ınbul 2004
ds
VA um
"P mh. M4 Ul
M
\v
re TTT hey
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
Plate 3: The fields on this hill, which is located in the
mountainous area just to the south of the Syrian
border, were the steepest surveyed in the verification
exercise. The steepest slope angle was 6.1?. These
fields probably represent the steepest in the entire
study area, and have only been brought into cultivation
in the last 3-4 years.
Plate 1: Sedimentation in the lower part of a field. The
buried stones show the impact of sediment
accumulation. Desiccation cracking is a result of the
high silt and clay content of these sediments.
The statistical method used to provide verification of
the application of RUSLE in the study area compares
predicted soil losses and evidence of rilling. The
predicted soil losses were derived from the RUSLE
for the geographical co-ordinates of the fields
surveyed. A Chi-squared analysis was applied to
contingency tables of predicted soil losses for
categories of individual fields and evidence of rilling.
Chi-square test was run on the above contingency
table after combining the categories. The value of the
chi-square is 12.175, this is, just, less than the critical
value of 13.28 of the chi-square with significance
level 0.01. The results of the chi-square therefore
provide weak support for veracity of the application
of the RUSLE in the study area. However more work
is required to provide convincing support.
3.3.2 Quality Assessment: R
factor. The rainfall station density in the study area (1
station per 200 km?), though low, meets the WMO
(World Meteorological Organization) standard of 1
station per 250 km? (Shaw, 1983). The locational
errors in this layer will be low given that it is a simple
linear extrapolation, and the accuracy of resulting
509
layer will be determined by the accuracy of the
computer and software.
K factor. The soil association maps were published by
the Ministry of Agriculture of Jordan in 1994 at
1:250.000. They were produced in digital form by the
same ministry in cooperation with the Royal
Jordanian Geographic Center.
LS factor. The accuracy of the DEM built by
interpolation from topographic maps with contours
drawn at 40 m interval, is determined by the x, y and
z accuracy of the topographic maps. The most
common measure for the quality of a map feature is
its relative and absolute positional accuracy. The
topographic maps available for this study conform to
the RJGC (Jordan) map accuracy standards.
C factor. For mapping the land cover in the 1970's,
Landsat Multi-spectral Scanner (MSS) data (56 x 79
m?) were available. However, land cover mapping for
the 1990 year was carried out by the processing of
finer resolution Landsat Thematic Mapper (TM) data
(30 x 30 m?) The calibration of the NDVI and
fractional vegetation cover images from spectral
unmixing with vegetation cover data relies on ground
sampling of vegetation that was not collected
simultaneously with the imagery. Calibration data
between remotely sensed data and biogeophysical
variables have relatively low accuracies, and this will
be the case in this study. It is likely that the greatest
errors like are in the C factor layers. However, as this
is a reconnaissance study the C factor layers probably
show the relative spatial and inter-annual variations in
vegetation cover that are appropriate to the
investigation.
P factor. The only error surrounding the P factor
would be that some management practices for soil
conservation had been omitted from the analysis.
This is extremely unlikely given the prevailing
farming systems.
4. CONCLUSION
A visual analysis of the series of maps of soil losses
indicates that the most critical factors in the RUSLE
were:
e The topographic (LS) factor; and
e The soil erodibility (K) factor.
Though in the case of irrigated agriculture the C
factor is also important.
Overall there was a 4.2% increase in the predicted
soil loss from the study area between 1972 and 1992.
There is not enough support for the contention that
rates of overall soil loss are accelerating. It is likely
that these numbers lie within the likely error margin
of the modeling.
Verification of the model was attempted by
comparing predicted soil losses with quantitative and
qualitative data obtained from 47 fields in the study