[pp.95-102]
Aki NAGANO (Kyushu University)
Abstract:
The aim of the present study is to apply a neural network model for landuse suitability analysis and to clarify what constitutes the evaluation of landuse suitability analysis in terms of the relationship between landuse suitability factors based on the strengths of neural network interconnection between associated neurons. The neural network model for landuse suitability analysis classifies evaluation results based on the physiographic and accessibility factors using geographic information system. The study area is located in southern Osaka, Japan. In this district, pressure for further urban development continues, especially within subdivision developments. As a result, some development areas have expanded into the hill sides. Consequently, development of such areas has associated risks of natural hazard such as land slides, slope failure and so forth. This study examines open space as potential residential development areas. Results have shown that this study enables identifying of the levels of costs for each factor that contributes in the evaluation for the site preferences. Furthermore, this study clarifies how influential factor generates suitability maps for landuse based on the ordinal combination method.
Key Word:
land use, suitability analysis, nueral network, relationship between landuse, suitability factors, decision making, GIS