Accuracy of the Land Use/Cover classification in the Oueme Basin of Benin (West Africa)

Abstract

In the scope of the European Union water initiative for developing countries, the research programme RIVERTWIN (A Regional Model for Integrated Water Management in Twinned River Basins) was initiated for three river basins, Neckar in Germany, Ouémé in Benin Republic and Chirchik in Uzbekistan. Our contribution to development of the model at regional level was the establishment of a Soil and Land Resources Information System for the Ouémé basin (SLISYS-Ouémé). The objective of the SLISYS was to provide information about soil, climate and land use in the Ouémé basin. As well as data on soil and climate, it was essential to have information on land cover and land management practice. Therefore, a land use/cover map was established at the scale 1/200.000 from satellite images LANDSAT ETM Plus of 2003 from 3 scenes. After image treatment, imaged maps were established and an interpretation of keys was defined. For efficacy reasons, the interpretation was carried out at the scale of 1/50.000 in order to get the maximum amount of information. Field controls were done for one month and more than 650 observation points were checked during the ground check process. Finally, 17 land use/cover classes were defined. The subsequent accuracy check showed that the overall interpretation precision was high (87 %). The land use unit “mosaic of cropland and fallow” had the most precision in its interpretation, and the classification of the unit “humid and dry dense forest” had the least precision.