Comparison of Quickbird-2 and Landsat-7 ETM+ data for mapping of vegetation cover in Fethiye-Kumluova coastal dune in the Mediterranean region of Turkey

Abstract

Dune vegetation plays an important socio-economic role in some coastal areas of Turkey's Mediterranean region as it prevents sand encroaching on residual areas, agricultural fields and sites of historical significance. These areas are also vital for biodiversity conservation due to the rare plant species and habitats found within dune ecosystems. In order to initiate the necessary protective measures to maintain these sensitive biotopes, dune areas must be carefully mapped and then monitored at regular, predefined time intervals. Vegetation mapping using remotely sensed data is often the primary source of information on current sand dune conditions and is used as a baseline for future management options. The objective of this study was to determine the effectiveness of Quickbird-2 data for the stratification of plant communities in the Kumluova sand dunes, and to compare the results with a classification map derived from Landsat-7 Enhanced Thematic Mapper (ETM+) data. The overall classification accuracy using Quickbird data with 10 classes was found to be 82.2%, with the Kappa coefficient of 0.80. Landsat classification produced an overall accuracy of 75.7% and a Kappa coefficient of 0.72 for 8 land use classes. The results indicated that Quickbird and Landsat ETM+ satellite data can be employed for different management purposes within dune areas. Quickbird imagery may be used for vegetation cover mapping for tactical forest management planning purposes as a substitute for aerial photography, while Landsat imagery may be used for the localisation of afforestation and deforested areas within dune ecosystems between successive years for strategic planning purposes.