancaster et al., 1992), in the SIR-C case the copolarized C-band data showed a greater contrast across the border.
To date, studies of the ecological change to the dune system
were mostly conducted in the Negev ecosystem (Breckle et al.,
2008), and comparative field studies between both sides of the
Fig. 4. Laboratory real dielectric permittivity measurements across the radar spectral region.
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
19
Fig. 5. Four pairs of images taken on 3 February 2011, representing views toward Sinai (1) and toward the Negev (2) at 4 different locations along the border, from the south
(A) to the center of the dune field (D) in leaps of 3–4 km. The rainfall gradient increases from south to north, and so does the vegetation cover, which increases from south
to north in the Negev. Grazing herbivore’s tracks are seen on a Sinai dune slope in A1. Grazing changes the vegetation patterns by reducing the prevalence of biocrusts and
annual herbaceous vegetation, and affects the spatial distribution of perennial shrubs. Large shrubs of Calligonum comosum L’Her only appear on the dune crests in Sinai, as
well as Tamarix aphylla (L.) Karsten trees, which were most likely planted.
border were not carried out, except by remote sensing tools (e.g.,
Qin et al., 2006; Seifan, 2009; Siegal et al., 2013; Rozenstein et al.,
2015; Rozenstein and Karnieli, 2015). Some of these remote sensing observations show a higher proportion in perennial vegetation
and microphytic cover in the Negev compared to the Sinai, but
neglect to note the differences in spatial vegetation patterns (e.g.,
Qin et al., 2006; Seifan, 2009). This aspect was only emphasized
in the current study, following the analysis of the SAR observations. Unlike coarse resolution optical remote sensing imagery that
are not sensitive to the signal of the higher vegetation in this area
(Karnieli and Tsoar, 1995), the SAR imagery is affected by the spatial patterns of this vegetation. The utilization of radar imagery
complements those studies, as it emphasizes the differences in the
spatial arrangement of the vegetation. Therefore it is shown that
combining multiple observations, using different sensors, which
are sensitive to different regions of the electromagnetic spectrum,
can produce new insight into old observations. However, extracting the vegetation height from an aerial image is difficult, especially
for low lying bushes. Future studies might benefit from the incorporation of Light Detection and Ranging (LiDAR) data over the area,
which may provide the means to assess the surface roughness
quantitatively (Sankey et al., 2010). LiDAR data, possibly in combination with aerial imagery, is expected to provide not only the
spatial distribution of the vegetation, but also its height, and the
dune topography (Schenk and Csathó, 2002). Such surface measurements may be desirable for modeling Aeolian erosion and
deposition (Wang et al., 2015).
20
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
Fig. 6. Vegetation density maps based on the areas marked by rectangles on the aerial image in Fig. 1A. The density variation in Sinai is greater than in the Negev. High
density values in Sinai mostly appear around patches of large shrubs and trees that are absent from the Negev landscape.
While our results support the consensus in the literature regarding the degradation of the Sinai ecosystem due to grazing (e.g.,
Otterman, 1974; Karnieli and Tsoar, 1995; Harrison et al., 1998),
this degradation is only demonstrated by a decrease in biocrusts
and vegetation cover, and by the Aeolian activity in the dunes.
Since no detailed flora and fauna surveys have been carried out
in the Sinai dunes, there is no information on biodiversity such as
species richness and composition for that ecosystem. The resilience
of the vegetation to anthropogenic pressures and its recovery
following their removal, suggest that pessimistic views of desertification and over use are not justified in this case, since the
degradation in this ecosystem is reversible (Meir and Tsoar, 1996;
Kidron et al., 2008). In fact, it was shown that moderate grazing
in the Negev dunes increased vegetation species richness (OlsvigWhittaker et al., 1993). Moreover, it was recently shown that
when disturbance to biocrusts occur, and dune mobility is restored,
the faunal communities’ response depends on the affinity of specific species to shifting sands (Columbus et al., 2012). Since some
rare and endangered species (e.g., Cerastes cerastes) seem to prefer the disturbed sands, it seems that in terms of conservation,
some disturbance may have a positive influence on faunal biodiversity, through maintaining habitat diversity. Therefore, conservation
efforts should focus on ensuring representation of all niches (active,
semi-stabilized, and stabilized sands) in the landscape, rather than
allowing for the extremes: uncontrolled intensive grazing on the
one hand, and complete removal of disturbance agents (e.g., trampling, grazing, and wood gathering) that lead to almost complete
stabilization on the other hand.
4. Summary and conclusion
Different land-use practices in Israel and Egypt are the dominant cause for differences in the vegetation structure and pattern
between Sinai and the Negev. This variation changes the surface
roughness, leading to a difference in radar backscatter across the
border. Dielectric permittivity differences between the two sides
of the border are small, and are likely not contributing much to the
difference in backscatter. Although this area was studied intensively using remote sensing means for many years, differences
in the vegetation structure across the border were not discussed
by previous studies. Therefore combining both SAR and optical
imagery into ecological studies can stress features otherwise overlooked.
Acknowledgements
Offer Rozenstein was supported by the Tim Casgrain Water
Management Fund from McGill University. Several individuals
from Ben-Gurion University of the Negev are acknowledged: Noa
Ohana-Levi is thanked for her advice on geo-statistical analysis and
comments on an early version of the manuscript. Professor Arnon
Karnieli is acknowledged for photographs in Fig. 5 and thanked for
helpful discussions about the research area. We thank Rachel Bernstein for her field assistance; Yehoshua Ratzon and Roman Joffe for
their technical assistance; and Dr. Shimrit Maman for her technical and administrative support. ASAR Data was provided by the
European Space Agency.
References
Belnap, J., Lange, O.L. (Eds.), 2001. Springer-Verlag, Berlin, p. 503.
Blom, R., Elachi, C., 1981. Spaceborne and airborne imaging radar observations of
sand dunes. J. Geophys. Res. 86, 3061–3073.
Blumberg, D.G., 1998. Remote sensing of desert dune forms by polarimetric
synthetic aperture radar (SAR). Remote Sens. Environ. 65, 204–216.
Blumberg, D., Greeley, R., 1993. Field studies of aerodynamic roughness length. J.
Arid Environ. 25, 39–48.
Breckle, S., Yair, A., Veste, M. (Eds.), 2008. Springer, New York: Berlin, Heidelberg,
p. 475.
Carmel, Y., Kadmon, R., 1998. Computerized classification of Mediterranean
vegetation using panchromatic aerial photographs. J. Veg. Sci. 9, 445–454.
Clark, P.J., Evans, F.C., 1954. Distance to nearest neighbor as a measure of spatial
relationships in populations. Ecology 35, 445–453.
Columbus, U., Ziv, Y., Tsoar, H., 2012. Ecological restoration of the western-Negev
dunes system in light of land-use changes: aeolian activity and faunal
response. 97th ESA Annual Convention.
Congalton, R.G., Green, K., 2008. Assessing the Accuracy of Remotely Sensed Data:
Principles and Practices. Lewis Publishers, Boca Raton, pp. 137.
O. Rozenstein et al. / International Journal of Applied Earth Observation and Geoinformation 46 (2016) 13–21
Danin, A., Ganor, E., 1991. Trapping of airborne dust by mosses in the Negev desert,
Israel. Earth Surf. Process. Landf. 16, 153–162.
Das, K., Paul, P.K., 2015. Present status of soil moisture estimation by microwave
remote sensing. Cogent Geosci. 1, http://dx.doi.org/10.1080/23312041.2015.
1084669.
Evans, D.L., Plaut, J.J., Stofan, E.R., 1997. Overview of the spaceborne imaging
radar-C/X-band synthetic aperture radar (SIR-C/X-SAR) missions. Remote Sens.
Environ. 59, 135–140.
Genis, A., Vulfson, L., Blumberg, D.G., Sprinstin, M., Kotlyar, A., Freilikher, V., et al.,
2013. Retrieving parameters of bare soil surface roughness and soil water
content under arid environment from ERS-1, -2 SAR data. Int. J. Remote Sens.
34, 6202–6215.
Harrison, J.B.J., Yair, A., Yair, A., 1998. Late Pleistocene aeolian and fluvial
interactions in the development of the Nizzana dune field, Negev desert, Israel.
Sedimentology 45, 507–518.
He, H.S., DeZonia, B.E., Mladenoff, D.J., 2000. An aggregation index (AI) to quantify
spatial patterns of landscapes. Landsc. Ecol. 15, 591–601.
Hess, L.L., Melack, J.M., Filoso, S., Yong Wang, 1995. Delineation of inundated area
and vegetation along the Amazon floodplain with the SIR-C synthetic aperture
radar. Geosci. Remote Sens. IEEE Trans. 33, 896–904.
Jaeger, J.A., 2000. Landscape division, splitting index: and effective mesh size: new
measures of landscape fragmentation. Landsc. Ecol. 15, 115–130.
Karnieli, A., Tsoar, H., 1995. Satellite spectral reflectance of biogenic crust
developed on desert dune sand along the Israel–Egypt border. Int. J. Remote
Sens. 16, 369–374.
Kidron, G.J., Yair, A., 1997. Rainfall-runoff relationship over encrusted dune
surfaces, Nizzana, western Negev, Israel. Earth Surf. Process. Landf. 22,
1169–1184.
Kidron, G.J., 2015. Dune crests serve as preferential habitats for perennial plants
during frequent drought years. J. Hydrol. 522, 295–304.
Kidron, G.J., 2014. The negative effect of biocrusts upon annual-plant growth on
sand dunes during extreme droughts. J. Hydrol. 508, 128–136.
Kidron, G.J., Vonshak, A., Abeliovich, A., 2008. Recovery rates of microbiotic crusts
within a dune ecosystem in the Negev desert. Geomorphology 100,
444–452.
Lancaster, N., Gaddis, L., Greeley, R., 1992. New airborne imaging radar
observations of sand dunes: Kelso dunes, California. Remote Sens. Environ. 39,
233–238.
Levin, N., 2015. Human factors explain the majority of MODIS-derived trends in
vegetation cover in Israel: a densely populated country in the eastern
Mediterranean. Reg. Environ. Change, 1–15 (in press).
Li, X.-R., Wang, X.-P., Li, T., Zhang, J.-G., 2002. Microbiotic soil crust and its effect on
vegetation and habitat on artificially stabilized desert dunes in Tengger desert,
North China. Biol. Fertil. Soils 35, 147–154.
Li, Z., Guo, X., 2015. Remote sensing of terrestrial non-photosynthetic vegetation
using hyperspectral, multispectral, SAR, and LiDAR data. Progr. Phys. Geogr. (in
press).
McGarigal, K., Cushman, S.A., Ene. E., 2012. FRAGSTATS v4: Spatial Pattern Analysis
Program for Categorical and Continuous Maps. Computer software program
produced by the authors at the University of Massachusetts, Amherst. Available
at: http://www.umass.edu/landeco/research/fragstats/fragstats.html.
Meir, A., Tsoar, H., 1996. International borders and range ecology: the case of
Bedouin transborder grazing. Hum. Ecol. 24, 39–64.
Mitchell, A., 2005. The ESRI Guide to GIS Analysis, Volume 2: Spatial
Measurements and Statistics. ESRI Press, Redlands, CA.
Mougin, E., Proisy, C., Marty, G., Fromard, F., Puig, H., Betoulle, J.L., et al., 1999.
Multifrequency and multipolarization radar backscattering from mangrove
forests. Geosci. Remote Sens. IEEE Trans. 37, 94–102.
Olsvig-Whittaker, L.S., Hosten, P.E., Marcus, I., Shochat, E., 1993. Influence of
grazing on sand field vegetation in the Negev desert. J. Arid Environ. 24,
81–93.
21
Otterman, J., 1981. Satellite and field studies of man’s impact on the surface in arid
regions. Tellus 33, 68–77.
Otterman, J., 1974. Baring high albedo soils by overgrazing: a hypothesized
desertification mechanism. Science 186, 531–533.
Pointing, S.B., Belnap, J., 2012. Microbial colonization and controls in dryland
systems. Nat. Rev. Microbiol. 10, 551–562.
Qin, Z., Karnieli, A., Berliner, P., 2001. A mono-window algorithm for retrieving
land surface temperature from Landsat TM data and its application to the
Israel–Egypt border region. Int. J. Remote Sens. 22, 3719–3746.
Qin, Z., Berliner, P., Karnieli, A., 2002. Micrometeorological modeling to understand
the thermal anomaly in the sand dunes across the Israel–Egypt border. J. Arid
Environ. 51, 281–318.
Qin, Z., Berliner, P., Karnieli, A., 2005. Ground temperature measurement and
emissivity determination to understand the thermal anomaly and its
significance on the development of an arid environmental ecosystem in the
sand dunes across the Israel–Egypt border. J. Arid Environ. 60, 27–52.
Qin, Z., Li, W., Burgheimer, J., Karnieli, A., 2006. Quantitative estimation of land
cover structure in an arid region across the Israel–Egypt border using remote
sensing data. J. Arid Environ. 66, 336–352.
Roskin, J., Porat, N., Tsoar, H., Blumberg, D.G., Zander, A.M., 2011. Age, origin and
climatic controls on vegetated linear dunes in the northwestern Negev desert
(Israel). Quat. Sci. Rev.
Roskin, J., Blumberg, D.G., Porat, N., Tsoar, H., Rozenstein, O., 2012. Do dune sands
redden with age? The case of the northwestern Negev dunefield, Israel.
Aeolian Res. 5, 63–75.
Rozenstein, O., Karnieli, A., 2015. Identification and characterization of biological
soil crusts in a sand dune desert environment across Israel–Egypt border using
LWIR emittance spectroscopy. J. Arid Environ. 112A, 75–86.
Rozenstein, O., Karnieli, A., 2011. Comparison of methods for land-use classification
incorporating remote sensing and GIS inputs. Appl. Geogr. 31, 533–544.
Rozenstein, O., Zaady, E., Katra, I., Karnieli, A., Adamowski, J., Yizhaq, H., 2014. The
effect of sand grain size on the development of cyanobacterial biocrusts.
Aeolian Res. 15, 217–226.
Rozenstein, O., Agam, N., Serio, C., Masiello, G., Venafra, S., Achal, S., et al., 2015.
Diurnal emissivity dynamics in bare versus biocrusted sand dunes. Sci. Total
Environ. 506–507, 422–429.
Sankey, J.B., Glenn, N.F., Germino, M.J., Gironella, A.I.N., Thackray, G.D., 2010.
Relationships of aeolian erosion and deposition with LiDAR-derived landscape
surface roughness following wildfire. Geomorphology 119, 135–145.
Schenk, T., Csathó, B., 2002. Fusion of LIDAR data and aerial imagery for a more
complete surface description. Int. Arch. Photogramm. Remote Sens. Spat. Inf.
Sci. 34, 310–317.
Schmidt, H., Karnieli, A., 2000. Remote sensing of the seasonal variability of
vegetation in a semi-arid environment. J. Arid Environ. 45, 43–59.
Schumaker, N.H., 1996. Using landscape indices to predict habitat connectivity.
Ecology 77, 1210–1225.
Scott, L.M., Janikas, M.V., 2010. Spatial statistics in ArcGIS. In: Anonymous (Ed.),
Handbook of Applied Spatial Analysis. Springer, pp. 27–41.
Seifan, M., 2009. Long-term effects of anthropogenic activities on semi-arid sand
dunes. J. Arid Environ. 73, 332–337.
Siegal, Z., Tsoar, H., Karnieli, A., 2013. Effects of prolonged drought on the
vegetation cover of sand dunes in the NW Negev desert: field survey, remote
sensing and conceptual modeling. Aeolian Res. 9, 161–173.
Tsoar, H., Karnieli, A., 1996. What determines the spectral reflectance of the
Negev–Sinai sand dunes. Int. J. Remote Sens. 17, 513–525.
Tsoar, H., Blumberg, D.G., Wenkart, R., 2008. Formation and geomorphology of the
north-western Negev sand dunes. In: Breckle, S., Yair, A., Veste, M. (Eds.), Arid
Dune Ecosystems: The Nizzana Sands in the Negev Desert. Springer, Berlin,
Heidelberg, pp. 25–48.
Wang, T., Xue, X., Zhou, L., Guo, J., 2015. Combating aeolian desertification in
Northern China. Land Degrad. Dev. 26, 118–132.