ZezoumaSANONPolytechnic University of Bobo-Dioulasso, EU project UNDESERT

01BP 1091 Bobo-Dioulasso 01Bobo-Dioulasso01Burkina Faso

ZezoumaSANONassociatedPartyYvonneBACHMANNCustodian/StewardIreneeSOMDACustodian/StewardMiproHIENCustodian/Steward2014-11-24The quantification of land cover/use changes in West African protected areas is essential for improving their management. The objective of this study was to identify and analyze the land cover changes of the Classified Forest of Koulbi in south-west Burkina Faso between 1986 and 2010 based on supervised image interpretation and the calculation of landscape metrics. In this classified forest the anthropogenic pressure of the last 30 years left significant traces. The following landscape metrics were calculated on the land cover/use map derived from Landsat images: number of patches per class, total area of each class and dominance of the area of the largest patch of the class. We found that the CFK has undergone a transformation from a mosaic of dry forests and tree savannas to dominance of tree savannas. The landscape metrics reveal a decrease of the dry forest area and the dry forest patch numbers. At the same time the dry forest class became more fragmented. The number of tree savanna patches increased and so did the total area of tree savanna.Obtain permission from data set owner(s)Classified forest of Koulbi South-West of BURKINA FASO-3.0092-2.73069.79899.44281986-11-182010-11-27ZezoumaSANONPolytechnic University of Bobo-Dioulasso, EU project UNDESERT

01BP 1091 Bobo-Dioulasso 01Bobo-Dioulasso01Burkina Faso

zezoumasanon@yahoo.fr

Analysis was conducted with ERDAS Imagine 9.1 on two Landsat images from the 18 of November 1986 and the 27 of November 2010, provided by the SP-CONEDD (BURKINA FASO). Geometric corrections were made on the basis of the PROGEREF (map of Vegetation South-West BURKINA FASO). A false colour composite was made for visual classification by combining the bands 4, 3 and 2 on red, green and blue. A supervised maximum likelihood classification was performed of the classified forest area. Altogether 10 training sites for each class were used in the classification. 06 land cover classes were identified for the image from 1986 and 06 for the image from 2010. For evaluation of the quality of the supervised classification we conducted a quantitative accuracy assessment. 153 random control points were selected on the satellite image and their reference value compared to field observations at these points. We calculated Producer's, User's, Overall Accuracy, and the Kappa coefficient. To evaluate the changes and fragmentation of the vegetation cover, we calculated three landscape metrics: a.) Total number of patches (this index provides information about the fragmentation process of a class between two or more dates), b.) Total class area and c.) Dominance of the patch of a class (this index gives the proportion of the dominant patch in the class to the entire class).

attached_fileNot authorized to access resourceOtherattached_fileNot authorized to access resourceOther timeseries other Monitoring of land cover by satellite image classification between 1986 and 2010.