Large scale mapping of actual Japanese beech forest distribution using remote sensing and natural environmental factors

In order to develop an accurate and quick method for drawing large-scale vegetation maps representing actual vegetation, a case study was implemented for a Japanese beech community in the Mt. Hyonosen and Mt. Ohginosen areas, of Tottori Prefecture, southwestern Japan. To map out the actual beech forest areas, as a first step, a vegetation prediction model was constructed to show the potential area of the natural beech community. The explanatory variables used in this model were GIS-based data layer sets, such as 10m resolution elevation maps, slope gradient, slope aspect, topographical configuration, soil moisture, accumulated amount of solar radiation and accumulated temperature. Next, areas representing both the broad-leafed forest and coniferous forest were interpreted from the 15 m resolution ASTER sensor satellite image. As a final step, the overlying of these maps produced the actualy area of beech community. Consequently, both the Data Producer's accuracy and the User's accuracy compared with natural beech community's area were approximately 50%. When compared with the natural beech communities and the broad-leafed forests area of the substitute plant communities the User's accuracy was 71%. According to these evaluations, the author has shown that this method is not sufficient for mapping the actual areas of the natural beech communities. However the results of this method were moderately satisfactory when used for large scale mapping of the actual area of both the natural beech communities and the broad-leafed forest of these substitute plant communities.