Download Vast amounts of data

Vast amounts
of data
The recent development of hyperspectral sensors is bringing the need
to explore a vital avenue of research:
the development of methods to process the vast amounts of data generated by this observation technique.
Two major projects have focused
on this methodological exploration,
bringing together scientific teams
with complementary expertise in
HYPERCRUNCH
HYPERWAVE
HYPERPEACH
Improving data mining
Applying and validating algorithms
Peach trees under the hyperspectral
The high spectral resolution that is intrinsic
HYPERWAVE’s aim is to validate and apply
magnifying glass
to hyperspectral remote sensing produces a
the algorithms developed by the HYPER-
The modelling of growth processes in fruit
vast quantity of data. It is therefore essential
CRUNCH project that developed a toolbox
trees could make it possible to develop ef-
to arrive at a better extraction of data – or
prototype making it possible to apply these
fective monitoring tools that are essential
data mining – from hyperspectral datacu-
algorithms whatever the subjects studied. To
to an effective management of this kind
bes. At an initial stage, the project sought
test and validate the prototype, it needs to
of intensive cultivation. In this context, the
ways of selecting a limited number of rele-
be confronted with numerous hyperspectral
project studied the biochemical parameters
vant bands for a specific application without
datacubes supplied by different sensors over
of peach tree leaves and in particular iron
loss of essential information.
a wide field of applications. The data and ex-
deficiency, a factor that inhibits plant growth
At a later stage, acquisition protocols were
pertise acquired by other STEREO programme
and reduces the production quantity.
developed to improve classification perfor-
projects or research carried out within VITO
In the Saragossa region of Spain, almost two
mance. The mathematical algorithms were
were used for this purpose. Initially develo-
hundred peach trees were given different
developed as independently as possible
ped for the ecosystem of orchards, the al-
amounts of iron supplements and then sub-
of applications and sensor specifications in
gorithms were integrated into the analysis
jected to hyperspectral reflectance measu-
such a way as to permit their implementa-
of aquatic environments or dune vegetation.
rements, both on the leaf itself and on the
tion in operational data processing chains,
This latter investigation made it possible to
canopy by means of field and airborne hyper-
for example in the future APEX sensor data
produce a map of the dune vegetation for
spectral measurements. This multiple ap-
chain. the Belgian coast (HYPERKART project).
proach enabled to test the traditional vegetation indexes and successfully generate
The data reduction techniques and mathe-
the pre-processing of data, classification methods and post-classification
processing in order to permit the
matical algorithms were made available to
Coordinator
new robust indexes designed to detect and
the scientific community. The field of ap-
- TAP, VITO
quantify growth anomalies. It allowed provi-
plication explored is precision farming and
Partners
ding a satisfactory estimate of chlorophyll a
more particularly stress monitoring (for mil-
-MUMM
and b in leaves on the basis of hyperspectral
dew, nitrogen deficiency) in Jonagold and
-Visie Lab, UA
data. The new indexes also made it possible
Golden Delicious apple orchards.
-Afdeling M3-BIORES, K.U.Leuven
to detect changes in the chlorophyll concen-
interpretation of images. The HYPERPEACH project is a concrete applica-
tration, such as a reduction in this pigment
Coordinator
due to iron deficiency, before it becomes vi-
-TAP, VITO
sible to the naked eye. Partners
tion of these investigations.
-Visie Lab, UA
Coordinator
-Afdeling M3-BIORES, K.U.Leuven
-Afdeling M3-BIORES, K.U.Leuven
Partners
-Visie Lab, UA
-TAP, VITO
-Instituto de Agricultura Sostenible, Spain
-Estación Experimental Aula Dei,
Spain
Hyperspectral technology
Federal Science Policy
Earth Observation Helpdesk STEREO 1 Programme