Abstract: The township population migration coefficient and the intensity of the rural population migration coefficient of Shizhu County in 2004 and 2014 are set as measurement indexes. By using GIS-ESDA spatial analysis method, the spatial correlation index, variation function and the dual Logistic regression model, we explore the characters of the spatial patterns of the population migration and its influence factors of the county between 2004 and 2014. The results show that: 1) During recent decade, the overall spatial pattern of population migration shows strong spatial positive correlation and the spatial aggregation degree increases. 2) Regarding to the hotspot, the concentration of population migration is obvious in the high value area, while that just happens in some places in the low value area. Looking at the revolution of the hotspot, the hotspot area of the population migration coefficient is expanded from agglomeration into the belt-shaped distribution along the river, while cold spot area is not changed significantly. The evolution of the intensity coefficient of rural population transfer in the hot zone has experienced development of “discrete-multi core”, and cold spot transition area tends to shrink. 3) The significant differences of randomness and the structure in the total variation of apparent indicate the imbalance of population migration and the increase of concentration year by year. 4) Internal factors drive population migration differently. Among them, the effects of the migration experience, the total income of the family, the number of labor force and the degree of road connection are most significant. Both of the family burdens and the topography of the area are characterized by their uniqueness and complexity.

Key words: population migration, Dual Logistic Regress Model, typical mountainous area, Chongqing