Abstract: Along with the concept of scientific development and constructing a socialism harmonious society being proposed in recent years, regional development issues have been becoming the focus of attention of many scholars. The applications of GIS in comprehensive research planning of ecological economy become more and more along with improvement of precision and the development of GIS technology. Gansu province is with long and narrow landform, in which the differences of location factor, economic basis and institutional settings between cells are large. This makes eco-economic regionalization inevitable. The paper proposed new research ideas——Grid-based eco-economic regionalization on the basis of the existing methods and theories of eco-economic regionalization. The study used Grid technology. It analyzed and compared data thought linking vector and raster data to fishnet-grid. Gansu was divided into 12 243 cells in 6 km×6 km regular girds exploring the coupled development mechanism of the ecological and the economic system in Gansu. The study chooses 26 indexes which conclude natural ecological, social and economic indexes from "Gansu statistical yearbook" in 2010 and 2011. This study mainly used GIS and relative statistical analysis softwares. First-degree ecological regions used cluster analysis (CA) methods based on ecological factors such as average annual rainfall, average annual temperature, forest cover and aridity index. Gansu was divided into four eco-economic regions, i.e. northwestern arid region (Ⅰ), central cold temperate semi-humid area (Ⅱ), southern paramos frigid humid region (Ⅲ).southern warm temperate humid region(Ⅳ). The data used in the second-degree divisions came from the result of first-degree ecological regions. The division used social and economic factors. In the second-degree division, these belts were also subdivided into 17 eco-economic regions using CA methods and principal component analysis (PCA). The third-degree divisions used the topography types which were got through DEM.  In the result, Gansu was divided into 21 eco-economic regions. The study had a large database based on GIS. All kinds of data were linked to the grid and the results were expressed in maps after spatial analysis. The results of first-degree and thirddegree regions were expressed with map using GIS, and second-degree was expressed with a table. The study achieved graphic visualization, got the data and the derived results through grid transformation based on statistical analysis softwares and integrated analysis with GIS technology. The results can be used to the study of spatial distribution properties on the base of different cells. The research shows that grid-based ecoeconomic regionalization method can translate result of regionalization into homogeneity fishnet from administration cell, increase the precision and speed of regionalization. It is the one of the optimal methods to eco-economic regionalization. The study expected to provide a scientific basis for forming the sustainable developing city pattern. The research came with some shortcomings: first, index system was not complete; second, research scale was large. If the data of villages could get, the result will be better; third, whether GIS fishnet technology could become an institutional model, this is worth carrying out the further study.

Key words: Gansu Province, eco-economic regionalization, grid-GIS, principal component, cluster analysis