ABSTRACT
Previous studies have been able to successfully detect changes in gently-sloping forested areas with low-diversity and homogeneous vegetation cover, using medium-resolution satellite data such as landsat. The aim of the present study is to examine the capacity of multi-temporal landsat data to identify changes in forested areas with mixed vegetation and generally located on steep slopes or non-uniform topography. landsat thematic mapper (TM)and landsat enhanced thematic mapper plus (ETM+) data for the years 1987-2000 was used to detect changes within a 19,500 ha forested area in the Western Black sea region of Turkey. The data comply with the forest cover type maps previously created for forest management plans of the research area. The methods used to detect changes were: postclassification comparison, image differencing, image rationing and NDVI (Normalized Difference Vegetation Index) differencing methods. Following the supervised classification process, error matrices were used to evaluate the accuracy of classified images obtained. The overall accuracy has been calculated as 87.59% for 1987 image and as 91.81% for 2000 image. General kappa statistics have been calculated as 0.8543 and 0.9038 for 1987 and 2000, respectively. The changes identified via the post-classification comparison method were compared with other change detetion methods. Maximum coherence was found to be 74.95% at 4/3 band rate. The NDVI difference and 3rd band difference methods achieved the same coherence with slight variations. The results suggest that landsat satellite data accurately conveys the temporal changes which occur on steeply-sloping forested areas with a mixed structure, providing a limited amount of detail but with a high level of accuracy. Moreover, it has been decided that the post-classification comparison method can meet the needs of forestry activities better than other methods as it provides information about the direction of these changes.
ABSTRACT
Forest roads have many kinds of direct and indirect effects such as physical, ecological, and socio-economic effects in landscape scale. In this study the relationship between forest road network and the structure of a multifunctionally planned forest landscape, in the south of Turkey was examined. The aim of the study was to determine whether or not the correlation between road network and landscape structure across a gradient of road density, categorical road/no road approach and landscape pattern data from the managed forest landscapes existed. It was concerned with three type polygons that were forest functioning patches, compartments, and stand patches and were associated with higher road density or existing of road network. To quantify the case, it was analyzed whether the spatial structure of forest patches adjacent to roads differs from the spatial structure of forest patches away from roads. GIS was used to integrate quantitative indices of landscape structure. There was little negatively significant relationship between road density and patch area/size, patch perimeter, and shape index of patton index (PI) and no correlation with patch number and Fractal Dimension (FD). The forest road network modified spatial pattern of patch shape through PI values. The road network increased the number of patch and decreased mean patch size. The existence of road network was more effective than topography index appearing of PI and FD in the roaded areas. It was indicated to potential reverse impacts of road network on spatial structure of forest compartments in terms of shape index (FD). Highest proportions of harvestable forest areas fall into roaded areas. The results define that road networks are effective on the spatial structure of forest landscape. The forest road network of study area has all probable ecological impacts of forest roads mentioned in context.