N'doumé, Claude Thierry
1280
ETUDE DE FAISABILITE POUR L'ELABORATION D'UNE CARTOGRAPHIE STATISTIQUE
D'INVENTAIRE DES VERGERS CAFE ET CACAO EN COTE D'IVOIRE PAR TELEDETECTION
SATELLITALE.
Claude N'DOUME*, Philippe LACHENAUD**, Alain HUSSARD***, Hugo NGUYEN V**
et Albert FLORI (**)
*Centre de Cartographie et de Télédétection (CCT)
Bureau National d'Etudes Techniques et de Développement (BNETD)
ndoumec@bnetd.sita.net
**Département Culture Pérenne (CP)
Centre de coopération Internationale en Recherche Agronomique pour le Développement (CIRAD)
lachenaud@cirad.fr
***Strategis SA
Montpellier
hussarda@strategis.fr
Inter-technical Commission IC-40
ABSTRACT
This study, carried out in Daloa region (western-center of Côte d’Ivoire), aims at implementing an inventory
statistical mapping of coffee and cacao plantations in Côte d’Ivoire by satellite remote sensing. The methodology
used resides in classifying a multi-spectral SPOT picture by the maximum likelihood method from a sampling of
coffee and cacao plantations. The first step concerned an in-depth agronomic survey (36 variables, including 19
agronomic) carried out with Plan Foncier Rural (PFR) and farmers on 358 plots in the PFR pilot zone of about
30.000 ha. The biometrical analysis of the data showed that, for both cash crops, the yield estimated by the
investigators constituted a factor summing up the key variables describing the state of the plantations (shade,
condition, maintenance, etc...). This parameter was thus taken into account in constituting the test-plots, and it
was observed that the approach yield-spatial delimitation of the parcels was giving better classification results
controlled over all the parcels surveyed. The test-zone classified image, cartographied with a unit of 1 ha is
presented. The global mapping precision (Pixels correctly classified) is 72 96, 69 96 for coffee and 66 % for
cacao. In the second phase, the methodology was applied to the whole scene (360,000 ha) and a validation
survey was carried out. The global mapping precision is 80% and the reconnaissance rate for coffee and cacao
are respectively 86 and 94%. The areas estimated might be more reliable than the estimates from land or farming
investigations used so far. It is to be noted that these results were achieved through test-plots which were limited
in number (ten per theme) and in area, hence low costs and duration in an operational phase.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 971