424
3. The ferrallitic pole is not expressed through the forest
cover.
4. The same image processing has been applied and give ac
ceptable results until a pixel size of 480 yi that is 9.60m. on the
ground.
2.3 Discussion
1. The image processing allows to show an information on
the forest cover which is related to soils organisation; this in
formation was'nt easy to read (heterogeneity of the light, dis
cordance with a topographic unit).
2. The forest cover expresses only the most transformed, that
is podzolic, part of the pedological system.
3. The limits of the two domains (vegetal and pedological)
are not precisely coincident; the shapes only may be compared.
This lack of precision suggests several remarks:
- the registration might not be good enough
- future studies may be done to explain these discrepancies
in terms of relative dynamics of soils and vegetation.
- the structural information on soils lying in this aerial
photography is more related to low-scale soils systems in
formation than to large scale soils characteristics.
4. This last remark will be tested when processing the whole
picture as invite us the results of changes in resolution.
3. CONCLUSION
Structural information about superficial formation is present in
the remotely sensed images studied. For the given examples,
this information is expressed through a natural vegetal cover
(primary or secondary) in terms of limits and domains. Using
structural maps as reference, it is shown that
1. The knowledge of deep weathered formations is necessary
to establish the significance and the hierarchy between the
different limits given by the vegetal cover.
2. For a given pedological system, the most evoluted part of
the mineral cover is visible, seen through the vegetal cover.
3. The overlay between the pedological structural limits and
the limits given by the vegetal cover is generally not precise; the
relation is a global one; only shapes may be compared first. The
study of the discrepancy can lead to further study of soils and
vegetation relative dynamics.
4. For the spectral bands used, the limits in the image define
domains whose contents are not homogeneous: the
methodology consists therefore on finding which combination
of reflectance and (or) texture can best describe them.
REFERENCES
Boulet R. et al. 1981. L'écosystème forestier Guyanais. Etude
et mise en valeur. № 4, CTFT. INRA, MUSEUM,
ORSTOM, 65p.
Dosso M. 1986. Analyse structurale d'une unité de modelé
latéritique comme référence pour la recherche d'informations
pédologiques d'ordre structural contenues dans les images de
télédétection correspondantes. Régional seminar on latentes.
Douala, Cameroun, Janvier 1986. MESRES, IBSRAM,
ORSTOM. A paraître.
Seyler F. 1986. Recherche méthodologique sur le traitement
d'images de télédétection aérienne, basée sur l'analyse
préalable des formations superficielles avec leur couverture
végétale. Etude d'un secteur de lande sur schistes en Bretagne
centrale. Thèse Doct. Université Paris VIL Mai 1986.
Veillon L. 1984. Etude tridimensionnelle d'un système de trans
formation de la couverture ferralitique sur SDB en podzol
dans la région d'ORGANABO. Rapport ORSTOM. Septem
bre 1984.