| You are currently reading our legacy website. You can view our current projects here: |
FREQUENTLY ASKED QUESTIONS
Where do you see soil reflectance spectroscopy
in five years' time?
Until now, reflectance spectroscopy has largely
been seen as a substitute for standard laboratory analyses. We believe
that the technique has the greatest potential for direct sensing of
soil functional capacity (the soil's capacity to perform specific,
production, environmental or engineering functions). The strength
of the technique lies in its ability to simultaneously sense a number
of fundamental soil properties, which are often closely interrelated
and together determine a soil's functional capacity. We have provided
examples of this in terms of ability of spectral screening tests to
discriminate soil fertility condition classes and soil erosion phases.
We expect to see a large increase in the use of reflectance spectroscopy
in risk-based soil management approaches, especially in large area
applications and precision agriculture. We see the development of
dedicated software that will allow 'criminal profiles' of soils to
be displayed directly on the computer screen and perhaps global standards
emerging. We also see reflectance spectra being entered directly into
computer simulation models (e.g. hydrological models) instead of conventional
data on soil attributes. There is also great opportunity for saving
costs by using the spectral library approach as a variance reduction
tool when selecting samples for conventional analyses. In summary,
we expect that reflectance spectroscopy will transform the way in
which soil measurements are done and permit much more reliable assessments
to be made than in the past.
Back to top 
In your studies, why does Landsat imagery
seem to work so well for spatial interpolation of soil properties,
despite the fact that the landscapes are highly vegetated?
Remote
sensing of soil properties direct from space platforms is hampered
by atmospheric interference, shade and shadow effects, mixtures of
materials within pixels, and variation in soil moisture content. However,
calibrations between soil properties and image reflectance can be
made because variation in soil properties across an image is related
to variation in vegetation, parent material, shade and shadow effects,
which all effect the image reflectance values. We are basically interpolating
ground observations relying on 'environmental' correlation. These
calibrations are not transferable to a different image. We expect
that where there is less vegetation and more soil signal that such
calibrations will improve and vice versa. We expect that new hyperspectral
imagery coming on line, which has a spectral resolution similar to
lab/field spectrometers, might enhance calibrations and even allow
direct unmixing of soil endmembers. There may also be scope for improving
spatial interpolation by combining soil spectral libraries with other
geo-referenced information, such as from digital terrain models.
Is assessment of soil quality at the spatial
resolution of satellite platforms useful for soil management recommendations
given the micro-variation that typically exists within farms?
With Landsat imagery we are working with a spatial resolution
of about 30 m. This is appropriate for planning of policies, targeting
of advisory services and monitoring of changes at a watershed scale.
This resolution is already vastly superior than could be derived from
say existing soil maps, which are typically only available at a scale
of 1:1 Million or 1:250,000 and assume homogeneity within mapping
units. Preliminary geo-statistical analysis in our western Kenya studies
also indicate that soil samples taken closer than between 250 m and
10 km (depending on the soil property) are not independent (they are
spatially correlated), which suggests that further increasing the
spatial resolution will not necessarily give more information. Having
identified potential trouble spots at the watershed scale, it would
then be appropriate to go on the ground at the farm scale and conduct
spectral screening tests to further asses micro-variation in soil
quality. This process can facilitate individual land users to learn
about the variability on their farms and to experiment on how best
to manage it according to their objectives.
Back to top
|
