Regionalization has been an instrument to describe the general attributes of the earth’s surface. As preparatory work for designing and implementing a socio-economic development strategy, regionalization may be based on a global, national or a local level. There has been considerable progress in studies of regionalization in physical geography, agricultural and integrated economic domains in China since 1949. The literature has developed from regionalization based on a single geographical factor to an integrated regionalization based on a cluster of spatial factors, and finally to regionalization approaches with the specific objective of serving as a means for locating agricultural sectors or geographical arrangement of socio-economic construction.
Eco-regionalization has gained importance within this evolving literature as a means for ensuring an appropriate management of ecosystems and sustainable use of natural resources. Based on theories and methods of general ecology and landscape ecology and taking the Upper Min River Basin, Sichuan, China as an example, this study tries to develop a new method for eco-regionalization. This new approach, the ‘landscape eco-regionalization method’ is a bottom-to-top method that differs from the traditional top-to-bottom eco-regionalization approach, in that it may use landscape indices such as mean patch size, fractal dimension, diversity, evenness, dominance，fragmentation and human disturbance. GIS and geo-statistics are also applied in the analysis of spatial disparities of the above indices for eco-regionalization in the Upper Min River Basin.
Physical and socio-economic data for 92 towns and townships in 5 counties in the Upper Min River Basin are used in the analysis and development of the index system. The Upper Min River Basin may fall into three eco-regions by using the traditional top-to-down method, which would include indices such as topographical variation, temperature, humidity, geomorphologic type and socio-economic activities. In contrast our study region may be divided into four ecological regions by using the landscape ecological bottom-to-top method. To identify similarities and disparities of the two methods, the results for division of eco-regions at the lower level are analyzed. Our analysis indicates that the two methods vary greatly in their use of the data, in the information identification and application scopes. The bottom-to-top method may have its advantages over the top-to-bottom method with respect to the division of eco-regions in comparatively small areas with substantial bio-physical heterogeneity.