Bi-temporal analysis of landscape changes in the easternmost mediterranean deltas using binary and classified change...

The aim of this study is (1) to quantify landscape changes in the easternmost Mediterranean deltas using bi-temporal binary change detection approach and (2) to analyze relationships between conservation/management designations and various categories of change that indicate type, degree and severity of human impact. For this purpose, image differencing and ratioing were applied to Landsat TM images of 1984 and 2006. A total of 136 candidate change images including normalized difference vegetation index (NDVI) and principal component analysis (PCA) difference images were tested to understand performance of bi-temporal pre-classification analysis procedures in the Mediterranean delta ecosystems. Results showed that visible image algebra provided high accuracies than did NDVI and PCA differencing. On the other hand, Band 5 differencing had one of the lowest change detection performances. Seven superclasses of change were identified using from/to change categories between the earlier and later dates. These classes were used to understand spatial character of anthropogenic impacts in the study area and derive qualitative and quantitative change information within and outside of the conservation/management areas. Change analysis indicated that natural site and wildlife reserve designations fell short of protecting sand dunes from agricultural expansion in the west. East of the study area, however, was exposed to least human impact owing to the fact that nature conservation status kept human interference at a minimum. Implications of these changes were discussed and solutions were proposed to deal with management problems leading to environmental change.