Multi-scale analysis of intrinsic soil factors from SAR-based mapping of drying rates Tal Svoray a, * , Maxim Shoshany b a Department of Geography and Environmental Development, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel b Department of Transportation and Geo-Information Engineering, Faculty of Civil and Environmental Engineering, Technion-Israel Institute of Technology, Haifa 32000, Israel Received 3 March 2003; received in revised form 1 April 2004; accepted 8 June 2004 Abstract Intrinsic soil factors affect and are affected by the spatial variation of soil properties. Therefore, intrinsic soil factors may both characterize and serve as an indicator for soil taxonomy. Difficulties in inferring intrinsic soil properties hamper attempts to assess their variability, on both local and regional/broad scales. Radar remote sensing might facilitate a breakthrough in this field, due to its sensitivity to the soil water content. In this research, a raster Geographic Information System (GIS) methodology is developed for combining multi-temporal ERS-2 SAR and Landsat TM data, which allows the estimation of drying rate patterns in bare soil surfaces. The drying rates provide further indication about intrinsic soil properties. The multi-scale behaviour of soil-drying rates is described using the richness – area curves and characteristic curves are determined to four soil formations typical to a climatic gradient between Mediterranean and semi-arid environments in Israel. To the best of our knowledge, this is one of the first attempts to document the effect of intrinsic soil factors on the soil system at the regional scale. The results achieved here demonstrate the connection between drying rates, richness – area variation and soil hydraulic conductivity of the four soil formations. D 2004 Elsevier Inc. All rights reserved. Keywords: SAR; Soil Moisture; Geographical Information Systems; Intrinsic Factors 1. Introduction Soil properties vary with place and time (Heuvelink & Webster, 2001). The magnitude and type of variation in soil properties affect the interpretation of observations regarding the evolution, diversity and dynamics of the soil system. An integration of various sources of information and synthesis of diverse approaches is required to study the coexistence of order, complexity, chaotic/nonchaotic, sta- ble/unstable and self-organization/non-self-organization in soil systems, as well as relationships between bio-, pedo-, topo-, hydro- and climodiversity (Phillips, 1999). It is the emergence of the science of complexity—contending with ‘‘how simple, fundamental processes, derived from reduc- tionism, can combine to produce complex holistic systems’’ (Malanson, 1999)—that might facilitate a better under- standing of soil processes and patterns. Within this frame of thought, there is a growing interest in examining Earth systems with a multi-scale approach (Coops & Waring, 2001; Zhang et al., 2002). It is anticipated that this will identify typologies of system behavior, from more complex or chaotic characteristics at a fine scale of examination to more regulated or ordered and stable characteristics at a broader scale. Water content is one of the most important soil properties that determine vegetation productivity in semi-arid regions (Kumar et al., 2002; Svoray et al., 2004). While it is difficult and sometimes even imprac- tical to map soil moisture in the field over large regions (Mahmood, 1996), remote sensing approaches, such as radar and thermal techniques are most adequate for this purpose. The advantage of Synthetic Aperture Radar (SAR) data over thermal data in determining soil mois- ture is its ability to penetrate the upper soil strata (Ulaby et al., 1986). This potential, however, is not fully exploited and spatial patterns of intrinsic soil properties derived from SAR data have not been presented yet in the scientific literature. In areas of large soil variability, 0034-4257/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.rse.2004.06.011 * Corresponding author. Tel.: +97-2-864-7379; fax: +97-2-864-2821. E-mail address: [email protected] (T. Svoray). www.elsevier.com/locate/rse Remote Sensing of Environment 92 (2004) 233 – 246
14
Embed
Multi-scale analysis of intrinsic soil factors from SAR … · · 2004-08-15Multi-scale analysis of intrinsic soil factors from SAR ... global and understanding global processes,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/rse
Remote Sensing of Environment 92 (2004) 233–246
Multi-scale analysis of intrinsic soil factors from SAR-based
mapping of drying rates
Tal Svoraya,*, Maxim Shoshanyb
aDepartment of Geography and Environmental Development, Ben-Gurion University of the Negev, Beer-Sheva 84105, IsraelbDepartment of Transportation and Geo-Information Engineering, Faculty of Civil and Environmental Engineering,
Technion-Israel Institute of Technology, Haifa 32000, Israel
Received 3 March 2003; received in revised form 1 April 2004; accepted 8 June 2004
Abstract
Intrinsic soil factors affect and are affected by the spatial variation of soil properties. Therefore, intrinsic soil factors may both characterize
and serve as an indicator for soil taxonomy. Difficulties in inferring intrinsic soil properties hamper attempts to assess their variability, on
both local and regional/broad scales. Radar remote sensing might facilitate a breakthrough in this field, due to its sensitivity to the soil water
content. In this research, a raster Geographic Information System (GIS) methodology is developed for combining multi-temporal ERS-2 SAR
and Landsat TM data, which allows the estimation of drying rate patterns in bare soil surfaces. The drying rates provide further indication
about intrinsic soil properties. The multi-scale behaviour of soil-drying rates is described using the richness–area curves and characteristic
curves are determined to four soil formations typical to a climatic gradient between Mediterranean and semi-arid environments in Israel. To
the best of our knowledge, this is one of the first attempts to document the effect of intrinsic soil factors on the soil system at the regional
scale. The results achieved here demonstrate the connection between drying rates, richness–area variation and soil hydraulic conductivity of
the four soil formations.
D 2004 Elsevier Inc. All rights reserved.
Keywords: SAR; Soil Moisture; Geographical Information Systems; Intrinsic Factors
1. Introduction
Soil properties vary with place and time (Heuvelink &
Webster, 2001). The magnitude and type of variation in soil
properties affect the interpretation of observations regarding
the evolution, diversity and dynamics of the soil system.
An integration of various sources of information and
synthesis of diverse approaches is required to study the
coexistence of order, complexity, chaotic/nonchaotic, sta-
ble/unstable and self-organization/non-self-organization in
soil systems, as well as relationships between bio-, pedo-,
topo-, hydro- and climodiversity (Phillips, 1999). It is the
emergence of the science of complexity—contending with
‘‘how simple, fundamental processes, derived from reduc-
tionism, can combine to produce complex holistic systems’’
(Malanson, 1999)—that might facilitate a better under-
standing of soil processes and patterns. Within this frame
0034-4257/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
(e.g., Prevot et al., 1993) have shown empirically that the
radar backscatter is more sensitive to volume scattering than
to changes in mass. To correspond to the linear soil moisture
model, the gravimetric soil moisture measurements were
converted into volumetric measures, based on bulk density
measurements for the specific soils of the study area
(Svoray, 2000).
Table 2 shows that the soil moisture values at both the
Avisur and Lehavim sites for February 1997 are the high-
est. The reason for this is illustrated in the accumulated
rainfall graph of 1997 (Fig. 5). Just before the field
campaign and satellite overpass of February 1997, a sig-
nificant rainfall event was recorded at both sites. In
contrast, the last rainfall with relevance to the field cam-
paign and satellite overpass of April 1997 was recorded on
Table 2
Field data of soil moisture concentration (%) in three habitats at the Avisur and L
Site Avisur
Habitat Height Slope Wadi
Measure Average Standard
deviation
Average Standard
deviation
Average Stand
devia
February 1997 27 – 26 – 28 –
April 1997 10 – 8 – 21 –
May 1997 12 1.64 12 2.03 13 0.34
23 March 1997—17 days before the campaign. The de-
crease over time of soil moisture concentration in soil under
similar climatic conditions in Israel’s east–west climatic
gradient was also determined by Svoray (1994). Water loss
of about 3% per day was measured in wet soil (25% of
gravimetric soil moisture) and about 0.5% in dry soil of 5%
soil moisture concentration. Thus, during the 17 days
before the April sampling, soil moisture had decreased
due to evaporation and infiltration, and was particularly
low on the slopes and heights. However, the soil moisture
concentration values in the valleys remained relatively
high, due to drainage accumulation and vegetation cover,
which reduces the sun’s intensity. During the May cam-
paign, soil moisture content values were relatively low and
more constant within the sites (it seems that the wadi had
dried and a moisture increase on slopes and heights might
have been due to an unrecorded small rainfall event). Table
2 gives the differences between the sites: Avisur soil
maintains a moderate moisture level while Lehavim soil
remains much drier.
ehavim sites
Lehavim
Height Slope Wadi
ard
tion
Average Standard
deviation
Average Standard
deviation
Average Standard
deviation
17 – 13 – 19 –
5 – 4 – 7 –
2 0.1 2 0.2 2 0.4
Fig. 5. The cumulative rainfall in the study area, represented by the meteorological stations of Beit-Gimal and Beer Sheva (the closest meteorological stations
with daily data available on the Avisur and Lehavim sites, respectively). Based on data from the Israel Meteorological Service.
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246 239
5.3. Satellite data
Five ERS-2 SAR images (Cvv band at mean incidence
angle of 230) were acquired from the study area in two sets of
measurements: (i) February, April and May 1997, represent-
ing the environment during the winter, spring and summer
seasons; and (ii) April and June 1999, representing the spring
and summer seasons of that year. The backscatter (r0) was
derived from the Digital Number (DN) of the ERS-2 SAR
PRI image, using the method by Laur et al. (1997). Since part
of the study area is hilly, backscatter was adjusted for the
variation in the local angle of incidence, based on a digital
elevation model (Svoray & Shoshany, 2003). The geometric
correction of the images used 250 ground control points and a
first-order algorithm with a root mean square error of less
than one pixel.
Implementations of the NBMI method on bare terrain
within the study area were achieved using the synergy of
ERS-2 SAR and Landsat TM images. The assessments were
applied on three dates: February, April and May 1997. The
ERS-2 SAR backscatter images were used to calculate
NBMI layers and the Landsat TM data to identify and mask
the vegetated surfaces, with the Normalised Difference
Vegetation Index (NDVI—the normalised ratio between
the red and infrared bands). The NDVI has shown a strong
correlation with vegetation cover in other environments
around the world and in the study region (Svoray et al.,
2003). The procedure was implemented by the ERDAS/
IMAGINES Model Maker in four stages (Fig. 6). At first,
NBMI layers were calculated from the ERS-2 SAR back-
scatter using Eq. (5). The value of dBt1 in this case is the
backscatter from the calibrated image pixel of each date; dBt2
is the average value of the intercept of soil moisture models
from various environments around the world (Shoshany et
al., 2000) and the soil moisture models calculated for the
present study area. At the second stage, the transformation of
the empirically derived relationship between the VSM con-
centration and NBMI was used as a model for calculating the
VSM layers from NBMI layers (Eq. (7)):
VSM ¼ NBMI þ 0:0204
0:0105: ð7Þ
At the third stage, NDVI was calculated for the study
area using Landsat TM images from February, April and
May 1997 (Eq. (8)):
NDVI ¼ band4� band3
band4þ band3; ð8Þ
where bands 4 and 3 denote the data recorded in the infrared
and red bands of the Landsat TM, respectively. A threshold
value of DN is selected to distinguish vegetated and non-
vegetated surfaces. The threshold value is set at 0.2, based
on spectral signatures of soil and vegetation in the study
area that had been measured previously (Svoray, 2000;
Svoray et al., 2003). The latter shows that under similar
ecological conditions the NDVI value of 0.2 is related to a
vegetation cover of less than 5%. Additional support for the
selected value is a random sampling of 90 plots with bare
soil and low herbaceous vegetation cover in the study area
(Fig. 7).
Use of this NDVI threshold value to distinguish between
vegetated and non-vegetated surfaces is described at the
fourth stage. The NDVI layer for each date represents the
status of vegetation cover at different phenological phases—
which is important given the wide cover of annual vegeta-
tion in the study area.
At this last stage, layers of VSM concentration and
NDVI were merged using a simple conditional model.
Fig. 7. The relationship between total vegetation cover and NDVI values in
the study area.
Fig. 6. The use of NDVI derived from Landsat TM images and the relationship between VSM and NBMI—for the assessment of moisture concentration in bare
soil regions.
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246240
The model traced pixels in the VSM layer with cor
responding NDVI values greater than 0.2 to be masked
at zero, and pixels with NDVI values smaller than 0.2 to
give their VSM value in the final layer. This stage
produced a VSM concentration layer that covered only
the bare soil surface of the study area for each of the
above-mentioned dates.
6. Results and discussion
6.1. Results
Fig. 8 shows the soil moisture images for February, April
and May 1997. In general, the February image presents the
largest VSM values (with mean values of 29% and 17% for
Fig. 8. VSM layers of bare soil plots along the climatic gradient of the study area at the three dates in the growing season of 1997. Reddish colors represent
lower VSM values and bluish colors higher VSM values.
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246 241
the northern and southern parts of the study area, respec-
tively), due to its relative proximity to the last rainfall event
(Fig. 5). The April image shows lower VSM values (with
mean values of 18% and 12% for the northern and southern
parts, respectively), while the May VSM values are the
smallest (with mean values of 13% and 9% for the northern
and southern parts, respectively). However, the spatial
variability in the VSM values in the three images is high
and a visual analysis shows that there is no continuous
decrease in VSM from the semi-humid area in the north to
the semi-arid area in the south.
The large degree of spatial variation is also observed in
the analysis of the drying rate patterns. The patchy pattern is
generalized by grouping the 125 optional combinations of
soil-drying rate into nine classes of similar temporal change
(see legend in Fig. 9). Thus, for example, the first category
(cyan) includes all of the groups that represent a decrease in
VSM from February to April and then an increase from
April to May; the second category (green) includes all
groups in which VSM increases from February to April
and also increases from April to May. As discussed earlier,
the output categories of this classification infer types of
potential water holding capacity in the cells. For example,
areas of unchanged low VSM levels (depicted in black)
suggest that there is no rainfall infiltration of the soil in these
cells and that they are runoff contributing areas. Identifica-
tion of such group is most important for modeling the
hydrology and erosion in these regions.
A frequency analysis of the nine categories in the entire
study area shows that in 73% of the area, there is a decrease
of VSM from February to April and then either another
decrease of VSM in May or the VSM level remains the
same between April and May. The categories of higher
VSM values in May (shown in Fig. 9 in shades of blue) are
mainly in the agricultural fields of the northern part of the
study area. These fields are cultivated with summer crops,
such as cotton, water melon and sunflower (Cohen &
Shoshany, 2002). The fields are irrigated in May, but the
vegetation cover is low at this date; thereby for these areas
represent an increase in VSM in May, in comparison with
April. Apart from this category which represents a distinc-
tive type of landuse, the other parts of the study area are
characterized by soil types dominated by natural hydrolog-
ical processes of infiltration and runoff.
The resulting heterogeneous and patchy pattern of soil-
drying rate groups can be attributed to the effect of intrinsic
Fig. 9. The four soil formations with 125 soil drying rate transition groups joined into nine categories.
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246242
factors on the soil. The coefficients of the richness–area
curves (Eq. (1)) applied to the four main soil formations of
the study area are given in Table 3. There are very strong
correlations (0.88 <R2 < 0.94) between the number of drying
groups and the areas they occupy (Fig. 10). This result
supports previous assertions that the power function of
richness–area curves represents intrinsic variation in soil
(Phillips, 2001) and further strengthens both the information
and functional aspects this multi-scale behavior at the
regional scale, where the validity of the richness–area
relationship has been questioned. The exponents (b coef-
ficients) of the four curves range from 0.25 to 0.37 with a
mean of 0.3, corresponding to values reported previously
(Phillips, 2001; Rosenzweig, 1995).
A bi/bt ratio was calculated between the mean exponent
of the four groups (bi) and the exponent of the richness–
area curve of the total study area (bt). The exponent of the
total area was calculated using bare soil pixels sampled from
the four soil formations that cover the study area. The
selected pixels were examined in the field to assure that
Fig. 10. Richness–area curves calculated to four soil formations of the
study area.
Table 3
Results of the coefficients achieved for the richness–area curves of the four
soil formations in the study area
Soil formation Polynomial model R2
c b
B3 5.99 0.37 0.91
M1 7.83 0.33 0.93
K3 12.32 0.25 0.88
N4 8.61 0.29 0.94
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246 243
areas covered by water sources and artificial structures are
excluded from the analysis.
The result (2.68) has an important pedologic implication
that will be discussed later. The largest exponent is for the
B3 soil formation and the second largest is for M1. The c
coefficients (with a range of 5.99–12.32 and a mean of
8.69) are much larger than observed in previous studies.
Another observation that stems from the application of the
richness–area curves in the current research is that there is a
strong correlation between the coefficients c and d of the
four soil formations examined for the power (R2 = 0.92)
model (Fig. 11). To our knowledge, this relationship has not
been reported previously.
6.2. Discussion
Differences in soil drying rates occur due to the com-
bined effect of intrinsic and extrinsic factors. A useful
method to determine which of the factors have more
influence is the analysis of the bi/bt ratio. The analysis is
based on the assumption that bt represents the overall
variance of four different soil formations and therefore
incorporates variation caused by extrinsic factors (since
the four soil formations vary in their extrinsic characters).
In contrast, bi represents the mean value of the exponent
coefficients of the four formations and is not likely to
represent variation associated with the extrinsic factors.
Therefore, if the bi/bt ratio is greater than 1, then the
variation within the unit and the effect of intrinsic factors
are greater than the variation between units that are caused
by extrinsic factors. As shown in Section 6.1, the bi/bt ratio
calculated here is much larger than one which implies that
the intrinsic factors have more influence on the spatial
pattern of soil-drying rate groups than extrinsic factors.
Furthermore, the large coefficients of determination for
the richness–area curves suggest that the heterogeneity of
the factors is involved with drying processes increase with
scale. This had already been observed by Seyfried (1998),
who, however, attributed the change in scale to extrinsic
factors: ‘‘. . .the increase of spatial variability with scale was
controlled by deterministic ‘sources’ such as soil series and
elevation induced climatic effects.’’ Similarly, a strong
relationship was found between the variation in soil mois-
ture and primary and secondary topographic attributes
(Sulebak et al., 2000). The two studies show increasing
variation in soil moisture with scale as a result of the effect
of environmental factors. However, it is important to note
that the two previous studies and the one reported here differ
in physiography and climate. Seyfried’s study was carried
out in a mountainous area between 1097 and 2237 m, with
snow precipitation. Under such severe topographic condi-
tions, extrinsic factors have a strong impact. Sulebak et al.
surveyed a wet area with large soil moisture values. In this
area, the water movement within (and upon) the soil plays a
much more important role than in our Mediterranean to
semi-arid climate. In the latter areas, water loss is mainly
Fig. 11. The relationship between c and b coefficients, determined to the
four soil formations of the study area.
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246244
due to evaporation and therefore the soil water holding
capacity (strongly dependent on the soil hydraulic proper-
ties) has more influence on changes in soil moisture storage.
The fact that the curves correspond well with the data not
only verifies the validity of the richness–area concept
beyond the plot scale but it also conform with previous
analyses of soil drying rates at much coarse scale. For
example studies such as Barros et al. (2000) who have
found that multi-temporal analysis of SAR images at the
Fig. 12. The relationship between c coefficients and the hydraulic conductivity o
conductivity was calculated based on the formula Ksoil/Kfe = 1�Rm where Rm is
Cousin et al., 2003).
scale of kilometers can be used to analyze spatial patterns of
intrinsic soil factors such as soil texture. In our study, the
link between soil moisture change, soil texture, hydraulic
conductivity and characteristics of the ‘‘richness–area’’
curves was examined and showed high correlation (Fig.
12) limited at this stage by the number of soil groups
presented here. Further insight would require more detailed
analysis with more detailed data of soil texture and hydrau-
lic conductivity.
As discussed earlier, the richness–area relationship
depends on the local or regional environmental conditions
and on the process studied. Nevertheless, it seems that the
soil variation due to long-term processes increases the
heterogeneity of soil moisture patterns with scale. Based
on these findings, we can conclude that intrinsic factors
affect the intra-unit variability. The changes in relationships
between the units (using exponents and, to some extent,
coefficients of determination) may be caused by differences
in extrinsic factors, such as parent material and topography,
between the units. For example, the exponents of the natural
areas (M1 and B3 soil formations) are larger than those of
agricultural areas (K3 and N4 soil formations). A surprising
result is the strong relation between the c and b coefficients
(Fig. 11). This implies that the two coefficients could be
integrated, as each exponent could have a predicted multi-
f the soils (Ksoil) and the fine earth hydraulic conductivity (Kfe). Hydraulic
the mass fraction of the rock fragments (based on Brakesniek et al., 1986;
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246 245
plier coefficient. The present study has focused on four
soil types at it provides convincing results, as illustrated in
Fig. 10. The final result, however, provides an indication
for further empirical study.
7. Summary and conclusions
Monitoring spatial patterns of soil attributes in general,
and the estimation of areal soil moisture in particular, is an
important task of environmental remote sensing (Schmugge
et al., 2002). The spatio-temporal dynamics of soil moisture
is especially needed in semi-arid regions, where water stress
determines the productivity of woody and herbaceous veg-
etation (Kumar et al., 2002; Tansey & Millington, 2001).
In the current research, we analysed spatio-temporal
patterns of soil moisture and examined the quantitative
effects of intrinsic soil factors on recurring soil-drying
processes at the scale of 12.5 m per pixel. The progress
achieved here with remote sensing and environmental
monitoring was facilitated by the multi-scale approach
(Walsh et al., 1998). Our results show strong relationship
between the size of the plot measured and the richness of the
drying rate groups within four Mediterranean soil forma-
tions. The richness–area curves enable an extension of the
remote sensing methodology beyond per-pixel estimates of
volumetric soil moisture concentration and thus offer an
alternative perspective to describe the multiscaling behavior
of soils. Furthermore, it may help to disaggregate general-
ized soil map polygons into more detailed landscape com-
ponents (Bui & Moran, 2001).
The relationship between the richness–area coefficients
and soil hydraulic conductivity at the 12.5 me scale as
presented in this study provides an important link between
the studies conducted at the broad scale and the multi-scale
behaviour of soils. Kim and Barros (2002a) show that in a
scale of 10 km, topography dominate soil moisture only
during and immediately after rain storm, while during the
interstorm periods in non-vegetated surfaces, soil moisture
is closely associated with the spatial variability of soil
hydraulic properties. Our study further supports this hy-
pothesis as it shows, in a much finer scale, that soil water
holding capacity has an important role in the determination
of scaling behavior of soil moisture fields.
Linking the local scale with the regional strengthens the
phenomenological basis of the remote sensing interpreta-
tion. It allows differentiation between areas that conform to
the intrinsic scale change and other areas that are represent-
ing local anomalies due to disturbance or error in the remote
sensing interpretation. Thus, the effect of the soil’s intrinsic
factors can be considered in soil (moisture) modeling,
thereby contributing to a gap in our knowledge.
The transition maps of soil-drying rates provided here
can serve as an important database for future work by
providing information on contributing areas, sources and
sinks in hydrological models.
Acknowledgements
This study was carried out under a JNF (Jewish National
Fund) research grant no. 190/9/328/8. We wish to
acknowledge Haim Katz for his help in the field campaign
and we thank the anonymous reviewers and Prof. Jonathan
D. Phillips who contributed significantly to the depth of the
soil information analysis.
References
Barros, A. P., Bindlish, R., & Rogowski, A. S. (2000). Soil hydrology and
spatial variability—perspectives on the interpretation of remotely
sensed dataTechnical report series 00-002. Pennsylvania, USA:
Environmental Institute, College of Earth and Mineral Sciences, PSU
(65 pp).
Bindlish, R., & Barros, A. P. (2001). Parameterization of vegetation back-
scatter in radar-based, soil moisture estimation. Remote Sensing of En-
vironment, 76, 130–137.
Birkeland, P. W. (1984). Soils and geomorphology. Oxford: Oxford Uni-
versity Press.
Brakesniek, D. L., Rawls, W. J., & Stephenson, G. R. (1986). Determining
the saturated hydraulic conductivity of a soil containing soil fragments.
Soil Science Society of America Journal, 50, 834.
Bui, E. N., & Moran, C. J. (2001). Disaggregation of polygons of surficial
geology and soil maps using spatial modelling and legacy data. Geo-
derma, 103, 79–94.
Burrough, P. A., Macmillan, R. A., & Deursen, W. V. (1992). Fuzzy clas-
sification methods for determining land suitability from soil profile
observations and topography. Journal of Soil Science, 43, 193–210.
Cohen, Y., & Shoshany, M. (2002). A national knowledge-based crop
recognition in Mediterranean environment. International Journal of
Applied Earth Observation and Geoinformation, 4, 75–87.
Coops, N. C., & Waring, R. H. (2001). The use of multiscale remote
sensing imagery to derive regional estimates of forest growth capacity
using 3-PGS. Remote Sensing of Environment, 75, 324–334.
Cousin, I., Nicoullaud, B., & Coutadeur, C. (2003). Influence of rock frag-
ments on the water retention and water percolation in a calcerous soil.
Catena, 53, 97–114.
Dan, Y. (1988). The soils of the southern Shefelah. In D. Urman, & E. Stern
(Eds.), Man and environment in the southern Shefelah—studies in re-
gional geography and history ( pp. 50–59). Givataim, Israel: Masada,
In Hebrew.
Dan, Y., & Raz, Z. (1970). The soil formations map of Israel. Beit Dagan,
Israel: The Volcani Centre.
Dobson, M. C., Ulaby, F. T., Hallikainen, M. T., & Elrayes, M. A. (1985).
Microwave dielectric behavior of wet soil: 2. Dielectric mixing models.
IEEE Transactions on Geoscience and Remote Sensing, 23, 35–46.
Griffiths, G. H., & Wooding, M. G. (1996). Temporal monitoring of
soil moisture using ERS-1 SAR data. Hydrological Processes, 10,
1127–1138.
Heuvelink, G. B. M., & Webster, R. (2001). Modeling soil variation: Past,
present, and future. Geoderma, 100, 269–301.
Hoosbeek, M. R., & Bryant, R. B. (1992). Towards the quantitative mode-
ling of pedogenesis—a review. Geoderma, 55, 183–210.
Ibanez, J. J., De Alba, S., Lobo, A., & Zucarello, B. (1998). Pedodiversity
and global soil patterns at coarse scales (with discussion). Geoderma,
83, 171–192.
Jenny, H. (1941). Factors of soil formation. New York: McGraw-Hill.
Jenny, H. (1961). Derivation of soil factor equations of soils and ecosys-
tems. Proceedings - Soil Science Society of America, 25, 385–388.
Kim, G., & Barros, A. P. (2002a). Space– time characterization of soil
moisture from passive microwave remotely sensed imagery and ancil-
lary data. Remote Sensing of Environment, 81, 393–403.
T. Svoray, M. Shoshany / Remote Sensing of Environment 92 (2004) 233–246246
Kim, G., & Barros, A. P. (2002b). Downscaling of remotely sensed soil
moisture with a modified fractal interpolation method using contrac-
tion mapping and ancillary data. Remote Sensing of Environment, 83,
400–413.
Kumar, L., Reitkerk, M., van Langevelde, F., van de Koppel, J., van Andel,
J., Hearne, J., de Ridder, N., Stroosnijder, L., Skidmore, A. K., & Prins,
H. H. T. (2002). Relationship between vegetation growth rates at the
onset of the wet season and soil type in the Sahel of Burkina Faso:
Implications for resource utilisation at large scales. Ecological Model-
ling, 149, 143–152.
Laur, H., Bally, P., Meadows, J., Sanchez, B., & Schaetter, E. (1997). ERS
SAR calibration—derivation of the backscattering coefficient sigma-
nought. In ESA ERS SAR PRI products. Document no: ES-TN-RS-
PM-HL09, ESA web site (http://uranus.esrin.esa.it/eeo/lft/ESC2).
Mahmood, R. (1996). Scale issues in soil moisture modeling: Problems and
prospects. Progress in Physical Geography, 20, 273–291.
Malanson, G. P. (1999). Considering complexity. Annals of the Association
of American Geographers, 89, 746–753.
Paton, T. R. (1978). The formation of soil material. London: Allen and
Unwin.
Phillips, J. D. (1998). On the relations between complex systems and the
factorial model of soil formation (with discussion).Geoderma, 86, 1–21.
Phillips, J. D. (1999). Earth surface systems: Complexity, order and scale.
Oxford: Blackwell.
Phillips, J. D. (2001). The relative importance of intrinsic and extrinsic
factors in pedodiversity. Annals of the Association of American Geog-
raphers, 91, 609–621.
Prevot, L., Champion, I., & Guyot, G. (1993). Estimating surface soil
moisture and leaf area index of a wheat canopy using a dual frequency
(C and X bands) scatterometer. Remote Sensing of Environment, 46,
331–339.
Ravikovitch, S. (1992). The soils of Israel: Formation, nature and prop-
erties. Tel Aviv, Israel: Hakibbutz Hameuchad Publishing House.
Rosenzweig, M. L. (1995). Species diversity in space and time. New York:
Cambridge University Press.
Schmugge, T. J., Kustas, W. P., Ritchie, J. C., Jackson, T. J., & Rango, A.
(2002). Remote sensing in hydrology. Advances in Water Resources,
25(8–12), 1367–1385.
Seydel, R. (1988). From equilibrium to chaos: Practical bifurcation and
stability analysis. Amsterdam: Elsevier.
Seyfried, M. (1998). Spatial variability constraints to modeling soil water at
different scales. Geoderma, 85, 231–254.
Shoshany, M., Kutiel, P., & Lavee, H. (1995). Seasonal vegetation cover
changes as indicators of soil types along a climatological gradient: A
mutual study of environmental patterns and controls using remote sens-
ing. International Journal of Remote Sensing, 16, 2137–2151.
Shoshany, M., & Svoray, T. (2002). Multidate adaptive unmixing and its
application to analysis of ecosystem transitions along a climatic gradi-
ent. Remote Sensing of Environment, 82, 5–20.
Shoshany, M., Svoray, T., Curran, P. J., Foody, G. M., & Perevolotsky, A.
(2000). The relationship between ERS-2 SAR backscatter and soil
moisture: Generalization from a humid to semi-arid transect. Interna-
tional Journal of Remote Sensing, 21, 2337–2343.
Skidmore, E. L., & Layton, J. B. (1992). Dry-soil aggregate stability as
influenced by selected soil properties. Soil Science Society of America
Journal, 56, 557–561.
Sulebak, J. R., Tallaksen, L. M., & Erichsen, B. (2000). Estimation of areal
soil moisture by use of terrain data. Geografiska Annaler. Series A, 82,
89–105.
Svoray, T. (1994). Characterization of hydraulic parameters—rainfall, run-
off and soil moisture—along a climatic gradient, in the geomorphic year
1992/3. Bachelor dissertation. Department of Geography, Bar-Ilan Uni-
versity, Israel (pp. 71) (in Hebrew).
Svoray, T. (2000). Radar remote sensing of Mediterranean vegetation: Es-
timating areal aboveground biomass of herbaceous vegetation in a het-
erogeneous environment. Ph.D. dissertation. Department of Geography,