Effects of spatial grain size on landscape pattern, especially the study of critical grain size, has been developed during the last two decades to describe and quantify complex landscape patterns and to link landscape patterns and ecological processes. Scale is an important issue in both landscape ecology and remote sensing. Scale in landscape ecology involves both grain size and extent. However, past studies mainly used low spatial resolution images. Thus, it is necessary to identify the scale characteristics and appropriate grain extent of farming-pastoral zone in higher resolution images. A high resolution image of a small watershed with areas of farming-pastoral zone was interpreted to create maps of landscape pattern. In the GIS and RS environment, the vector drawings were transformed into shutter drawings according to different spatial grain sizes. The landscape pattern was then converted to ArcGrid formats with grain sizes of 1, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50 m using ArcGIS Spatial Analyst. Landscape indexes were chosen and the relations between spatial grain size and these landscape indexes were determined by correlation analysis. Landscape indexes were examined at the class-level including five Area/Perimeter/Density indexes, seven Shape indexes and eight Contag indexes; and landscape indexes were also chosen at the landscape-level including four Area/Perimeter/Density indexes, four Shape indexes, three Contag indexes and four Diversity indexes. The results showed that the scale effects of selected landscape indexes were different in the farming-pastoral zone. The scale effects were apparent at both the class and the landscape levels. Critical grain sizes of landscape indexes of the farming-pastoral zone were evident, such as the grain size of 10 m for the PD, IJI and FARC-MN indexes in the watershed. An appropriate grain extent (5-10 m) was identified for calculating the landscape indexes of the farming-pastoral zone for the high resolution image. For these reasons, the scaling conversion should be taken into account in order to provide reference for studies such as forecasting landscape changes, comparison analyses and evaluations of the land use types based on landscape metrics. The various landscape type patches in the small watershed had a fractal character and the fractal dimensions of landscapes responded to scaling differently, such that fractal dimensions decreased nonlinearly with increasing grain size. Moreover, it indicated that at the border of landscape types there was a tendency towards simplification. This study should provide useful information about farming-pastoral zone effects on landscape mechanisms and the importance of scaling to these areas in high resolution image spatial models in future research.